Technology Governance Wiki - User contributions [en]

Governance by markets

2026-10-07T11:22:04Z

Robert Florijn:

= Glossary =
= [https://wiki.ivir.nl/index.php?title=Data_as_an_input_to_market_power Data as an input to market power] (Sergey) =In the present digital economy, data has emerged as a valuable economic commodity that entities leverage to gain a competitive edge over their rivals [[https://repository.nls.ac.in/cgi/viewcontent.cgi?article=1478&context=nlsblr 1]]. '''Data''' is an input into production, so businesses that control data can have market power and impede the entry of potential competitors. '''Market power''' refers to a firm's ability to influence the price, output, or other market conditions in its favour, without being constrained by competitive forces [[https://www.tutor2u.net/economics/reference/what-is-meant-by-market-power 2]].'''Data as an input to market power''' means that access to information can help a business limit the competitive pressure it faces. Information about searches, clicks or purchases can improve services and advertising. If competitors cannot obtain comparable information at a reasonable cost, they may struggle to offer an attractive alternative. This can allow a platform to raise prices or worsen quality or privacy protections without losing enough business to make doing so unprofitable. This matters even when a service is free, because competition also concerns advertising charges, personal information and the choices available. Data can be used repeatedly and by several businesses without being used up. However, access can be restricted through contracts and technical controls. Nevertheless, holding a large dataset alone does not establish market power.'''Google Search''' illustrates this mechanism. Google controls over 91% of global search engine traffic [[https://www.statista.com/statistics/1381664/worldwide-all-devices-market-share-of-search-engines/ 3]] and uses information about users’ searches and the results they click to improve the relevance of its rankings. A larger volume of searches gives it more opportunities to learn, including about unusual queries that smaller search engines encounter less often. Better results can attract further use, producing additional information and reinforcing the advantage. A new search engine may therefore struggle to attract users while lacking the information needed to improve its service. Data access can constitute a significant barrier to entry when this type of analytics is necessary to participate in a market, particularly when alternatives are limited. A platform’s control over data access can shape who is able to compete with it. By determining whether, and on what terms, rivals can access that information, a platform can influence the competitive opportunities available to others. This suggests that regulation should examine access to strategically important data alongside prices and market shares. Article 6(11) of the EU Digital Markets Act [[https://eur-lex.europa.eu/eli/reg/2022/1925/oj/eng#:~:text=giving%20their%20consent.-,11.%C2%A0,-The%20gatekeeper%20shall 4]] reflects this approach: certain large search providers must, on request, share specified search data with rival search providers on fair and non-discriminatory terms, with relevant personal data anonymised. The aim is to help rivals compete while protecting users. A data advantage does not, by itself, establish unlawful conduct.Would access to equivalent data enable a rival to compete effectively? The concept cannot answer this on its own. Additional data may eventually add little value, become outdated, or have substitutes. Firms also need the skills and computing resources to use it. Google’s default positions in browsers and devices can further restrict rivals’ access to users even if data become available. The unresolved issue is how much competitive advantage comes from data itself and how much comes from these other resources and arrangements. 

= = Two Sided Markets (Ceren) = =In two-sided markets, two sides of a market interact through an intermediary. These markets typically have some externality,<code class="mwt-code">Carl Shapiro and Hal R Varian, ''Information Rules: A Strategic Guide to the Network Economy'' (Harvard Business School Press 2010), 183.</code> where the decisions of the sides affect the other. The intermediary also has multitudes of effects on both sides, mainly identified in the related literature through its pricing.An example of a two sided market is payment systems, whereby the credit card providing bank is the intermediary between the customers and merchants. In this example, banks provide credit cards to customers to be utilized in their payments, and merchants accept payment via multiple credit card systems to accommodate payments from customers holding credit cards from such providers[https://doi.org/10.1257/jep.23.3.125 .]The main power of the intermediary materializes in their pricing toward the merchant (and the merchant’s bank in interchange fees in case of payment being transferred through different banks). Customers will be expected to be more loyal, holding a few credit cards, whereas merchants accept payment from many banks to accommodate all customers’ payment methods. Banks will provide more subsidies to customers to attract them and compete aggressively on the customer side, making their platform more desirable to work with by merchants, who will in turn be more accepting of the platform’s fees to gain the business of the larger volume of customers. However, dynamic pricing (increasing fees after gaining entry and loyalty) may be a way for these platforms to abuse their power.The concept may be derailed from these logical / analytical systems of operation or expected outcomes when public policy interferes with the market. For example, net neutrality rules regulate what internet service providers (the intermediary) can charge content providers (one side) who aim to reach the consumers (the other side). In this case the above considerations and the concept are halted.

= Switching costs and lock-in (Kate) =
'''Definition '''When you move from one product or service to another, switching costs are everything you have to either give up or spend. This can include money, time, lost contacts, etc. Lock-in is when you feel (and are) trapped by these costs that add up and you decide instead of moving products or services, it’s just easier to stay with a service, not because you want to but because it’s just too much of a hassle. This is true even if a less expensive or better service exists all because leaving the product/service just costs too much in terms of time, effort, and money. Shapiro and Varian (2010) argue that switching costs are normal in information markets and that the full cost of a switch includes what you as the customer pay and what a new supplier spends to win that customer over.  '''Example: Apple''' The more Apple products you own, the harder it is to leave. As an avid Apple user, I am absolutely stuck in their ecosystem. I have my photos in Apple’s iCloud, my group chats in iMessage, and my Apple Watch only works with an iPhone. Leaving Apple is not easy. So much so that in March 2024, the US Department of Justice sued Apple alleging it raises costs of leaving/switching phones by making messaging worse and less secure when you’re messaging with non iPhones. In June 2025, Judge Julien Neals denied Apple’s motion to dismiss and the case is still in pre-trial proceedings.  '''Governance implication''' Lock-in makes users dependent on a platform and because of that, once we’re locked in, the platform (Meta, Apple, Google, etc) can raise prices, show more ads or even provide lower quality terms. So in this way, the platform can control users because if a user wants to leave, it’s going to cost you in that the platform has already designed your exit including how you can export your data, licenses and APIs. Under the EU’s Digital Markets Act, the European Commission adopted decisions in March 2025 making Apple open nine iOS connectivity features, mostly used by devices like smartwatches, headphones and televisions. These decisions are legally binding and they focus on lock-in by lowering the cost of choosing non-Apple products. It’s worth noting, however, that they only apply in the EU. When the Commission adopted the decisions, Apple said (reported by MacTech) “It’s bad for our products and for our European users.” '''Open Question''' So what happens when the cost of leaving isn’t just your own costs as a user but also the cost of leaving family and friends behind?  If a user leaves, let’s say, Instagram, they won’t be charged to leave (they will, however, have some friction to leaving in that they’ll be asked why they’re leaving and if there’s anything Instagram can do to make them stay), but they will be leaving their family and friends that stay on the app. This is where data portability can’t really help because while you can export your photos, you can’t make your friends leave. This is part of what makes leaving a network or platform so hard. '''References:''' European Commission (2025, March 19). ''Commission provides guidance under Digital Markets Act to facilitate development of innovative products on Apple’s platforms''. https://digital-markets-act.ec.europa.eu/commission-provides-guidance-under-digital-markets-act-facilitate-development-innovative-products-2025-03-19_en MacTech (2025, March 19). ''Apple isn’t happy with the latest decisions regarding the DMA''. https://www.mactech.com/2025/03/19/apple-isnt-happy-with-the-latest-decisions-regarding-the-dma Shapiro, C. & Varian, H.R. (2010). ''Information Rules: A Strategic Guide to the Network Economy.'' Boston, MA: Harvard Business School Press. The Hill. (2025, June 30). ''Judge rejects Apple’s bid to dismiss DOJ antitrust case''. https://thehill.com/policy/technology/5380007-apple-federal-judge-antitrust-case/ The Mac Observer. (2026, September 12). ''Apple’s other antitrust fights: Where the DOJ case and the UK’s CMA stand now''. https://www.macobserver.com/news/apple-doj-antitrust-case-uk-cma-status/ U.S. Department of Justice (2024, March 21). ''Justice Department Sues Apple for Monopolizing Smartphone Markets'' [Press Release]. https://www.justice.gov/archives/opa/pr/justice-department-sues-apple-monopolizing-smartphone-markets

= Excludability and rivalrousness (Paul)  =
=== Excludability and Rivalrousness ===The terms excludability and rivalrousness both refer to consumption characteristics that can be assigned to- or identified in goods. '''Excludability''': Goods to which economic actors can technically or legally restrict access are considered to have ‘high excludability’. Conversely, goods to which access cannot be restricted are of ‘low excludability’, or, in absolute terms, ‘non-excludable’.'''Rivalrousness''': Rivalrousness on the other hand refers to the finitude of the good in question. The greater the depletion of the stock, quantity, or amount is on account of one party’s consumption, the more rivalrous a good is. Tabulating the characteristics against one another allows for inference and definition of the type of good an object, service, or resource may constitute (see fig. 1). Excludability and Rivalrousness of goods can therethrough serve as a predictor for certain types of market failures: Goods which are non-excludable or of low excludability, but nonetheless costly to provide, are vulnerable to so-called free rider problem. Goods which are non-excludable, but highly rivalrous (common goods), may lead to the tragedy of the commons, that is over-exploitation.

[[image:Untitled Diagram.drawio (1).png|300px|thumb]]
=== AI Training and the (non-)excludability of information ===The unusual nature of excludability and rivalrousness of data further compromises the applicability of the goods classification built thereon. In traditional economics, a non-excludable and non-rivalrous good would be seen as a public good. This does apply insofar as data is vulnerable to corporate exploitation without commensurate indemnification. Data usage may however actually be anti-rival in the sense that increasing aggregate usage increases the value of the data (better labeling, easier access, previously undrawn conclusions, etc.), rather than diminish the ‘stock’ of data. Conversely, data may be rivalrous not in its absolute depletion by consumption, but rather the benefit or competitive advantage usage of that data may confer: If all AI firms are able to train on the exact same data (all available data), the utility of that data may still increase, but the competitive advantage of possessing that data is exhausted.

==== Free-Rider Problems & Public Data ====As any party can benefit from publicly available data at no cost or low cost, the incentive for private market actors to fund its creation is comparatively weak. Consequently, public bodies (and less frequently open-source foundations and non-profits) often assume the responsibility for the provision of this type of data. A classical example is weather data and its creation by national weather agencies.

= Interoperability (Sisi) =

#What is interoperability?[[https://www.ibm.com/think/topics/interoperability 1]]Interoperability is a “standards-based” approach that enables different IT systems to communicate with minimal end-user interaction to exchange data and “share functionality”. Each system also allows the other system to “read, update, modify, and analyze” data with minimal human interaction. [[https://www.oracle.com/nl/interoperability/ 2]]Interoperability works by having different systems share data by employing the same standards for data formats (e.g., Extensible Markup Language or JavaScript Object Notation), exchange, and vocabulary to title data elements. These standards are agreed upon “in advance” by organizations, industry types, and user groups. Then, software developers build the system based on the model agreed upon.

#Concrete example from a digital technology or platformInteroperable systems are vital in healthcare, commerce, and public safety. A notable example includes the integration of Electronic Health Records (HER). This system allows different healthcare providers to securely access and share personal patient information regardless of the specific EHR system used.[[https://data.europa.eu/en/news-events/news/interoperability-solutions-bridging-systems-connected-future 3]] An example of such a system is the GNU Health Hospital Information System, which is a model employed by small private offices and large national public health systems.[[https://interoperable-europe.ec.europa.eu/collection/gnu-health/solution/gnu-health-hospital-information-system 4]]For example, a health care professional may access lab results and patient history data from various platforms to enable them to make a diagnosis.

#One governance implication: what does this concept explain about how platforms exercise power?Interoperability is a powerful tool because it helps provide a “seamless user experience” and optimizes data sharing between separate systems. It helps companies and organizations access data from multiple sources and improve efficiency, such as accelerating workflows and reducing the possibility of errors by pooling data from its sources.Interoperability standards are mandated by law on the technical side to ensure “critical systems” are compatible, especially in public safety and health. [[https://www.ibm.com/think/topics/interoperability 5]]In Europe, interoperability is governed by the European Health Data Space regulation (EHDS); the “first common EU data space dedicated to a specific sector as part of the European strategy for data.”This Regulation aims to benefit EU citizens such as patients, researchers, healthcare professionals, and industry players by establishing a common framework for the exchange of electronic health data within the EU. In addition, it enables certain data to be “reused” for public purpose and innovative scientific research. [[https://health.ec.europa.eu/ehealth-digital-health-and-care/european-health-data-space-regulation-ehds_en 6]]

#Where does the concept break down or fail to explain somethingWhile the EDHS harmonizes a legal and technical framework for sharing EU data, its shortcoming is its gradual implementation. The EDHS entered into force in 2025; however, most of its obligations will not apply until 2029. This leaves four years for health care companies to use interoperability in a deregulated manner. [[https://health.ec.europa.eu/ehealth-digital-health-and-care/european-health-data-space-regulation-ehds_en 7]]Furthermore, there are actual barriers to achieving full interoperability in healthcare, including data silos (which isolate data collections and hinder data exchange), lack of standardization (where one system records blood pressure as free text, another system stores it as a structured numeric value), and deficient infrastructure, which requires significant technological upgrading.The burden remains on health companies to decide which standards to adopt and how much to invest in stable technological infrastructure to strengthen their data governance and interoperability. [[https://www.tigahealth.com/how-to-overcome-healthcare-interoperability-challenges/ 8]]

= [[Economies of Scale (Jihae)]] =
= [[Two and Multi-Sided Markets Robert]] =

= [[Economies of Scope (Mariona)]] =
= [[Network Effects (Indirect) - Sebastián]] =
= [[Monopsony - (Atmika)]] =
== Definition: ==
 A Monopsony is a market condition where there is one buyer and multiple sellers; the opposite of the better known non-competitive market - Monopoly. Here, the buyer has the power to control the price and set market conditions. 
== Digital platform: Uber ==
'''Definition:''' 
 Uber was released into the market as an online platform to connect drivers with customers - but it’s business model created a huge social and economic impact in the labor sector. (1) Uber engaged drivers on a contract basis and offered short-term, low commitment and flexible jobs, which we characterise as the ‘[https://en.wikipedia.org/wiki/Gig_economy Gig economy]’. Uber offered a better transportation network alternative - dynamic pricing, more operating efficiency and on-demand mobility in contrast to the rigid capacities and fares of conventional transportational means available (mainly taxicabs), which would soon be replaced.(2) Additional social perspectives showed fair, non-discriminatory and safer practices - such as access and availability to lower economic groups.   

 '''How power is exercised '''

 1. Capturing labour force: The primary economic impact in the labour space is with regard to human capital mobilisation; a 2014 survey showed that 36% of 600 Uber drivers reported neither having a job, nor actively seeking work, before signing up to drive for Uber. (3) The presence of Uber resulted in increased regional employment. Drivers valued the flexibility and self-supervision enough to accept lower earnings than in previous jobs.

 2. Control of price and contractual terms: Uber exercises vertical control by dictating terms on which drivers transact and fixed prices - independent contractors are not allowed liberty to negotiate. 

 3. Horizontal friction: Assimilation and use of dominant position in one sectoral space to be able to expand into another such as in the case of Uber Eats. While Uber has since started to reduce it’s region specific operations by selling their holdings to the local leading brand, such as in China with Didi, they maintain equity shares as a control mechanism. 

 '''Where does the concept break down or fail to explain?'''

 A highly non-competitive market structure, is one wherein the buyer and seller cannot freely move within the market, curbed by inefficiency, high barrier of entry and lack of innovation. But modern day monopsony looks alot more different than this traditional concept. For one in this case, Uber was a new innovation brought into the market to outdate the old inefficient model of taxis with more control by its users. Moreover, monopsony can exist even in settings that appear easy to enter or leave.(4)

 While it may not be a pure monopsony, Uber does hold 71% market shares in the TNC space in the United States. However, the lack of barriers for entry saw Lyft introduced at the same time, where both provided similar models and impacts, with consumers switching between platforms. 

'''References:'''

 (1) Koling, A., Michalek, J., Armanios, D. et al. Effects of Uber and Lyft on jobs, wages and GDP. Nat Cities 3, 700–707 (2026). https://doi.org/10.1038/s44284-026-00478-0 (2) Zha, L., Yin, Y. & Du, Y. Surge pricing and labor supply in the ridesourcing market. Transport. Res. Proc. 23, 2–21 (2017). (3) Levin, A. The Driver Roadmap: Where Uber Driver-Partners Have Been, and Where They’re Going (Benenson Strategy Group, 2014). (4) Manning, A. (2021). Monopsony in Labor Markets: A Review. ILR Review, 74(1), 3-26. https://doi.org/10.1177/0019793920922499.

===== [https://wiki.ivir.nl/index.php?title=Market_tipping_-_Rodrigue Market Tipping - Rodrigue] =====

= Multi-homing (Juliette) =

1/ Plain-language definition

Multi-homing means using two or more competing platforms at the same time, for the same thing. A shop that sells on both Amazon and Bol.com is multi-homing. A person who has both WhatsApp and Signal on their phone is multi-homing too. The opposite is single-homing, when you use only one platform. When people multi-home, it is easier for them to leave a platform, because they already have another one ready (1).

2/ Concrete example

Taxi apps are a good example. In many cities, drivers have Uber and Bolt open at the same time. They take the first good ride that comes. Many passengers, however, use only one app because they are used to it. So drivers can move between apps, but each app keeps its own group of loyal passengers. If a driver wants those passengers, he has to stay on that app. This means the app still has power over drivers, even if they use several apps (2).

3/ Governance implication

Multi-homing shows why platforms try hard to keep users on one service only. They do not usually ban other apps. Instead, they make using them less attractive. Taxi apps give drivers bonuses if they do a certain number of rides on the same app. Online shops and hotel booking sites have stopped sellers from offering lower prices on other websites. In this way, the platform makes its own rules. It controls what users do without a formal contract. Regulators now want to protect multi-homing. The Digital Markets Act’s article 5.3 says large platforms cannot stop businesses from offering better prices elsewhere. It also says users must be able to take their data with them (article 6.9) and leave a service easily (article 6.13) (3). The idea is that if people can use many platforms, no single platform can control them.

4/ Open question

We often think that multi-homing reduces the power of platforms, even if it is not always true. If only one side multi-homes, the platform can ask that side to pay more, because it controls access to the loyal side. Research shows that the result depends on who multi-homes. Sometimes the group that should win actually loses (2). The authors Bryan and Gans even find that in some cities, one taxi app can be better for society than several (4). Also, multi-homing does not show how much people really use each platform. Someone can have two accounts but use only one. While multi-homing is easy to see, his understanding remains complex.

5/ Bibliography

(1) Belleflamme P and Peitz M, 'Platform Competition: Who Benefits from Multihoming?' (2019) 64 International Journal of Industrial Organization 1

(2) Bryan KA and Gans JS, 'A Theory of Multihoming in Rideshare Competition' (2019) 28 Journal of Economics & Management Strategy 89

(3) Regulation (EU) 2022/1925 (Digital Markets Act) [2022] OJ L265/1

(4) Zhu F and Iansiti M, 'Why Some Platforms Thrive and Others Don't' (2019) Harvard Business Review (January–February) 118

= Vertical Integration by Pablo Latorre ='''1. Definition'''“Vertical Integration” is the control by a single corporation of two or more consecutive steps of the same production chain of activity in such a manner that said corporation does not need to rely on others for them (1). In the context of digital markets, “Vertical Integration” usually refers to the position of a platform that operates a marketplace or infrastructure that other enterprises depend on and, at the same time, commercializes products or services in that same space (2). Therefore, in a “Vertical Integration” the platform is both, provider and competitor, of other enterprises. '''2. Example: Amazon as a Marketplace'''Amazon operates the largest online marketplace (3) in the world and, thus, thousands of small and medium size enterprises depend on it to connect with their costumers. At the same time, Amazon commercializes its own products on its marketplace (4), acting as a retailer and competitor to those enterprises. Such dual role allows Amazon to exercise competitive advantages: first, it can access detailed, non-public data on the enterprises’ sales, including products, prices, and costumer behavior; and second, it can decide the rules that define the products that appear first as the result of a costumer search.In 2022 Amazon had to commit to the European Commission to not use the enterprises’ non-public data for its retail business and to treat equally rival offers in its “Buy Box” (5). Currently, the Digital Markets Act (DMA) establishes matching general obligations for designated gatekeepers (6). '''3. Governance Implication'''“Vertical Integration” allows platforms to exercise power over enterprises through digital architecture. When a platform operates a marketplace and competes in it, its ordinary decisions on digital architecture (i.e. on algorithms, searches, products visibility) can become tools for power and competition. In a sense, platforms are the ''de facto'' private regulators of markets in which they have clear commercial interests. Such state of affairs allows to understand a strategy change in antitrust cases. Traditionally, antitrust intervenes after the harm, however regulators are moving towards ''ex ante'' conduct rules (i.e. the DMA) (7). '''4. Open Question'''“Vertical Integration” clearly explains the incentive structure that drives platforms to favor their own products through the exercise of digital architecture mediated competitive advantages. However, “Vertical Integration” assumes that the line between one product (i.e. the marketplace, the search engine) and the other (i.e. the products, the websites competing on the platform) is clear when, in practice, this is not always so. For instance, nowadays when a user search something in Google Search, Google not only provides links to websites that contain the desired information but also the desired information itself with AI-generated summaries. In this case, two possible perspectives arise: on the one hand, that Google could be abusing its position to capture the online information market by displacing websites that depend on it; on the other hand, that Google could be simply improving its search engine and search itself because what users actually want is information and not links to it. In this example, “Vertical Integration” is incapable of settling the issue because it depends on a clear distinction between one product and the other (or one market and the other) and that is absent in this case. Moreover, even if in the presented case was possible to separate one product from the other (or one market from the other), “Vertical Integration” does not provide a clear criteria to establish when an “integration” is abusive exclusion of competitors and when it is legitimate and necessary innovation. '''References'''(1) “Vertical integration” (''Cambridge Dictionary'') [https://dictionary.cambridge.org/dictionary/english/vertical-integration https://dictionary.cambridge.org/dictionary/english/vertical-integration] accessed 4 October 2026.(2) [https://academic.oup.com/antitrust/article/6/3/355/5034073?itm_medium=sidebar&itm_source=trendmd-widget&itm_campaign=Journal_of_Antitrust_Enforcement&itm_content=Journal_of_Antitrust_Enforcement_0 Bostoen Friso, “Online Platforms and Vertical Integration: The Return of Margin Squeeze” (2018) (6) Journal of Antitrust Enforcement 355.](3) Lovell Keir, “Top 10 Ecommerce Marketplaces in the World: 2026 Rankings by GMV, Users & Reach” (''ServeRetail'', 2 January 2026) [https://www.serveretail.com/blog/marketplace/the-10-largest-ecommerce-marketplaces-in-the-world-size-reach-and-cx-strategy/ https://www.serveretail.com/blog/marketplace/the-10-largest-ecommerce-marketplaces-in-the-world-size-reach-and-cx-strategy/] accessed 4 October 2026.(4) Ellison Mara, “What are Amazon's Brands, Explained” (''Ir Mountaintopuniversity Ng'') [https://ir.mountaintopuniversity.edu.ng/article/what-are-amazon-s-brands-explained https://ir.mountaintopuniversity.edu.ng/article/what-are-amazon-s-brands-explained] accessed 4 October 2026.(5) [https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=oj:JOC_2023_087_R_0005 Summary of Commission Decision of 20 December 2022 Relating to a Proceeding under Article 102 of the Treaty on the Functioning of the European Union and Article 54 of the EEA Agreement (Cases AT.40462 – Amazon Marketplace, and AT.40703 – Amazon Buy Box) (2023) OJ C87.](6) [https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=uriserv%3AOJ.L_.2022.265.01.0001.01.ENG&toc=OJ%3AL%3A2022%3A265%3ATOC Regulation (EU) 2022/1925 of the European Parliament and of the Council of 14 September 2022 on contestable and fair markets in the digital sector and amending Directives (EU) 2019/1937 and (EU) 2020/1828 (Digital Markets Act) (2022).](7) OECD, "Ex Ante Regulation and Competition in Digital Markets" (''OECD Roundtables on Competition Policy Papers, OECD Publishing 2021'') [https://doi.org/10.1787/c83e178d-en https://doi.org/10.1787/c83e178d-en] accessed 4 October 2026.

Two- and Multi-Sided Markets Robert

2026-10-07T11:16:57Z

Robert Florijn: Robert Florijn moved page Two- and Multi-Sided Markets Robert to Two and Multi-Sided Markets Robert

#REDIRECT [[Two and Multi-Sided Markets Robert]]

Two and Multi-Sided Markets Robert

2026-10-07T11:16:57Z

Robert Florijn: Robert Florijn moved page Two- and Multi-Sided Markets Robert to Two and Multi-Sided Markets Robert

= '''Two-Sided and Multi-Sided Markets: How Platforms Govern Exchange''' =

#''' What does the concept mean?'''
A '''two-sided market''' connects two distinct groups through an intermediary, with participation or decisions on one side affecting outcomes or value on the other. Each group provides the other with network benefits, making the platform more valuable as participation grows (Katz & Shapiro, 1985). Price and openness shape ''particip''a''tion'' (Rysman, 2009). A '''multi-sided platform''' generates value primarily by facilitating direct interactions between more distinct types of customers (Haigiu & Wright, 2011). Google Pay, for example, requires both consumer usage and merchant acceptance. Platforms often charge the groups they support differently: fees from one side may fund free access for another.
#''' Example: Google Search'''
Google Search connects users seeking information with advertisers seeking their attention. Users pay no monetary fee, while advertisers pay for advertising placements. More users attract advertisers, but advertising can inconvenience users. Google must manage these competing interests. Publishers add another relationship: their content makes search useful, while search visibility influences their traffic.Google also determines rankings, data collection, and visibility. On 23 July 2026, the European Commission fined Google €890 million for favouring its own services in Google Search, in breach of Article 6(5) of the Digital Markets Act (DMA). Google was. Separately, it specified measures to implement Article 6(11): access to specified search data for eligible rival search engines on fair, reasonable, and non-discriminatory terms, with anonymisation safeguards. Even after anonymising information identifying search users, the dataset may still contain personal information about the people they searched for. The GDPR therefore still applies to that information.
#''' What does this explain about governance?'''
Two- or multi-sided market theory helps explain how commercial design choices govern exchange. Pricing, access rules, ease of use, rankings, technical standards, and data policies determine who can participate and on what terms. Growing participation can attract further participants and make entry harder for competitors. Such feedback can reinforce concentration, although monopoly is not inevitable (Shapiro & Varian, 1998; Crémer et al., 2019).The governance concern is dependence: businesses may need a platform to reach customers while competing with its own services. Article 102 TFEU addresses abuse by dominant undertakings; the DMA imposes specified obligations on designated gatekeepers to promote contestability and fairness. Dependence alone does not establish an infringement. Platform governance therefore combines private rule-setting with public oversight.
#''' Where does the concept fall short?'''
The model explains interdependence, but whose interests count as a “side”? Users are also data sources and rights-holders; publishers and workers face different dependencies. Crawford (2021) examines AI’s reliance on human labour, energy and extracted materials. These inputs entail operating costs, but associated social and environmental harms are not reflected in prices (Crawford, 2021)! The question is whether analysis centred on transactions and participation adequately captures externalities like privacy, information power, labour conditions and environmental harm. Governance must also consider human rights, public institutions, physical supply chains and environmental costs.

Governance by markets

2026-10-07T11:15:51Z

Robert Florijn:

= Glossary =
= [https://wiki.ivir.nl/index.php?title=Data_as_an_input_to_market_power Data as an input to market power] (Sergey) =In the present digital economy, data has emerged as a valuable economic commodity that entities leverage to gain a competitive edge over their rivals [[https://repository.nls.ac.in/cgi/viewcontent.cgi?article=1478&context=nlsblr 1]]. '''Data''' is an input into production, so businesses that control data can have market power and impede the entry of potential competitors. '''Market power''' refers to a firm's ability to influence the price, output, or other market conditions in its favour, without being constrained by competitive forces [[https://www.tutor2u.net/economics/reference/what-is-meant-by-market-power 2]].'''Data as an input to market power''' means that access to information can help a business limit the competitive pressure it faces. Information about searches, clicks or purchases can improve services and advertising. If competitors cannot obtain comparable information at a reasonable cost, they may struggle to offer an attractive alternative. This can allow a platform to raise prices or worsen quality or privacy protections without losing enough business to make doing so unprofitable. This matters even when a service is free, because competition also concerns advertising charges, personal information and the choices available. Data can be used repeatedly and by several businesses without being used up. However, access can be restricted through contracts and technical controls. Nevertheless, holding a large dataset alone does not establish market power.'''Google Search''' illustrates this mechanism. Google controls over 91% of global search engine traffic [[https://www.statista.com/statistics/1381664/worldwide-all-devices-market-share-of-search-engines/ 3]] and uses information about users’ searches and the results they click to improve the relevance of its rankings. A larger volume of searches gives it more opportunities to learn, including about unusual queries that smaller search engines encounter less often. Better results can attract further use, producing additional information and reinforcing the advantage. A new search engine may therefore struggle to attract users while lacking the information needed to improve its service. Data access can constitute a significant barrier to entry when this type of analytics is necessary to participate in a market, particularly when alternatives are limited. A platform’s control over data access can shape who is able to compete with it. By determining whether, and on what terms, rivals can access that information, a platform can influence the competitive opportunities available to others. This suggests that regulation should examine access to strategically important data alongside prices and market shares. Article 6(11) of the EU Digital Markets Act [[https://eur-lex.europa.eu/eli/reg/2022/1925/oj/eng#:~:text=giving%20their%20consent.-,11.%C2%A0,-The%20gatekeeper%20shall 4]] reflects this approach: certain large search providers must, on request, share specified search data with rival search providers on fair and non-discriminatory terms, with relevant personal data anonymised. The aim is to help rivals compete while protecting users. A data advantage does not, by itself, establish unlawful conduct.Would access to equivalent data enable a rival to compete effectively? The concept cannot answer this on its own. Additional data may eventually add little value, become outdated, or have substitutes. Firms also need the skills and computing resources to use it. Google’s default positions in browsers and devices can further restrict rivals’ access to users even if data become available. The unresolved issue is how much competitive advantage comes from data itself and how much comes from these other resources and arrangements. 

= = Two Sided Markets (Ceren) = =In two-sided markets, two sides of a market interact through an intermediary. These markets typically have some externality,<code class="mwt-code">Carl Shapiro and Hal R Varian, ''Information Rules: A Strategic Guide to the Network Economy'' (Harvard Business School Press 2010), 183.</code> where the decisions of the sides affect the other. The intermediary also has multitudes of effects on both sides, mainly identified in the related literature through its pricing.An example of a two sided market is payment systems, whereby the credit card providing bank is the intermediary between the customers and merchants. In this example, banks provide credit cards to customers to be utilized in their payments, and merchants accept payment via multiple credit card systems to accommodate payments from customers holding credit cards from such providers[https://doi.org/10.1257/jep.23.3.125 .]The main power of the intermediary materializes in their pricing toward the merchant (and the merchant’s bank in interchange fees in case of payment being transferred through different banks). Customers will be expected to be more loyal, holding a few credit cards, whereas merchants accept payment from many banks to accommodate all customers’ payment methods. Banks will provide more subsidies to customers to attract them and compete aggressively on the customer side, making their platform more desirable to work with by merchants, who will in turn be more accepting of the platform’s fees to gain the business of the larger volume of customers. However, dynamic pricing (increasing fees after gaining entry and loyalty) may be a way for these platforms to abuse their power.The concept may be derailed from these logical / analytical systems of operation or expected outcomes when public policy interferes with the market. For example, net neutrality rules regulate what internet service providers (the intermediary) can charge content providers (one side) who aim to reach the consumers (the other side). In this case the above considerations and the concept are halted.

= Switching costs and lock-in (Kate) =
'''Definition '''When you move from one product or service to another, switching costs are everything you have to either give up or spend. This can include money, time, lost contacts, etc. Lock-in is when you feel (and are) trapped by these costs that add up and you decide instead of moving products or services, it’s just easier to stay with a service, not because you want to but because it’s just too much of a hassle. This is true even if a less expensive or better service exists all because leaving the product/service just costs too much in terms of time, effort, and money. Shapiro and Varian (2010) argue that switching costs are normal in information markets and that the full cost of a switch includes what you as the customer pay and what a new supplier spends to win that customer over.  '''Example: Apple''' The more Apple products you own, the harder it is to leave. As an avid Apple user, I am absolutely stuck in their ecosystem. I have my photos in Apple’s iCloud, my group chats in iMessage, and my Apple Watch only works with an iPhone. Leaving Apple is not easy. So much so that in March 2024, the US Department of Justice sued Apple alleging it raises costs of leaving/switching phones by making messaging worse and less secure when you’re messaging with non iPhones. In June 2025, Judge Julien Neals denied Apple’s motion to dismiss and the case is still in pre-trial proceedings.  '''Governance implication''' Lock-in makes users dependent on a platform and because of that, once we’re locked in, the platform (Meta, Apple, Google, etc) can raise prices, show more ads or even provide lower quality terms. So in this way, the platform can control users because if a user wants to leave, it’s going to cost you in that the platform has already designed your exit including how you can export your data, licenses and APIs. Under the EU’s Digital Markets Act, the European Commission adopted decisions in March 2025 making Apple open nine iOS connectivity features, mostly used by devices like smartwatches, headphones and televisions. These decisions are legally binding and they focus on lock-in by lowering the cost of choosing non-Apple products. It’s worth noting, however, that they only apply in the EU. When the Commission adopted the decisions, Apple said (reported by MacTech) “It’s bad for our products and for our European users.” '''Open Question''' So what happens when the cost of leaving isn’t just your own costs as a user but also the cost of leaving family and friends behind?  If a user leaves, let’s say, Instagram, they won’t be charged to leave (they will, however, have some friction to leaving in that they’ll be asked why they’re leaving and if there’s anything Instagram can do to make them stay), but they will be leaving their family and friends that stay on the app. This is where data portability can’t really help because while you can export your photos, you can’t make your friends leave. This is part of what makes leaving a network or platform so hard. '''References:''' European Commission (2025, March 19). ''Commission provides guidance under Digital Markets Act to facilitate development of innovative products on Apple’s platforms''. https://digital-markets-act.ec.europa.eu/commission-provides-guidance-under-digital-markets-act-facilitate-development-innovative-products-2025-03-19_en MacTech (2025, March 19). ''Apple isn’t happy with the latest decisions regarding the DMA''. https://www.mactech.com/2025/03/19/apple-isnt-happy-with-the-latest-decisions-regarding-the-dma Shapiro, C. & Varian, H.R. (2010). ''Information Rules: A Strategic Guide to the Network Economy.'' Boston, MA: Harvard Business School Press. The Hill. (2025, June 30). ''Judge rejects Apple’s bid to dismiss DOJ antitrust case''. https://thehill.com/policy/technology/5380007-apple-federal-judge-antitrust-case/ The Mac Observer. (2026, September 12). ''Apple’s other antitrust fights: Where the DOJ case and the UK’s CMA stand now''. https://www.macobserver.com/news/apple-doj-antitrust-case-uk-cma-status/ U.S. Department of Justice (2024, March 21). ''Justice Department Sues Apple for Monopolizing Smartphone Markets'' [Press Release]. https://www.justice.gov/archives/opa/pr/justice-department-sues-apple-monopolizing-smartphone-markets

= Excludability and rivalrousness (Paul)  =
=== Excludability and Rivalrousness ===The terms excludability and rivalrousness both refer to consumption characteristics that can be assigned to- or identified in goods. '''Excludability''': Goods to which economic actors can technically or legally restrict access are considered to have ‘high excludability’. Conversely, goods to which access cannot be restricted are of ‘low excludability’, or, in absolute terms, ‘non-excludable’.'''Rivalrousness''': Rivalrousness on the other hand refers to the finitude of the good in question. The greater the depletion of the stock, quantity, or amount is on account of one party’s consumption, the more rivalrous a good is. Tabulating the characteristics against one another allows for inference and definition of the type of good an object, service, or resource may constitute (see fig. 1). Excludability and Rivalrousness of goods can therethrough serve as a predictor for certain types of market failures: Goods which are non-excludable or of low excludability, but nonetheless costly to provide, are vulnerable to so-called free rider problem. Goods which are non-excludable, but highly rivalrous (common goods), may lead to the tragedy of the commons, that is over-exploitation.

[[image:Untitled Diagram.drawio (1).png|300px|thumb]]
=== AI Training and the (non-)excludability of information ===The unusual nature of excludability and rivalrousness of data further compromises the applicability of the goods classification built thereon. In traditional economics, a non-excludable and non-rivalrous good would be seen as a public good. This does apply insofar as data is vulnerable to corporate exploitation without commensurate indemnification. Data usage may however actually be anti-rival in the sense that increasing aggregate usage increases the value of the data (better labeling, easier access, previously undrawn conclusions, etc.), rather than diminish the ‘stock’ of data. Conversely, data may be rivalrous not in its absolute depletion by consumption, but rather the benefit or competitive advantage usage of that data may confer: If all AI firms are able to train on the exact same data (all available data), the utility of that data may still increase, but the competitive advantage of possessing that data is exhausted.

==== Free-Rider Problems & Public Data ====As any party can benefit from publicly available data at no cost or low cost, the incentive for private market actors to fund its creation is comparatively weak. Consequently, public bodies (and less frequently open-source foundations and non-profits) often assume the responsibility for the provision of this type of data. A classical example is weather data and its creation by national weather agencies.

= Interoperability (Sisi) =

#What is interoperability?[[https://www.ibm.com/think/topics/interoperability 1]]Interoperability is a “standards-based” approach that enables different IT systems to communicate with minimal end-user interaction to exchange data and “share functionality”. Each system also allows the other system to “read, update, modify, and analyze” data with minimal human interaction. [[https://www.oracle.com/nl/interoperability/ 2]]Interoperability works by having different systems share data by employing the same standards for data formats (e.g., Extensible Markup Language or JavaScript Object Notation), exchange, and vocabulary to title data elements. These standards are agreed upon “in advance” by organizations, industry types, and user groups. Then, software developers build the system based on the model agreed upon.

#Concrete example from a digital technology or platformInteroperable systems are vital in healthcare, commerce, and public safety. A notable example includes the integration of Electronic Health Records (HER). This system allows different healthcare providers to securely access and share personal patient information regardless of the specific EHR system used.[[https://data.europa.eu/en/news-events/news/interoperability-solutions-bridging-systems-connected-future 3]] An example of such a system is the GNU Health Hospital Information System, which is a model employed by small private offices and large national public health systems.[[https://interoperable-europe.ec.europa.eu/collection/gnu-health/solution/gnu-health-hospital-information-system 4]]For example, a health care professional may access lab results and patient history data from various platforms to enable them to make a diagnosis.

#One governance implication: what does this concept explain about how platforms exercise power?Interoperability is a powerful tool because it helps provide a “seamless user experience” and optimizes data sharing between separate systems. It helps companies and organizations access data from multiple sources and improve efficiency, such as accelerating workflows and reducing the possibility of errors by pooling data from its sources.Interoperability standards are mandated by law on the technical side to ensure “critical systems” are compatible, especially in public safety and health. [[https://www.ibm.com/think/topics/interoperability 5]]In Europe, interoperability is governed by the European Health Data Space regulation (EHDS); the “first common EU data space dedicated to a specific sector as part of the European strategy for data.”This Regulation aims to benefit EU citizens such as patients, researchers, healthcare professionals, and industry players by establishing a common framework for the exchange of electronic health data within the EU. In addition, it enables certain data to be “reused” for public purpose and innovative scientific research. [[https://health.ec.europa.eu/ehealth-digital-health-and-care/european-health-data-space-regulation-ehds_en 6]]

#Where does the concept break down or fail to explain somethingWhile the EDHS harmonizes a legal and technical framework for sharing EU data, its shortcoming is its gradual implementation. The EDHS entered into force in 2025; however, most of its obligations will not apply until 2029. This leaves four years for health care companies to use interoperability in a deregulated manner. [[https://health.ec.europa.eu/ehealth-digital-health-and-care/european-health-data-space-regulation-ehds_en 7]]Furthermore, there are actual barriers to achieving full interoperability in healthcare, including data silos (which isolate data collections and hinder data exchange), lack of standardization (where one system records blood pressure as free text, another system stores it as a structured numeric value), and deficient infrastructure, which requires significant technological upgrading.The burden remains on health companies to decide which standards to adopt and how much to invest in stable technological infrastructure to strengthen their data governance and interoperability. [[https://www.tigahealth.com/how-to-overcome-healthcare-interoperability-challenges/ 8]]

= [[Economies of Scale (Jihae)]] =

= [[Economies of Scope (Mariona)]] =
= [[Network Effects (Indirect) - Sebastián]] =
= [[Monopsony - (Atmika)]] =
== Definition: ==
 A Monopsony is a market condition where there is one buyer and multiple sellers; the opposite of the better known non-competitive market - Monopoly. Here, the buyer has the power to control the price and set market conditions. 
== Digital platform: Uber ==
'''Definition:''' 
 Uber was released into the market as an online platform to connect drivers with customers - but it’s business model created a huge social and economic impact in the labor sector. (1) Uber engaged drivers on a contract basis and offered short-term, low commitment and flexible jobs, which we characterise as the ‘[https://en.wikipedia.org/wiki/Gig_economy Gig economy]’. Uber offered a better transportation network alternative - dynamic pricing, more operating efficiency and on-demand mobility in contrast to the rigid capacities and fares of conventional transportational means available (mainly taxicabs), which would soon be replaced.(2) Additional social perspectives showed fair, non-discriminatory and safer practices - such as access and availability to lower economic groups.   

 '''How power is exercised '''

 1. Capturing labour force: The primary economic impact in the labour space is with regard to human capital mobilisation; a 2014 survey showed that 36% of 600 Uber drivers reported neither having a job, nor actively seeking work, before signing up to drive for Uber. (3) The presence of Uber resulted in increased regional employment. Drivers valued the flexibility and self-supervision enough to accept lower earnings than in previous jobs.

 2. Control of price and contractual terms: Uber exercises vertical control by dictating terms on which drivers transact and fixed prices - independent contractors are not allowed liberty to negotiate. 

 3. Horizontal friction: Assimilation and use of dominant position in one sectoral space to be able to expand into another such as in the case of Uber Eats. While Uber has since started to reduce it’s region specific operations by selling their holdings to the local leading brand, such as in China with Didi, they maintain equity shares as a control mechanism. 

 '''Where does the concept break down or fail to explain?'''

 A highly non-competitive market structure, is one wherein the buyer and seller cannot freely move within the market, curbed by inefficiency, high barrier of entry and lack of innovation. But modern day monopsony looks alot more different than this traditional concept. For one in this case, Uber was a new innovation brought into the market to outdate the old inefficient model of taxis with more control by its users. Moreover, monopsony can exist even in settings that appear easy to enter or leave.(4)

 While it may not be a pure monopsony, Uber does hold 71% market shares in the TNC space in the United States. However, the lack of barriers for entry saw Lyft introduced at the same time, where both provided similar models and impacts, with consumers switching between platforms. 

'''References:'''

 (1) Koling, A., Michalek, J., Armanios, D. et al. Effects of Uber and Lyft on jobs, wages and GDP. Nat Cities 3, 700–707 (2026). https://doi.org/10.1038/s44284-026-00478-0 (2) Zha, L., Yin, Y. & Du, Y. Surge pricing and labor supply in the ridesourcing market. Transport. Res. Proc. 23, 2–21 (2017). (3) Levin, A. The Driver Roadmap: Where Uber Driver-Partners Have Been, and Where They’re Going (Benenson Strategy Group, 2014). (4) Manning, A. (2021). Monopsony in Labor Markets: A Review. ILR Review, 74(1), 3-26. https://doi.org/10.1177/0019793920922499.

===== [https://wiki.ivir.nl/index.php?title=Market_tipping_-_Rodrigue Market Tipping - Rodrigue] =====

= Multi-homing (Juliette) =

1/ Plain-language definition

Multi-homing means using two or more competing platforms at the same time, for the same thing. A shop that sells on both Amazon and Bol.com is multi-homing. A person who has both WhatsApp and Signal on their phone is multi-homing too. The opposite is single-homing, when you use only one platform. When people multi-home, it is easier for them to leave a platform, because they already have another one ready (1).

2/ Concrete example

Taxi apps are a good example. In many cities, drivers have Uber and Bolt open at the same time. They take the first good ride that comes. Many passengers, however, use only one app because they are used to it. So drivers can move between apps, but each app keeps its own group of loyal passengers. If a driver wants those passengers, he has to stay on that app. This means the app still has power over drivers, even if they use several apps (2).

3/ Governance implication

Multi-homing shows why platforms try hard to keep users on one service only. They do not usually ban other apps. Instead, they make using them less attractive. Taxi apps give drivers bonuses if they do a certain number of rides on the same app. Online shops and hotel booking sites have stopped sellers from offering lower prices on other websites. In this way, the platform makes its own rules. It controls what users do without a formal contract. Regulators now want to protect multi-homing. The Digital Markets Act’s article 5.3 says large platforms cannot stop businesses from offering better prices elsewhere. It also says users must be able to take their data with them (article 6.9) and leave a service easily (article 6.13) (3). The idea is that if people can use many platforms, no single platform can control them.

4/ Open question

We often think that multi-homing reduces the power of platforms, even if it is not always true. If only one side multi-homes, the platform can ask that side to pay more, because it controls access to the loyal side. Research shows that the result depends on who multi-homes. Sometimes the group that should win actually loses (2). The authors Bryan and Gans even find that in some cities, one taxi app can be better for society than several (4). Also, multi-homing does not show how much people really use each platform. Someone can have two accounts but use only one. While multi-homing is easy to see, his understanding remains complex.

5/ Bibliography

(1) Belleflamme P and Peitz M, 'Platform Competition: Who Benefits from Multihoming?' (2019) 64 International Journal of Industrial Organization 1

(2) Bryan KA and Gans JS, 'A Theory of Multihoming in Rideshare Competition' (2019) 28 Journal of Economics & Management Strategy 89

(3) Regulation (EU) 2022/1925 (Digital Markets Act) [2022] OJ L265/1

(4) Zhu F and Iansiti M, 'Why Some Platforms Thrive and Others Don't' (2019) Harvard Business Review (January–February) 118

= Vertical Integration by Pablo Latorre ='''1. Definition'''“Vertical Integration” is the control by a single corporation of two or more consecutive steps of the same production chain of activity in such a manner that said corporation does not need to rely on others for them (1). In the context of digital markets, “Vertical Integration” usually refers to the position of a platform that operates a marketplace or infrastructure that other enterprises depend on and, at the same time, commercializes products or services in that same space (2). Therefore, in a “Vertical Integration” the platform is both, provider and competitor, of other enterprises. '''2. Example: Amazon as a Marketplace'''Amazon operates the largest online marketplace (3) in the world and, thus, thousands of small and medium size enterprises depend on it to connect with their costumers. At the same time, Amazon commercializes its own products on its marketplace (4), acting as a retailer and competitor to those enterprises. Such dual role allows Amazon to exercise competitive advantages: first, it can access detailed, non-public data on the enterprises’ sales, including products, prices, and costumer behavior; and second, it can decide the rules that define the products that appear first as the result of a costumer search.In 2022 Amazon had to commit to the European Commission to not use the enterprises’ non-public data for its retail business and to treat equally rival offers in its “Buy Box” (5). Currently, the Digital Markets Act (DMA) establishes matching general obligations for designated gatekeepers (6). '''3. Governance Implication'''“Vertical Integration” allows platforms to exercise power over enterprises through digital architecture. When a platform operates a marketplace and competes in it, its ordinary decisions on digital architecture (i.e. on algorithms, searches, products visibility) can become tools for power and competition. In a sense, platforms are the ''de facto'' private regulators of markets in which they have clear commercial interests. Such state of affairs allows to understand a strategy change in antitrust cases. Traditionally, antitrust intervenes after the harm, however regulators are moving towards ''ex ante'' conduct rules (i.e. the DMA) (7). '''4. Open Question'''“Vertical Integration” clearly explains the incentive structure that drives platforms to favor their own products through the exercise of digital architecture mediated competitive advantages. However, “Vertical Integration” assumes that the line between one product (i.e. the marketplace, the search engine) and the other (i.e. the products, the websites competing on the platform) is clear when, in practice, this is not always so. For instance, nowadays when a user search something in Google Search, Google not only provides links to websites that contain the desired information but also the desired information itself with AI-generated summaries. In this case, two possible perspectives arise: on the one hand, that Google could be abusing its position to capture the online information market by displacing websites that depend on it; on the other hand, that Google could be simply improving its search engine and search itself because what users actually want is information and not links to it. In this example, “Vertical Integration” is incapable of settling the issue because it depends on a clear distinction between one product and the other (or one market and the other) and that is absent in this case. Moreover, even if in the presented case was possible to separate one product from the other (or one market from the other), “Vertical Integration” does not provide a clear criteria to establish when an “integration” is abusive exclusion of competitors and when it is legitimate and necessary innovation. '''References'''(1) “Vertical integration” (''Cambridge Dictionary'') [https://dictionary.cambridge.org/dictionary/english/vertical-integration https://dictionary.cambridge.org/dictionary/english/vertical-integration] accessed 4 October 2026.(2) [https://academic.oup.com/antitrust/article/6/3/355/5034073?itm_medium=sidebar&itm_source=trendmd-widget&itm_campaign=Journal_of_Antitrust_Enforcement&itm_content=Journal_of_Antitrust_Enforcement_0 Bostoen Friso, “Online Platforms and Vertical Integration: The Return of Margin Squeeze” (2018) (6) Journal of Antitrust Enforcement 355.](3) Lovell Keir, “Top 10 Ecommerce Marketplaces in the World: 2026 Rankings by GMV, Users & Reach” (''ServeRetail'', 2 January 2026) [https://www.serveretail.com/blog/marketplace/the-10-largest-ecommerce-marketplaces-in-the-world-size-reach-and-cx-strategy/ https://www.serveretail.com/blog/marketplace/the-10-largest-ecommerce-marketplaces-in-the-world-size-reach-and-cx-strategy/] accessed 4 October 2026.(4) Ellison Mara, “What are Amazon's Brands, Explained” (''Ir Mountaintopuniversity Ng'') [https://ir.mountaintopuniversity.edu.ng/article/what-are-amazon-s-brands-explained https://ir.mountaintopuniversity.edu.ng/article/what-are-amazon-s-brands-explained] accessed 4 October 2026.(5) [https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=oj:JOC_2023_087_R_0005 Summary of Commission Decision of 20 December 2022 Relating to a Proceeding under Article 102 of the Treaty on the Functioning of the European Union and Article 54 of the EEA Agreement (Cases AT.40462 – Amazon Marketplace, and AT.40703 – Amazon Buy Box) (2023) OJ C87.](6) [https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=uriserv%3AOJ.L_.2022.265.01.0001.01.ENG&toc=OJ%3AL%3A2022%3A265%3ATOC Regulation (EU) 2022/1925 of the European Parliament and of the Council of 14 September 2022 on contestable and fair markets in the digital sector and amending Directives (EU) 2019/1937 and (EU) 2020/1828 (Digital Markets Act) (2022).](7) OECD, "Ex Ante Regulation and Competition in Digital Markets" (''OECD Roundtables on Competition Policy Papers, OECD Publishing 2021'') [https://doi.org/10.1787/c83e178d-en https://doi.org/10.1787/c83e178d-en] accessed 4 October 2026.

Governance by markets

2026-10-07T11:09:36Z

Robert Florijn:

= Glossary =
= [https://wiki.ivir.nl/index.php?title=Data_as_an_input_to_market_power Data as an input to market power] (Sergey) =In the present digital economy, data has emerged as a valuable economic commodity that entities leverage to gain a competitive edge over their rivals [[https://repository.nls.ac.in/cgi/viewcontent.cgi?article=1478&context=nlsblr 1]]. '''Data''' is an input into production, so businesses that control data can have market power and impede the entry of potential competitors. '''Market power''' refers to a firm's ability to influence the price, output, or other market conditions in its favour, without being constrained by competitive forces [[https://www.tutor2u.net/economics/reference/what-is-meant-by-market-power 2]].'''Data as an input to market power''' means that access to information can help a business limit the competitive pressure it faces. Information about searches, clicks or purchases can improve services and advertising. If competitors cannot obtain comparable information at a reasonable cost, they may struggle to offer an attractive alternative. This can allow a platform to raise prices or worsen quality or privacy protections without losing enough business to make doing so unprofitable. This matters even when a service is free, because competition also concerns advertising charges, personal information and the choices available. Data can be used repeatedly and by several businesses without being used up. However, access can be restricted through contracts and technical controls. Nevertheless, holding a large dataset alone does not establish market power.'''Google Search''' illustrates this mechanism. Google controls over 91% of global search engine traffic [[https://www.statista.com/statistics/1381664/worldwide-all-devices-market-share-of-search-engines/ 3]] and uses information about users’ searches and the results they click to improve the relevance of its rankings. A larger volume of searches gives it more opportunities to learn, including about unusual queries that smaller search engines encounter less often. Better results can attract further use, producing additional information and reinforcing the advantage. A new search engine may therefore struggle to attract users while lacking the information needed to improve its service. Data access can constitute a significant barrier to entry when this type of analytics is necessary to participate in a market, particularly when alternatives are limited. A platform’s control over data access can shape who is able to compete with it. By determining whether, and on what terms, rivals can access that information, a platform can influence the competitive opportunities available to others. This suggests that regulation should examine access to strategically important data alongside prices and market shares. Article 6(11) of the EU Digital Markets Act [[https://eur-lex.europa.eu/eli/reg/2022/1925/oj/eng#:~:text=giving%20their%20consent.-,11.%C2%A0,-The%20gatekeeper%20shall 4]] reflects this approach: certain large search providers must, on request, share specified search data with rival search providers on fair and non-discriminatory terms, with relevant personal data anonymised. The aim is to help rivals compete while protecting users. A data advantage does not, by itself, establish unlawful conduct.Would access to equivalent data enable a rival to compete effectively? The concept cannot answer this on its own. Additional data may eventually add little value, become outdated, or have substitutes. Firms also need the skills and computing resources to use it. Google’s default positions in browsers and devices can further restrict rivals’ access to users even if data become available. The unresolved issue is how much competitive advantage comes from data itself and how much comes from these other resources and arrangements. 

= = Two Sided Markets (Ceren) = =In two-sided markets, two sides of a market interact through an intermediary. These markets typically have some externality,<code class="mwt-code">Carl Shapiro and Hal R Varian, ''Information Rules: A Strategic Guide to the Network Economy'' (Harvard Business School Press 2010), 183.</code> where the decisions of the sides affect the other. The intermediary also has multitudes of effects on both sides, mainly identified in the related literature through its pricing.An example of a two sided market is payment systems, whereby the credit card providing bank is the intermediary between the customers and merchants. In this example, banks provide credit cards to customers to be utilized in their payments, and merchants accept payment via multiple credit card systems to accommodate payments from customers holding credit cards from such providers[https://doi.org/10.1257/jep.23.3.125 .]The main power of the intermediary materializes in their pricing toward the merchant (and the merchant’s bank in interchange fees in case of payment being transferred through different banks). Customers will be expected to be more loyal, holding a few credit cards, whereas merchants accept payment from many banks to accommodate all customers’ payment methods. Banks will provide more subsidies to customers to attract them and compete aggressively on the customer side, making their platform more desirable to work with by merchants, who will in turn be more accepting of the platform’s fees to gain the business of the larger volume of customers. However, dynamic pricing (increasing fees after gaining entry and loyalty) may be a way for these platforms to abuse their power.The concept may be derailed from these logical / analytical systems of operation or expected outcomes when public policy interferes with the market. For example, net neutrality rules regulate what internet service providers (the intermediary) can charge content providers (one side) who aim to reach the consumers (the other side). In this case the above considerations and the concept are halted.

= Switching costs and lock-in (Kate) =
'''Definition '''When you move from one product or service to another, switching costs are everything you have to either give up or spend. This can include money, time, lost contacts, etc. Lock-in is when you feel (and are) trapped by these costs that add up and you decide instead of moving products or services, it’s just easier to stay with a service, not because you want to but because it’s just too much of a hassle. This is true even if a less expensive or better service exists all because leaving the product/service just costs too much in terms of time, effort, and money. Shapiro and Varian (2010) argue that switching costs are normal in information markets and that the full cost of a switch includes what you as the customer pay and what a new supplier spends to win that customer over.  '''Example: Apple''' The more Apple products you own, the harder it is to leave. As an avid Apple user, I am absolutely stuck in their ecosystem. I have my photos in Apple’s iCloud, my group chats in iMessage, and my Apple Watch only works with an iPhone. Leaving Apple is not easy. So much so that in March 2024, the US Department of Justice sued Apple alleging it raises costs of leaving/switching phones by making messaging worse and less secure when you’re messaging with non iPhones. In June 2025, Judge Julien Neals denied Apple’s motion to dismiss and the case is still in pre-trial proceedings.  '''Governance implication''' Lock-in makes users dependent on a platform and because of that, once we’re locked in, the platform (Meta, Apple, Google, etc) can raise prices, show more ads or even provide lower quality terms. So in this way, the platform can control users because if a user wants to leave, it’s going to cost you in that the platform has already designed your exit including how you can export your data, licenses and APIs. Under the EU’s Digital Markets Act, the European Commission adopted decisions in March 2025 making Apple open nine iOS connectivity features, mostly used by devices like smartwatches, headphones and televisions. These decisions are legally binding and they focus on lock-in by lowering the cost of choosing non-Apple products. It’s worth noting, however, that they only apply in the EU. When the Commission adopted the decisions, Apple said (reported by MacTech) “It’s bad for our products and for our European users.” '''Open Question''' So what happens when the cost of leaving isn’t just your own costs as a user but also the cost of leaving family and friends behind?  If a user leaves, let’s say, Instagram, they won’t be charged to leave (they will, however, have some friction to leaving in that they’ll be asked why they’re leaving and if there’s anything Instagram can do to make them stay), but they will be leaving their family and friends that stay on the app. This is where data portability can’t really help because while you can export your photos, you can’t make your friends leave. This is part of what makes leaving a network or platform so hard. '''References:''' European Commission (2025, March 19). ''Commission provides guidance under Digital Markets Act to facilitate development of innovative products on Apple’s platforms''. https://digital-markets-act.ec.europa.eu/commission-provides-guidance-under-digital-markets-act-facilitate-development-innovative-products-2025-03-19_en MacTech (2025, March 19). ''Apple isn’t happy with the latest decisions regarding the DMA''. https://www.mactech.com/2025/03/19/apple-isnt-happy-with-the-latest-decisions-regarding-the-dma Shapiro, C. & Varian, H.R. (2010). ''Information Rules: A Strategic Guide to the Network Economy.'' Boston, MA: Harvard Business School Press. The Hill. (2025, June 30). ''Judge rejects Apple’s bid to dismiss DOJ antitrust case''. https://thehill.com/policy/technology/5380007-apple-federal-judge-antitrust-case/ The Mac Observer. (2026, September 12). ''Apple’s other antitrust fights: Where the DOJ case and the UK’s CMA stand now''. https://www.macobserver.com/news/apple-doj-antitrust-case-uk-cma-status/ U.S. Department of Justice (2024, March 21). ''Justice Department Sues Apple for Monopolizing Smartphone Markets'' [Press Release]. https://www.justice.gov/archives/opa/pr/justice-department-sues-apple-monopolizing-smartphone-markets

= Excludability and rivalrousness (Paul)  =
=== Excludability and Rivalrousness ===The terms excludability and rivalrousness both refer to consumption characteristics that can be assigned to- or identified in goods. '''Excludability''': Goods to which economic actors can technically or legally restrict access are considered to have ‘high excludability’. Conversely, goods to which access cannot be restricted are of ‘low excludability’, or, in absolute terms, ‘non-excludable’.'''Rivalrousness''': Rivalrousness on the other hand refers to the finitude of the good in question. The greater the depletion of the stock, quantity, or amount is on account of one party’s consumption, the more rivalrous a good is. Tabulating the characteristics against one another allows for inference and definition of the type of good an object, service, or resource may constitute (see fig. 1). Excludability and Rivalrousness of goods can therethrough serve as a predictor for certain types of market failures: Goods which are non-excludable or of low excludability, but nonetheless costly to provide, are vulnerable to so-called free rider problem. Goods which are non-excludable, but highly rivalrous (common goods), may lead to the tragedy of the commons, that is over-exploitation.

[[image:Untitled Diagram.drawio (1).png|300px|thumb]]
=== AI Training and the (non-)excludability of information ===The unusual nature of excludability and rivalrousness of data further compromises the applicability of the goods classification built thereon. In traditional economics, a non-excludable and non-rivalrous good would be seen as a public good. This does apply insofar as data is vulnerable to corporate exploitation without commensurate indemnification. Data usage may however actually be anti-rival in the sense that increasing aggregate usage increases the value of the data (better labeling, easier access, previously undrawn conclusions, etc.), rather than diminish the ‘stock’ of data. Conversely, data may be rivalrous not in its absolute depletion by consumption, but rather the benefit or competitive advantage usage of that data may confer: If all AI firms are able to train on the exact same data (all available data), the utility of that data may still increase, but the competitive advantage of possessing that data is exhausted.

==== Free-Rider Problems & Public Data ====As any party can benefit from publicly available data at no cost or low cost, the incentive for private market actors to fund its creation is comparatively weak. Consequently, public bodies (and less frequently open-source foundations and non-profits) often assume the responsibility for the provision of this type of data. A classical example is weather data and its creation by national weather agencies.

= Interoperability (Sisi) =

#What is interoperability?[[https://www.ibm.com/think/topics/interoperability 1]]Interoperability is a “standards-based” approach that enables different IT systems to communicate with minimal end-user interaction to exchange data and “share functionality”. Each system also allows the other system to “read, update, modify, and analyze” data with minimal human interaction. [[https://www.oracle.com/nl/interoperability/ 2]]Interoperability works by having different systems share data by employing the same standards for data formats (e.g., Extensible Markup Language or JavaScript Object Notation), exchange, and vocabulary to title data elements. These standards are agreed upon “in advance” by organizations, industry types, and user groups. Then, software developers build the system based on the model agreed upon.

#Concrete example from a digital technology or platformInteroperable systems are vital in healthcare, commerce, and public safety. A notable example includes the integration of Electronic Health Records (HER). This system allows different healthcare providers to securely access and share personal patient information regardless of the specific EHR system used.[[https://data.europa.eu/en/news-events/news/interoperability-solutions-bridging-systems-connected-future 3]] An example of such a system is the GNU Health Hospital Information System, which is a model employed by small private offices and large national public health systems.[[https://interoperable-europe.ec.europa.eu/collection/gnu-health/solution/gnu-health-hospital-information-system 4]]For example, a health care professional may access lab results and patient history data from various platforms to enable them to make a diagnosis.

#One governance implication: what does this concept explain about how platforms exercise power?Interoperability is a powerful tool because it helps provide a “seamless user experience” and optimizes data sharing between separate systems. It helps companies and organizations access data from multiple sources and improve efficiency, such as accelerating workflows and reducing the possibility of errors by pooling data from its sources.Interoperability standards are mandated by law on the technical side to ensure “critical systems” are compatible, especially in public safety and health. [[https://www.ibm.com/think/topics/interoperability 5]]In Europe, interoperability is governed by the European Health Data Space regulation (EHDS); the “first common EU data space dedicated to a specific sector as part of the European strategy for data.”This Regulation aims to benefit EU citizens such as patients, researchers, healthcare professionals, and industry players by establishing a common framework for the exchange of electronic health data within the EU. In addition, it enables certain data to be “reused” for public purpose and innovative scientific research. [[https://health.ec.europa.eu/ehealth-digital-health-and-care/european-health-data-space-regulation-ehds_en 6]]

#Where does the concept break down or fail to explain somethingWhile the EDHS harmonizes a legal and technical framework for sharing EU data, its shortcoming is its gradual implementation. The EDHS entered into force in 2025; however, most of its obligations will not apply until 2029. This leaves four years for health care companies to use interoperability in a deregulated manner. [[https://health.ec.europa.eu/ehealth-digital-health-and-care/european-health-data-space-regulation-ehds_en 7]]Furthermore, there are actual barriers to achieving full interoperability in healthcare, including data silos (which isolate data collections and hinder data exchange), lack of standardization (where one system records blood pressure as free text, another system stores it as a structured numeric value), and deficient infrastructure, which requires significant technological upgrading.The burden remains on health companies to decide which standards to adopt and how much to invest in stable technological infrastructure to strengthen their data governance and interoperability. [[https://www.tigahealth.com/how-to-overcome-healthcare-interoperability-challenges/ 8]]

= [[Economies of Scale (Jihae)]] =

= [[Economies of Scope (Mariona)]] =
= [[Network Effects (Indirect) - Sebastián]] =
= [[Monopsony - (Atmika)]] =
== Definition: ==
 A Monopsony is a market condition where there is one buyer and multiple sellers; the opposite of the better known non-competitive market - Monopoly. Here, the buyer has the power to control the price and set market conditions. 
== Digital platform: Uber ==
'''Definition:''' 
 Uber was released into the market as an online platform to connect drivers with customers - but it’s business model created a huge social and economic impact in the labor sector. (1) Uber engaged drivers on a contract basis and offered short-term, low commitment and flexible jobs, which we characterise as the ‘[https://en.wikipedia.org/wiki/Gig_economy Gig economy]’. Uber offered a better transportation network alternative - dynamic pricing, more operating efficiency and on-demand mobility in contrast to the rigid capacities and fares of conventional transportational means available (mainly taxicabs), which would soon be replaced.(2) Additional social perspectives showed fair, non-discriminatory and safer practices - such as access and availability to lower economic groups.   

 '''How power is exercised '''

 1. Capturing labour force: The primary economic impact in the labour space is with regard to human capital mobilisation; a 2014 survey showed that 36% of 600 Uber drivers reported neither having a job, nor actively seeking work, before signing up to drive for Uber. (3) The presence of Uber resulted in increased regional employment. Drivers valued the flexibility and self-supervision enough to accept lower earnings than in previous jobs.

 2. Control of price and contractual terms: Uber exercises vertical control by dictating terms on which drivers transact and fixed prices - independent contractors are not allowed liberty to negotiate. 

 3. Horizontal friction: Assimilation and use of dominant position in one sectoral space to be able to expand into another such as in the case of Uber Eats. While Uber has since started to reduce it’s region specific operations by selling their holdings to the local leading brand, such as in China with Didi, they maintain equity shares as a control mechanism. 

 '''Where does the concept break down or fail to explain?'''

 A highly non-competitive market structure, is one wherein the buyer and seller cannot freely move within the market, curbed by inefficiency, high barrier of entry and lack of innovation. But modern day monopsony looks alot more different than this traditional concept. For one in this case, Uber was a new innovation brought into the market to outdate the old inefficient model of taxis with more control by its users. Moreover, monopsony can exist even in settings that appear easy to enter or leave.(4)

 While it may not be a pure monopsony, Uber does hold 71% market shares in the TNC space in the United States. However, the lack of barriers for entry saw Lyft introduced at the same time, where both provided similar models and impacts, with consumers switching between platforms. 

'''References:'''

 (1) Koling, A., Michalek, J., Armanios, D. et al. Effects of Uber and Lyft on jobs, wages and GDP. Nat Cities 3, 700–707 (2026). https://doi.org/10.1038/s44284-026-00478-0 (2) Zha, L., Yin, Y. & Du, Y. Surge pricing and labor supply in the ridesourcing market. Transport. Res. Proc. 23, 2–21 (2017). (3) Levin, A. The Driver Roadmap: Where Uber Driver-Partners Have Been, and Where They’re Going (Benenson Strategy Group, 2014). (4) Manning, A. (2021). Monopsony in Labor Markets: A Review. ILR Review, 74(1), 3-26. https://doi.org/10.1177/0019793920922499.

===== [https://wiki.ivir.nl/index.php?title=Market_tipping_-_Rodrigue Market Tipping - Rodrigue] =====

= Multi-homing (Juliette) =

1/ Plain-language definition

Multi-homing means using two or more competing platforms at the same time, for the same thing. A shop that sells on both Amazon and Bol.com is multi-homing. A person who has both WhatsApp and Signal on their phone is multi-homing too. The opposite is single-homing, when you use only one platform. When people multi-home, it is easier for them to leave a platform, because they already have another one ready (1).

2/ Concrete example

Taxi apps are a good example. In many cities, drivers have Uber and Bolt open at the same time. They take the first good ride that comes. Many passengers, however, use only one app because they are used to it. So drivers can move between apps, but each app keeps its own group of loyal passengers. If a driver wants those passengers, he has to stay on that app. This means the app still has power over drivers, even if they use several apps (2).

3/ Governance implication

Multi-homing shows why platforms try hard to keep users on one service only. They do not usually ban other apps. Instead, they make using them less attractive. Taxi apps give drivers bonuses if they do a certain number of rides on the same app. Online shops and hotel booking sites have stopped sellers from offering lower prices on other websites. In this way, the platform makes its own rules. It controls what users do without a formal contract. Regulators now want to protect multi-homing. The Digital Markets Act’s article 5.3 says large platforms cannot stop businesses from offering better prices elsewhere. It also says users must be able to take their data with them (article 6.9) and leave a service easily (article 6.13) (3). The idea is that if people can use many platforms, no single platform can control them.

4/ Open question

We often think that multi-homing reduces the power of platforms, even if it is not always true. If only one side multi-homes, the platform can ask that side to pay more, because it controls access to the loyal side. Research shows that the result depends on who multi-homes. Sometimes the group that should win actually loses (2). The authors Bryan and Gans even find that in some cities, one taxi app can be better for society than several (4). Also, multi-homing does not show how much people really use each platform. Someone can have two accounts but use only one. While multi-homing is easy to see, his understanding remains complex.

5/ Bibliography

(1) Belleflamme P and Peitz M, 'Platform Competition: Who Benefits from Multihoming?' (2019) 64 International Journal of Industrial Organization 1

(2) Bryan KA and Gans JS, 'A Theory of Multihoming in Rideshare Competition' (2019) 28 Journal of Economics & Management Strategy 89

(3) Regulation (EU) 2022/1925 (Digital Markets Act) [2022] OJ L265/1

(4) Zhu F and Iansiti M, 'Why Some Platforms Thrive and Others Don't' (2019) Harvard Business Review (January–February) 118

= Vertical Integration by Pablo Latorre ='''1. Definition'''“Vertical Integration” is the control by a single corporation of two or more consecutive steps of the same production chain of activity in such a manner that said corporation does not need to rely on others for them (1). In the context of digital markets, “Vertical Integration” usually refers to the position of a platform that operates a marketplace or infrastructure that other enterprises depend on and, at the same time, commercializes products or services in that same space (2). Therefore, in a “Vertical Integration” the platform is both, provider and competitor, of other enterprises. '''2. Example: Amazon as a Marketplace'''Amazon operates the largest online marketplace (3) in the world and, thus, thousands of small and medium size enterprises depend on it to connect with their costumers. At the same time, Amazon commercializes its own products on its marketplace (4), acting as a retailer and competitor to those enterprises. Such dual role allows Amazon to exercise competitive advantages: first, it can access detailed, non-public data on the enterprises’ sales, including products, prices, and costumer behavior; and second, it can decide the rules that define the products that appear first as the result of a costumer search.In 2022 Amazon had to commit to the European Commission to not use the enterprises’ non-public data for its retail business and to treat equally rival offers in its “Buy Box” (5). Currently, the Digital Markets Act (DMA) establishes matching general obligations for designated gatekeepers (6). '''3. Governance Implication'''“Vertical Integration” allows platforms to exercise power over enterprises through digital architecture. When a platform operates a marketplace and competes in it, its ordinary decisions on digital architecture (i.e. on algorithms, searches, products visibility) can become tools for power and competition. In a sense, platforms are the ''de facto'' private regulators of markets in which they have clear commercial interests. Such state of affairs allows to understand a strategy change in antitrust cases. Traditionally, antitrust intervenes after the harm, however regulators are moving towards ''ex ante'' conduct rules (i.e. the DMA) (7). '''4. Open Question'''“Vertical Integration” clearly explains the incentive structure that drives platforms to favor their own products through the exercise of digital architecture mediated competitive advantages. However, “Vertical Integration” assumes that the line between one product (i.e. the marketplace, the search engine) and the other (i.e. the products, the websites competing on the platform) is clear when, in practice, this is not always so. For instance, nowadays when a user search something in Google Search, Google not only provides links to websites that contain the desired information but also the desired information itself with AI-generated summaries. In this case, two possible perspectives arise: on the one hand, that Google could be abusing its position to capture the online information market by displacing websites that depend on it; on the other hand, that Google could be simply improving its search engine and search itself because what users actually want is information and not links to it. In this example, “Vertical Integration” is incapable of settling the issue because it depends on a clear distinction between one product and the other (or one market and the other) and that is absent in this case. Moreover, even if in the presented case was possible to separate one product from the other (or one market from the other), “Vertical Integration” does not provide a clear criteria to establish when an “integration” is abusive exclusion of competitors and when it is legitimate and necessary innovation. '''References'''(1) “Vertical integration” (''Cambridge Dictionary'') [https://dictionary.cambridge.org/dictionary/english/vertical-integration https://dictionary.cambridge.org/dictionary/english/vertical-integration] accessed 4 October 2026.(2) [https://academic.oup.com/antitrust/article/6/3/355/5034073?itm_medium=sidebar&itm_source=trendmd-widget&itm_campaign=Journal_of_Antitrust_Enforcement&itm_content=Journal_of_Antitrust_Enforcement_0 Bostoen Friso, “Online Platforms and Vertical Integration: The Return of Margin Squeeze” (2018) (6) Journal of Antitrust Enforcement 355.](3) Lovell Keir, “Top 10 Ecommerce Marketplaces in the World: 2026 Rankings by GMV, Users & Reach” (''ServeRetail'', 2 January 2026) [https://www.serveretail.com/blog/marketplace/the-10-largest-ecommerce-marketplaces-in-the-world-size-reach-and-cx-strategy/ https://www.serveretail.com/blog/marketplace/the-10-largest-ecommerce-marketplaces-in-the-world-size-reach-and-cx-strategy/] accessed 4 October 2026.(4) Ellison Mara, “What are Amazon's Brands, Explained” (''Ir Mountaintopuniversity Ng'') [https://ir.mountaintopuniversity.edu.ng/article/what-are-amazon-s-brands-explained https://ir.mountaintopuniversity.edu.ng/article/what-are-amazon-s-brands-explained] accessed 4 October 2026.(5) [https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=oj:JOC_2023_087_R_0005 Summary of Commission Decision of 20 December 2022 Relating to a Proceeding under Article 102 of the Treaty on the Functioning of the European Union and Article 54 of the EEA Agreement (Cases AT.40462 – Amazon Marketplace, and AT.40703 – Amazon Buy Box) (2023) OJ C87.](6) [https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=uriserv%3AOJ.L_.2022.265.01.0001.01.ENG&toc=OJ%3AL%3A2022%3A265%3ATOC Regulation (EU) 2022/1925 of the European Parliament and of the Council of 14 September 2022 on contestable and fair markets in the digital sector and amending Directives (EU) 2019/1937 and (EU) 2020/1828 (Digital Markets Act) (2022).](7) OECD, "Ex Ante Regulation and Competition in Digital Markets" (''OECD Roundtables on Competition Policy Papers, OECD Publishing 2021'') [https://doi.org/10.1787/c83e178d-en https://doi.org/10.1787/c83e178d-en] accessed 4 October 2026.

    [[https://wiki.ivir.nl/index.php?title=Two- and_Multi-Sided_Markets_Robert|Two- and Multi-Sided Markets]]

Two- en Multi-Sided Markets Robert

2026-10-07T11:03:04Z

Robert Florijn: Robert Florijn moved page Two- en Multi-Sided Markets Robert to Two- and Multi-Sided Markets Robert

#REDIRECT [[Two- and Multi-Sided Markets Robert]]

Two and Multi-Sided Markets Robert

2026-10-07T11:03:04Z

Robert Florijn: Robert Florijn moved page Two- en Multi-Sided Markets Robert to Two- and Multi-Sided Markets Robert

Governance by markets

2026-10-07T11:02:31Z

Robert Florijn:

= Glossary =
= [https://wiki.ivir.nl/index.php?title=Data_as_an_input_to_market_power Data as an input to market power] (Sergey) =In the present digital economy, data has emerged as a valuable economic commodity that entities leverage to gain a competitive edge over their rivals [[https://repository.nls.ac.in/cgi/viewcontent.cgi?article=1478&context=nlsblr 1]]. '''Data''' is an input into production, so businesses that control data can have market power and impede the entry of potential competitors. '''Market power''' refers to a firm's ability to influence the price, output, or other market conditions in its favour, without being constrained by competitive forces [[https://www.tutor2u.net/economics/reference/what-is-meant-by-market-power 2]].'''Data as an input to market power''' means that access to information can help a business limit the competitive pressure it faces. Information about searches, clicks or purchases can improve services and advertising. If competitors cannot obtain comparable information at a reasonable cost, they may struggle to offer an attractive alternative. This can allow a platform to raise prices or worsen quality or privacy protections without losing enough business to make doing so unprofitable. This matters even when a service is free, because competition also concerns advertising charges, personal information and the choices available. Data can be used repeatedly and by several businesses without being used up. However, access can be restricted through contracts and technical controls. Nevertheless, holding a large dataset alone does not establish market power.'''Google Search''' illustrates this mechanism. Google controls over 91% of global search engine traffic [[https://www.statista.com/statistics/1381664/worldwide-all-devices-market-share-of-search-engines/ 3]] and uses information about users’ searches and the results they click to improve the relevance of its rankings. A larger volume of searches gives it more opportunities to learn, including about unusual queries that smaller search engines encounter less often. Better results can attract further use, producing additional information and reinforcing the advantage. A new search engine may therefore struggle to attract users while lacking the information needed to improve its service. Data access can constitute a significant barrier to entry when this type of analytics is necessary to participate in a market, particularly when alternatives are limited. A platform’s control over data access can shape who is able to compete with it. By determining whether, and on what terms, rivals can access that information, a platform can influence the competitive opportunities available to others. This suggests that regulation should examine access to strategically important data alongside prices and market shares. Article 6(11) of the EU Digital Markets Act [[https://eur-lex.europa.eu/eli/reg/2022/1925/oj/eng#:~:text=giving%20their%20consent.-,11.%C2%A0,-The%20gatekeeper%20shall 4]] reflects this approach: certain large search providers must, on request, share specified search data with rival search providers on fair and non-discriminatory terms, with relevant personal data anonymised. The aim is to help rivals compete while protecting users. A data advantage does not, by itself, establish unlawful conduct.Would access to equivalent data enable a rival to compete effectively? The concept cannot answer this on its own. Additional data may eventually add little value, become outdated, or have substitutes. Firms also need the skills and computing resources to use it. Google’s default positions in browsers and devices can further restrict rivals’ access to users even if data become available. The unresolved issue is how much competitive advantage comes from data itself and how much comes from these other resources and arrangements. 

= = Two Sided Markets (Ceren) = =In two-sided markets, two sides of a market interact through an intermediary. These markets typically have some externality,<code class="mwt-code">Carl Shapiro and Hal R Varian, ''Information Rules: A Strategic Guide to the Network Economy'' (Harvard Business School Press 2010), 183.</code> where the decisions of the sides affect the other. The intermediary also has multitudes of effects on both sides, mainly identified in the related literature through its pricing.An example of a two sided market is payment systems, whereby the credit card providing bank is the intermediary between the customers and merchants. In this example, banks provide credit cards to customers to be utilized in their payments, and merchants accept payment via multiple credit card systems to accommodate payments from customers holding credit cards from such providers[https://doi.org/10.1257/jep.23.3.125 .]The main power of the intermediary materializes in their pricing toward the merchant (and the merchant’s bank in interchange fees in case of payment being transferred through different banks). Customers will be expected to be more loyal, holding a few credit cards, whereas merchants accept payment from many banks to accommodate all customers’ payment methods. Banks will provide more subsidies to customers to attract them and compete aggressively on the customer side, making their platform more desirable to work with by merchants, who will in turn be more accepting of the platform’s fees to gain the business of the larger volume of customers. However, dynamic pricing (increasing fees after gaining entry and loyalty) may be a way for these platforms to abuse their power.The concept may be derailed from these logical / analytical systems of operation or expected outcomes when public policy interferes with the market. For example, net neutrality rules regulate what internet service providers (the intermediary) can charge content providers (one side) who aim to reach the consumers (the other side). In this case the above considerations and the concept are halted.

= Switching costs and lock-in (Kate) =
'''Definition '''When you move from one product or service to another, switching costs are everything you have to either give up or spend. This can include money, time, lost contacts, etc. Lock-in is when you feel (and are) trapped by these costs that add up and you decide instead of moving products or services, it’s just easier to stay with a service, not because you want to but because it’s just too much of a hassle. This is true even if a less expensive or better service exists all because leaving the product/service just costs too much in terms of time, effort, and money. Shapiro and Varian (2010) argue that switching costs are normal in information markets and that the full cost of a switch includes what you as the customer pay and what a new supplier spends to win that customer over.  '''Example: Apple''' The more Apple products you own, the harder it is to leave. As an avid Apple user, I am absolutely stuck in their ecosystem. I have my photos in Apple’s iCloud, my group chats in iMessage, and my Apple Watch only works with an iPhone. Leaving Apple is not easy. So much so that in March 2024, the US Department of Justice sued Apple alleging it raises costs of leaving/switching phones by making messaging worse and less secure when you’re messaging with non iPhones. In June 2025, Judge Julien Neals denied Apple’s motion to dismiss and the case is still in pre-trial proceedings.  '''Governance implication''' Lock-in makes users dependent on a platform and because of that, once we’re locked in, the platform (Meta, Apple, Google, etc) can raise prices, show more ads or even provide lower quality terms. So in this way, the platform can control users because if a user wants to leave, it’s going to cost you in that the platform has already designed your exit including how you can export your data, licenses and APIs. Under the EU’s Digital Markets Act, the European Commission adopted decisions in March 2025 making Apple open nine iOS connectivity features, mostly used by devices like smartwatches, headphones and televisions. These decisions are legally binding and they focus on lock-in by lowering the cost of choosing non-Apple products. It’s worth noting, however, that they only apply in the EU. When the Commission adopted the decisions, Apple said (reported by MacTech) “It’s bad for our products and for our European users.” '''Open Question''' So what happens when the cost of leaving isn’t just your own costs as a user but also the cost of leaving family and friends behind?  If a user leaves, let’s say, Instagram, they won’t be charged to leave (they will, however, have some friction to leaving in that they’ll be asked why they’re leaving and if there’s anything Instagram can do to make them stay), but they will be leaving their family and friends that stay on the app. This is where data portability can’t really help because while you can export your photos, you can’t make your friends leave. This is part of what makes leaving a network or platform so hard. '''References:''' European Commission (2025, March 19). ''Commission provides guidance under Digital Markets Act to facilitate development of innovative products on Apple’s platforms''. https://digital-markets-act.ec.europa.eu/commission-provides-guidance-under-digital-markets-act-facilitate-development-innovative-products-2025-03-19_en MacTech (2025, March 19). ''Apple isn’t happy with the latest decisions regarding the DMA''. https://www.mactech.com/2025/03/19/apple-isnt-happy-with-the-latest-decisions-regarding-the-dma Shapiro, C. & Varian, H.R. (2010). ''Information Rules: A Strategic Guide to the Network Economy.'' Boston, MA: Harvard Business School Press. The Hill. (2025, June 30). ''Judge rejects Apple’s bid to dismiss DOJ antitrust case''. https://thehill.com/policy/technology/5380007-apple-federal-judge-antitrust-case/ The Mac Observer. (2026, September 12). ''Apple’s other antitrust fights: Where the DOJ case and the UK’s CMA stand now''. https://www.macobserver.com/news/apple-doj-antitrust-case-uk-cma-status/ U.S. Department of Justice (2024, March 21). ''Justice Department Sues Apple for Monopolizing Smartphone Markets'' [Press Release]. https://www.justice.gov/archives/opa/pr/justice-department-sues-apple-monopolizing-smartphone-markets

= Excludability and rivalrousness (Paul)  =
=== Excludability and Rivalrousness ===The terms excludability and rivalrousness both refer to consumption characteristics that can be assigned to- or identified in goods. '''Excludability''': Goods to which economic actors can technically or legally restrict access are considered to have ‘high excludability’. Conversely, goods to which access cannot be restricted are of ‘low excludability’, or, in absolute terms, ‘non-excludable’.'''Rivalrousness''': Rivalrousness on the other hand refers to the finitude of the good in question. The greater the depletion of the stock, quantity, or amount is on account of one party’s consumption, the more rivalrous a good is. Tabulating the characteristics against one another allows for inference and definition of the type of good an object, service, or resource may constitute (see fig. 1). Excludability and Rivalrousness of goods can therethrough serve as a predictor for certain types of market failures: Goods which are non-excludable or of low excludability, but nonetheless costly to provide, are vulnerable to so-called free rider problem. Goods which are non-excludable, but highly rivalrous (common goods), may lead to the tragedy of the commons, that is over-exploitation.

[[image:Untitled Diagram.drawio (1).png|300px|thumb]]
=== AI Training and the (non-)excludability of information ===The unusual nature of excludability and rivalrousness of data further compromises the applicability of the goods classification built thereon. In traditional economics, a non-excludable and non-rivalrous good would be seen as a public good. This does apply insofar as data is vulnerable to corporate exploitation without commensurate indemnification. Data usage may however actually be anti-rival in the sense that increasing aggregate usage increases the value of the data (better labeling, easier access, previously undrawn conclusions, etc.), rather than diminish the ‘stock’ of data. Conversely, data may be rivalrous not in its absolute depletion by consumption, but rather the benefit or competitive advantage usage of that data may confer: If all AI firms are able to train on the exact same data (all available data), the utility of that data may still increase, but the competitive advantage of possessing that data is exhausted.

==== Free-Rider Problems & Public Data ====As any party can benefit from publicly available data at no cost or low cost, the incentive for private market actors to fund its creation is comparatively weak. Consequently, public bodies (and less frequently open-source foundations and non-profits) often assume the responsibility for the provision of this type of data. A classical example is weather data and its creation by national weather agencies.

= Interoperability (Sisi) =

#What is interoperability?[[https://www.ibm.com/think/topics/interoperability 1]]Interoperability is a “standards-based” approach that enables different IT systems to communicate with minimal end-user interaction to exchange data and “share functionality”. Each system also allows the other system to “read, update, modify, and analyze” data with minimal human interaction. [[https://www.oracle.com/nl/interoperability/ 2]]Interoperability works by having different systems share data by employing the same standards for data formats (e.g., Extensible Markup Language or JavaScript Object Notation), exchange, and vocabulary to title data elements. These standards are agreed upon “in advance” by organizations, industry types, and user groups. Then, software developers build the system based on the model agreed upon.

#Concrete example from a digital technology or platformInteroperable systems are vital in healthcare, commerce, and public safety. A notable example includes the integration of Electronic Health Records (HER). This system allows different healthcare providers to securely access and share personal patient information regardless of the specific EHR system used.[[https://data.europa.eu/en/news-events/news/interoperability-solutions-bridging-systems-connected-future 3]] An example of such a system is the GNU Health Hospital Information System, which is a model employed by small private offices and large national public health systems.[[https://interoperable-europe.ec.europa.eu/collection/gnu-health/solution/gnu-health-hospital-information-system 4]]For example, a health care professional may access lab results and patient history data from various platforms to enable them to make a diagnosis.

#One governance implication: what does this concept explain about how platforms exercise power?Interoperability is a powerful tool because it helps provide a “seamless user experience” and optimizes data sharing between separate systems. It helps companies and organizations access data from multiple sources and improve efficiency, such as accelerating workflows and reducing the possibility of errors by pooling data from its sources.Interoperability standards are mandated by law on the technical side to ensure “critical systems” are compatible, especially in public safety and health. [[https://www.ibm.com/think/topics/interoperability 5]]In Europe, interoperability is governed by the European Health Data Space regulation (EHDS); the “first common EU data space dedicated to a specific sector as part of the European strategy for data.”This Regulation aims to benefit EU citizens such as patients, researchers, healthcare professionals, and industry players by establishing a common framework for the exchange of electronic health data within the EU. In addition, it enables certain data to be “reused” for public purpose and innovative scientific research. [[https://health.ec.europa.eu/ehealth-digital-health-and-care/european-health-data-space-regulation-ehds_en 6]]

#Where does the concept break down or fail to explain somethingWhile the EDHS harmonizes a legal and technical framework for sharing EU data, its shortcoming is its gradual implementation. The EDHS entered into force in 2025; however, most of its obligations will not apply until 2029. This leaves four years for health care companies to use interoperability in a deregulated manner. [[https://health.ec.europa.eu/ehealth-digital-health-and-care/european-health-data-space-regulation-ehds_en 7]]Furthermore, there are actual barriers to achieving full interoperability in healthcare, including data silos (which isolate data collections and hinder data exchange), lack of standardization (where one system records blood pressure as free text, another system stores it as a structured numeric value), and deficient infrastructure, which requires significant technological upgrading.The burden remains on health companies to decide which standards to adopt and how much to invest in stable technological infrastructure to strengthen their data governance and interoperability. [[https://www.tigahealth.com/how-to-overcome-healthcare-interoperability-challenges/ 8]]

= [[Economies of Scale (Jihae)]] =

= [[Economies of Scope (Mariona)]] =
= [[Network Effects (Indirect) - Sebastián]] =
= [[Monopsony - (Atmika)]] =
== Definition: ==
 A Monopsony is a market condition where there is one buyer and multiple sellers; the opposite of the better known non-competitive market - Monopoly. Here, the buyer has the power to control the price and set market conditions. 
== Digital platform: Uber ==
'''Definition:''' 
 Uber was released into the market as an online platform to connect drivers with customers - but it’s business model created a huge social and economic impact in the labor sector. (1) Uber engaged drivers on a contract basis and offered short-term, low commitment and flexible jobs, which we characterise as the ‘[https://en.wikipedia.org/wiki/Gig_economy Gig economy]’. Uber offered a better transportation network alternative - dynamic pricing, more operating efficiency and on-demand mobility in contrast to the rigid capacities and fares of conventional transportational means available (mainly taxicabs), which would soon be replaced.(2) Additional social perspectives showed fair, non-discriminatory and safer practices - such as access and availability to lower economic groups.   

 '''How power is exercised '''

 1. Capturing labour force: The primary economic impact in the labour space is with regard to human capital mobilisation; a 2014 survey showed that 36% of 600 Uber drivers reported neither having a job, nor actively seeking work, before signing up to drive for Uber. (3) The presence of Uber resulted in increased regional employment. Drivers valued the flexibility and self-supervision enough to accept lower earnings than in previous jobs.

 2. Control of price and contractual terms: Uber exercises vertical control by dictating terms on which drivers transact and fixed prices - independent contractors are not allowed liberty to negotiate. 

 3. Horizontal friction: Assimilation and use of dominant position in one sectoral space to be able to expand into another such as in the case of Uber Eats. While Uber has since started to reduce it’s region specific operations by selling their holdings to the local leading brand, such as in China with Didi, they maintain equity shares as a control mechanism. 

 '''Where does the concept break down or fail to explain?'''

 A highly non-competitive market structure, is one wherein the buyer and seller cannot freely move within the market, curbed by inefficiency, high barrier of entry and lack of innovation. But modern day monopsony looks alot more different than this traditional concept. For one in this case, Uber was a new innovation brought into the market to outdate the old inefficient model of taxis with more control by its users. Moreover, monopsony can exist even in settings that appear easy to enter or leave.(4)

 While it may not be a pure monopsony, Uber does hold 71% market shares in the TNC space in the United States. However, the lack of barriers for entry saw Lyft introduced at the same time, where both provided similar models and impacts, with consumers switching between platforms. 

'''References:'''

 (1) Koling, A., Michalek, J., Armanios, D. et al. Effects of Uber and Lyft on jobs, wages and GDP. Nat Cities 3, 700–707 (2026). https://doi.org/10.1038/s44284-026-00478-0 (2) Zha, L., Yin, Y. & Du, Y. Surge pricing and labor supply in the ridesourcing market. Transport. Res. Proc. 23, 2–21 (2017). (3) Levin, A. The Driver Roadmap: Where Uber Driver-Partners Have Been, and Where They’re Going (Benenson Strategy Group, 2014). (4) Manning, A. (2021). Monopsony in Labor Markets: A Review. ILR Review, 74(1), 3-26. https://doi.org/10.1177/0019793920922499.

===== [https://wiki.ivir.nl/index.php?title=Market_tipping_-_Rodrigue Market Tipping - Rodrigue] =====

= Multi-homing (Juliette) =

1/ Plain-language definition

Multi-homing means using two or more competing platforms at the same time, for the same thing. A shop that sells on both Amazon and Bol.com is multi-homing. A person who has both WhatsApp and Signal on their phone is multi-homing too. The opposite is single-homing, when you use only one platform. When people multi-home, it is easier for them to leave a platform, because they already have another one ready (1).

2/ Concrete example

Taxi apps are a good example. In many cities, drivers have Uber and Bolt open at the same time. They take the first good ride that comes. Many passengers, however, use only one app because they are used to it. So drivers can move between apps, but each app keeps its own group of loyal passengers. If a driver wants those passengers, he has to stay on that app. This means the app still has power over drivers, even if they use several apps (2).

3/ Governance implication

Multi-homing shows why platforms try hard to keep users on one service only. They do not usually ban other apps. Instead, they make using them less attractive. Taxi apps give drivers bonuses if they do a certain number of rides on the same app. Online shops and hotel booking sites have stopped sellers from offering lower prices on other websites. In this way, the platform makes its own rules. It controls what users do without a formal contract. Regulators now want to protect multi-homing. The Digital Markets Act’s article 5.3 says large platforms cannot stop businesses from offering better prices elsewhere. It also says users must be able to take their data with them (article 6.9) and leave a service easily (article 6.13) (3). The idea is that if people can use many platforms, no single platform can control them.

4/ Open question

We often think that multi-homing reduces the power of platforms, even if it is not always true. If only one side multi-homes, the platform can ask that side to pay more, because it controls access to the loyal side. Research shows that the result depends on who multi-homes. Sometimes the group that should win actually loses (2). The authors Bryan and Gans even find that in some cities, one taxi app can be better for society than several (4). Also, multi-homing does not show how much people really use each platform. Someone can have two accounts but use only one. While multi-homing is easy to see, his understanding remains complex.

5/ Bibliography

(1) Belleflamme P and Peitz M, 'Platform Competition: Who Benefits from Multihoming?' (2019) 64 International Journal of Industrial Organization 1

(2) Bryan KA and Gans JS, 'A Theory of Multihoming in Rideshare Competition' (2019) 28 Journal of Economics & Management Strategy 89

(3) Regulation (EU) 2022/1925 (Digital Markets Act) [2022] OJ L265/1

(4) Zhu F and Iansiti M, 'Why Some Platforms Thrive and Others Don't' (2019) Harvard Business Review (January–February) 118

= Vertical Integration by Pablo Latorre ='''1. Definition'''“Vertical Integration” is the control by a single corporation of two or more consecutive steps of the same production chain of activity in such a manner that said corporation does not need to rely on others for them (1). In the context of digital markets, “Vertical Integration” usually refers to the position of a platform that operates a marketplace or infrastructure that other enterprises depend on and, at the same time, commercializes products or services in that same space (2). Therefore, in a “Vertical Integration” the platform is both, provider and competitor, of other enterprises. '''2. Example: Amazon as a Marketplace'''Amazon operates the largest online marketplace (3) in the world and, thus, thousands of small and medium size enterprises depend on it to connect with their costumers. At the same time, Amazon commercializes its own products on its marketplace (4), acting as a retailer and competitor to those enterprises. Such dual role allows Amazon to exercise competitive advantages: first, it can access detailed, non-public data on the enterprises’ sales, including products, prices, and costumer behavior; and second, it can decide the rules that define the products that appear first as the result of a costumer search.In 2022 Amazon had to commit to the European Commission to not use the enterprises’ non-public data for its retail business and to treat equally rival offers in its “Buy Box” (5). Currently, the Digital Markets Act (DMA) establishes matching general obligations for designated gatekeepers (6). '''3. Governance Implication'''“Vertical Integration” allows platforms to exercise power over enterprises through digital architecture. When a platform operates a marketplace and competes in it, its ordinary decisions on digital architecture (i.e. on algorithms, searches, products visibility) can become tools for power and competition. In a sense, platforms are the ''de facto'' private regulators of markets in which they have clear commercial interests. Such state of affairs allows to understand a strategy change in antitrust cases. Traditionally, antitrust intervenes after the harm, however regulators are moving towards ''ex ante'' conduct rules (i.e. the DMA) (7). '''4. Open Question'''“Vertical Integration” clearly explains the incentive structure that drives platforms to favor their own products through the exercise of digital architecture mediated competitive advantages. However, “Vertical Integration” assumes that the line between one product (i.e. the marketplace, the search engine) and the other (i.e. the products, the websites competing on the platform) is clear when, in practice, this is not always so. For instance, nowadays when a user search something in Google Search, Google not only provides links to websites that contain the desired information but also the desired information itself with AI-generated summaries. In this case, two possible perspectives arise: on the one hand, that Google could be abusing its position to capture the online information market by displacing websites that depend on it; on the other hand, that Google could be simply improving its search engine and search itself because what users actually want is information and not links to it. In this example, “Vertical Integration” is incapable of settling the issue because it depends on a clear distinction between one product and the other (or one market and the other) and that is absent in this case. Moreover, even if in the presented case was possible to separate one product from the other (or one market from the other), “Vertical Integration” does not provide a clear criteria to establish when an “integration” is abusive exclusion of competitors and when it is legitimate and necessary innovation. '''References'''(1) “Vertical integration” (''Cambridge Dictionary'') [https://dictionary.cambridge.org/dictionary/english/vertical-integration https://dictionary.cambridge.org/dictionary/english/vertical-integration] accessed 4 October 2026.(2) [https://academic.oup.com/antitrust/article/6/3/355/5034073?itm_medium=sidebar&itm_source=trendmd-widget&itm_campaign=Journal_of_Antitrust_Enforcement&itm_content=Journal_of_Antitrust_Enforcement_0 Bostoen Friso, “Online Platforms and Vertical Integration: The Return of Margin Squeeze” (2018) (6) Journal of Antitrust Enforcement 355.](3) Lovell Keir, “Top 10 Ecommerce Marketplaces in the World: 2026 Rankings by GMV, Users & Reach” (''ServeRetail'', 2 January 2026) [https://www.serveretail.com/blog/marketplace/the-10-largest-ecommerce-marketplaces-in-the-world-size-reach-and-cx-strategy/ https://www.serveretail.com/blog/marketplace/the-10-largest-ecommerce-marketplaces-in-the-world-size-reach-and-cx-strategy/] accessed 4 October 2026.(4) Ellison Mara, “What are Amazon's Brands, Explained” (''Ir Mountaintopuniversity Ng'') [https://ir.mountaintopuniversity.edu.ng/article/what-are-amazon-s-brands-explained https://ir.mountaintopuniversity.edu.ng/article/what-are-amazon-s-brands-explained] accessed 4 October 2026.(5) [https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=oj:JOC_2023_087_R_0005 Summary of Commission Decision of 20 December 2022 Relating to a Proceeding under Article 102 of the Treaty on the Functioning of the European Union and Article 54 of the EEA Agreement (Cases AT.40462 – Amazon Marketplace, and AT.40703 – Amazon Buy Box) (2023) OJ C87.](6) [https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=uriserv%3AOJ.L_.2022.265.01.0001.01.ENG&toc=OJ%3AL%3A2022%3A265%3ATOC Regulation (EU) 2022/1925 of the European Parliament and of the Council of 14 September 2022 on contestable and fair markets in the digital sector and amending Directives (EU) 2019/1937 and (EU) 2020/1828 (Digital Markets Act) (2022).](7) OECD, "Ex Ante Regulation and Competition in Digital Markets" (''OECD Roundtables on Competition Policy Papers, OECD Publishing 2021'') [https://doi.org/10.1787/c83e178d-en https://doi.org/10.1787/c83e178d-en] accessed 4 October 2026.

Two- and Multi-Sided Markets

Two and Multi-Sided Markets Robert

2026-10-07T11:00:36Z

Robert Florijn: Created page with "= '''Two-Sided and Multi-Sided Markets: How Platforms Govern Exchange''' = #''' What does the concept mean?''' A '''two-sided market''' connects two distinct groups through an intermediary, with participation or decisions on one side affecting outcomes or value on the other. Each group provides the other with network benefits, making the platform more valuable as participation grows (Katz & Shapiro, 1985). <span style="box-sizing: border-box; margin: 0px; paddin..."

Civilian AI at the Front - Robert Florijn

2026-09-24T07:46:54Z

Robert Florijn: Created contribution page and linked uploaded PDF

'''Robert Florijn'''

This contribution shows how Civilian AI and commercial digital infrastructure have become increasingly embedded in military capability, while the EU, in practice, seems blind to its own dual-use procurement regulation and technological dependency. 

[[:file:Civilian AI at the Front Final.pdf|Civilian AI at the Front Final (pdf)]]

Governance by history

2026-09-24T07:34:34Z

Robert Florijn:

[[File:Old internet.jpg|300px|thumb]]

= contributions =
*can
*you 
*please
*create
*individual pages
*from the contributions below
*and link them back here? 
*[[Information Control and Workplace Surveillance - Jihae Park and Mariona Fortuny]]
*[https://wiki.ivir.nl/index.php?title=Centralization_of_Compute_and_Data Centralization of AI Compute and Data - by Sergey Abramov and Paul Stein]
==[[Civilian AI at the Front - Robert Florijn]]==

= links things that popped up during class =
*[https://craphound.com/ Cory Doctorow]
*[https://dougengelbart.org/content/view/209/ The mother of All demos] (first mouse, graphical interface, word processor, etc)
*Can a simple sensor be political? [https://gizmodo.com/why-cant-this-soap-dispenser-identify-dark-skin-1797931773 well, this "racist" soap dispenser made headlines a couple of years ago.]
*Podcast series: [https://www.the-santiago-boys.com/ Morozov: the Santiago Boys.]
*Anatomy on AN AI. [[File:Ai-anatomy-map.pdf|check out the pdf here]]

= Workplace surveillance =
== Information Control and Workplace Surveillance - by Jihae Park and Mariona Fortuny Jané ==

==== The Current Governance Issue ====
There are numerous current governance problems arising from technological innovations, yet here we will specifically focus on the information control and workplace surveillance. Whether intended or unintended, there were certain historical incidents which resulted in social and political consequences. As noticed from the Control Revolution, the dramatic evolution in mass media and telecommunications technologies has led to the reinforcement of the economic and political control during the Industrial Revolution. ([https://archive.org/details/beniger-james-the-control-revolution Beniger, 1986]) By applying this logic to a modern example, this is clearly how many corporations are getting in control of their employees through surveillance technologies. Along with the emergence of remote and hybrid work after Covid-19, employers are now relying on “bossware” to analyze the emails, chats, and messages of their employees on platforms such as Teams or Slack. ([https://ojs.library.queensu.ca/index.php/surveillance-and-society/article/view/19100/13263 Ollier-Malaterre et al., 2026, p.237]) Some companies have strict restrictions on transferring data both internally and externally in protection of confidential information and consistently monitor their workers to the certain extent which could slow down their efficiency. This is problematic as it’s not only violating the privacy of their employees but also depriving employees of rights they should be naturally be able to exercise.



==== Historical Lineage ====
In relation to the current issue of the information control and workplace surveillance, Max Weber’s argument of bureaucracy and rationalization in the middle and late nineteenth century are good historical examples which are still frequently visible nowadays. Weber stated how the range of control could extend not only by expanding the capability to process information but also by limiting the amount of information to be processed. Here, the former is directly related to the control implemented by bureaucratization (in the mid 19th century) and computerization (in the late 20th century), whereas the latter addresses rationalization. An example from bureaucracy is the expansion of paperwork in response to the growth in information. We can still note these traces of bureaucracy from numerous procedures that we have to go through for a decision-making to take place in corporations. He brings up the example of standardized time zones for rationalization which is simply an example for an omission or a breakdown of information to facilitate its processing. ([https://archive.org/details/beniger-james-the-control-revolution Beniger, 1986, p.15-16]) This is also commonly seen from nowadays as some companies may use internal terminologies for the sake of both confidentiality and efficiency.

==== Mechanism of Path Dependence ====Following along the line with the historical lineage, there are certain mechanisms at work of path dependence for information control and workplace surveillance. Bureaucracy is closely related with organizational routines. Many companies have already settled down with a fixed workflow and decision-making process, which makes it more costly for one to change from its conventional way of working. This means switching costs from an old system to a new one is clearly high considering its time and capital, not necessarily will guarantee increasing returns. Hence, many firms end up maintaining its old strategy, reluctant to make any modifications or changes. 

==== Politics and Distribution ====

As businesses grew too large for managers to oversee everything directly, companies replaced human judgment with formal rule-systems, as we see explained by Beniger's 'intellectual technology' ([https://archive.org/details/beniger-james-the-control-revolution/page/65/mode/2up Beniger, 1986, pp. 17, 21]). Algorithmic management continues this: a call once made by a supervisor's judgment is now a rule built into software that runs without one, extending a logic Beniger traces as far back as to Taylorism and the so called 'assembly line'. Winner ([https://www.jstor.org/stable/20024652 1980]) asks whether such systems must concentrate power this way or were driven there, when other designs were equally possible; nothing required shutting workers out of the decision, so leaving them out was a political choice, and not necessarily a technical necessity. Barbrook and Cameron ([https://doi.org/10.1080/09505439609526455 1996]) explain the reasoning why some can defend this 'domination': they show how 1990s tech culture sold new technologies as freedom even as it tightened control, so the rhetoric could become one of closer oversight as flexibility and autonomy. Together, authority accrues to employers, who in some systems can set and revise these oversight rules unilaterally and face less resistance because the arrangement is framed as freedom; workers bear the cost, losing the informal give-and-take they had with a human supervisor for oversight that is harder to see, question, or contest, while being told it grants them more freedom, not less.

==== Contingency and Alternatives ====

A plausible alternative happened on U.S. post WWII: in some workplaces, labor and its pacing was demanded to be set through collective bargaining and enforced jointly, by both management and union stewards monitoring compliance on the line ([https://doi.org/10.1093/icc/dtaf038 Breznitz et al., 2025, p. 211]). So, in this context, workplace surveillance could have developed as a bargained, dual-administered practice rather than a management monopoly. The 1950 [https://en.wikipedia.org/wiki/Reuther%27s_Treaty_of_Detroit Reuther's Treaty of Detroit ] cut that possibility off: unions won guaranteed wages and benefits in exchange for surrendering any general claim over how work was supervised ([https://doi.org/10.1093/icc/dtaf038 Breznitz et al., 2025, pp. 210–211]). Workers depended on wages to survive, and employers controlled the wage relationship. That asymmetry meant employers, given their position of power, were able to arrange things so that, at the very least, workers would have far less capacity to sustain a fight over control itself. Thus, the "trade" wasn't a fair exchange between equals. This reflects which side could afford to hold out on which issue. Hypothetically speaking, had that joint administration model prevailed instead, this could have greatly impacted norms internationally and good practices with worker-union agreed surveillance would have become a widely accepted standard practice.

==== Governance Implication ====

An interesting proposition for present-day governance that wants to fill a gap in workplace surveillance is what [https://doi.org/10.1016/j.digbus.2025.100120 Kayas, Ong and Belal (2025)] are proposing: They argue that EU law, through the [https://eur-lex.europa.eu/eli/reg/2016/679/oj/eng GDPR] and its required [https://en.wikipedia.org/wiki/Data_protection_impact_assessment data protection impact assessments (DPIAs)], already offers meaningful protection for workers in terms of lawful high-risk data collection and data processing. Nevertheless, these DPIAs are built around data and privacy, so they leave out several dimensions of monitoring itself: its purpose, how intrusive it is, how visible it is to the people being watched, and its effects on workers' autonomy, trust, and standing relative to management. These authors propose a "surveillance impact assessment" scoring a system against these characteristics directly, before any surveillance system is put in place. Although this mechanism "is constrained by its reliance on existing literature and the lack of empirical validation," it feels like a genuine alternative for two reasons: 1. it targets the actual mechanism at issue, and 2. at the same time it builds upon a mechanism of active participatory approach where employees have a direct say in how surveillance is designed ([https://doi.org/10.1016/j.digbus.2025.100120 Kayas et al., 2025, p. 11]), which, combined with the actual assessment, would put workers' own judgment at the centre of the one decision that affects them. It's important to note that this alternative is proposed as worth piloting, not as a proven fix, and it is fair to ask whether another mandatory assessment stacked on top of DPIAs would function as intended or simply read to workers as more compliance paperwork.

==== References ====
 Barbrook, R., & Cameron, A. (1996). The Californian Ideology. ''Science as Culture'', 6(1), 44–72. [https://doi.org/10.1080/09505439609526455 https://doi.org/10.1080/09505439609526455] Beniger, J. R. (1986). ''The Control Revolution: Technological and Economic Origins of the Information Society''. Harvard University Press. [https://archive.org/details/beniger-james-the-control-revolution https://archive.org/details/beniger-james-the-control-revolution]  Breznitz, D., Levy, K., Lipartito, K., & Zehavi, A. (2025). An equity-focused research agenda for workplace surveillance. ''Industrial and Corporate Change'', 35(1), 206–221. [https://doi.org/10.1093/icc/dtaf038 https://doi.org/10.1093/icc/dtaf038] Kayas, O. G., Ong, C. E., & Belal, H. M. (2025). From humans to algorithms: A sociotechnical framework of workplace surveillance. ''Digital Business'', 5, 100120. [https://doi.org/10.1016/j.digbus.2025.100120 https://doi.org/10.1016/j.digbus.2025.100120] Ollier-Malaterre, A., Parent-Rocheleau, X., Benkrid, K., & Provost Savard, Y. (2026). Making Sense of Workplace Surveillance: How Workers and Union Representatives Experience, Imagine, and Resist Surveillance Exposure. ''Surveillance & Society'', 24(3), 236–254. [https://ojs.library.queensu.ca/index.php/surveillance-and-society/article/view/19100/13263 https://ojs.library.queensu.ca/index.php/surveillance-and-society/article/view/19100/13263] Winner, L. (1980). Do Artifacts Have Politics? ''Daedalus'', 109(1), 121–136. [https://www.jstor.org/stable/20024652 https://www.jstor.org/stable/20024652]

 
== [[Algorithmic management and workplace surveillance - by Sebastián Muller and Kate Davey''' ''']] ==

=== 1. The current issue. ===
 Algorithmic management and workplace surveillance can be defined as the use of computer-programmed procedures to coordinate labor input in an organization ([https://joint-research-centre.ec.europa.eu/scientific-activities/employment/algorithmic-management-and-digital-monitoring-work_en 1]). Simply put, it is a way to control workers, and workers generally have no say in how the system that tracks them is created. As Winner says "different people are differently situated and possess unequal degrees of power as well as unequal levels of awareness" ([https://www.jstor.org/stable/20024652?seq=7 2]).
The population affected is workers; both “white collar” and “blue collar” workers are impacted. Whether it’s Amazon tracking warehouse workers’ “Time Off Task” with handheld scanners ([https://www.wired.com/story/amazon-worker-tracking-details-revealed 3]) or Meta employees having their every click tracked ([https://www.techpolicy.press/metas-worker-surveillance-tests-eu-rules-on-ai-and-labor/ 4]) to train AI, employers have the power to watch and control their employees in an unprecedented way.
It matters legally because workers have very few rights to fight back against these surveillance programs; even the The Electronic Communications Privacy Act of 1986 allows exceptions to surveillance of employees ([https://bja.ojp.gov/program/it/privacy-civil-liberties/authorities/statutes/1285 5]). 
=== 2. Historical lineage. === 2.1 Panopticon: it is an architectural apparatus developed by Jeremy Bentham in 1791. The design of this building effectively generated the sense of constant monitoring, which in turn motivated the prisoners to behave as if they were being watched at all times. As human labor became commodified, capitalists began implementing measures to ensure the maximization of productivity, including the replication of this design ([https://doi.org/10.14452/MR-070-09-2019-02_2 6]). While unused today, the panopticon is widely considered as both a theoretical and practical inspiration of other forms of surveillance in the workplace.
 
2.2 Scientific management: this theory attempts to analyze and standardize workflows to improve labor productivity. Frederick Winslow Taylor, its main pioneer, used early forms of data analysis during the XX century to decrease walking distances, eliminating unnecessary movements and preventing deliberate slow-walking among workers. ([https://doi.org/10.14452/MR-070-09-2019-02_2 7]). There are many modern forms of data-driven decision-making processes in the workforce. For instance, Amazon warehouses track individual workers' scan rates in real time through handheld scanners, generating data feeds to track and evaluate performance ([https://www.theverge.com/2019/4/25/18516004/amazon-warehouse-fulfillment-centers-productivity-firing-terminations 8]).
 
2.3 PC in the workplace: John Blankenbaker designed the first commercially available personal computer in 1971 ([https://www.computerhistory.org/revolution/personal-computers/17/297 9]), however, it wasn’t until the mid-1980s that the use of personal computers in the workplace became an industry standard. This achievement gave rise to the digital monitoring of work, which in turn allowed employers to monitor employees in real-time, effectively replacing the human eye with a more efficient surveillance mechanism. This represents one of the key aspects of modern-day workplace monitoring. 
 

=== 3. Mechanism of path dependence. ===
''' '''When analyzing the historic lineage of work surveillance, it is possible to point out two main path dependencies mechanisms in place.Firstly, it is clear that many of the technological developments that arose through these events have one major goal, which is to increase wealth within a hierarchical ecosystem. Employers resort to several mechanisms to maximize profits and cut inefficiencies at the expense of workers. 
In addition, standardization can also be singled out as a key mechanism at play. Once one corporation implements any of these developments (such as the digital monitoring of the workplace), the market incentives would force the rest to do the same in order to be profitable and remain competitive.

=== 4. Politics and distribution. ===
 One simple answer to this question would be that capitalists implement and leverage these mechanisms to increase their wealth while the working class remains powerless and unable to withdraw itself from the system in force. However, if we resort to Winner's framework, we see that the technical elements of these technologies are not intrinsically capitalist, since workplace surveillance can also be observed in alternative regimes (including chattel slavery and the Soviet regime). It seems that, rather than being exclusively tied to capitalistic values, the methods in question respond to strong hierarchical structures.
In addition, even under a libertarian framework, structural inequalities are still present, allowing a privileged class to enjoy individual freedoms while segregated groups keep performing low-income jobs ([https://www.metamute.org/editorial/articles/californian-ideology 10]), arguably still subject to surveillance. Thus, not only are these technologies compatible with hierarchical top-down forms of government, but also with structures that allow such hierarchies to form without state intervention.
That said, this form of management seems less compatible with democratic societies, which tend not only to value individual liberties but to turn them into enforceable rights (including workers' privacy, health and integrity), thus establishing boundaries for workplace surveillance. This could explain why studies of American business leaders have found them unhappy with democratic scruples ([https://www.jstor.org/stable/20024652 11]).
 
=== 5. Contingency and alternatives. === 
Bush writes about the memex, a machine that could extend a person’s memory and that the memory would belong to the person who owned it. As Bush expands, “For mature thought there is no mechanical substitute" ([https://dl.acm.org/doi/pdf/10.1145/227181.227186 12]). His idea was that a machine could take over the repetitive nature of someone’s job, while the person could record their own thoughts for their own use. Bush’s memex was never created, but the idea inspired products like Microsoft’s Productivity Score that were made for employers (the buyers) not for the person whose memory it used (the worker).
This immediately brings to mind Meta's Model Capability Initiative, which was done to train AI agents the company hopes to sell to other employers as well as watch their employees ([https://www.reuters.com/sustainability/boards-policy-regulation/meta-start-capturing-employee-mouse-movements-keystrokes-ai-training-data-2026-04-21/ 13]). If this was Bush’s memex, it would belong to the employee using it, not Meta. 

=== 6. Governance implication. ===
Democratic oversight of the use of algorithms as a form of surveillance is essential to ensure that workers’ rights are not overstepped in the name of efficiency. As such we propose: all workers should have a legal right to be consulted and approve a surveillance program before a company installs and uses it. A tracking system should be a group decision, not an inevitable, unilateral one. While the United States has very little protections for workers in terms of surveillance, the EU leads the way with the GDPR, specifically Article 22 ([http://data.europa.eu/eli/reg/2016/679/oj 14]), and the AI Act Article 5 ([http://data.europa.eu/eli/reg/2024/1689/oj 15]). It’s also worth noting that in December 2026, the Platform Work Directive ([https://eur-lex.europa.eu/eli/dir/2024/2831/oj/eng 16]) will come into effect which will ban automated termination.

= [[Platform Dependence in Public Communication (News) - Ceren Isiklar & Rodrigue Kayembe|Platform Dependence in Public Communication (News)]] Ceren Isiklar & Rodrigue Kayembe =

= Centralization of AI Compute and Data =

== Centralization of AI Compute and Data - by Sisi Hammoudi & Juliette Chamot ==
=== Governance issue ===
Today, several path-dependence phenomena are emerging across advanced computing technologies. This assignment addresses one of them: the centralisation of AI compute and data. Training and running frontier AI systems now depends on a handful of hyperscale providers, namely Amazon, Microsoft, Google, and Nvidia, which dominates the GPU market. In other words, near monopolies exist throughout the supply chain. This concentration creates a governance problem affecting public and private actors alike, both legally and institutionally. On the institutional side, it prevents public administrations, and states other than the US or China, from reaching sovereign AI capacity. It also raises competition-law concerns, since dominant market positions can prevent the market from delivering the best products at the best prices to consumers. Finally, AI developers, like states, are often forced into contracts with hyperscale providers, creating a structural dependence on them.


=== Historical lineage ===
==== 1945 ====
Wartime scientific instruments should have been redirected to extend human command over knowledge, helping establish the postwar model of sustained federal funding for computing research (Vannevar Bush's, "As We May Think », 1945).


==== 1990s ====
The privatisation of the NSFNET established the idea that critical digital infrastructure, even though publicly subsidised, could legitimately be privately owned. The EU Commission followed the US's lead, recommending privatisation and deregulation of telecommunications and media as the best way to build the "infobahn" though EU members could instead have chosen nationalisation of the infrastructure (Barbrook & Cameron, "The Californian Ideology," 1996).


==== 2006 ====
<div class="s13">Hyperscale cloud computing turned massive fixed capital investment in data centresinto a rentable service, favouring firms able to invest at scale.</div>
<div class="s13">Same period: Nvidia's proprietary CUDA software, released alongside its GPUs, became the de facto standard for machine learning, locking in a decade of tooling and expertise around one firm's hardware.</div>
<div class="s13">
==== From 2022 ====
Semiconductor export controls and the CHIPS Act turned compute into an explicit instrument of geopolitical strategy.</div>



=== Mechanisms of path dependance ===
Several mechanisms drive this centralisation, but path-dependence ones dominate. Because AI compute is a capital-intensive industry, new entrants struggle to compete with hyperscalers that have already sunk huge sums into infrastructure, a barrier that produces monopoly-like situations. This dynamic, known as increasing returns, generates system-scale economies that harden into a near-irreversibility of investment (David, "Clio and the Economics of QWERTY," 1985). It overlaps with a second mechanism, standardisation, itself a form of technical interrelatedness: researchers' investment in Nvidia's CUDA-based tooling locks the field into that ecosystem regardless of whether better alternatives exist. A further mechanism is legal and contractual entrenchment. For example, Microsoft's stake in OpenAI binds developers to a specific provider by agreement rather than technical necessity.=== Politics and distribution ===Technology does not have a history of distributing power equally. Rather, its development has strengthened the interests and authority of powerful groups that already possessed economic and political resources, and it is yet to be determined whether AI compute and data will follow suit. 
Author Langdon Winner illustrates the historical constraints of technological development with the case of New York from the 1920s to the 1970s. He argued that the city’s roads, bridges, and parks’ physical arrangement contained “explicit and implicit political purpose.” These infrastructures privileged wealthy and white communities while restricting poor and Black populations’ choice of mobility. Similarly, the development of industrial machinery increased productivity, simultaneously weakening union and non-union workers’ bargaining power.
In the twentieth century, following the Industrial Revolution, a Control Revolution emerged from the increasing need to manage expanding production, transportation, and information flows requiring strengthened organizational bureaucratic control. This was heavily shaped by the political context of post-World War II warfare.
Therefore, the bearer of the initial cost of technology is the person putting in the capital to erect it, but ultimately, the person bearing the long-term cost of technology, whose choices diminish, is not the rich and privileged, but the everyday person subject to technologies. Forced to adapt, to reshape their professional skills to adapt to an environment trying to make them obsolete.
AI compute and data are at risk of creating the same effect. In their critique of dotcom neoliberalism, authors Barbrook and Cameron exposed a similar socio-economic tension. The Californian Ideology claims digital technology empowers individuals; however, perpetual unequal access to it creates an apartheid between “information-rich” and “information-poor” populations.
Finally, the development of AI compute and data depends on expensive and scarce computing infrastructure and a highly concentrated semiconductor supply chain. Today, this is one of the most concentrated industries, with single companies having near-total monopolies over the production of chips necessary for the functioning of AI. This monopoly model threatens to reinforce the existing economic hierarchy.

=== Contingency and alternatives ===If a different technological trajectory was plausible, the most apt time for it to have occurred would be during the transition from the Industrial Revolution to the “Control Revolution”. When industrial production accelerated, our society experienced a “crisis of control”, creating opportunities to organize information and communication differently.
Instead of allowing control technologies to be monopolized within businesses and states, our societies could have pursued a more democratic and decentralized system. Barbrook and Cameron provide such an alternative path through their idea of the “electronic agora,” where digital networks could encourage community participation and direct democracy, where government could actively engage with the people to shape digital infrastructure and expand access. These alternatives did not actually prevail because corporate and state market participants possessed greater resources and authority to shape technological development at the end of the twentieth century. 
Inherently, technological development is not predetermined; it can support either authoritarian or democratic arrangements. Today, monopolies control chip production needed for AI compute and data, but other technologies like blockchain permit democratic participation.

=== Governance implication ===History always threatens to repeat itself. As Barbrook discussed, digital societies are shaped by the political context in which they develop. It is a valid concern, then, that the development of information technology, through AI computing, would follow the same trajectory- that is, be constrained to benefit the wealthy who funded its development, and ultimately harm the middle and lower classes enabling its functioning. 
However, AI’s place in our society has not yet crystallized. Technology governance does not have to adhere to this system. Such a seemingly sustainable approach to AI compute and data governance has already been broached. In 2022, the Notre Dame Journal on Emerging Technologies published an article in its issue on Artificial Intelligence where the authors proposed an AI compute system based on a “dual investment strategy”. Firstly, the model would be launched by “subsidizing and negotiating” cloud computing for human AI researchers through “existing” cloud vendors, which would expand on existing AI compute initiatives. Secondly, the compute system would be specifically invested in as a publicly owned infrastructure that would be built and operate similarly to national libraries which own “academic supercomputing facilities”. 
The pilot program would have a “base level of compute access” which would keep costs of processing down and enable it to meet the compute needs of a majority of academic researchers. Absent any strengthened legislation on the use of AI compute and data, the hardware would be accompanied by an AI ethics framework to enable sustainable technological governance. In addition to an ethical staff, the human resources needed to keep an AI compute model running would include data officers, scientists, data center and software engineers, research developers, and system administrators. These positions require specialized training, allowing new types of professionals, from any background, to have access to professional prosperity in the field of AI. 
By implementing these frameworks, never neglecting the human element, and the economic and intellectual impact AI has on our society, we move from a historical determinist perspective to one acknowledging historical constraints and learning from them.


=== Bibliography ===
<div><div>[1] Langdon Winner, Do Artifacts Have Politics, Daedalus, Winter, 1980, Vol. 109, No. 1, Modern Technology: Problem or Opportunity? pp. 121-136, 124, (1980)</div>
<div>[2] Langdon Winner, Do Artifacts Have Politics, Daedalus, Winter, 1980, Vol. 109, No. 1, Modern Technology: Problem or Opportunity? pp. 121-136, 124, (1980)</div>
<div>[3] James R. Beniger, The Control Revolution: Technological and Economic Origins of the Information Society, Harvard University Press, p. 443, 450 (1986)</div>
<div>[4] Richard Barbrook, Andy Cameron, The Californian Ideology, ''Science as Culture'', ''6''(1), 44–72, 61 (1996)</div>
<div>[5] Anton Korinek, Jai Vipra, Concentrating Intelligence: Scaling and Market Structure in Artificial Intelligence, National Bureau of Economic Research, 10.3386/w33139, p. 11, 13-15 (2024)</div></div>
<div><div>[6] James R. Beniger, The Control Revolution: Technological and Economic Origins of the Information Society, Harvard University Press, p. 443, 442 (1986)</div>
<div>[7] Richard Barbrook, Andy Cameron, The Californian Ideology, ''Science as Culture'', ''6''(1), 44–72 (1996)</div>
<div>[8] Langdon Winner, Do Artifacts Have Politics, Daedalus, Winter, 1980, Vol. 109, No. 1, Modern Technology: Problem or Opportunity? pp. 121-136, 130 (1980)</div>
<div>[9] Daniel E. Ho, Jennifer King, Russell C. Wald & Christopher Wan, The Centrality of Data and Compute for AI Innovation: A Blueprint for the National Research Cloud, 3 Notre Dame J. on Emerging Tech. 71, 81(2022)</div>
<div>[10] Daniel E. Ho, Jennifer King, Russell C. Wald & Christopher Wan, The Centrality of Data and Compute for AI Innovation: A Blueprint for the National Research Cloud, 3 Notre Dame J. on Emerging Tech. 71, 82 (2022)</div></div>

== Centralization of AI compute and Data - by Atmika Dechamma & Pablo Latorre ==
=== Governance Issue ===
AI computing capacity and data presents the governance problem of an extreme concentration of technological infrastructure and information in a restricted number of companies that privately and globally dominate the market. This affects public and private actors - such as small and medium enterprises, potential new competitors, and Global South countries - that increasingly depend on AI systems outside their control and/or are deterred from competing with established companies because of impenetrable entry barriers - most notably initial investments and operational costs (1). The institutional and legal relevance of this is that it consolidates, reproduces and amplifies asymmetries of power in favor of the “Big Tech”, reverting the technological trend from digital decentralization to physical infrastructure dependency and curbing innovation and market efficiency.

=== Historical Lineage ===
==== Limitation of Machine Learning ====The AI revolution was built on the limitation of human capabilities to process large information and derive efficient results - where before there existed only facilitative tools to allow for human-enabled processes, AI now offered derivations and cognitive reasoning in moments and at large scales.

 ==== Revert to Physical Infrastructure Dependency in the rush to be the best ====The present AI model heavily relies on hardware in order to perform complex data processing for deep learning. The data centre models followed require larger model sizes (which require higher resource, information and capital investment) to be able to yield better results. Due to the limitation of resources and a rush to establish the best working model of an AI agent, data centres are built on a rudimentary model heavily dependent on massive amounts of electricity and fresh water (2).

 ==== Regulation and Decentralization of Information ====Regulation and influence of sovereigns and institutions has resulted in data sharing and information exchange structurally through open source and cloud computing. But, while AI models claims their selling point to be accessibility for all, their information and derivations remain in a closed server, stored for the purpose of training their AI agents to be the best (3). Big AI contenders attract higher consumer and information flows overall resulting in a better developed AI model, proving disadvantageous in any market where early contenders have an advantage and accumulation of power.
=== Mechanisms of Path Dependence ===Early on in the industry there is an expectation of human and capital investment from all stakeholders involved, this high cost for entry into the market and expected capital to sustain and develop a model by both consumer and enterprise makes for a high switching cost. For the AI model to run and scale there is a presumption that the rate of growth is dependent on the size of data fed, resulting in a better model and hence, more consumer usage. A closed centralised data loop is created but the requirement for storage and processing far surpasses current infrastructures capabilities - heavily relying on early market entry advantage and sustained capital investments taking away from smaller aspirants and hoarding of power within closed circles. Despite its physical dependency, AI data processing has unique requirements beyond legacy infrastructures (4) - larger demand and production of GPUs and designed to handle immense thermal heat generated while consuming electricity beyond current capabilities. 

 The AI space largely remains unregulated due to its undefined form and fluid nature, however, this has given way for unethical and morality issues regarding data privacy, exploitation of labour and environmental harms.

=== Politics and Distribution ===
Concentration strengthens the values, interests, capacities, and authority of the “Big Tech” over other actors of public and private nature. In this context, Barbrook and Cameron (5) illuminate the ethos behind AI technology development while Winner (6) makes observable the power relations surrounding it: concentration is the consequence of an AI development anchored in the belief that it must be free from all institutional and normative restriction since it is inherently liberating and non-political which, in turn, provides a political/moral justification for the concentration (if technology is inherently liberating and non-political then there is no power relations to observe or question) that translates itself into power dynamics where the AI developer not only owns the AI technology but also decides how it is trained and deployed and which models the user can or cannot access.

=== Contingency and Alternatives ===
Concentration is due, first and foremost, to the fact that AI computing capacity requires massive and cutting-edge physical infrastructure - such as data centers and microprocessors - which, in turn, requires upfront investments that are unaffordable to most companies. In David’s (7) language, companies that could finance the infrastructure and position themselves first obtained self-reinforcing advantages and a “lock-in” effect in the market. In other words, concentration is the consequence of the development of AI physical infrastructure privately with no mixed or public alternatives as a public service. Therefore, there was a space for an alternative future at the end of the 1980s or the beginning of the 1990s when the privatization model of the technological infrastructure was consolidated in line with the “Californian Ideology”. In that moment governments had the chance to treat technological infrastructure as a mixed or public good rendering it universally accessible as it was the case of the Internet; this would have permitted companies of all sizes to profit from economies of scale, reducing entry barriers and fostering competition. However, this alternative path did not prevail since the Californian Ideology provided the moral/political justification to exclude governments from the development of technological infrastructure in the exact moment in which such a decision would produce structural consequences. If that path had been followed it could have contributed to collective access to economies of scale, lower entry barriers, and a more competitive AI market.

=== Governance Implication ===
Considering the above, public or mixed investment on AI physical infrastructure is a necessary governance measure for market decentralization and collective access to AI computing capacity and data. From an historical perspective, the proposed measure is not merely a corrective action for a market failure but for a contingent and ideological decision made between the late 1980s and early 1990s that produced a structural “path dependence” that now is necessary to change.
==== References ====
(1) The Quiet Re-Centralization of Tech: Why AI Is Reversing 20 Years of Platform Decentralization. (2026, January 10). Retrieved on 8 September 2026, from Medium website : [https://medium.com/@sumitbhardwaj1357/the-quiet-re-centralization-of-tech-why-ai-is-reversing-20-years-of-platform-decentralization-ce15962b6a13 https://medium.com/@sumitbhardwaj1357/the-quiet-re-centralization-of-tech-why-ai-is-reversing-20-years-of-platform-decentralization-ce15962b6a13]

(2) [https://www.mdpi.com/1996-1073/19/1/137 Ginzburg-Ganz, E., Lifshits, P., Machlev, R., Belikov, J., Krieger, Z., & Levron, Y. (2025). Technical challenges of AI data center integration into power grids—A survey. ''Energies'', ''19''(1), 137.]

(3) [https://decai-resources.s3.us-east-1.amazonaws.com/decentralized_ML_perspective-16.pdf Singh, A., Lu, C., Gupta, G., Chopra, A., Blanc, J., Klinghoffer, T., ... & Raskar, R. (2024). A perspective on decentralizing AI.]

(4) The Rise of AI Data Center Models and the Decline of the General-Purpose Data Center. (Circa 2026). Retrieved on 8 September 2026, from SubZero Engineering website : [https://www.subzeroeng.com/the-decline-of-the-general-purpose-data-center/ https://www.subzeroeng.com/the-decline-of-the-general-purpose-data-center/]

(5) [https://git.xpub.nl/onebigear/SI_09_reference_library/raw/commit/22bf61e1dcfed35d58fba23808212bd00584be47/Richard_Barbrook_Andy_Cameron_Californian_ideology.pdf Barbrook, R., & Cameron, A. (1996). The Californian Ideology. ''Science as Culture'', ''6''(1), 44-72].

(6) [https://d1wqtxts1xzle7.cloudfront.net/87708502/langdon_winner_do_artifacts_have_politics-libre.pdf?1655588533=&response-content-disposition=inline%3B+filename%3DDo_Artifacts_Have_Politics.pdf&Expires=1788890463&Signature=UBBiMqkip5MfYZbkbLNAN6aE2hyZdKkER-HFcmyxXkwIZuTnMKdOGoao8ddFr-rQ5T55A8NWUOlnZkGYmi66e7XR9h99DgTELMCrh48sXjOb8KYgA48r~SUUymt1IW~R~aOKJzitMIUHcrc6yEGKK1OuaA~mrEIVOXpUjQMYjqOd26Xmk9YJEglTkqaxeR47VCCth5FvozNQIowdfjzeu8aNHnI8V7Yur1IiXY~PGvYRcN6wBtcgpVMHZUtNaI8NjmYbp2p5TBYrPMF9MCLtvpGhPlsCfsvA8Bdm6XulGwRlksZBKWvVUQ8D~M8oLu12V39YaiETIObyJqbE7ctIcQ__&Key-Pair-Id=APKAJLOHF5GGSLRBV4ZA Winner, L. (2017). Do artifacts have politics?. In ''Computer ethics''(pp. 177-192). Routledge].

(7) [https://sjbae.pbworks.com/f/David+1985.pdf David, P. A. (1985). Clio and the Economics of QWERTY. ''The American economic review'', ''75''(2), 332-337].

== Centralization of AI compute and Data - by Sergey Abramov and Paul Stein ==<div>

=== The current governance issue ===
Advanced AI currently depends on a narrow group of tech giants1 and semiconductor oligopolies controlling microchips and software, hyperscale cloud infrastructure, large datasets, frontier models, and distribution platforms. This presents a governance problem where this narrow group exercises control over foundational digital infrastructure. This directly affects consumers, businesses, and academia out of AI. Legally and institutionally, this matters because it creates a profound democratic deficit: this neutralizes antitrust frameworks designed for traditional markets, often bypassing conventional copyright and data privacy protections through unchecked data harvesting thereby complicating enforcement.This ultimately might affect national sovereignty as the countries that host the buildout will shape the future of AI: who controls it, what values it contains, how it is used2.
 
=== Historical lineage ===
In 1987, Taiwan Semiconductor Manufacturing Company (TSMC) was founded, establishing the world’s first semiconductor fab. Instead of designing and making its own chips, TSMC only manufactured chips designed by other companies. Today, nearly 100% of the world's most advanced AI processors are manufactured by TSMC, leaving the entire global AI compute supply chain centralized under one company.
 In 2006, Nvidia launched CUDA (Compute Unified Device Architecture), a free software layer that allowed developers to use Nvidia's graphics chips (GPUs) which later became perfect for training deep neural networks. Today, this creates a near-monopoly on AI compute hardware.
 Soon after Amazon launched AWS, Microsoft and Google launched their own cloud platforms marking the structural pivot to the cloud service provider model, shifting computing from a distributed architecture (where companies bought and managed their own physical servers) to a centralized infrastructure. As training modern AI models requires tens of thousands of GPUs, today’s frontier AI companies are entirely dependent on the extremely expensive physical infrastructure of a handful of cloud hyper-scalers (Microsoft, Google, AWS, and Meta).
 In 2009, Fei-Fei Li launched ImageNet, a massive database of over 14 million images, which proved that AI model performance scales directly with data volume. This created an immediate institutional advantage for consumer internet platforms (like Google and Meta), because they were already architected to collect, centralize, and store masses of user data through search engines and social networks, they possessed the unique data stack required to train high-performing AI.
 
=== Mechanism of path dependence ===
The centralization of AI compute and data is a classic case of technological lock-in. When structural choices were made decades ago, they triggered powerful, self-reinforcing loops. One such loop is the switching costs and software standardisation. As developers have spent more than a decade working with CUDA, moving to a rival hardware provider (or a decentralized compute network) requires an enterprise to completely rewrite its foundational software stack. The cost of code migration, developer retraining, and potential hardware downtime force companies to stay tied to dominant hardware monopolies such as Nvidia. Commitments, licensing conditions, and exclusivity provisions add contractual costs. Fewer than 1% of cloud customers switch provider annually, identifying technical differentiation and data-transfer charges as important barriers4.
 Network effects also matter for computing power. More customers generate more revenue, usage information, and opportunities to improve infrastructure and models. Better services attract additional users and developers, whose applications make the platform more valuable. In cloud–AI partnerships, providers may additionally obtain information about model performance and future infrastructure demand. These feedback loops strengthen firms already operating across cloud, models, software, and distribution.
 
=== Politics and distribution ===
According to Langdon Winner's definition, modern AI computing infrastructure may qualify as an "inherently political technology" because its large scale and significant financial requirements not only lend themselves to, but necessitate,5 centralized, hierarchical control. The physical siting of this infrastructure shows evidence of Winner’s second category of politicization of technologies, the ‘political arrangement’:6 Of the five largest U.S. AI firms, four are headquartered in California,7 while just one is (partially) domiciled in Texas. Texas, however, hosts twice the GPU clusters.8 This placement geographically insulates beneficiaries of the data infrastructure boom from its negative externalities. The specific choice of location may itself be a deliberate preference on account of Texas’ unique approach to water usage rights.9 Juxtaposition of this unequal and even exploitative balance of interests with the pontificating rhetoric of the urban, information rich, beneficiaries is evocative of the internal contradictions of the Californian Ideology - the virtual artisans existence and privileged position is still based on the exploitation of a vulnerable out-group, the "information-rich" consuming the output, the "information-poor" hosting the extraction.10
 
=== Contingency and Alternatives ===
Profit motive and economies of scale now concentrate investment among a few hyperscalers,11 but this reflects policy choice, not government incapacity: Compute infrastructure may be inherently political, but the technology is agnostic as to where the organizing authority originates, and states have historically funded infrastructure of comparable or greater scale, while facing less pressure to show short-term returns. The clearest turning point was 1995, when the National Science Foundation retired NSFNET, the publicly funded backbone that had connected U.S. research and supercomputing centers since the mid-1980s,12 handing internet infrastructure to commercial providers with no conditions attached. This was a deliberate policy decision made amid the deregulatory push of the Clinton era. The privatization that followed ushered in the private industrial dominance that persists today in the form of hyperscalers. Had a publicly governed backbone instead been scaled up, today's compute infrastructure might rest on a different foundation entirely.
 
=== Governance implication ===
Despite private hyperscaler dominance, the U.S. does have a nominal public alternative: the National AI Research Resource (NAIRR), established following the National AI Initiative Act of 2020 and piloted from January 2024, which explicitly aims to democratize access to the computational resources essential to AI research.13In Europe, a comparable program is the EuroHPCJU,14 an initiative intended to pool resources to build supercomputing capacity. The undertaking’s legislative basis has since been extended to include networked ‘AI factories', as well as the so-called “Gigafactories”.15 The framing in language of sovereignty indicates intent to at least compete with private infrastructure initiatives. Though government-led measures cannot reverse the centralization of compute, they can democratize access - provided they are intensified. The European Commission has “set a goal of investing more than €1 billion per year in AI research and innovation”.16 While laudable in its own right, it pales in comparison to the investments of industry leaders, which are approaching, if not exceeding, trillions of Dollars.

=== References & Notes ===
1.Amazon, Google, Meta, Microsoft, and Oracle collectively hold an estimated 71% of the world’s cumulative AI compute as of Q4 2025, measured in H100-equivalents of computing power. Luke Emberson, Josh You and Venkat Somala (2026), "Five hyperscalers now own over two-thirds of global AI compute". Published online at epoch.ai. Retrieved from 'https://epoch.ai/data-insights/hyperscalers-control-most-compute' [online resource]. Accessed 7 Sept 2026.2. Alasdair Phillips-Robins, Teddy Tawil and Sam Winter-Levy, ‘The Compute Coalition: How to Build the Future of AI in the Free World’ (Carnegie Endowment for International Peace, 8 June 2026) https://carnegieendowment.org/research/2026/06/the-compute-coalition-how-to-build-the-future-of-ai-in-the-free-world accessed 7 September 2026.

3. Amazon Web Services (AWS), a cloud computing platform

4. Competition and Markets Authority, Cloud Infrastructure Services: Final Decision Report (31 July 2025) https://assets.publishing.service.gov.uk/media/688b8891fdde2b8f73469544/final_decision_report.pdf accessed 7 September 2026.

5. Langdon Winner, 'Do Artifacts Have Politics?' (1980) 109(1) ''Daedalus'' 121, 128.

6. Ibid. 123.

7. OpenAI, Anthropic, Google DeepMind, Meta AI. xAI is an exception to the rule, domiciled between Palo Alto, CA and San Antonio, TX.

8. Texas’ 17 to California’s 8, cf. Texas Royalty Brokers, 'AI Clusters by State' (''Texas Royalty Brokers'', 2025) https://texasroyaltybrokers.com/ai-clusters-by-state/ accessed 7 September 2026.

9. The ‘rule of capture’, established in Houston & TC Railway Co v East (Tex 1904), permits landowners to extract groundwater beneath their land without liability for depleting adjacent wells or aquifers, a doctrine now exploited by hyperscale AI data centres to draw large cooling-water volumes with minimal regulatory constraint.

10. Richard Barbrook and Andy Cameron, 'The Californian Ideology' (1996) 6(1) ''Science as Culture'' 44.

11. Goldman Sachs, 'Global AI Investment Is Forecast to Exceed $1 Trillion in 2026' (''Goldman Sachs Insights'', 7 August 2026) https://www.goldmansachs.com/insights/articles/global-investment-is-forecast-to-exceed-1-trillion-in-2026 accessed 7 September 2026.

12.  National Science Foundation, 'Birth of the Commercial Internet' (''NSF Impacts'') https://www.nsf.gov/impacts/internet accessed 7 September 2026.

13. CASRAI, 'NAIRR: National AI Research Resource, Explained' (''CASRAI'') https://casrai.org/dictionary/term/national-ai-research-resource-nairr accessed 7 September 2026.

14. Council Regulation (EU) 2021/1173 of 13 July 2021 on establishing the European High Performance Computing Joint Undertaking and repealing Regulation (EU) 2018/1488 [2021] OJ L256/1.

15. Council Regulation (EU) 2026/150 of 16 January 2026 amending Regulation (EU) 2021/1173 on establishing the European High Performance Computing Joint Undertaking [2026] OJ L 2026/150.

16. European Commission, 'Realising Europe's Digital Decade' (''European Commission'', 11 October 2024) https://commission.europa.eu/strategy-and-policy/priorities-2019-2024/story-von-der-leyen-commission/realising-europes-digital-decade_en accessed 7 September 2026.</div>

Governance by history

2026-09-24T06:40:01Z

Robert Florijn: Civilian AI ois already at the front; the EU doen't see it!

[[File:Old internet.jpg|300px|thumb]]

= contributions =
*can
*you 
*please
*create
*individual pages
*from the contributions below
*and link them back here? 
*[[Information Control and Workplace Surveillance - Jihae Park and Mariona Fortuny]]
*[https://wiki.ivir.nl/index.php?title=Centralization_of_Compute_and_Data Centralization of AI Compute and Data - by Sergey Abramov and Paul Stein]
*'''Civilian AI at the Front Final'''-Robert Florijn: [[Media:Civilian AI at the Front Final.pdf|PDF]]

= links things that popped up during class =
*[https://craphound.com/ Cory Doctorow]
*[https://dougengelbart.org/content/view/209/ The mother of All demos] (first mouse, graphical interface, word processor, etc)
*Can a simple sensor be political? [https://gizmodo.com/why-cant-this-soap-dispenser-identify-dark-skin-1797931773 well, this "racist" soap dispenser made headlines a couple of years ago.]
*Podcast series: [https://www.the-santiago-boys.com/ Morozov: the Santiago Boys.]
*Anatomy on AN AI. [[File:Ai-anatomy-map.pdf|check out the pdf here]]

= Workplace surveillance =
== Information Control and Workplace Surveillance - by Jihae Park and Mariona Fortuny Jané ==

==== The Current Governance Issue ====
There are numerous current governance problems arising from technological innovations, yet here we will specifically focus on the information control and workplace surveillance. Whether intended or unintended, there were certain historical incidents which resulted in social and political consequences. As noticed from the Control Revolution, the dramatic evolution in mass media and telecommunications technologies has led to the reinforcement of the economic and political control during the Industrial Revolution. ([https://archive.org/details/beniger-james-the-control-revolution Beniger, 1986]) By applying this logic to a modern example, this is clearly how many corporations are getting in control of their employees through surveillance technologies. Along with the emergence of remote and hybrid work after Covid-19, employers are now relying on “bossware” to analyze the emails, chats, and messages of their employees on platforms such as Teams or Slack. ([https://ojs.library.queensu.ca/index.php/surveillance-and-society/article/view/19100/13263 Ollier-Malaterre et al., 2026, p.237]) Some companies have strict restrictions on transferring data both internally and externally in protection of confidential information and consistently monitor their workers to the certain extent which could slow down their efficiency. This is problematic as it’s not only violating the privacy of their employees but also depriving employees of rights they should be naturally be able to exercise.



==== Historical Lineage ====
In relation to the current issue of the information control and workplace surveillance, Max Weber’s argument of bureaucracy and rationalization in the middle and late nineteenth century are good historical examples which are still frequently visible nowadays. Weber stated how the range of control could extend not only by expanding the capability to process information but also by limiting the amount of information to be processed. Here, the former is directly related to the control implemented by bureaucratization (in the mid 19th century) and computerization (in the late 20th century), whereas the latter addresses rationalization. An example from bureaucracy is the expansion of paperwork in response to the growth in information. We can still note these traces of bureaucracy from numerous procedures that we have to go through for a decision-making to take place in corporations. He brings up the example of standardized time zones for rationalization which is simply an example for an omission or a breakdown of information to facilitate its processing. ([https://archive.org/details/beniger-james-the-control-revolution Beniger, 1986, p.15-16]) This is also commonly seen from nowadays as some companies may use internal terminologies for the sake of both confidentiality and efficiency.

==== Mechanism of Path Dependence ====Following along the line with the historical lineage, there are certain mechanisms at work of path dependence for information control and workplace surveillance. Bureaucracy is closely related with organizational routines. Many companies have already settled down with a fixed workflow and decision-making process, which makes it more costly for one to change from its conventional way of working. This means switching costs from an old system to a new one is clearly high considering its time and capital, not necessarily will guarantee increasing returns. Hence, many firms end up maintaining its old strategy, reluctant to make any modifications or changes. 

==== Politics and Distribution ====

As businesses grew too large for managers to oversee everything directly, companies replaced human judgment with formal rule-systems, as we see explained by Beniger's 'intellectual technology' ([https://archive.org/details/beniger-james-the-control-revolution/page/65/mode/2up Beniger, 1986, pp. 17, 21]). Algorithmic management continues this: a call once made by a supervisor's judgment is now a rule built into software that runs without one, extending a logic Beniger traces as far back as to Taylorism and the so called 'assembly line'. Winner ([https://www.jstor.org/stable/20024652 1980]) asks whether such systems must concentrate power this way or were driven there, when other designs were equally possible; nothing required shutting workers out of the decision, so leaving them out was a political choice, and not necessarily a technical necessity. Barbrook and Cameron ([https://doi.org/10.1080/09505439609526455 1996]) explain the reasoning why some can defend this 'domination': they show how 1990s tech culture sold new technologies as freedom even as it tightened control, so the rhetoric could become one of closer oversight as flexibility and autonomy. Together, authority accrues to employers, who in some systems can set and revise these oversight rules unilaterally and face less resistance because the arrangement is framed as freedom; workers bear the cost, losing the informal give-and-take they had with a human supervisor for oversight that is harder to see, question, or contest, while being told it grants them more freedom, not less.

==== Contingency and Alternatives ====

A plausible alternative happened on U.S. post WWII: in some workplaces, labor and its pacing was demanded to be set through collective bargaining and enforced jointly, by both management and union stewards monitoring compliance on the line ([https://doi.org/10.1093/icc/dtaf038 Breznitz et al., 2025, p. 211]). So, in this context, workplace surveillance could have developed as a bargained, dual-administered practice rather than a management monopoly. The 1950 [https://en.wikipedia.org/wiki/Reuther%27s_Treaty_of_Detroit Reuther's Treaty of Detroit ] cut that possibility off: unions won guaranteed wages and benefits in exchange for surrendering any general claim over how work was supervised ([https://doi.org/10.1093/icc/dtaf038 Breznitz et al., 2025, pp. 210–211]). Workers depended on wages to survive, and employers controlled the wage relationship. That asymmetry meant employers, given their position of power, were able to arrange things so that, at the very least, workers would have far less capacity to sustain a fight over control itself. Thus, the "trade" wasn't a fair exchange between equals. This reflects which side could afford to hold out on which issue. Hypothetically speaking, had that joint administration model prevailed instead, this could have greatly impacted norms internationally and good practices with worker-union agreed surveillance would have become a widely accepted standard practice.

==== Governance Implication ====

An interesting proposition for present-day governance that wants to fill a gap in workplace surveillance is what [https://doi.org/10.1016/j.digbus.2025.100120 Kayas, Ong and Belal (2025)] are proposing: They argue that EU law, through the [https://eur-lex.europa.eu/eli/reg/2016/679/oj/eng GDPR] and its required [https://en.wikipedia.org/wiki/Data_protection_impact_assessment data protection impact assessments (DPIAs)], already offers meaningful protection for workers in terms of lawful high-risk data collection and data processing. Nevertheless, these DPIAs are built around data and privacy, so they leave out several dimensions of monitoring itself: its purpose, how intrusive it is, how visible it is to the people being watched, and its effects on workers' autonomy, trust, and standing relative to management. These authors propose a "surveillance impact assessment" scoring a system against these characteristics directly, before any surveillance system is put in place. Although this mechanism "is constrained by its reliance on existing literature and the lack of empirical validation," it feels like a genuine alternative for two reasons: 1. it targets the actual mechanism at issue, and 2. at the same time it builds upon a mechanism of active participatory approach where employees have a direct say in how surveillance is designed ([https://doi.org/10.1016/j.digbus.2025.100120 Kayas et al., 2025, p. 11]), which, combined with the actual assessment, would put workers' own judgment at the centre of the one decision that affects them. It's important to note that this alternative is proposed as worth piloting, not as a proven fix, and it is fair to ask whether another mandatory assessment stacked on top of DPIAs would function as intended or simply read to workers as more compliance paperwork.

==== References ====
 Barbrook, R., & Cameron, A. (1996). The Californian Ideology. ''Science as Culture'', 6(1), 44–72. [https://doi.org/10.1080/09505439609526455 https://doi.org/10.1080/09505439609526455] Beniger, J. R. (1986). ''The Control Revolution: Technological and Economic Origins of the Information Society''. Harvard University Press. [https://archive.org/details/beniger-james-the-control-revolution https://archive.org/details/beniger-james-the-control-revolution]  Breznitz, D., Levy, K., Lipartito, K., & Zehavi, A. (2025). An equity-focused research agenda for workplace surveillance. ''Industrial and Corporate Change'', 35(1), 206–221. [https://doi.org/10.1093/icc/dtaf038 https://doi.org/10.1093/icc/dtaf038] Kayas, O. G., Ong, C. E., & Belal, H. M. (2025). From humans to algorithms: A sociotechnical framework of workplace surveillance. ''Digital Business'', 5, 100120. [https://doi.org/10.1016/j.digbus.2025.100120 https://doi.org/10.1016/j.digbus.2025.100120] Ollier-Malaterre, A., Parent-Rocheleau, X., Benkrid, K., & Provost Savard, Y. (2026). Making Sense of Workplace Surveillance: How Workers and Union Representatives Experience, Imagine, and Resist Surveillance Exposure. ''Surveillance & Society'', 24(3), 236–254. [https://ojs.library.queensu.ca/index.php/surveillance-and-society/article/view/19100/13263 https://ojs.library.queensu.ca/index.php/surveillance-and-society/article/view/19100/13263] Winner, L. (1980). Do Artifacts Have Politics? ''Daedalus'', 109(1), 121–136. [https://www.jstor.org/stable/20024652 https://www.jstor.org/stable/20024652]

 
== [[Algorithmic management and workplace surveillance - by Sebastián Muller and Kate Davey''' ''']] ==

=== 1. The current issue. ===
 Algorithmic management and workplace surveillance can be defined as the use of computer-programmed procedures to coordinate labor input in an organization ([https://joint-research-centre.ec.europa.eu/scientific-activities/employment/algorithmic-management-and-digital-monitoring-work_en 1]). Simply put, it is a way to control workers, and workers generally have no say in how the system that tracks them is created. As Winner says "different people are differently situated and possess unequal degrees of power as well as unequal levels of awareness" ([https://www.jstor.org/stable/20024652?seq=7 2]).
The population affected is workers; both “white collar” and “blue collar” workers are impacted. Whether it’s Amazon tracking warehouse workers’ “Time Off Task” with handheld scanners ([https://www.wired.com/story/amazon-worker-tracking-details-revealed 3]) or Meta employees having their every click tracked ([https://www.techpolicy.press/metas-worker-surveillance-tests-eu-rules-on-ai-and-labor/ 4]) to train AI, employers have the power to watch and control their employees in an unprecedented way.
It matters legally because workers have very few rights to fight back against these surveillance programs; even the The Electronic Communications Privacy Act of 1986 allows exceptions to surveillance of employees ([https://bja.ojp.gov/program/it/privacy-civil-liberties/authorities/statutes/1285 5]). 
=== 2. Historical lineage. === 2.1 Panopticon: it is an architectural apparatus developed by Jeremy Bentham in 1791. The design of this building effectively generated the sense of constant monitoring, which in turn motivated the prisoners to behave as if they were being watched at all times. As human labor became commodified, capitalists began implementing measures to ensure the maximization of productivity, including the replication of this design ([https://doi.org/10.14452/MR-070-09-2019-02_2 6]). While unused today, the panopticon is widely considered as both a theoretical and practical inspiration of other forms of surveillance in the workplace.
 
2.2 Scientific management: this theory attempts to analyze and standardize workflows to improve labor productivity. Frederick Winslow Taylor, its main pioneer, used early forms of data analysis during the XX century to decrease walking distances, eliminating unnecessary movements and preventing deliberate slow-walking among workers. ([https://doi.org/10.14452/MR-070-09-2019-02_2 7]). There are many modern forms of data-driven decision-making processes in the workforce. For instance, Amazon warehouses track individual workers' scan rates in real time through handheld scanners, generating data feeds to track and evaluate performance ([https://www.theverge.com/2019/4/25/18516004/amazon-warehouse-fulfillment-centers-productivity-firing-terminations 8]).
 
2.3 PC in the workplace: John Blankenbaker designed the first commercially available personal computer in 1971 ([https://www.computerhistory.org/revolution/personal-computers/17/297 9]), however, it wasn’t until the mid-1980s that the use of personal computers in the workplace became an industry standard. This achievement gave rise to the digital monitoring of work, which in turn allowed employers to monitor employees in real-time, effectively replacing the human eye with a more efficient surveillance mechanism. This represents one of the key aspects of modern-day workplace monitoring. 
 

=== 3. Mechanism of path dependence. ===
''' '''When analyzing the historic lineage of work surveillance, it is possible to point out two main path dependencies mechanisms in place.Firstly, it is clear that many of the technological developments that arose through these events have one major goal, which is to increase wealth within a hierarchical ecosystem. Employers resort to several mechanisms to maximize profits and cut inefficiencies at the expense of workers. 
In addition, standardization can also be singled out as a key mechanism at play. Once one corporation implements any of these developments (such as the digital monitoring of the workplace), the market incentives would force the rest to do the same in order to be profitable and remain competitive.

=== 4. Politics and distribution. ===
 One simple answer to this question would be that capitalists implement and leverage these mechanisms to increase their wealth while the working class remains powerless and unable to withdraw itself from the system in force. However, if we resort to Winner's framework, we see that the technical elements of these technologies are not intrinsically capitalist, since workplace surveillance can also be observed in alternative regimes (including chattel slavery and the Soviet regime). It seems that, rather than being exclusively tied to capitalistic values, the methods in question respond to strong hierarchical structures.
In addition, even under a libertarian framework, structural inequalities are still present, allowing a privileged class to enjoy individual freedoms while segregated groups keep performing low-income jobs ([https://www.metamute.org/editorial/articles/californian-ideology 10]), arguably still subject to surveillance. Thus, not only are these technologies compatible with hierarchical top-down forms of government, but also with structures that allow such hierarchies to form without state intervention.
That said, this form of management seems less compatible with democratic societies, which tend not only to value individual liberties but to turn them into enforceable rights (including workers' privacy, health and integrity), thus establishing boundaries for workplace surveillance. This could explain why studies of American business leaders have found them unhappy with democratic scruples ([https://www.jstor.org/stable/20024652 11]).
 
=== 5. Contingency and alternatives. === 
Bush writes about the memex, a machine that could extend a person’s memory and that the memory would belong to the person who owned it. As Bush expands, “For mature thought there is no mechanical substitute" ([https://dl.acm.org/doi/pdf/10.1145/227181.227186 12]). His idea was that a machine could take over the repetitive nature of someone’s job, while the person could record their own thoughts for their own use. Bush’s memex was never created, but the idea inspired products like Microsoft’s Productivity Score that were made for employers (the buyers) not for the person whose memory it used (the worker).
This immediately brings to mind Meta's Model Capability Initiative, which was done to train AI agents the company hopes to sell to other employers as well as watch their employees ([https://www.reuters.com/sustainability/boards-policy-regulation/meta-start-capturing-employee-mouse-movements-keystrokes-ai-training-data-2026-04-21/ 13]). If this was Bush’s memex, it would belong to the employee using it, not Meta. 

=== 6. Governance implication. ===
Democratic oversight of the use of algorithms as a form of surveillance is essential to ensure that workers’ rights are not overstepped in the name of efficiency. As such we propose: all workers should have a legal right to be consulted and approve a surveillance program before a company installs and uses it. A tracking system should be a group decision, not an inevitable, unilateral one. While the United States has very little protections for workers in terms of surveillance, the EU leads the way with the GDPR, specifically Article 22 ([http://data.europa.eu/eli/reg/2016/679/oj 14]), and the AI Act Article 5 ([http://data.europa.eu/eli/reg/2024/1689/oj 15]). It’s also worth noting that in December 2026, the Platform Work Directive ([https://eur-lex.europa.eu/eli/dir/2024/2831/oj/eng 16]) will come into effect which will ban automated termination.

= [[Platform Dependence in Public Communication (News) - Ceren Isiklar & Rodrigue Kayembe|Platform Dependence in Public Communication (News)]] Ceren Isiklar & Rodrigue Kayembe =

= Centralization of AI Compute and Data =

== Centralization of AI Compute and Data - by Sisi Hammoudi & Juliette Chamot ==
=== Governance issue ===
Today, several path-dependence phenomena are emerging across advanced computing technologies. This assignment addresses one of them: the centralisation of AI compute and data. Training and running frontier AI systems now depends on a handful of hyperscale providers, namely Amazon, Microsoft, Google, and Nvidia, which dominates the GPU market. In other words, near monopolies exist throughout the supply chain. This concentration creates a governance problem affecting public and private actors alike, both legally and institutionally. On the institutional side, it prevents public administrations, and states other than the US or China, from reaching sovereign AI capacity. It also raises competition-law concerns, since dominant market positions can prevent the market from delivering the best products at the best prices to consumers. Finally, AI developers, like states, are often forced into contracts with hyperscale providers, creating a structural dependence on them.


=== Historical lineage ===
==== 1945 ====
Wartime scientific instruments should have been redirected to extend human command over knowledge, helping establish the postwar model of sustained federal funding for computing research (Vannevar Bush's, "As We May Think », 1945).


==== 1990s ====
The privatisation of the NSFNET established the idea that critical digital infrastructure, even though publicly subsidised, could legitimately be privately owned. The EU Commission followed the US's lead, recommending privatisation and deregulation of telecommunications and media as the best way to build the "infobahn" though EU members could instead have chosen nationalisation of the infrastructure (Barbrook & Cameron, "The Californian Ideology," 1996).


==== 2006 ====
<div class="s13">Hyperscale cloud computing turned massive fixed capital investment in data centresinto a rentable service, favouring firms able to invest at scale.</div>
<div class="s13">Same period: Nvidia's proprietary CUDA software, released alongside its GPUs, became the de facto standard for machine learning, locking in a decade of tooling and expertise around one firm's hardware.</div>
<div class="s13">
==== From 2022 ====
Semiconductor export controls and the CHIPS Act turned compute into an explicit instrument of geopolitical strategy.</div>



=== Mechanisms of path dependance ===
Several mechanisms drive this centralisation, but path-dependence ones dominate. Because AI compute is a capital-intensive industry, new entrants struggle to compete with hyperscalers that have already sunk huge sums into infrastructure, a barrier that produces monopoly-like situations. This dynamic, known as increasing returns, generates system-scale economies that harden into a near-irreversibility of investment (David, "Clio and the Economics of QWERTY," 1985). It overlaps with a second mechanism, standardisation, itself a form of technical interrelatedness: researchers' investment in Nvidia's CUDA-based tooling locks the field into that ecosystem regardless of whether better alternatives exist. A further mechanism is legal and contractual entrenchment. For example, Microsoft's stake in OpenAI binds developers to a specific provider by agreement rather than technical necessity.=== Politics and distribution ===Technology does not have a history of distributing power equally. Rather, its development has strengthened the interests and authority of powerful groups that already possessed economic and political resources, and it is yet to be determined whether AI compute and data will follow suit. 
Author Langdon Winner illustrates the historical constraints of technological development with the case of New York from the 1920s to the 1970s. He argued that the city’s roads, bridges, and parks’ physical arrangement contained “explicit and implicit political purpose.” These infrastructures privileged wealthy and white communities while restricting poor and Black populations’ choice of mobility. Similarly, the development of industrial machinery increased productivity, simultaneously weakening union and non-union workers’ bargaining power.
In the twentieth century, following the Industrial Revolution, a Control Revolution emerged from the increasing need to manage expanding production, transportation, and information flows requiring strengthened organizational bureaucratic control. This was heavily shaped by the political context of post-World War II warfare.
Therefore, the bearer of the initial cost of technology is the person putting in the capital to erect it, but ultimately, the person bearing the long-term cost of technology, whose choices diminish, is not the rich and privileged, but the everyday person subject to technologies. Forced to adapt, to reshape their professional skills to adapt to an environment trying to make them obsolete.
AI compute and data are at risk of creating the same effect. In their critique of dotcom neoliberalism, authors Barbrook and Cameron exposed a similar socio-economic tension. The Californian Ideology claims digital technology empowers individuals; however, perpetual unequal access to it creates an apartheid between “information-rich” and “information-poor” populations.
Finally, the development of AI compute and data depends on expensive and scarce computing infrastructure and a highly concentrated semiconductor supply chain. Today, this is one of the most concentrated industries, with single companies having near-total monopolies over the production of chips necessary for the functioning of AI. This monopoly model threatens to reinforce the existing economic hierarchy.

=== Contingency and alternatives ===If a different technological trajectory was plausible, the most apt time for it to have occurred would be during the transition from the Industrial Revolution to the “Control Revolution”. When industrial production accelerated, our society experienced a “crisis of control”, creating opportunities to organize information and communication differently.
Instead of allowing control technologies to be monopolized within businesses and states, our societies could have pursued a more democratic and decentralized system. Barbrook and Cameron provide such an alternative path through their idea of the “electronic agora,” where digital networks could encourage community participation and direct democracy, where government could actively engage with the people to shape digital infrastructure and expand access. These alternatives did not actually prevail because corporate and state market participants possessed greater resources and authority to shape technological development at the end of the twentieth century. 
Inherently, technological development is not predetermined; it can support either authoritarian or democratic arrangements. Today, monopolies control chip production needed for AI compute and data, but other technologies like blockchain permit democratic participation.

=== Governance implication ===History always threatens to repeat itself. As Barbrook discussed, digital societies are shaped by the political context in which they develop. It is a valid concern, then, that the development of information technology, through AI computing, would follow the same trajectory- that is, be constrained to benefit the wealthy who funded its development, and ultimately harm the middle and lower classes enabling its functioning. 
However, AI’s place in our society has not yet crystallized. Technology governance does not have to adhere to this system. Such a seemingly sustainable approach to AI compute and data governance has already been broached. In 2022, the Notre Dame Journal on Emerging Technologies published an article in its issue on Artificial Intelligence where the authors proposed an AI compute system based on a “dual investment strategy”. Firstly, the model would be launched by “subsidizing and negotiating” cloud computing for human AI researchers through “existing” cloud vendors, which would expand on existing AI compute initiatives. Secondly, the compute system would be specifically invested in as a publicly owned infrastructure that would be built and operate similarly to national libraries which own “academic supercomputing facilities”. 
The pilot program would have a “base level of compute access” which would keep costs of processing down and enable it to meet the compute needs of a majority of academic researchers. Absent any strengthened legislation on the use of AI compute and data, the hardware would be accompanied by an AI ethics framework to enable sustainable technological governance. In addition to an ethical staff, the human resources needed to keep an AI compute model running would include data officers, scientists, data center and software engineers, research developers, and system administrators. These positions require specialized training, allowing new types of professionals, from any background, to have access to professional prosperity in the field of AI. 
By implementing these frameworks, never neglecting the human element, and the economic and intellectual impact AI has on our society, we move from a historical determinist perspective to one acknowledging historical constraints and learning from them.


=== Bibliography ===
<div><div>[1] Langdon Winner, Do Artifacts Have Politics, Daedalus, Winter, 1980, Vol. 109, No. 1, Modern Technology: Problem or Opportunity? pp. 121-136, 124, (1980)</div>
<div>[2] Langdon Winner, Do Artifacts Have Politics, Daedalus, Winter, 1980, Vol. 109, No. 1, Modern Technology: Problem or Opportunity? pp. 121-136, 124, (1980)</div>
<div>[3] James R. Beniger, The Control Revolution: Technological and Economic Origins of the Information Society, Harvard University Press, p. 443, 450 (1986)</div>
<div>[4] Richard Barbrook, Andy Cameron, The Californian Ideology, ''Science as Culture'', ''6''(1), 44–72, 61 (1996)</div>
<div>[5] Anton Korinek, Jai Vipra, Concentrating Intelligence: Scaling and Market Structure in Artificial Intelligence, National Bureau of Economic Research, 10.3386/w33139, p. 11, 13-15 (2024)</div></div>
<div><div>[6] James R. Beniger, The Control Revolution: Technological and Economic Origins of the Information Society, Harvard University Press, p. 443, 442 (1986)</div>
<div>[7] Richard Barbrook, Andy Cameron, The Californian Ideology, ''Science as Culture'', ''6''(1), 44–72 (1996)</div>
<div>[8] Langdon Winner, Do Artifacts Have Politics, Daedalus, Winter, 1980, Vol. 109, No. 1, Modern Technology: Problem or Opportunity? pp. 121-136, 130 (1980)</div>
<div>[9] Daniel E. Ho, Jennifer King, Russell C. Wald & Christopher Wan, The Centrality of Data and Compute for AI Innovation: A Blueprint for the National Research Cloud, 3 Notre Dame J. on Emerging Tech. 71, 81(2022)</div>
<div>[10] Daniel E. Ho, Jennifer King, Russell C. Wald & Christopher Wan, The Centrality of Data and Compute for AI Innovation: A Blueprint for the National Research Cloud, 3 Notre Dame J. on Emerging Tech. 71, 82 (2022)</div></div>

== Centralization of AI compute and Data - by Atmika Dechamma & Pablo Latorre ==
=== Governance Issue ===
AI computing capacity and data presents the governance problem of an extreme concentration of technological infrastructure and information in a restricted number of companies that privately and globally dominate the market. This affects public and private actors - such as small and medium enterprises, potential new competitors, and Global South countries - that increasingly depend on AI systems outside their control and/or are deterred from competing with established companies because of impenetrable entry barriers - most notably initial investments and operational costs (1). The institutional and legal relevance of this is that it consolidates, reproduces and amplifies asymmetries of power in favor of the “Big Tech”, reverting the technological trend from digital decentralization to physical infrastructure dependency and curbing innovation and market efficiency.

=== Historical Lineage ===
==== Limitation of Machine Learning ====The AI revolution was built on the limitation of human capabilities to process large information and derive efficient results - where before there existed only facilitative tools to allow for human-enabled processes, AI now offered derivations and cognitive reasoning in moments and at large scales.

 ==== Revert to Physical Infrastructure Dependency in the rush to be the best ====The present AI model heavily relies on hardware in order to perform complex data processing for deep learning. The data centre models followed require larger model sizes (which require higher resource, information and capital investment) to be able to yield better results. Due to the limitation of resources and a rush to establish the best working model of an AI agent, data centres are built on a rudimentary model heavily dependent on massive amounts of electricity and fresh water (2).

 ==== Regulation and Decentralization of Information ====Regulation and influence of sovereigns and institutions has resulted in data sharing and information exchange structurally through open source and cloud computing. But, while AI models claims their selling point to be accessibility for all, their information and derivations remain in a closed server, stored for the purpose of training their AI agents to be the best (3). Big AI contenders attract higher consumer and information flows overall resulting in a better developed AI model, proving disadvantageous in any market where early contenders have an advantage and accumulation of power.
=== Mechanisms of Path Dependence ===Early on in the industry there is an expectation of human and capital investment from all stakeholders involved, this high cost for entry into the market and expected capital to sustain and develop a model by both consumer and enterprise makes for a high switching cost. For the AI model to run and scale there is a presumption that the rate of growth is dependent on the size of data fed, resulting in a better model and hence, more consumer usage. A closed centralised data loop is created but the requirement for storage and processing far surpasses current infrastructures capabilities - heavily relying on early market entry advantage and sustained capital investments taking away from smaller aspirants and hoarding of power within closed circles. Despite its physical dependency, AI data processing has unique requirements beyond legacy infrastructures (4) - larger demand and production of GPUs and designed to handle immense thermal heat generated while consuming electricity beyond current capabilities. 

 The AI space largely remains unregulated due to its undefined form and fluid nature, however, this has given way for unethical and morality issues regarding data privacy, exploitation of labour and environmental harms.

=== Politics and Distribution ===
Concentration strengthens the values, interests, capacities, and authority of the “Big Tech” over other actors of public and private nature. In this context, Barbrook and Cameron (5) illuminate the ethos behind AI technology development while Winner (6) makes observable the power relations surrounding it: concentration is the consequence of an AI development anchored in the belief that it must be free from all institutional and normative restriction since it is inherently liberating and non-political which, in turn, provides a political/moral justification for the concentration (if technology is inherently liberating and non-political then there is no power relations to observe or question) that translates itself into power dynamics where the AI developer not only owns the AI technology but also decides how it is trained and deployed and which models the user can or cannot access.

=== Contingency and Alternatives ===
Concentration is due, first and foremost, to the fact that AI computing capacity requires massive and cutting-edge physical infrastructure - such as data centers and microprocessors - which, in turn, requires upfront investments that are unaffordable to most companies. In David’s (7) language, companies that could finance the infrastructure and position themselves first obtained self-reinforcing advantages and a “lock-in” effect in the market. In other words, concentration is the consequence of the development of AI physical infrastructure privately with no mixed or public alternatives as a public service. Therefore, there was a space for an alternative future at the end of the 1980s or the beginning of the 1990s when the privatization model of the technological infrastructure was consolidated in line with the “Californian Ideology”. In that moment governments had the chance to treat technological infrastructure as a mixed or public good rendering it universally accessible as it was the case of the Internet; this would have permitted companies of all sizes to profit from economies of scale, reducing entry barriers and fostering competition. However, this alternative path did not prevail since the Californian Ideology provided the moral/political justification to exclude governments from the development of technological infrastructure in the exact moment in which such a decision would produce structural consequences. If that path had been followed it could have contributed to collective access to economies of scale, lower entry barriers, and a more competitive AI market.

=== Governance Implication ===
Considering the above, public or mixed investment on AI physical infrastructure is a necessary governance measure for market decentralization and collective access to AI computing capacity and data. From an historical perspective, the proposed measure is not merely a corrective action for a market failure but for a contingent and ideological decision made between the late 1980s and early 1990s that produced a structural “path dependence” that now is necessary to change.
==== References ====
(1) The Quiet Re-Centralization of Tech: Why AI Is Reversing 20 Years of Platform Decentralization. (2026, January 10). Retrieved on 8 September 2026, from Medium website : [https://medium.com/@sumitbhardwaj1357/the-quiet-re-centralization-of-tech-why-ai-is-reversing-20-years-of-platform-decentralization-ce15962b6a13 https://medium.com/@sumitbhardwaj1357/the-quiet-re-centralization-of-tech-why-ai-is-reversing-20-years-of-platform-decentralization-ce15962b6a13]

(2) [https://www.mdpi.com/1996-1073/19/1/137 Ginzburg-Ganz, E., Lifshits, P., Machlev, R., Belikov, J., Krieger, Z., & Levron, Y. (2025). Technical challenges of AI data center integration into power grids—A survey. ''Energies'', ''19''(1), 137.]

(3) [https://decai-resources.s3.us-east-1.amazonaws.com/decentralized_ML_perspective-16.pdf Singh, A., Lu, C., Gupta, G., Chopra, A., Blanc, J., Klinghoffer, T., ... & Raskar, R. (2024). A perspective on decentralizing AI.]

(4) The Rise of AI Data Center Models and the Decline of the General-Purpose Data Center. (Circa 2026). Retrieved on 8 September 2026, from SubZero Engineering website : [https://www.subzeroeng.com/the-decline-of-the-general-purpose-data-center/ https://www.subzeroeng.com/the-decline-of-the-general-purpose-data-center/]

(5) [https://git.xpub.nl/onebigear/SI_09_reference_library/raw/commit/22bf61e1dcfed35d58fba23808212bd00584be47/Richard_Barbrook_Andy_Cameron_Californian_ideology.pdf Barbrook, R., & Cameron, A. (1996). The Californian Ideology. ''Science as Culture'', ''6''(1), 44-72].

(6) [https://d1wqtxts1xzle7.cloudfront.net/87708502/langdon_winner_do_artifacts_have_politics-libre.pdf?1655588533=&response-content-disposition=inline%3B+filename%3DDo_Artifacts_Have_Politics.pdf&Expires=1788890463&Signature=UBBiMqkip5MfYZbkbLNAN6aE2hyZdKkER-HFcmyxXkwIZuTnMKdOGoao8ddFr-rQ5T55A8NWUOlnZkGYmi66e7XR9h99DgTELMCrh48sXjOb8KYgA48r~SUUymt1IW~R~aOKJzitMIUHcrc6yEGKK1OuaA~mrEIVOXpUjQMYjqOd26Xmk9YJEglTkqaxeR47VCCth5FvozNQIowdfjzeu8aNHnI8V7Yur1IiXY~PGvYRcN6wBtcgpVMHZUtNaI8NjmYbp2p5TBYrPMF9MCLtvpGhPlsCfsvA8Bdm6XulGwRlksZBKWvVUQ8D~M8oLu12V39YaiETIObyJqbE7ctIcQ__&Key-Pair-Id=APKAJLOHF5GGSLRBV4ZA Winner, L. (2017). Do artifacts have politics?. In ''Computer ethics''(pp. 177-192). Routledge].

(7) [https://sjbae.pbworks.com/f/David+1985.pdf David, P. A. (1985). Clio and the Economics of QWERTY. ''The American economic review'', ''75''(2), 332-337].

== Centralization of AI compute and Data - by Sergey Abramov and Paul Stein ==<div>

=== The current governance issue ===
Advanced AI currently depends on a narrow group of tech giants1 and semiconductor oligopolies controlling microchips and software, hyperscale cloud infrastructure, large datasets, frontier models, and distribution platforms. This presents a governance problem where this narrow group exercises control over foundational digital infrastructure. This directly affects consumers, businesses, and academia out of AI. Legally and institutionally, this matters because it creates a profound democratic deficit: this neutralizes antitrust frameworks designed for traditional markets, often bypassing conventional copyright and data privacy protections through unchecked data harvesting thereby complicating enforcement.This ultimately might affect national sovereignty as the countries that host the buildout will shape the future of AI: who controls it, what values it contains, how it is used2.
 
=== Historical lineage ===
In 1987, Taiwan Semiconductor Manufacturing Company (TSMC) was founded, establishing the world’s first semiconductor fab. Instead of designing and making its own chips, TSMC only manufactured chips designed by other companies. Today, nearly 100% of the world's most advanced AI processors are manufactured by TSMC, leaving the entire global AI compute supply chain centralized under one company.
 In 2006, Nvidia launched CUDA (Compute Unified Device Architecture), a free software layer that allowed developers to use Nvidia's graphics chips (GPUs) which later became perfect for training deep neural networks. Today, this creates a near-monopoly on AI compute hardware.
 Soon after Amazon launched AWS, Microsoft and Google launched their own cloud platforms marking the structural pivot to the cloud service provider model, shifting computing from a distributed architecture (where companies bought and managed their own physical servers) to a centralized infrastructure. As training modern AI models requires tens of thousands of GPUs, today’s frontier AI companies are entirely dependent on the extremely expensive physical infrastructure of a handful of cloud hyper-scalers (Microsoft, Google, AWS, and Meta).
 In 2009, Fei-Fei Li launched ImageNet, a massive database of over 14 million images, which proved that AI model performance scales directly with data volume. This created an immediate institutional advantage for consumer internet platforms (like Google and Meta), because they were already architected to collect, centralize, and store masses of user data through search engines and social networks, they possessed the unique data stack required to train high-performing AI.
 
=== Mechanism of path dependence ===
The centralization of AI compute and data is a classic case of technological lock-in. When structural choices were made decades ago, they triggered powerful, self-reinforcing loops. One such loop is the switching costs and software standardisation. As developers have spent more than a decade working with CUDA, moving to a rival hardware provider (or a decentralized compute network) requires an enterprise to completely rewrite its foundational software stack. The cost of code migration, developer retraining, and potential hardware downtime force companies to stay tied to dominant hardware monopolies such as Nvidia. Commitments, licensing conditions, and exclusivity provisions add contractual costs. Fewer than 1% of cloud customers switch provider annually, identifying technical differentiation and data-transfer charges as important barriers4.
 Network effects also matter for computing power. More customers generate more revenue, usage information, and opportunities to improve infrastructure and models. Better services attract additional users and developers, whose applications make the platform more valuable. In cloud–AI partnerships, providers may additionally obtain information about model performance and future infrastructure demand. These feedback loops strengthen firms already operating across cloud, models, software, and distribution.
 
=== Politics and distribution ===
According to Langdon Winner's definition, modern AI computing infrastructure may qualify as an "inherently political technology" because its large scale and significant financial requirements not only lend themselves to, but necessitate,5 centralized, hierarchical control. The physical siting of this infrastructure shows evidence of Winner’s second category of politicization of technologies, the ‘political arrangement’:6 Of the five largest U.S. AI firms, four are headquartered in California,7 while just one is (partially) domiciled in Texas. Texas, however, hosts twice the GPU clusters.8 This placement geographically insulates beneficiaries of the data infrastructure boom from its negative externalities. The specific choice of location may itself be a deliberate preference on account of Texas’ unique approach to water usage rights.9 Juxtaposition of this unequal and even exploitative balance of interests with the pontificating rhetoric of the urban, information rich, beneficiaries is evocative of the internal contradictions of the Californian Ideology - the virtual artisans existence and privileged position is still based on the exploitation of a vulnerable out-group, the "information-rich" consuming the output, the "information-poor" hosting the extraction.10
 
=== Contingency and Alternatives ===
Profit motive and economies of scale now concentrate investment among a few hyperscalers,11 but this reflects policy choice, not government incapacity: Compute infrastructure may be inherently political, but the technology is agnostic as to where the organizing authority originates, and states have historically funded infrastructure of comparable or greater scale, while facing less pressure to show short-term returns. The clearest turning point was 1995, when the National Science Foundation retired NSFNET, the publicly funded backbone that had connected U.S. research and supercomputing centers since the mid-1980s,12 handing internet infrastructure to commercial providers with no conditions attached. This was a deliberate policy decision made amid the deregulatory push of the Clinton era. The privatization that followed ushered in the private industrial dominance that persists today in the form of hyperscalers. Had a publicly governed backbone instead been scaled up, today's compute infrastructure might rest on a different foundation entirely.
 
=== Governance implication ===
Despite private hyperscaler dominance, the U.S. does have a nominal public alternative: the National AI Research Resource (NAIRR), established following the National AI Initiative Act of 2020 and piloted from January 2024, which explicitly aims to democratize access to the computational resources essential to AI research.13In Europe, a comparable program is the EuroHPCJU,14 an initiative intended to pool resources to build supercomputing capacity. The undertaking’s legislative basis has since been extended to include networked ‘AI factories', as well as the so-called “Gigafactories”.15 The framing in language of sovereignty indicates intent to at least compete with private infrastructure initiatives. Though government-led measures cannot reverse the centralization of compute, they can democratize access - provided they are intensified. The European Commission has “set a goal of investing more than €1 billion per year in AI research and innovation”.16 While laudable in its own right, it pales in comparison to the investments of industry leaders, which are approaching, if not exceeding, trillions of Dollars.

=== References & Notes ===
1.Amazon, Google, Meta, Microsoft, and Oracle collectively hold an estimated 71% of the world’s cumulative AI compute as of Q4 2025, measured in H100-equivalents of computing power. Luke Emberson, Josh You and Venkat Somala (2026), "Five hyperscalers now own over two-thirds of global AI compute". Published online at epoch.ai. Retrieved from 'https://epoch.ai/data-insights/hyperscalers-control-most-compute' [online resource]. Accessed 7 Sept 2026.2. Alasdair Phillips-Robins, Teddy Tawil and Sam Winter-Levy, ‘The Compute Coalition: How to Build the Future of AI in the Free World’ (Carnegie Endowment for International Peace, 8 June 2026) https://carnegieendowment.org/research/2026/06/the-compute-coalition-how-to-build-the-future-of-ai-in-the-free-world accessed 7 September 2026.

3. Amazon Web Services (AWS), a cloud computing platform

4. Competition and Markets Authority, Cloud Infrastructure Services: Final Decision Report (31 July 2025) https://assets.publishing.service.gov.uk/media/688b8891fdde2b8f73469544/final_decision_report.pdf accessed 7 September 2026.

5. Langdon Winner, 'Do Artifacts Have Politics?' (1980) 109(1) ''Daedalus'' 121, 128.

6. Ibid. 123.

7. OpenAI, Anthropic, Google DeepMind, Meta AI. xAI is an exception to the rule, domiciled between Palo Alto, CA and San Antonio, TX.

8. Texas’ 17 to California’s 8, cf. Texas Royalty Brokers, 'AI Clusters by State' (''Texas Royalty Brokers'', 2025) https://texasroyaltybrokers.com/ai-clusters-by-state/ accessed 7 September 2026.

9. The ‘rule of capture’, established in Houston & TC Railway Co v East (Tex 1904), permits landowners to extract groundwater beneath their land without liability for depleting adjacent wells or aquifers, a doctrine now exploited by hyperscale AI data centres to draw large cooling-water volumes with minimal regulatory constraint.

10. Richard Barbrook and Andy Cameron, 'The Californian Ideology' (1996) 6(1) ''Science as Culture'' 44.

11. Goldman Sachs, 'Global AI Investment Is Forecast to Exceed $1 Trillion in 2026' (''Goldman Sachs Insights'', 7 August 2026) https://www.goldmansachs.com/insights/articles/global-investment-is-forecast-to-exceed-1-trillion-in-2026 accessed 7 September 2026.

12.  National Science Foundation, 'Birth of the Commercial Internet' (''NSF Impacts'') https://www.nsf.gov/impacts/internet accessed 7 September 2026.

13. CASRAI, 'NAIRR: National AI Research Resource, Explained' (''CASRAI'') https://casrai.org/dictionary/term/national-ai-research-resource-nairr accessed 7 September 2026.

14. Council Regulation (EU) 2021/1173 of 13 July 2021 on establishing the European High Performance Computing Joint Undertaking and repealing Regulation (EU) 2018/1488 [2021] OJ L256/1.

15. Council Regulation (EU) 2026/150 of 16 January 2026 amending Regulation (EU) 2021/1173 on establishing the European High Performance Computing Joint Undertaking [2026] OJ L 2026/150.

16. European Commission, 'Realising Europe's Digital Decade' (''European Commission'', 11 October 2024) https://commission.europa.eu/strategy-and-policy/priorities-2019-2024/story-von-der-leyen-commission/realising-europes-digital-decade_en accessed 7 September 2026.</div>

File:Civilian AI Front Final.PDF.pdf

2026-09-24T06:32:49Z

Robert Florijn: Course contribution Robert Florijn for Ass #1 Governance by history

== Summary ==
Course contribution Robert Florijn for Ass #1 Governance by history

Governance by history

2026-09-23T20:51:30Z

Robert Florijn:

[[File:Old internet.jpg|300px|thumb]]

= contributions =
*can
*you 
*please
*create
*individual pages
*from the contributions below
*and link them back here? 
*[[Information Control and Workplace Surveillance - Jihae Park and Mariona Fortuny]]
*[https://wiki.ivir.nl/index.php?title=Centralization_of_Compute_and_Data Centralization of AI Compute and Data - by Sergey Abramov and Paul Stein]
*[[:file:Civilian AI at the Front Final]]

= links things that popped up during class =
*[https://craphound.com/ Cory Doctorow]
*[https://dougengelbart.org/content/view/209/ The mother of All demos] (first mouse, graphical interface, word processor, etc)
*Can a simple sensor be political? [https://gizmodo.com/why-cant-this-soap-dispenser-identify-dark-skin-1797931773 well, this "racist" soap dispenser made headlines a couple of years ago.]
*Podcast series: [https://www.the-santiago-boys.com/ Morozov: the Santiago Boys.]
*Anatomy on AN AI. [[File:Ai-anatomy-map.pdf|check out the pdf here]]

= Workplace surveillance =
== Information Control and Workplace Surveillance - by Jihae Park and Mariona Fortuny Jané ==

==== The Current Governance Issue ====
There are numerous current governance problems arising from technological innovations, yet here we will specifically focus on the information control and workplace surveillance. Whether intended or unintended, there were certain historical incidents which resulted in social and political consequences. As noticed from the Control Revolution, the dramatic evolution in mass media and telecommunications technologies has led to the reinforcement of the economic and political control during the Industrial Revolution. ([https://archive.org/details/beniger-james-the-control-revolution Beniger, 1986]) By applying this logic to a modern example, this is clearly how many corporations are getting in control of their employees through surveillance technologies. Along with the emergence of remote and hybrid work after Covid-19, employers are now relying on “bossware” to analyze the emails, chats, and messages of their employees on platforms such as Teams or Slack. ([https://ojs.library.queensu.ca/index.php/surveillance-and-society/article/view/19100/13263 Ollier-Malaterre et al., 2026, p.237]) Some companies have strict restrictions on transferring data both internally and externally in protection of confidential information and consistently monitor their workers to the certain extent which could slow down their efficiency. This is problematic as it’s not only violating the privacy of their employees but also depriving employees of rights they should be naturally be able to exercise.



==== Historical Lineage ====
In relation to the current issue of the information control and workplace surveillance, Max Weber’s argument of bureaucracy and rationalization in the middle and late nineteenth century are good historical examples which are still frequently visible nowadays. Weber stated how the range of control could extend not only by expanding the capability to process information but also by limiting the amount of information to be processed. Here, the former is directly related to the control implemented by bureaucratization (in the mid 19th century) and computerization (in the late 20th century), whereas the latter addresses rationalization. An example from bureaucracy is the expansion of paperwork in response to the growth in information. We can still note these traces of bureaucracy from numerous procedures that we have to go through for a decision-making to take place in corporations. He brings up the example of standardized time zones for rationalization which is simply an example for an omission or a breakdown of information to facilitate its processing. ([https://archive.org/details/beniger-james-the-control-revolution Beniger, 1986, p.15-16]) This is also commonly seen from nowadays as some companies may use internal terminologies for the sake of both confidentiality and efficiency.

==== Mechanism of Path Dependence ====Following along the line with the historical lineage, there are certain mechanisms at work of path dependence for information control and workplace surveillance. Bureaucracy is closely related with organizational routines. Many companies have already settled down with a fixed workflow and decision-making process, which makes it more costly for one to change from its conventional way of working. This means switching costs from an old system to a new one is clearly high considering its time and capital, not necessarily will guarantee increasing returns. Hence, many firms end up maintaining its old strategy, reluctant to make any modifications or changes. 

==== Politics and Distribution ====

As businesses grew too large for managers to oversee everything directly, companies replaced human judgment with formal rule-systems, as we see explained by Beniger's 'intellectual technology' ([https://archive.org/details/beniger-james-the-control-revolution/page/65/mode/2up Beniger, 1986, pp. 17, 21]). Algorithmic management continues this: a call once made by a supervisor's judgment is now a rule built into software that runs without one, extending a logic Beniger traces as far back as to Taylorism and the so called 'assembly line'. Winner ([https://www.jstor.org/stable/20024652 1980]) asks whether such systems must concentrate power this way or were driven there, when other designs were equally possible; nothing required shutting workers out of the decision, so leaving them out was a political choice, and not necessarily a technical necessity. Barbrook and Cameron ([https://doi.org/10.1080/09505439609526455 1996]) explain the reasoning why some can defend this 'domination': they show how 1990s tech culture sold new technologies as freedom even as it tightened control, so the rhetoric could become one of closer oversight as flexibility and autonomy. Together, authority accrues to employers, who in some systems can set and revise these oversight rules unilaterally and face less resistance because the arrangement is framed as freedom; workers bear the cost, losing the informal give-and-take they had with a human supervisor for oversight that is harder to see, question, or contest, while being told it grants them more freedom, not less.

==== Contingency and Alternatives ====

A plausible alternative happened on U.S. post WWII: in some workplaces, labor and its pacing was demanded to be set through collective bargaining and enforced jointly, by both management and union stewards monitoring compliance on the line ([https://doi.org/10.1093/icc/dtaf038 Breznitz et al., 2025, p. 211]). So, in this context, workplace surveillance could have developed as a bargained, dual-administered practice rather than a management monopoly. The 1950 [https://en.wikipedia.org/wiki/Reuther%27s_Treaty_of_Detroit Reuther's Treaty of Detroit ] cut that possibility off: unions won guaranteed wages and benefits in exchange for surrendering any general claim over how work was supervised ([https://doi.org/10.1093/icc/dtaf038 Breznitz et al., 2025, pp. 210–211]). Workers depended on wages to survive, and employers controlled the wage relationship. That asymmetry meant employers, given their position of power, were able to arrange things so that, at the very least, workers would have far less capacity to sustain a fight over control itself. Thus, the "trade" wasn't a fair exchange between equals. This reflects which side could afford to hold out on which issue. Hypothetically speaking, had that joint administration model prevailed instead, this could have greatly impacted norms internationally and good practices with worker-union agreed surveillance would have become a widely accepted standard practice.

==== Governance Implication ====

An interesting proposition for present-day governance that wants to fill a gap in workplace surveillance is what [https://doi.org/10.1016/j.digbus.2025.100120 Kayas, Ong and Belal (2025)] are proposing: They argue that EU law, through the [https://eur-lex.europa.eu/eli/reg/2016/679/oj/eng GDPR] and its required [https://en.wikipedia.org/wiki/Data_protection_impact_assessment data protection impact assessments (DPIAs)], already offers meaningful protection for workers in terms of lawful high-risk data collection and data processing. Nevertheless, these DPIAs are built around data and privacy, so they leave out several dimensions of monitoring itself: its purpose, how intrusive it is, how visible it is to the people being watched, and its effects on workers' autonomy, trust, and standing relative to management. These authors propose a "surveillance impact assessment" scoring a system against these characteristics directly, before any surveillance system is put in place. Although this mechanism "is constrained by its reliance on existing literature and the lack of empirical validation," it feels like a genuine alternative for two reasons: 1. it targets the actual mechanism at issue, and 2. at the same time it builds upon a mechanism of active participatory approach where employees have a direct say in how surveillance is designed ([https://doi.org/10.1016/j.digbus.2025.100120 Kayas et al., 2025, p. 11]), which, combined with the actual assessment, would put workers' own judgment at the centre of the one decision that affects them. It's important to note that this alternative is proposed as worth piloting, not as a proven fix, and it is fair to ask whether another mandatory assessment stacked on top of DPIAs would function as intended or simply read to workers as more compliance paperwork.

==== References ====
 Barbrook, R., & Cameron, A. (1996). The Californian Ideology. ''Science as Culture'', 6(1), 44–72. [https://doi.org/10.1080/09505439609526455 https://doi.org/10.1080/09505439609526455] Beniger, J. R. (1986). ''The Control Revolution: Technological and Economic Origins of the Information Society''. Harvard University Press. [https://archive.org/details/beniger-james-the-control-revolution https://archive.org/details/beniger-james-the-control-revolution]  Breznitz, D., Levy, K., Lipartito, K., & Zehavi, A. (2025). An equity-focused research agenda for workplace surveillance. ''Industrial and Corporate Change'', 35(1), 206–221. [https://doi.org/10.1093/icc/dtaf038 https://doi.org/10.1093/icc/dtaf038] Kayas, O. G., Ong, C. E., & Belal, H. M. (2025). From humans to algorithms: A sociotechnical framework of workplace surveillance. ''Digital Business'', 5, 100120. [https://doi.org/10.1016/j.digbus.2025.100120 https://doi.org/10.1016/j.digbus.2025.100120] Ollier-Malaterre, A., Parent-Rocheleau, X., Benkrid, K., & Provost Savard, Y. (2026). Making Sense of Workplace Surveillance: How Workers and Union Representatives Experience, Imagine, and Resist Surveillance Exposure. ''Surveillance & Society'', 24(3), 236–254. [https://ojs.library.queensu.ca/index.php/surveillance-and-society/article/view/19100/13263 https://ojs.library.queensu.ca/index.php/surveillance-and-society/article/view/19100/13263] Winner, L. (1980). Do Artifacts Have Politics? ''Daedalus'', 109(1), 121–136. [https://www.jstor.org/stable/20024652 https://www.jstor.org/stable/20024652]

 
== [[Algorithmic management and workplace surveillance - by Sebastián Muller and Kate Davey''' ''']] ==

=== 1. The current issue. ===
 Algorithmic management and workplace surveillance can be defined as the use of computer-programmed procedures to coordinate labor input in an organization ([https://joint-research-centre.ec.europa.eu/scientific-activities/employment/algorithmic-management-and-digital-monitoring-work_en 1]). Simply put, it is a way to control workers, and workers generally have no say in how the system that tracks them is created. As Winner says "different people are differently situated and possess unequal degrees of power as well as unequal levels of awareness" ([https://www.jstor.org/stable/20024652?seq=7 2]).
The population affected is workers; both “white collar” and “blue collar” workers are impacted. Whether it’s Amazon tracking warehouse workers’ “Time Off Task” with handheld scanners ([https://www.wired.com/story/amazon-worker-tracking-details-revealed 3]) or Meta employees having their every click tracked ([https://www.techpolicy.press/metas-worker-surveillance-tests-eu-rules-on-ai-and-labor/ 4]) to train AI, employers have the power to watch and control their employees in an unprecedented way.
It matters legally because workers have very few rights to fight back against these surveillance programs; even the The Electronic Communications Privacy Act of 1986 allows exceptions to surveillance of employees ([https://bja.ojp.gov/program/it/privacy-civil-liberties/authorities/statutes/1285 5]). 
=== 2. Historical lineage. === 2.1 Panopticon: it is an architectural apparatus developed by Jeremy Bentham in 1791. The design of this building effectively generated the sense of constant monitoring, which in turn motivated the prisoners to behave as if they were being watched at all times. As human labor became commodified, capitalists began implementing measures to ensure the maximization of productivity, including the replication of this design ([https://doi.org/10.14452/MR-070-09-2019-02_2 6]). While unused today, the panopticon is widely considered as both a theoretical and practical inspiration of other forms of surveillance in the workplace.
 
2.2 Scientific management: this theory attempts to analyze and standardize workflows to improve labor productivity. Frederick Winslow Taylor, its main pioneer, used early forms of data analysis during the XX century to decrease walking distances, eliminating unnecessary movements and preventing deliberate slow-walking among workers. ([https://doi.org/10.14452/MR-070-09-2019-02_2 7]). There are many modern forms of data-driven decision-making processes in the workforce. For instance, Amazon warehouses track individual workers' scan rates in real time through handheld scanners, generating data feeds to track and evaluate performance ([https://www.theverge.com/2019/4/25/18516004/amazon-warehouse-fulfillment-centers-productivity-firing-terminations 8]).
 
2.3 PC in the workplace: John Blankenbaker designed the first commercially available personal computer in 1971 ([https://www.computerhistory.org/revolution/personal-computers/17/297 9]), however, it wasn’t until the mid-1980s that the use of personal computers in the workplace became an industry standard. This achievement gave rise to the digital monitoring of work, which in turn allowed employers to monitor employees in real-time, effectively replacing the human eye with a more efficient surveillance mechanism. This represents one of the key aspects of modern-day workplace monitoring. 
 

=== 3. Mechanism of path dependence. ===
''' '''When analyzing the historic lineage of work surveillance, it is possible to point out two main path dependencies mechanisms in place.Firstly, it is clear that many of the technological developments that arose through these events have one major goal, which is to increase wealth within a hierarchical ecosystem. Employers resort to several mechanisms to maximize profits and cut inefficiencies at the expense of workers. 
In addition, standardization can also be singled out as a key mechanism at play. Once one corporation implements any of these developments (such as the digital monitoring of the workplace), the market incentives would force the rest to do the same in order to be profitable and remain competitive.

=== 4. Politics and distribution. ===
 One simple answer to this question would be that capitalists implement and leverage these mechanisms to increase their wealth while the working class remains powerless and unable to withdraw itself from the system in force. However, if we resort to Winner's framework, we see that the technical elements of these technologies are not intrinsically capitalist, since workplace surveillance can also be observed in alternative regimes (including chattel slavery and the Soviet regime). It seems that, rather than being exclusively tied to capitalistic values, the methods in question respond to strong hierarchical structures.
In addition, even under a libertarian framework, structural inequalities are still present, allowing a privileged class to enjoy individual freedoms while segregated groups keep performing low-income jobs ([https://www.metamute.org/editorial/articles/californian-ideology 10]), arguably still subject to surveillance. Thus, not only are these technologies compatible with hierarchical top-down forms of government, but also with structures that allow such hierarchies to form without state intervention.
That said, this form of management seems less compatible with democratic societies, which tend not only to value individual liberties but to turn them into enforceable rights (including workers' privacy, health and integrity), thus establishing boundaries for workplace surveillance. This could explain why studies of American business leaders have found them unhappy with democratic scruples ([https://www.jstor.org/stable/20024652 11]).
 
=== 5. Contingency and alternatives. === 
Bush writes about the memex, a machine that could extend a person’s memory and that the memory would belong to the person who owned it. As Bush expands, “For mature thought there is no mechanical substitute" ([https://dl.acm.org/doi/pdf/10.1145/227181.227186 12]). His idea was that a machine could take over the repetitive nature of someone’s job, while the person could record their own thoughts for their own use. Bush’s memex was never created, but the idea inspired products like Microsoft’s Productivity Score that were made for employers (the buyers) not for the person whose memory it used (the worker).
This immediately brings to mind Meta's Model Capability Initiative, which was done to train AI agents the company hopes to sell to other employers as well as watch their employees ([https://www.reuters.com/sustainability/boards-policy-regulation/meta-start-capturing-employee-mouse-movements-keystrokes-ai-training-data-2026-04-21/ 13]). If this was Bush’s memex, it would belong to the employee using it, not Meta. 

=== 6. Governance implication. ===
Democratic oversight of the use of algorithms as a form of surveillance is essential to ensure that workers’ rights are not overstepped in the name of efficiency. As such we propose: all workers should have a legal right to be consulted and approve a surveillance program before a company installs and uses it. A tracking system should be a group decision, not an inevitable, unilateral one. While the United States has very little protections for workers in terms of surveillance, the EU leads the way with the GDPR, specifically Article 22 ([http://data.europa.eu/eli/reg/2016/679/oj 14]), and the AI Act Article 5 ([http://data.europa.eu/eli/reg/2024/1689/oj 15]). It’s also worth noting that in December 2026, the Platform Work Directive ([https://eur-lex.europa.eu/eli/dir/2024/2831/oj/eng 16]) will come into effect which will ban automated termination.

= [[Platform Dependence in Public Communication (News) - Ceren Isiklar & Rodrigue Kayembe|Platform Dependence in Public Communication (News)]] Ceren Isiklar & Rodrigue Kayembe =

= Centralization of AI Compute and Data =

== Centralization of AI Compute and Data - by Sisi Hammoudi & Juliette Chamot ==
=== Governance issue ===
Today, several path-dependence phenomena are emerging across advanced computing technologies. This assignment addresses one of them: the centralisation of AI compute and data. Training and running frontier AI systems now depends on a handful of hyperscale providers, namely Amazon, Microsoft, Google, and Nvidia, which dominates the GPU market. In other words, near monopolies exist throughout the supply chain. This concentration creates a governance problem affecting public and private actors alike, both legally and institutionally. On the institutional side, it prevents public administrations, and states other than the US or China, from reaching sovereign AI capacity. It also raises competition-law concerns, since dominant market positions can prevent the market from delivering the best products at the best prices to consumers. Finally, AI developers, like states, are often forced into contracts with hyperscale providers, creating a structural dependence on them.


=== Historical lineage ===
==== 1945 ====
Wartime scientific instruments should have been redirected to extend human command over knowledge, helping establish the postwar model of sustained federal funding for computing research (Vannevar Bush's, "As We May Think », 1945).


==== 1990s ====
The privatisation of the NSFNET established the idea that critical digital infrastructure, even though publicly subsidised, could legitimately be privately owned. The EU Commission followed the US's lead, recommending privatisation and deregulation of telecommunications and media as the best way to build the "infobahn" though EU members could instead have chosen nationalisation of the infrastructure (Barbrook & Cameron, "The Californian Ideology," 1996).


==== 2006 ====
<div class="s13">Hyperscale cloud computing turned massive fixed capital investment in data centresinto a rentable service, favouring firms able to invest at scale.</div>
<div class="s13">Same period: Nvidia's proprietary CUDA software, released alongside its GPUs, became the de facto standard for machine learning, locking in a decade of tooling and expertise around one firm's hardware.</div>
<div class="s13">
==== From 2022 ====
Semiconductor export controls and the CHIPS Act turned compute into an explicit instrument of geopolitical strategy.</div>



=== Mechanisms of path dependance ===
Several mechanisms drive this centralisation, but path-dependence ones dominate. Because AI compute is a capital-intensive industry, new entrants struggle to compete with hyperscalers that have already sunk huge sums into infrastructure, a barrier that produces monopoly-like situations. This dynamic, known as increasing returns, generates system-scale economies that harden into a near-irreversibility of investment (David, "Clio and the Economics of QWERTY," 1985). It overlaps with a second mechanism, standardisation, itself a form of technical interrelatedness: researchers' investment in Nvidia's CUDA-based tooling locks the field into that ecosystem regardless of whether better alternatives exist. A further mechanism is legal and contractual entrenchment. For example, Microsoft's stake in OpenAI binds developers to a specific provider by agreement rather than technical necessity.=== Politics and distribution ===Technology does not have a history of distributing power equally. Rather, its development has strengthened the interests and authority of powerful groups that already possessed economic and political resources, and it is yet to be determined whether AI compute and data will follow suit. 
Author Langdon Winner illustrates the historical constraints of technological development with the case of New York from the 1920s to the 1970s. He argued that the city’s roads, bridges, and parks’ physical arrangement contained “explicit and implicit political purpose.” These infrastructures privileged wealthy and white communities while restricting poor and Black populations’ choice of mobility. Similarly, the development of industrial machinery increased productivity, simultaneously weakening union and non-union workers’ bargaining power.
In the twentieth century, following the Industrial Revolution, a Control Revolution emerged from the increasing need to manage expanding production, transportation, and information flows requiring strengthened organizational bureaucratic control. This was heavily shaped by the political context of post-World War II warfare.
Therefore, the bearer of the initial cost of technology is the person putting in the capital to erect it, but ultimately, the person bearing the long-term cost of technology, whose choices diminish, is not the rich and privileged, but the everyday person subject to technologies. Forced to adapt, to reshape their professional skills to adapt to an environment trying to make them obsolete.
AI compute and data are at risk of creating the same effect. In their critique of dotcom neoliberalism, authors Barbrook and Cameron exposed a similar socio-economic tension. The Californian Ideology claims digital technology empowers individuals; however, perpetual unequal access to it creates an apartheid between “information-rich” and “information-poor” populations.
Finally, the development of AI compute and data depends on expensive and scarce computing infrastructure and a highly concentrated semiconductor supply chain. Today, this is one of the most concentrated industries, with single companies having near-total monopolies over the production of chips necessary for the functioning of AI. This monopoly model threatens to reinforce the existing economic hierarchy.

=== Contingency and alternatives ===If a different technological trajectory was plausible, the most apt time for it to have occurred would be during the transition from the Industrial Revolution to the “Control Revolution”. When industrial production accelerated, our society experienced a “crisis of control”, creating opportunities to organize information and communication differently.
Instead of allowing control technologies to be monopolized within businesses and states, our societies could have pursued a more democratic and decentralized system. Barbrook and Cameron provide such an alternative path through their idea of the “electronic agora,” where digital networks could encourage community participation and direct democracy, where government could actively engage with the people to shape digital infrastructure and expand access. These alternatives did not actually prevail because corporate and state market participants possessed greater resources and authority to shape technological development at the end of the twentieth century. 
Inherently, technological development is not predetermined; it can support either authoritarian or democratic arrangements. Today, monopolies control chip production needed for AI compute and data, but other technologies like blockchain permit democratic participation.

=== Governance implication ===History always threatens to repeat itself. As Barbrook discussed, digital societies are shaped by the political context in which they develop. It is a valid concern, then, that the development of information technology, through AI computing, would follow the same trajectory- that is, be constrained to benefit the wealthy who funded its development, and ultimately harm the middle and lower classes enabling its functioning. 
However, AI’s place in our society has not yet crystallized. Technology governance does not have to adhere to this system. Such a seemingly sustainable approach to AI compute and data governance has already been broached. In 2022, the Notre Dame Journal on Emerging Technologies published an article in its issue on Artificial Intelligence where the authors proposed an AI compute system based on a “dual investment strategy”. Firstly, the model would be launched by “subsidizing and negotiating” cloud computing for human AI researchers through “existing” cloud vendors, which would expand on existing AI compute initiatives. Secondly, the compute system would be specifically invested in as a publicly owned infrastructure that would be built and operate similarly to national libraries which own “academic supercomputing facilities”. 
The pilot program would have a “base level of compute access” which would keep costs of processing down and enable it to meet the compute needs of a majority of academic researchers. Absent any strengthened legislation on the use of AI compute and data, the hardware would be accompanied by an AI ethics framework to enable sustainable technological governance. In addition to an ethical staff, the human resources needed to keep an AI compute model running would include data officers, scientists, data center and software engineers, research developers, and system administrators. These positions require specialized training, allowing new types of professionals, from any background, to have access to professional prosperity in the field of AI. 
By implementing these frameworks, never neglecting the human element, and the economic and intellectual impact AI has on our society, we move from a historical determinist perspective to one acknowledging historical constraints and learning from them.


=== Bibliography ===
<div><div>[1] Langdon Winner, Do Artifacts Have Politics, Daedalus, Winter, 1980, Vol. 109, No. 1, Modern Technology: Problem or Opportunity? pp. 121-136, 124, (1980)</div>
<div>[2] Langdon Winner, Do Artifacts Have Politics, Daedalus, Winter, 1980, Vol. 109, No. 1, Modern Technology: Problem or Opportunity? pp. 121-136, 124, (1980)</div>
<div>[3] James R. Beniger, The Control Revolution: Technological and Economic Origins of the Information Society, Harvard University Press, p. 443, 450 (1986)</div>
<div>[4] Richard Barbrook, Andy Cameron, The Californian Ideology, ''Science as Culture'', ''6''(1), 44–72, 61 (1996)</div>
<div>[5] Anton Korinek, Jai Vipra, Concentrating Intelligence: Scaling and Market Structure in Artificial Intelligence, National Bureau of Economic Research, 10.3386/w33139, p. 11, 13-15 (2024)</div></div>
<div><div>[6] James R. Beniger, The Control Revolution: Technological and Economic Origins of the Information Society, Harvard University Press, p. 443, 442 (1986)</div>
<div>[7] Richard Barbrook, Andy Cameron, The Californian Ideology, ''Science as Culture'', ''6''(1), 44–72 (1996)</div>
<div>[8] Langdon Winner, Do Artifacts Have Politics, Daedalus, Winter, 1980, Vol. 109, No. 1, Modern Technology: Problem or Opportunity? pp. 121-136, 130 (1980)</div>
<div>[9] Daniel E. Ho, Jennifer King, Russell C. Wald & Christopher Wan, The Centrality of Data and Compute for AI Innovation: A Blueprint for the National Research Cloud, 3 Notre Dame J. on Emerging Tech. 71, 81(2022)</div>
<div>[10] Daniel E. Ho, Jennifer King, Russell C. Wald & Christopher Wan, The Centrality of Data and Compute for AI Innovation: A Blueprint for the National Research Cloud, 3 Notre Dame J. on Emerging Tech. 71, 82 (2022)</div></div>

== Centralization of AI compute and Data - by Atmika Dechamma & Pablo Latorre ==
=== Governance Issue ===
AI computing capacity and data presents the governance problem of an extreme concentration of technological infrastructure and information in a restricted number of companies that privately and globally dominate the market. This affects public and private actors - such as small and medium enterprises, potential new competitors, and Global South countries - that increasingly depend on AI systems outside their control and/or are deterred from competing with established companies because of impenetrable entry barriers - most notably initial investments and operational costs (1). The institutional and legal relevance of this is that it consolidates, reproduces and amplifies asymmetries of power in favor of the “Big Tech”, reverting the technological trend from digital decentralization to physical infrastructure dependency and curbing innovation and market efficiency.

=== Historical Lineage ===
==== Limitation of Machine Learning ====The AI revolution was built on the limitation of human capabilities to process large information and derive efficient results - where before there existed only facilitative tools to allow for human-enabled processes, AI now offered derivations and cognitive reasoning in moments and at large scales.

 ==== Revert to Physical Infrastructure Dependency in the rush to be the best ====The present AI model heavily relies on hardware in order to perform complex data processing for deep learning. The data centre models followed require larger model sizes (which require higher resource, information and capital investment) to be able to yield better results. Due to the limitation of resources and a rush to establish the best working model of an AI agent, data centres are built on a rudimentary model heavily dependent on massive amounts of electricity and fresh water (2).

 ==== Regulation and Decentralization of Information ====Regulation and influence of sovereigns and institutions has resulted in data sharing and information exchange structurally through open source and cloud computing. But, while AI models claims their selling point to be accessibility for all, their information and derivations remain in a closed server, stored for the purpose of training their AI agents to be the best (3). Big AI contenders attract higher consumer and information flows overall resulting in a better developed AI model, proving disadvantageous in any market where early contenders have an advantage and accumulation of power.
=== Mechanisms of Path Dependence ===Early on in the industry there is an expectation of human and capital investment from all stakeholders involved, this high cost for entry into the market and expected capital to sustain and develop a model by both consumer and enterprise makes for a high switching cost. For the AI model to run and scale there is a presumption that the rate of growth is dependent on the size of data fed, resulting in a better model and hence, more consumer usage. A closed centralised data loop is created but the requirement for storage and processing far surpasses current infrastructures capabilities - heavily relying on early market entry advantage and sustained capital investments taking away from smaller aspirants and hoarding of power within closed circles. Despite its physical dependency, AI data processing has unique requirements beyond legacy infrastructures (4) - larger demand and production of GPUs and designed to handle immense thermal heat generated while consuming electricity beyond current capabilities. 

 The AI space largely remains unregulated due to its undefined form and fluid nature, however, this has given way for unethical and morality issues regarding data privacy, exploitation of labour and environmental harms.

=== Politics and Distribution ===
Concentration strengthens the values, interests, capacities, and authority of the “Big Tech” over other actors of public and private nature. In this context, Barbrook and Cameron (5) illuminate the ethos behind AI technology development while Winner (6) makes observable the power relations surrounding it: concentration is the consequence of an AI development anchored in the belief that it must be free from all institutional and normative restriction since it is inherently liberating and non-political which, in turn, provides a political/moral justification for the concentration (if technology is inherently liberating and non-political then there is no power relations to observe or question) that translates itself into power dynamics where the AI developer not only owns the AI technology but also decides how it is trained and deployed and which models the user can or cannot access.

=== Contingency and Alternatives ===
Concentration is due, first and foremost, to the fact that AI computing capacity requires massive and cutting-edge physical infrastructure - such as data centers and microprocessors - which, in turn, requires upfront investments that are unaffordable to most companies. In David’s (7) language, companies that could finance the infrastructure and position themselves first obtained self-reinforcing advantages and a “lock-in” effect in the market. In other words, concentration is the consequence of the development of AI physical infrastructure privately with no mixed or public alternatives as a public service. Therefore, there was a space for an alternative future at the end of the 1980s or the beginning of the 1990s when the privatization model of the technological infrastructure was consolidated in line with the “Californian Ideology”. In that moment governments had the chance to treat technological infrastructure as a mixed or public good rendering it universally accessible as it was the case of the Internet; this would have permitted companies of all sizes to profit from economies of scale, reducing entry barriers and fostering competition. However, this alternative path did not prevail since the Californian Ideology provided the moral/political justification to exclude governments from the development of technological infrastructure in the exact moment in which such a decision would produce structural consequences. If that path had been followed it could have contributed to collective access to economies of scale, lower entry barriers, and a more competitive AI market.

=== Governance Implication ===
Considering the above, public or mixed investment on AI physical infrastructure is a necessary governance measure for market decentralization and collective access to AI computing capacity and data. From an historical perspective, the proposed measure is not merely a corrective action for a market failure but for a contingent and ideological decision made between the late 1980s and early 1990s that produced a structural “path dependence” that now is necessary to change.
==== References ====
(1) The Quiet Re-Centralization of Tech: Why AI Is Reversing 20 Years of Platform Decentralization. (2026, January 10). Retrieved on 8 September 2026, from Medium website : [https://medium.com/@sumitbhardwaj1357/the-quiet-re-centralization-of-tech-why-ai-is-reversing-20-years-of-platform-decentralization-ce15962b6a13 https://medium.com/@sumitbhardwaj1357/the-quiet-re-centralization-of-tech-why-ai-is-reversing-20-years-of-platform-decentralization-ce15962b6a13]

(2) [https://www.mdpi.com/1996-1073/19/1/137 Ginzburg-Ganz, E., Lifshits, P., Machlev, R., Belikov, J., Krieger, Z., & Levron, Y. (2025). Technical challenges of AI data center integration into power grids—A survey. ''Energies'', ''19''(1), 137.]

(3) [https://decai-resources.s3.us-east-1.amazonaws.com/decentralized_ML_perspective-16.pdf Singh, A., Lu, C., Gupta, G., Chopra, A., Blanc, J., Klinghoffer, T., ... & Raskar, R. (2024). A perspective on decentralizing AI.]

(4) The Rise of AI Data Center Models and the Decline of the General-Purpose Data Center. (Circa 2026). Retrieved on 8 September 2026, from SubZero Engineering website : [https://www.subzeroeng.com/the-decline-of-the-general-purpose-data-center/ https://www.subzeroeng.com/the-decline-of-the-general-purpose-data-center/]

(5) [https://git.xpub.nl/onebigear/SI_09_reference_library/raw/commit/22bf61e1dcfed35d58fba23808212bd00584be47/Richard_Barbrook_Andy_Cameron_Californian_ideology.pdf Barbrook, R., & Cameron, A. (1996). The Californian Ideology. ''Science as Culture'', ''6''(1), 44-72].

(6) [https://d1wqtxts1xzle7.cloudfront.net/87708502/langdon_winner_do_artifacts_have_politics-libre.pdf?1655588533=&response-content-disposition=inline%3B+filename%3DDo_Artifacts_Have_Politics.pdf&Expires=1788890463&Signature=UBBiMqkip5MfYZbkbLNAN6aE2hyZdKkER-HFcmyxXkwIZuTnMKdOGoao8ddFr-rQ5T55A8NWUOlnZkGYmi66e7XR9h99DgTELMCrh48sXjOb8KYgA48r~SUUymt1IW~R~aOKJzitMIUHcrc6yEGKK1OuaA~mrEIVOXpUjQMYjqOd26Xmk9YJEglTkqaxeR47VCCth5FvozNQIowdfjzeu8aNHnI8V7Yur1IiXY~PGvYRcN6wBtcgpVMHZUtNaI8NjmYbp2p5TBYrPMF9MCLtvpGhPlsCfsvA8Bdm6XulGwRlksZBKWvVUQ8D~M8oLu12V39YaiETIObyJqbE7ctIcQ__&Key-Pair-Id=APKAJLOHF5GGSLRBV4ZA Winner, L. (2017). Do artifacts have politics?. In ''Computer ethics''(pp. 177-192). Routledge].

(7) [https://sjbae.pbworks.com/f/David+1985.pdf David, P. A. (1985). Clio and the Economics of QWERTY. ''The American economic review'', ''75''(2), 332-337].

== Centralization of AI compute and Data - by Sergey Abramov and Paul Stein ==<div>

=== The current governance issue ===
Advanced AI currently depends on a narrow group of tech giants1 and semiconductor oligopolies controlling microchips and software, hyperscale cloud infrastructure, large datasets, frontier models, and distribution platforms. This presents a governance problem where this narrow group exercises control over foundational digital infrastructure. This directly affects consumers, businesses, and academia out of AI. Legally and institutionally, this matters because it creates a profound democratic deficit: this neutralizes antitrust frameworks designed for traditional markets, often bypassing conventional copyright and data privacy protections through unchecked data harvesting thereby complicating enforcement.This ultimately might affect national sovereignty as the countries that host the buildout will shape the future of AI: who controls it, what values it contains, how it is used2.
 
=== Historical lineage ===
In 1987, Taiwan Semiconductor Manufacturing Company (TSMC) was founded, establishing the world’s first semiconductor fab. Instead of designing and making its own chips, TSMC only manufactured chips designed by other companies. Today, nearly 100% of the world's most advanced AI processors are manufactured by TSMC, leaving the entire global AI compute supply chain centralized under one company.
 In 2006, Nvidia launched CUDA (Compute Unified Device Architecture), a free software layer that allowed developers to use Nvidia's graphics chips (GPUs) which later became perfect for training deep neural networks. Today, this creates a near-monopoly on AI compute hardware.
 Soon after Amazon launched AWS, Microsoft and Google launched their own cloud platforms marking the structural pivot to the cloud service provider model, shifting computing from a distributed architecture (where companies bought and managed their own physical servers) to a centralized infrastructure. As training modern AI models requires tens of thousands of GPUs, today’s frontier AI companies are entirely dependent on the extremely expensive physical infrastructure of a handful of cloud hyper-scalers (Microsoft, Google, AWS, and Meta).
 In 2009, Fei-Fei Li launched ImageNet, a massive database of over 14 million images, which proved that AI model performance scales directly with data volume. This created an immediate institutional advantage for consumer internet platforms (like Google and Meta), because they were already architected to collect, centralize, and store masses of user data through search engines and social networks, they possessed the unique data stack required to train high-performing AI.
 
=== Mechanism of path dependence ===
The centralization of AI compute and data is a classic case of technological lock-in. When structural choices were made decades ago, they triggered powerful, self-reinforcing loops. One such loop is the switching costs and software standardisation. As developers have spent more than a decade working with CUDA, moving to a rival hardware provider (or a decentralized compute network) requires an enterprise to completely rewrite its foundational software stack. The cost of code migration, developer retraining, and potential hardware downtime force companies to stay tied to dominant hardware monopolies such as Nvidia. Commitments, licensing conditions, and exclusivity provisions add contractual costs. Fewer than 1% of cloud customers switch provider annually, identifying technical differentiation and data-transfer charges as important barriers4.
 Network effects also matter for computing power. More customers generate more revenue, usage information, and opportunities to improve infrastructure and models. Better services attract additional users and developers, whose applications make the platform more valuable. In cloud–AI partnerships, providers may additionally obtain information about model performance and future infrastructure demand. These feedback loops strengthen firms already operating across cloud, models, software, and distribution.
 
=== Politics and distribution ===
According to Langdon Winner's definition, modern AI computing infrastructure may qualify as an "inherently political technology" because its large scale and significant financial requirements not only lend themselves to, but necessitate,5 centralized, hierarchical control. The physical siting of this infrastructure shows evidence of Winner’s second category of politicization of technologies, the ‘political arrangement’:6 Of the five largest U.S. AI firms, four are headquartered in California,7 while just one is (partially) domiciled in Texas. Texas, however, hosts twice the GPU clusters.8 This placement geographically insulates beneficiaries of the data infrastructure boom from its negative externalities. The specific choice of location may itself be a deliberate preference on account of Texas’ unique approach to water usage rights.9 Juxtaposition of this unequal and even exploitative balance of interests with the pontificating rhetoric of the urban, information rich, beneficiaries is evocative of the internal contradictions of the Californian Ideology - the virtual artisans existence and privileged position is still based on the exploitation of a vulnerable out-group, the "information-rich" consuming the output, the "information-poor" hosting the extraction.10
 
=== Contingency and Alternatives ===
Profit motive and economies of scale now concentrate investment among a few hyperscalers,11 but this reflects policy choice, not government incapacity: Compute infrastructure may be inherently political, but the technology is agnostic as to where the organizing authority originates, and states have historically funded infrastructure of comparable or greater scale, while facing less pressure to show short-term returns. The clearest turning point was 1995, when the National Science Foundation retired NSFNET, the publicly funded backbone that had connected U.S. research and supercomputing centers since the mid-1980s,12 handing internet infrastructure to commercial providers with no conditions attached. This was a deliberate policy decision made amid the deregulatory push of the Clinton era. The privatization that followed ushered in the private industrial dominance that persists today in the form of hyperscalers. Had a publicly governed backbone instead been scaled up, today's compute infrastructure might rest on a different foundation entirely.
 
=== Governance implication ===
Despite private hyperscaler dominance, the U.S. does have a nominal public alternative: the National AI Research Resource (NAIRR), established following the National AI Initiative Act of 2020 and piloted from January 2024, which explicitly aims to democratize access to the computational resources essential to AI research.13In Europe, a comparable program is the EuroHPCJU,14 an initiative intended to pool resources to build supercomputing capacity. The undertaking’s legislative basis has since been extended to include networked ‘AI factories', as well as the so-called “Gigafactories”.15 The framing in language of sovereignty indicates intent to at least compete with private infrastructure initiatives. Though government-led measures cannot reverse the centralization of compute, they can democratize access - provided they are intensified. The European Commission has “set a goal of investing more than €1 billion per year in AI research and innovation”.16 While laudable in its own right, it pales in comparison to the investments of industry leaders, which are approaching, if not exceeding, trillions of Dollars.

=== References & Notes ===
1.Amazon, Google, Meta, Microsoft, and Oracle collectively hold an estimated 71% of the world’s cumulative AI compute as of Q4 2025, measured in H100-equivalents of computing power. Luke Emberson, Josh You and Venkat Somala (2026), "Five hyperscalers now own over two-thirds of global AI compute". Published online at epoch.ai. Retrieved from 'https://epoch.ai/data-insights/hyperscalers-control-most-compute' [online resource]. Accessed 7 Sept 2026.2. Alasdair Phillips-Robins, Teddy Tawil and Sam Winter-Levy, ‘The Compute Coalition: How to Build the Future of AI in the Free World’ (Carnegie Endowment for International Peace, 8 June 2026) https://carnegieendowment.org/research/2026/06/the-compute-coalition-how-to-build-the-future-of-ai-in-the-free-world accessed 7 September 2026.

3. Amazon Web Services (AWS), a cloud computing platform

4. Competition and Markets Authority, Cloud Infrastructure Services: Final Decision Report (31 July 2025) https://assets.publishing.service.gov.uk/media/688b8891fdde2b8f73469544/final_decision_report.pdf accessed 7 September 2026.

5. Langdon Winner, 'Do Artifacts Have Politics?' (1980) 109(1) ''Daedalus'' 121, 128.

6. Ibid. 123.

7. OpenAI, Anthropic, Google DeepMind, Meta AI. xAI is an exception to the rule, domiciled between Palo Alto, CA and San Antonio, TX.

8. Texas’ 17 to California’s 8, cf. Texas Royalty Brokers, 'AI Clusters by State' (''Texas Royalty Brokers'', 2025) https://texasroyaltybrokers.com/ai-clusters-by-state/ accessed 7 September 2026.

9. The ‘rule of capture’, established in Houston & TC Railway Co v East (Tex 1904), permits landowners to extract groundwater beneath their land without liability for depleting adjacent wells or aquifers, a doctrine now exploited by hyperscale AI data centres to draw large cooling-water volumes with minimal regulatory constraint.

10. Richard Barbrook and Andy Cameron, 'The Californian Ideology' (1996) 6(1) ''Science as Culture'' 44.

11. Goldman Sachs, 'Global AI Investment Is Forecast to Exceed $1 Trillion in 2026' (''Goldman Sachs Insights'', 7 August 2026) https://www.goldmansachs.com/insights/articles/global-investment-is-forecast-to-exceed-1-trillion-in-2026 accessed 7 September 2026.

12.  National Science Foundation, 'Birth of the Commercial Internet' (''NSF Impacts'') https://www.nsf.gov/impacts/internet accessed 7 September 2026.

13. CASRAI, 'NAIRR: National AI Research Resource, Explained' (''CASRAI'') https://casrai.org/dictionary/term/national-ai-research-resource-nairr accessed 7 September 2026.

14. Council Regulation (EU) 2021/1173 of 13 July 2021 on establishing the European High Performance Computing Joint Undertaking and repealing Regulation (EU) 2018/1488 [2021] OJ L256/1.

15. Council Regulation (EU) 2026/150 of 16 January 2026 amending Regulation (EU) 2021/1173 on establishing the European High Performance Computing Joint Undertaking [2026] OJ L 2026/150.

16. European Commission, 'Realising Europe's Digital Decade' (''European Commission'', 11 October 2024) https://commission.europa.eu/strategy-and-policy/priorities-2019-2024/story-von-der-leyen-commission/realising-europes-digital-decade_en accessed 7 September 2026.</div>

Governance by history

2026-09-23T20:11:21Z

Robert Florijn: /* contributions */

[[File:Old internet.jpg|300px|thumb]]

= contributions =
*can
*you 
*please
*create
*individual pages
*from the contributions below
*and link them back here? 
*[[Information Control and Workplace Surveillance - Jihae Park and Mariona Fortuny]]
*[https://wiki.ivir.nl/index.php?title=Centralization_of_Compute_and_Data Centralization of AI Compute and Data - by Sergey Abramov and Paul Stein]
*[[Media:Civilian At The Front Final.pdf]]

= links things that popped up during class =
*[https://craphound.com/ Cory Doctorow]
*[https://dougengelbart.org/content/view/209/ The mother of All demos] (first mouse, graphical interface, word processor, etc)
*Can a simple sensor be political? [https://gizmodo.com/why-cant-this-soap-dispenser-identify-dark-skin-1797931773 well, this "racist" soap dispenser made headlines a couple of years ago.]
*Podcast series: [https://www.the-santiago-boys.com/ Morozov: the Santiago Boys.]
*Anatomy on AN AI. [[File:Ai-anatomy-map.pdf|check out the pdf here]]

= Workplace surveillance =
== Information Control and Workplace Surveillance - by Jihae Park and Mariona Fortuny Jané ==

==== The Current Governance Issue ====
There are numerous current governance problems arising from technological innovations, yet here we will specifically focus on the information control and workplace surveillance. Whether intended or unintended, there were certain historical incidents which resulted in social and political consequences. As noticed from the Control Revolution, the dramatic evolution in mass media and telecommunications technologies has led to the reinforcement of the economic and political control during the Industrial Revolution. ([https://archive.org/details/beniger-james-the-control-revolution Beniger, 1986]) By applying this logic to a modern example, this is clearly how many corporations are getting in control of their employees through surveillance technologies. Along with the emergence of remote and hybrid work after Covid-19, employers are now relying on “bossware” to analyze the emails, chats, and messages of their employees on platforms such as Teams or Slack. ([https://ojs.library.queensu.ca/index.php/surveillance-and-society/article/view/19100/13263 Ollier-Malaterre et al., 2026, p.237]) Some companies have strict restrictions on transferring data both internally and externally in protection of confidential information and consistently monitor their workers to the certain extent which could slow down their efficiency. This is problematic as it’s not only violating the privacy of their employees but also depriving employees of rights they should be naturally be able to exercise.



==== Historical Lineage ====
In relation to the current issue of the information control and workplace surveillance, Max Weber’s argument of bureaucracy and rationalization in the middle and late nineteenth century are good historical examples which are still frequently visible nowadays. Weber stated how the range of control could extend not only by expanding the capability to process information but also by limiting the amount of information to be processed. Here, the former is directly related to the control implemented by bureaucratization (in the mid 19th century) and computerization (in the late 20th century), whereas the latter addresses rationalization. An example from bureaucracy is the expansion of paperwork in response to the growth in information. We can still note these traces of bureaucracy from numerous procedures that we have to go through for a decision-making to take place in corporations. He brings up the example of standardized time zones for rationalization which is simply an example for an omission or a breakdown of information to facilitate its processing. ([https://archive.org/details/beniger-james-the-control-revolution Beniger, 1986, p.15-16]) This is also commonly seen from nowadays as some companies may use internal terminologies for the sake of both confidentiality and efficiency.

==== Mechanism of Path Dependence ====Following along the line with the historical lineage, there are certain mechanisms at work of path dependence for information control and workplace surveillance. Bureaucracy is closely related with organizational routines. Many companies have already settled down with a fixed workflow and decision-making process, which makes it more costly for one to change from its conventional way of working. This means switching costs from an old system to a new one is clearly high considering its time and capital, not necessarily will guarantee increasing returns. Hence, many firms end up maintaining its old strategy, reluctant to make any modifications or changes. 

==== Politics and Distribution ====

As businesses grew too large for managers to oversee everything directly, companies replaced human judgment with formal rule-systems, as we see explained by Beniger's 'intellectual technology' ([https://archive.org/details/beniger-james-the-control-revolution/page/65/mode/2up Beniger, 1986, pp. 17, 21]). Algorithmic management continues this: a call once made by a supervisor's judgment is now a rule built into software that runs without one, extending a logic Beniger traces as far back as to Taylorism and the so called 'assembly line'. Winner ([https://www.jstor.org/stable/20024652 1980]) asks whether such systems must concentrate power this way or were driven there, when other designs were equally possible; nothing required shutting workers out of the decision, so leaving them out was a political choice, and not necessarily a technical necessity. Barbrook and Cameron ([https://doi.org/10.1080/09505439609526455 1996]) explain the reasoning why some can defend this 'domination': they show how 1990s tech culture sold new technologies as freedom even as it tightened control, so the rhetoric could become one of closer oversight as flexibility and autonomy. Together, authority accrues to employers, who in some systems can set and revise these oversight rules unilaterally and face less resistance because the arrangement is framed as freedom; workers bear the cost, losing the informal give-and-take they had with a human supervisor for oversight that is harder to see, question, or contest, while being told it grants them more freedom, not less.

==== Contingency and Alternatives ====

A plausible alternative happened on U.S. post WWII: in some workplaces, labor and its pacing was demanded to be set through collective bargaining and enforced jointly, by both management and union stewards monitoring compliance on the line ([https://doi.org/10.1093/icc/dtaf038 Breznitz et al., 2025, p. 211]). So, in this context, workplace surveillance could have developed as a bargained, dual-administered practice rather than a management monopoly. The 1950 [https://en.wikipedia.org/wiki/Reuther%27s_Treaty_of_Detroit Reuther's Treaty of Detroit ] cut that possibility off: unions won guaranteed wages and benefits in exchange for surrendering any general claim over how work was supervised ([https://doi.org/10.1093/icc/dtaf038 Breznitz et al., 2025, pp. 210–211]). Workers depended on wages to survive, and employers controlled the wage relationship. That asymmetry meant employers, given their position of power, were able to arrange things so that, at the very least, workers would have far less capacity to sustain a fight over control itself. Thus, the "trade" wasn't a fair exchange between equals. This reflects which side could afford to hold out on which issue. Hypothetically speaking, had that joint administration model prevailed instead, this could have greatly impacted norms internationally and good practices with worker-union agreed surveillance would have become a widely accepted standard practice.

==== Governance Implication ====

An interesting proposition for present-day governance that wants to fill a gap in workplace surveillance is what [https://doi.org/10.1016/j.digbus.2025.100120 Kayas, Ong and Belal (2025)] are proposing: They argue that EU law, through the [https://eur-lex.europa.eu/eli/reg/2016/679/oj/eng GDPR] and its required [https://en.wikipedia.org/wiki/Data_protection_impact_assessment data protection impact assessments (DPIAs)], already offers meaningful protection for workers in terms of lawful high-risk data collection and data processing. Nevertheless, these DPIAs are built around data and privacy, so they leave out several dimensions of monitoring itself: its purpose, how intrusive it is, how visible it is to the people being watched, and its effects on workers' autonomy, trust, and standing relative to management. These authors propose a "surveillance impact assessment" scoring a system against these characteristics directly, before any surveillance system is put in place. Although this mechanism "is constrained by its reliance on existing literature and the lack of empirical validation," it feels like a genuine alternative for two reasons: 1. it targets the actual mechanism at issue, and 2. at the same time it builds upon a mechanism of active participatory approach where employees have a direct say in how surveillance is designed ([https://doi.org/10.1016/j.digbus.2025.100120 Kayas et al., 2025, p. 11]), which, combined with the actual assessment, would put workers' own judgment at the centre of the one decision that affects them. It's important to note that this alternative is proposed as worth piloting, not as a proven fix, and it is fair to ask whether another mandatory assessment stacked on top of DPIAs would function as intended or simply read to workers as more compliance paperwork.

==== References ====
 Barbrook, R., & Cameron, A. (1996). The Californian Ideology. ''Science as Culture'', 6(1), 44–72. [https://doi.org/10.1080/09505439609526455 https://doi.org/10.1080/09505439609526455] Beniger, J. R. (1986). ''The Control Revolution: Technological and Economic Origins of the Information Society''. Harvard University Press. [https://archive.org/details/beniger-james-the-control-revolution https://archive.org/details/beniger-james-the-control-revolution]  Breznitz, D., Levy, K., Lipartito, K., & Zehavi, A. (2025). An equity-focused research agenda for workplace surveillance. ''Industrial and Corporate Change'', 35(1), 206–221. [https://doi.org/10.1093/icc/dtaf038 https://doi.org/10.1093/icc/dtaf038] Kayas, O. G., Ong, C. E., & Belal, H. M. (2025). From humans to algorithms: A sociotechnical framework of workplace surveillance. ''Digital Business'', 5, 100120. [https://doi.org/10.1016/j.digbus.2025.100120 https://doi.org/10.1016/j.digbus.2025.100120] Ollier-Malaterre, A., Parent-Rocheleau, X., Benkrid, K., & Provost Savard, Y. (2026). Making Sense of Workplace Surveillance: How Workers and Union Representatives Experience, Imagine, and Resist Surveillance Exposure. ''Surveillance & Society'', 24(3), 236–254. [https://ojs.library.queensu.ca/index.php/surveillance-and-society/article/view/19100/13263 https://ojs.library.queensu.ca/index.php/surveillance-and-society/article/view/19100/13263] Winner, L. (1980). Do Artifacts Have Politics? ''Daedalus'', 109(1), 121–136. [https://www.jstor.org/stable/20024652 https://www.jstor.org/stable/20024652]

 
== [[Algorithmic management and workplace surveillance - by Sebastián Muller and Kate Davey''' ''']] ==

=== 1. The current issue. ===
 Algorithmic management and workplace surveillance can be defined as the use of computer-programmed procedures to coordinate labor input in an organization ([https://joint-research-centre.ec.europa.eu/scientific-activities/employment/algorithmic-management-and-digital-monitoring-work_en 1]). Simply put, it is a way to control workers, and workers generally have no say in how the system that tracks them is created. As Winner says "different people are differently situated and possess unequal degrees of power as well as unequal levels of awareness" ([https://www.jstor.org/stable/20024652?seq=7 2]).
The population affected is workers; both “white collar” and “blue collar” workers are impacted. Whether it’s Amazon tracking warehouse workers’ “Time Off Task” with handheld scanners ([https://www.wired.com/story/amazon-worker-tracking-details-revealed 3]) or Meta employees having their every click tracked ([https://www.techpolicy.press/metas-worker-surveillance-tests-eu-rules-on-ai-and-labor/ 4]) to train AI, employers have the power to watch and control their employees in an unprecedented way.
It matters legally because workers have very few rights to fight back against these surveillance programs; even the The Electronic Communications Privacy Act of 1986 allows exceptions to surveillance of employees ([https://bja.ojp.gov/program/it/privacy-civil-liberties/authorities/statutes/1285 5]). 
=== 2. Historical lineage. === 2.1 Panopticon: it is an architectural apparatus developed by Jeremy Bentham in 1791. The design of this building effectively generated the sense of constant monitoring, which in turn motivated the prisoners to behave as if they were being watched at all times. As human labor became commodified, capitalists began implementing measures to ensure the maximization of productivity, including the replication of this design ([https://doi.org/10.14452/MR-070-09-2019-02_2 6]). While unused today, the panopticon is widely considered as both a theoretical and practical inspiration of other forms of surveillance in the workplace.
 
2.2 Scientific management: this theory attempts to analyze and standardize workflows to improve labor productivity. Frederick Winslow Taylor, its main pioneer, used early forms of data analysis during the XX century to decrease walking distances, eliminating unnecessary movements and preventing deliberate slow-walking among workers. ([https://doi.org/10.14452/MR-070-09-2019-02_2 7]). There are many modern forms of data-driven decision-making processes in the workforce. For instance, Amazon warehouses track individual workers' scan rates in real time through handheld scanners, generating data feeds to track and evaluate performance ([https://www.theverge.com/2019/4/25/18516004/amazon-warehouse-fulfillment-centers-productivity-firing-terminations 8]).
 
2.3 PC in the workplace: John Blankenbaker designed the first commercially available personal computer in 1971 ([https://www.computerhistory.org/revolution/personal-computers/17/297 9]), however, it wasn’t until the mid-1980s that the use of personal computers in the workplace became an industry standard. This achievement gave rise to the digital monitoring of work, which in turn allowed employers to monitor employees in real-time, effectively replacing the human eye with a more efficient surveillance mechanism. This represents one of the key aspects of modern-day workplace monitoring. 
 

=== 3. Mechanism of path dependence. ===
''' '''When analyzing the historic lineage of work surveillance, it is possible to point out two main path dependencies mechanisms in place.Firstly, it is clear that many of the technological developments that arose through these events have one major goal, which is to increase wealth within a hierarchical ecosystem. Employers resort to several mechanisms to maximize profits and cut inefficiencies at the expense of workers. 
In addition, standardization can also be singled out as a key mechanism at play. Once one corporation implements any of these developments (such as the digital monitoring of the workplace), the market incentives would force the rest to do the same in order to be profitable and remain competitive.

=== 4. Politics and distribution. ===
 One simple answer to this question would be that capitalists implement and leverage these mechanisms to increase their wealth while the working class remains powerless and unable to withdraw itself from the system in force. However, if we resort to Winner's framework, we see that the technical elements of these technologies are not intrinsically capitalist, since workplace surveillance can also be observed in alternative regimes (including chattel slavery and the Soviet regime). It seems that, rather than being exclusively tied to capitalistic values, the methods in question respond to strong hierarchical structures.
In addition, even under a libertarian framework, structural inequalities are still present, allowing a privileged class to enjoy individual freedoms while segregated groups keep performing low-income jobs ([https://www.metamute.org/editorial/articles/californian-ideology 10]), arguably still subject to surveillance. Thus, not only are these technologies compatible with hierarchical top-down forms of government, but also with structures that allow such hierarchies to form without state intervention.
That said, this form of management seems less compatible with democratic societies, which tend not only to value individual liberties but to turn them into enforceable rights (including workers' privacy, health and integrity), thus establishing boundaries for workplace surveillance. This could explain why studies of American business leaders have found them unhappy with democratic scruples ([https://www.jstor.org/stable/20024652 11]).
 
=== 5. Contingency and alternatives. === 
Bush writes about the memex, a machine that could extend a person’s memory and that the memory would belong to the person who owned it. As Bush expands, “For mature thought there is no mechanical substitute" ([https://dl.acm.org/doi/pdf/10.1145/227181.227186 12]). His idea was that a machine could take over the repetitive nature of someone’s job, while the person could record their own thoughts for their own use. Bush’s memex was never created, but the idea inspired products like Microsoft’s Productivity Score that were made for employers (the buyers) not for the person whose memory it used (the worker).
This immediately brings to mind Meta's Model Capability Initiative, which was done to train AI agents the company hopes to sell to other employers as well as watch their employees ([https://www.reuters.com/sustainability/boards-policy-regulation/meta-start-capturing-employee-mouse-movements-keystrokes-ai-training-data-2026-04-21/ 13]). If this was Bush’s memex, it would belong to the employee using it, not Meta. 

=== 6. Governance implication. ===
Democratic oversight of the use of algorithms as a form of surveillance is essential to ensure that workers’ rights are not overstepped in the name of efficiency. As such we propose: all workers should have a legal right to be consulted and approve a surveillance program before a company installs and uses it. A tracking system should be a group decision, not an inevitable, unilateral one. While the United States has very little protections for workers in terms of surveillance, the EU leads the way with the GDPR, specifically Article 22 ([http://data.europa.eu/eli/reg/2016/679/oj 14]), and the AI Act Article 5 ([http://data.europa.eu/eli/reg/2024/1689/oj 15]). It’s also worth noting that in December 2026, the Platform Work Directive ([https://eur-lex.europa.eu/eli/dir/2024/2831/oj/eng 16]) will come into effect which will ban automated termination.

= [[Platform Dependence in Public Communication (News) - Ceren Isiklar & Rodrigue Kayembe|Platform Dependence in Public Communication (News)]] Ceren Isiklar & Rodrigue Kayembe =

= Centralization of AI Compute and Data =

== Centralization of AI Compute and Data - by Sisi Hammoudi & Juliette Chamot ==
=== Governance issue ===
Today, several path-dependence phenomena are emerging across advanced computing technologies. This assignment addresses one of them: the centralisation of AI compute and data. Training and running frontier AI systems now depends on a handful of hyperscale providers, namely Amazon, Microsoft, Google, and Nvidia, which dominates the GPU market. In other words, near monopolies exist throughout the supply chain. This concentration creates a governance problem affecting public and private actors alike, both legally and institutionally. On the institutional side, it prevents public administrations, and states other than the US or China, from reaching sovereign AI capacity. It also raises competition-law concerns, since dominant market positions can prevent the market from delivering the best products at the best prices to consumers. Finally, AI developers, like states, are often forced into contracts with hyperscale providers, creating a structural dependence on them.


=== Historical lineage ===
==== 1945 ====
Wartime scientific instruments should have been redirected to extend human command over knowledge, helping establish the postwar model of sustained federal funding for computing research (Vannevar Bush's, "As We May Think », 1945).


==== 1990s ====
The privatisation of the NSFNET established the idea that critical digital infrastructure, even though publicly subsidised, could legitimately be privately owned. The EU Commission followed the US's lead, recommending privatisation and deregulation of telecommunications and media as the best way to build the "infobahn" though EU members could instead have chosen nationalisation of the infrastructure (Barbrook & Cameron, "The Californian Ideology," 1996).


==== 2006 ====
<div class="s13">Hyperscale cloud computing turned massive fixed capital investment in data centresinto a rentable service, favouring firms able to invest at scale.</div>
<div class="s13">Same period: Nvidia's proprietary CUDA software, released alongside its GPUs, became the de facto standard for machine learning, locking in a decade of tooling and expertise around one firm's hardware.</div>
<div class="s13">
==== From 2022 ====
Semiconductor export controls and the CHIPS Act turned compute into an explicit instrument of geopolitical strategy.</div>



=== Mechanisms of path dependance ===
Several mechanisms drive this centralisation, but path-dependence ones dominate. Because AI compute is a capital-intensive industry, new entrants struggle to compete with hyperscalers that have already sunk huge sums into infrastructure, a barrier that produces monopoly-like situations. This dynamic, known as increasing returns, generates system-scale economies that harden into a near-irreversibility of investment (David, "Clio and the Economics of QWERTY," 1985). It overlaps with a second mechanism, standardisation, itself a form of technical interrelatedness: researchers' investment in Nvidia's CUDA-based tooling locks the field into that ecosystem regardless of whether better alternatives exist. A further mechanism is legal and contractual entrenchment. For example, Microsoft's stake in OpenAI binds developers to a specific provider by agreement rather than technical necessity.=== Politics and distribution ===Technology does not have a history of distributing power equally. Rather, its development has strengthened the interests and authority of powerful groups that already possessed economic and political resources, and it is yet to be determined whether AI compute and data will follow suit. 
Author Langdon Winner illustrates the historical constraints of technological development with the case of New York from the 1920s to the 1970s. He argued that the city’s roads, bridges, and parks’ physical arrangement contained “explicit and implicit political purpose.” These infrastructures privileged wealthy and white communities while restricting poor and Black populations’ choice of mobility. Similarly, the development of industrial machinery increased productivity, simultaneously weakening union and non-union workers’ bargaining power.
In the twentieth century, following the Industrial Revolution, a Control Revolution emerged from the increasing need to manage expanding production, transportation, and information flows requiring strengthened organizational bureaucratic control. This was heavily shaped by the political context of post-World War II warfare.
Therefore, the bearer of the initial cost of technology is the person putting in the capital to erect it, but ultimately, the person bearing the long-term cost of technology, whose choices diminish, is not the rich and privileged, but the everyday person subject to technologies. Forced to adapt, to reshape their professional skills to adapt to an environment trying to make them obsolete.
AI compute and data are at risk of creating the same effect. In their critique of dotcom neoliberalism, authors Barbrook and Cameron exposed a similar socio-economic tension. The Californian Ideology claims digital technology empowers individuals; however, perpetual unequal access to it creates an apartheid between “information-rich” and “information-poor” populations.
Finally, the development of AI compute and data depends on expensive and scarce computing infrastructure and a highly concentrated semiconductor supply chain. Today, this is one of the most concentrated industries, with single companies having near-total monopolies over the production of chips necessary for the functioning of AI. This monopoly model threatens to reinforce the existing economic hierarchy.

=== Contingency and alternatives ===If a different technological trajectory was plausible, the most apt time for it to have occurred would be during the transition from the Industrial Revolution to the “Control Revolution”. When industrial production accelerated, our society experienced a “crisis of control”, creating opportunities to organize information and communication differently.
Instead of allowing control technologies to be monopolized within businesses and states, our societies could have pursued a more democratic and decentralized system. Barbrook and Cameron provide such an alternative path through their idea of the “electronic agora,” where digital networks could encourage community participation and direct democracy, where government could actively engage with the people to shape digital infrastructure and expand access. These alternatives did not actually prevail because corporate and state market participants possessed greater resources and authority to shape technological development at the end of the twentieth century. 
Inherently, technological development is not predetermined; it can support either authoritarian or democratic arrangements. Today, monopolies control chip production needed for AI compute and data, but other technologies like blockchain permit democratic participation.

=== Governance implication ===History always threatens to repeat itself. As Barbrook discussed, digital societies are shaped by the political context in which they develop. It is a valid concern, then, that the development of information technology, through AI computing, would follow the same trajectory- that is, be constrained to benefit the wealthy who funded its development, and ultimately harm the middle and lower classes enabling its functioning. 
However, AI’s place in our society has not yet crystallized. Technology governance does not have to adhere to this system. Such a seemingly sustainable approach to AI compute and data governance has already been broached. In 2022, the Notre Dame Journal on Emerging Technologies published an article in its issue on Artificial Intelligence where the authors proposed an AI compute system based on a “dual investment strategy”. Firstly, the model would be launched by “subsidizing and negotiating” cloud computing for human AI researchers through “existing” cloud vendors, which would expand on existing AI compute initiatives. Secondly, the compute system would be specifically invested in as a publicly owned infrastructure that would be built and operate similarly to national libraries which own “academic supercomputing facilities”. 
The pilot program would have a “base level of compute access” which would keep costs of processing down and enable it to meet the compute needs of a majority of academic researchers. Absent any strengthened legislation on the use of AI compute and data, the hardware would be accompanied by an AI ethics framework to enable sustainable technological governance. In addition to an ethical staff, the human resources needed to keep an AI compute model running would include data officers, scientists, data center and software engineers, research developers, and system administrators. These positions require specialized training, allowing new types of professionals, from any background, to have access to professional prosperity in the field of AI. 
By implementing these frameworks, never neglecting the human element, and the economic and intellectual impact AI has on our society, we move from a historical determinist perspective to one acknowledging historical constraints and learning from them.


=== Bibliography ===
<div><div>[1] Langdon Winner, Do Artifacts Have Politics, Daedalus, Winter, 1980, Vol. 109, No. 1, Modern Technology: Problem or Opportunity? pp. 121-136, 124, (1980)</div>
<div>[2] Langdon Winner, Do Artifacts Have Politics, Daedalus, Winter, 1980, Vol. 109, No. 1, Modern Technology: Problem or Opportunity? pp. 121-136, 124, (1980)</div>
<div>[3] James R. Beniger, The Control Revolution: Technological and Economic Origins of the Information Society, Harvard University Press, p. 443, 450 (1986)</div>
<div>[4] Richard Barbrook, Andy Cameron, The Californian Ideology, ''Science as Culture'', ''6''(1), 44–72, 61 (1996)</div>
<div>[5] Anton Korinek, Jai Vipra, Concentrating Intelligence: Scaling and Market Structure in Artificial Intelligence, National Bureau of Economic Research, 10.3386/w33139, p. 11, 13-15 (2024)</div></div>
<div><div>[6] James R. Beniger, The Control Revolution: Technological and Economic Origins of the Information Society, Harvard University Press, p. 443, 442 (1986)</div>
<div>[7] Richard Barbrook, Andy Cameron, The Californian Ideology, ''Science as Culture'', ''6''(1), 44–72 (1996)</div>
<div>[8] Langdon Winner, Do Artifacts Have Politics, Daedalus, Winter, 1980, Vol. 109, No. 1, Modern Technology: Problem or Opportunity? pp. 121-136, 130 (1980)</div>
<div>[9] Daniel E. Ho, Jennifer King, Russell C. Wald & Christopher Wan, The Centrality of Data and Compute for AI Innovation: A Blueprint for the National Research Cloud, 3 Notre Dame J. on Emerging Tech. 71, 81(2022)</div>
<div>[10] Daniel E. Ho, Jennifer King, Russell C. Wald & Christopher Wan, The Centrality of Data and Compute for AI Innovation: A Blueprint for the National Research Cloud, 3 Notre Dame J. on Emerging Tech. 71, 82 (2022)</div></div>

== Centralization of AI compute and Data - by Atmika Dechamma & Pablo Latorre ==
=== Governance Issue ===
AI computing capacity and data presents the governance problem of an extreme concentration of technological infrastructure and information in a restricted number of companies that privately and globally dominate the market. This affects public and private actors - such as small and medium enterprises, potential new competitors, and Global South countries - that increasingly depend on AI systems outside their control and/or are deterred from competing with established companies because of impenetrable entry barriers - most notably initial investments and operational costs (1). The institutional and legal relevance of this is that it consolidates, reproduces and amplifies asymmetries of power in favor of the “Big Tech”, reverting the technological trend from digital decentralization to physical infrastructure dependency and curbing innovation and market efficiency.

=== Historical Lineage ===
==== Limitation of Machine Learning ====The AI revolution was built on the limitation of human capabilities to process large information and derive efficient results - where before there existed only facilitative tools to allow for human-enabled processes, AI now offered derivations and cognitive reasoning in moments and at large scales.

 ==== Revert to Physical Infrastructure Dependency in the rush to be the best ====The present AI model heavily relies on hardware in order to perform complex data processing for deep learning. The data centre models followed require larger model sizes (which require higher resource, information and capital investment) to be able to yield better results. Due to the limitation of resources and a rush to establish the best working model of an AI agent, data centres are built on a rudimentary model heavily dependent on massive amounts of electricity and fresh water (2).

 ==== Regulation and Decentralization of Information ====Regulation and influence of sovereigns and institutions has resulted in data sharing and information exchange structurally through open source and cloud computing. But, while AI models claims their selling point to be accessibility for all, their information and derivations remain in a closed server, stored for the purpose of training their AI agents to be the best (3). Big AI contenders attract higher consumer and information flows overall resulting in a better developed AI model, proving disadvantageous in any market where early contenders have an advantage and accumulation of power.
=== Mechanisms of Path Dependence ===Early on in the industry there is an expectation of human and capital investment from all stakeholders involved, this high cost for entry into the market and expected capital to sustain and develop a model by both consumer and enterprise makes for a high switching cost. For the AI model to run and scale there is a presumption that the rate of growth is dependent on the size of data fed, resulting in a better model and hence, more consumer usage. A closed centralised data loop is created but the requirement for storage and processing far surpasses current infrastructures capabilities - heavily relying on early market entry advantage and sustained capital investments taking away from smaller aspirants and hoarding of power within closed circles. Despite its physical dependency, AI data processing has unique requirements beyond legacy infrastructures (4) - larger demand and production of GPUs and designed to handle immense thermal heat generated while consuming electricity beyond current capabilities. 

 The AI space largely remains unregulated due to its undefined form and fluid nature, however, this has given way for unethical and morality issues regarding data privacy, exploitation of labour and environmental harms.

=== Politics and Distribution ===
Concentration strengthens the values, interests, capacities, and authority of the “Big Tech” over other actors of public and private nature. In this context, Barbrook and Cameron (5) illuminate the ethos behind AI technology development while Winner (6) makes observable the power relations surrounding it: concentration is the consequence of an AI development anchored in the belief that it must be free from all institutional and normative restriction since it is inherently liberating and non-political which, in turn, provides a political/moral justification for the concentration (if technology is inherently liberating and non-political then there is no power relations to observe or question) that translates itself into power dynamics where the AI developer not only owns the AI technology but also decides how it is trained and deployed and which models the user can or cannot access.

=== Contingency and Alternatives ===
Concentration is due, first and foremost, to the fact that AI computing capacity requires massive and cutting-edge physical infrastructure - such as data centers and microprocessors - which, in turn, requires upfront investments that are unaffordable to most companies. In David’s (7) language, companies that could finance the infrastructure and position themselves first obtained self-reinforcing advantages and a “lock-in” effect in the market. In other words, concentration is the consequence of the development of AI physical infrastructure privately with no mixed or public alternatives as a public service. Therefore, there was a space for an alternative future at the end of the 1980s or the beginning of the 1990s when the privatization model of the technological infrastructure was consolidated in line with the “Californian Ideology”. In that moment governments had the chance to treat technological infrastructure as a mixed or public good rendering it universally accessible as it was the case of the Internet; this would have permitted companies of all sizes to profit from economies of scale, reducing entry barriers and fostering competition. However, this alternative path did not prevail since the Californian Ideology provided the moral/political justification to exclude governments from the development of technological infrastructure in the exact moment in which such a decision would produce structural consequences. If that path had been followed it could have contributed to collective access to economies of scale, lower entry barriers, and a more competitive AI market.

=== Governance Implication ===
Considering the above, public or mixed investment on AI physical infrastructure is a necessary governance measure for market decentralization and collective access to AI computing capacity and data. From an historical perspective, the proposed measure is not merely a corrective action for a market failure but for a contingent and ideological decision made between the late 1980s and early 1990s that produced a structural “path dependence” that now is necessary to change.
==== References ====
(1) The Quiet Re-Centralization of Tech: Why AI Is Reversing 20 Years of Platform Decentralization. (2026, January 10). Retrieved on 8 September 2026, from Medium website : [https://medium.com/@sumitbhardwaj1357/the-quiet-re-centralization-of-tech-why-ai-is-reversing-20-years-of-platform-decentralization-ce15962b6a13 https://medium.com/@sumitbhardwaj1357/the-quiet-re-centralization-of-tech-why-ai-is-reversing-20-years-of-platform-decentralization-ce15962b6a13]

(2) [https://www.mdpi.com/1996-1073/19/1/137 Ginzburg-Ganz, E., Lifshits, P., Machlev, R., Belikov, J., Krieger, Z., & Levron, Y. (2025). Technical challenges of AI data center integration into power grids—A survey. ''Energies'', ''19''(1), 137.]

(3) [https://decai-resources.s3.us-east-1.amazonaws.com/decentralized_ML_perspective-16.pdf Singh, A., Lu, C., Gupta, G., Chopra, A., Blanc, J., Klinghoffer, T., ... & Raskar, R. (2024). A perspective on decentralizing AI.]

(4) The Rise of AI Data Center Models and the Decline of the General-Purpose Data Center. (Circa 2026). Retrieved on 8 September 2026, from SubZero Engineering website : [https://www.subzeroeng.com/the-decline-of-the-general-purpose-data-center/ https://www.subzeroeng.com/the-decline-of-the-general-purpose-data-center/]

(5) [https://git.xpub.nl/onebigear/SI_09_reference_library/raw/commit/22bf61e1dcfed35d58fba23808212bd00584be47/Richard_Barbrook_Andy_Cameron_Californian_ideology.pdf Barbrook, R., & Cameron, A. (1996). The Californian Ideology. ''Science as Culture'', ''6''(1), 44-72].

(6) [https://d1wqtxts1xzle7.cloudfront.net/87708502/langdon_winner_do_artifacts_have_politics-libre.pdf?1655588533=&response-content-disposition=inline%3B+filename%3DDo_Artifacts_Have_Politics.pdf&Expires=1788890463&Signature=UBBiMqkip5MfYZbkbLNAN6aE2hyZdKkER-HFcmyxXkwIZuTnMKdOGoao8ddFr-rQ5T55A8NWUOlnZkGYmi66e7XR9h99DgTELMCrh48sXjOb8KYgA48r~SUUymt1IW~R~aOKJzitMIUHcrc6yEGKK1OuaA~mrEIVOXpUjQMYjqOd26Xmk9YJEglTkqaxeR47VCCth5FvozNQIowdfjzeu8aNHnI8V7Yur1IiXY~PGvYRcN6wBtcgpVMHZUtNaI8NjmYbp2p5TBYrPMF9MCLtvpGhPlsCfsvA8Bdm6XulGwRlksZBKWvVUQ8D~M8oLu12V39YaiETIObyJqbE7ctIcQ__&Key-Pair-Id=APKAJLOHF5GGSLRBV4ZA Winner, L. (2017). Do artifacts have politics?. In ''Computer ethics''(pp. 177-192). Routledge].

(7) [https://sjbae.pbworks.com/f/David+1985.pdf David, P. A. (1985). Clio and the Economics of QWERTY. ''The American economic review'', ''75''(2), 332-337].

== Centralization of AI compute and Data - by Sergey Abramov and Paul Stein ==<div>

=== The current governance issue ===
Advanced AI currently depends on a narrow group of tech giants1 and semiconductor oligopolies controlling microchips and software, hyperscale cloud infrastructure, large datasets, frontier models, and distribution platforms. This presents a governance problem where this narrow group exercises control over foundational digital infrastructure. This directly affects consumers, businesses, and academia out of AI. Legally and institutionally, this matters because it creates a profound democratic deficit: this neutralizes antitrust frameworks designed for traditional markets, often bypassing conventional copyright and data privacy protections through unchecked data harvesting thereby complicating enforcement.This ultimately might affect national sovereignty as the countries that host the buildout will shape the future of AI: who controls it, what values it contains, how it is used2.
 
=== Historical lineage ===
In 1987, Taiwan Semiconductor Manufacturing Company (TSMC) was founded, establishing the world’s first semiconductor fab. Instead of designing and making its own chips, TSMC only manufactured chips designed by other companies. Today, nearly 100% of the world's most advanced AI processors are manufactured by TSMC, leaving the entire global AI compute supply chain centralized under one company.
 In 2006, Nvidia launched CUDA (Compute Unified Device Architecture), a free software layer that allowed developers to use Nvidia's graphics chips (GPUs) which later became perfect for training deep neural networks. Today, this creates a near-monopoly on AI compute hardware.
 Soon after Amazon launched AWS, Microsoft and Google launched their own cloud platforms marking the structural pivot to the cloud service provider model, shifting computing from a distributed architecture (where companies bought and managed their own physical servers) to a centralized infrastructure. As training modern AI models requires tens of thousands of GPUs, today’s frontier AI companies are entirely dependent on the extremely expensive physical infrastructure of a handful of cloud hyper-scalers (Microsoft, Google, AWS, and Meta).
 In 2009, Fei-Fei Li launched ImageNet, a massive database of over 14 million images, which proved that AI model performance scales directly with data volume. This created an immediate institutional advantage for consumer internet platforms (like Google and Meta), because they were already architected to collect, centralize, and store masses of user data through search engines and social networks, they possessed the unique data stack required to train high-performing AI.
 
=== Mechanism of path dependence ===
The centralization of AI compute and data is a classic case of technological lock-in. When structural choices were made decades ago, they triggered powerful, self-reinforcing loops. One such loop is the switching costs and software standardisation. As developers have spent more than a decade working with CUDA, moving to a rival hardware provider (or a decentralized compute network) requires an enterprise to completely rewrite its foundational software stack. The cost of code migration, developer retraining, and potential hardware downtime force companies to stay tied to dominant hardware monopolies such as Nvidia. Commitments, licensing conditions, and exclusivity provisions add contractual costs. Fewer than 1% of cloud customers switch provider annually, identifying technical differentiation and data-transfer charges as important barriers4.
 Network effects also matter for computing power. More customers generate more revenue, usage information, and opportunities to improve infrastructure and models. Better services attract additional users and developers, whose applications make the platform more valuable. In cloud–AI partnerships, providers may additionally obtain information about model performance and future infrastructure demand. These feedback loops strengthen firms already operating across cloud, models, software, and distribution.
 
=== Politics and distribution ===
According to Langdon Winner's definition, modern AI computing infrastructure may qualify as an "inherently political technology" because its large scale and significant financial requirements not only lend themselves to, but necessitate,5 centralized, hierarchical control. The physical siting of this infrastructure shows evidence of Winner’s second category of politicization of technologies, the ‘political arrangement’:6 Of the five largest U.S. AI firms, four are headquartered in California,7 while just one is (partially) domiciled in Texas. Texas, however, hosts twice the GPU clusters.8 This placement geographically insulates beneficiaries of the data infrastructure boom from its negative externalities. The specific choice of location may itself be a deliberate preference on account of Texas’ unique approach to water usage rights.9 Juxtaposition of this unequal and even exploitative balance of interests with the pontificating rhetoric of the urban, information rich, beneficiaries is evocative of the internal contradictions of the Californian Ideology - the virtual artisans existence and privileged position is still based on the exploitation of a vulnerable out-group, the "information-rich" consuming the output, the "information-poor" hosting the extraction.10
 
=== Contingency and Alternatives ===
Profit motive and economies of scale now concentrate investment among a few hyperscalers,11 but this reflects policy choice, not government incapacity: Compute infrastructure may be inherently political, but the technology is agnostic as to where the organizing authority originates, and states have historically funded infrastructure of comparable or greater scale, while facing less pressure to show short-term returns. The clearest turning point was 1995, when the National Science Foundation retired NSFNET, the publicly funded backbone that had connected U.S. research and supercomputing centers since the mid-1980s,12 handing internet infrastructure to commercial providers with no conditions attached. This was a deliberate policy decision made amid the deregulatory push of the Clinton era. The privatization that followed ushered in the private industrial dominance that persists today in the form of hyperscalers. Had a publicly governed backbone instead been scaled up, today's compute infrastructure might rest on a different foundation entirely.
 
=== Governance implication ===
Despite private hyperscaler dominance, the U.S. does have a nominal public alternative: the National AI Research Resource (NAIRR), established following the National AI Initiative Act of 2020 and piloted from January 2024, which explicitly aims to democratize access to the computational resources essential to AI research.13In Europe, a comparable program is the EuroHPCJU,14 an initiative intended to pool resources to build supercomputing capacity. The undertaking’s legislative basis has since been extended to include networked ‘AI factories', as well as the so-called “Gigafactories”.15 The framing in language of sovereignty indicates intent to at least compete with private infrastructure initiatives. Though government-led measures cannot reverse the centralization of compute, they can democratize access - provided they are intensified. The European Commission has “set a goal of investing more than €1 billion per year in AI research and innovation”.16 While laudable in its own right, it pales in comparison to the investments of industry leaders, which are approaching, if not exceeding, trillions of Dollars.

=== References & Notes ===
1.Amazon, Google, Meta, Microsoft, and Oracle collectively hold an estimated 71% of the world’s cumulative AI compute as of Q4 2025, measured in H100-equivalents of computing power. Luke Emberson, Josh You and Venkat Somala (2026), "Five hyperscalers now own over two-thirds of global AI compute". Published online at epoch.ai. Retrieved from 'https://epoch.ai/data-insights/hyperscalers-control-most-compute' [online resource]. Accessed 7 Sept 2026.2. Alasdair Phillips-Robins, Teddy Tawil and Sam Winter-Levy, ‘The Compute Coalition: How to Build the Future of AI in the Free World’ (Carnegie Endowment for International Peace, 8 June 2026) https://carnegieendowment.org/research/2026/06/the-compute-coalition-how-to-build-the-future-of-ai-in-the-free-world accessed 7 September 2026.

3. Amazon Web Services (AWS), a cloud computing platform

4. Competition and Markets Authority, Cloud Infrastructure Services: Final Decision Report (31 July 2025) https://assets.publishing.service.gov.uk/media/688b8891fdde2b8f73469544/final_decision_report.pdf accessed 7 September 2026.

5. Langdon Winner, 'Do Artifacts Have Politics?' (1980) 109(1) ''Daedalus'' 121, 128.

6. Ibid. 123.

7. OpenAI, Anthropic, Google DeepMind, Meta AI. xAI is an exception to the rule, domiciled between Palo Alto, CA and San Antonio, TX.

8. Texas’ 17 to California’s 8, cf. Texas Royalty Brokers, 'AI Clusters by State' (''Texas Royalty Brokers'', 2025) https://texasroyaltybrokers.com/ai-clusters-by-state/ accessed 7 September 2026.

9. The ‘rule of capture’, established in Houston & TC Railway Co v East (Tex 1904), permits landowners to extract groundwater beneath their land without liability for depleting adjacent wells or aquifers, a doctrine now exploited by hyperscale AI data centres to draw large cooling-water volumes with minimal regulatory constraint.

10. Richard Barbrook and Andy Cameron, 'The Californian Ideology' (1996) 6(1) ''Science as Culture'' 44.

11. Goldman Sachs, 'Global AI Investment Is Forecast to Exceed $1 Trillion in 2026' (''Goldman Sachs Insights'', 7 August 2026) https://www.goldmansachs.com/insights/articles/global-investment-is-forecast-to-exceed-1-trillion-in-2026 accessed 7 September 2026.

12.  National Science Foundation, 'Birth of the Commercial Internet' (''NSF Impacts'') https://www.nsf.gov/impacts/internet accessed 7 September 2026.

13. CASRAI, 'NAIRR: National AI Research Resource, Explained' (''CASRAI'') https://casrai.org/dictionary/term/national-ai-research-resource-nairr accessed 7 September 2026.

14. Council Regulation (EU) 2021/1173 of 13 July 2021 on establishing the European High Performance Computing Joint Undertaking and repealing Regulation (EU) 2018/1488 [2021] OJ L256/1.

15. Council Regulation (EU) 2026/150 of 16 January 2026 amending Regulation (EU) 2021/1173 on establishing the European High Performance Computing Joint Undertaking [2026] OJ L 2026/150.

16. European Commission, 'Realising Europe's Digital Decade' (''European Commission'', 11 October 2024) https://commission.europa.eu/strategy-and-policy/priorities-2019-2024/story-von-der-leyen-commission/realising-europes-digital-decade_en accessed 7 September 2026.</div>

Governance by history

2026-09-23T20:07:09Z

Robert Florijn:

[[File:Old internet.jpg|300px|thumb]]

= contributions =
*can
*you 
*please
*create
*individual pages
*from the contributions below
*and link them back here? 
*[[Information Control and Workplace Surveillance - Jihae Park and Mariona Fortuny]]
*[https://wiki.ivir.nl/index.php?title=Centralization_of_Compute_and_Data Centralization of AI Compute and Data - by Sergey Abramov and Paul Stein]
*[[Media:Civilian At The Front.pdf]]

= links things that popped up during class =
*[https://craphound.com/ Cory Doctorow]
*[https://dougengelbart.org/content/view/209/ The mother of All demos] (first mouse, graphical interface, word processor, etc)
*Can a simple sensor be political? [https://gizmodo.com/why-cant-this-soap-dispenser-identify-dark-skin-1797931773 well, this "racist" soap dispenser made headlines a couple of years ago.]
*Podcast series: [https://www.the-santiago-boys.com/ Morozov: the Santiago Boys.]
*Anatomy on AN AI. [[File:Ai-anatomy-map.pdf|check out the pdf here]]

= Workplace surveillance =
== Information Control and Workplace Surveillance - by Jihae Park and Mariona Fortuny Jané ==

==== The Current Governance Issue ====
There are numerous current governance problems arising from technological innovations, yet here we will specifically focus on the information control and workplace surveillance. Whether intended or unintended, there were certain historical incidents which resulted in social and political consequences. As noticed from the Control Revolution, the dramatic evolution in mass media and telecommunications technologies has led to the reinforcement of the economic and political control during the Industrial Revolution. ([https://archive.org/details/beniger-james-the-control-revolution Beniger, 1986]) By applying this logic to a modern example, this is clearly how many corporations are getting in control of their employees through surveillance technologies. Along with the emergence of remote and hybrid work after Covid-19, employers are now relying on “bossware” to analyze the emails, chats, and messages of their employees on platforms such as Teams or Slack. ([https://ojs.library.queensu.ca/index.php/surveillance-and-society/article/view/19100/13263 Ollier-Malaterre et al., 2026, p.237]) Some companies have strict restrictions on transferring data both internally and externally in protection of confidential information and consistently monitor their workers to the certain extent which could slow down their efficiency. This is problematic as it’s not only violating the privacy of their employees but also depriving employees of rights they should be naturally be able to exercise.



==== Historical Lineage ====
In relation to the current issue of the information control and workplace surveillance, Max Weber’s argument of bureaucracy and rationalization in the middle and late nineteenth century are good historical examples which are still frequently visible nowadays. Weber stated how the range of control could extend not only by expanding the capability to process information but also by limiting the amount of information to be processed. Here, the former is directly related to the control implemented by bureaucratization (in the mid 19th century) and computerization (in the late 20th century), whereas the latter addresses rationalization. An example from bureaucracy is the expansion of paperwork in response to the growth in information. We can still note these traces of bureaucracy from numerous procedures that we have to go through for a decision-making to take place in corporations. He brings up the example of standardized time zones for rationalization which is simply an example for an omission or a breakdown of information to facilitate its processing. ([https://archive.org/details/beniger-james-the-control-revolution Beniger, 1986, p.15-16]) This is also commonly seen from nowadays as some companies may use internal terminologies for the sake of both confidentiality and efficiency.

==== Mechanism of Path Dependence ====Following along the line with the historical lineage, there are certain mechanisms at work of path dependence for information control and workplace surveillance. Bureaucracy is closely related with organizational routines. Many companies have already settled down with a fixed workflow and decision-making process, which makes it more costly for one to change from its conventional way of working. This means switching costs from an old system to a new one is clearly high considering its time and capital, not necessarily will guarantee increasing returns. Hence, many firms end up maintaining its old strategy, reluctant to make any modifications or changes. 

==== Politics and Distribution ====

As businesses grew too large for managers to oversee everything directly, companies replaced human judgment with formal rule-systems, as we see explained by Beniger's 'intellectual technology' ([https://archive.org/details/beniger-james-the-control-revolution/page/65/mode/2up Beniger, 1986, pp. 17, 21]). Algorithmic management continues this: a call once made by a supervisor's judgment is now a rule built into software that runs without one, extending a logic Beniger traces as far back as to Taylorism and the so called 'assembly line'. Winner ([https://www.jstor.org/stable/20024652 1980]) asks whether such systems must concentrate power this way or were driven there, when other designs were equally possible; nothing required shutting workers out of the decision, so leaving them out was a political choice, and not necessarily a technical necessity. Barbrook and Cameron ([https://doi.org/10.1080/09505439609526455 1996]) explain the reasoning why some can defend this 'domination': they show how 1990s tech culture sold new technologies as freedom even as it tightened control, so the rhetoric could become one of closer oversight as flexibility and autonomy. Together, authority accrues to employers, who in some systems can set and revise these oversight rules unilaterally and face less resistance because the arrangement is framed as freedom; workers bear the cost, losing the informal give-and-take they had with a human supervisor for oversight that is harder to see, question, or contest, while being told it grants them more freedom, not less.

==== Contingency and Alternatives ====

A plausible alternative happened on U.S. post WWII: in some workplaces, labor and its pacing was demanded to be set through collective bargaining and enforced jointly, by both management and union stewards monitoring compliance on the line ([https://doi.org/10.1093/icc/dtaf038 Breznitz et al., 2025, p. 211]). So, in this context, workplace surveillance could have developed as a bargained, dual-administered practice rather than a management monopoly. The 1950 [https://en.wikipedia.org/wiki/Reuther%27s_Treaty_of_Detroit Reuther's Treaty of Detroit ] cut that possibility off: unions won guaranteed wages and benefits in exchange for surrendering any general claim over how work was supervised ([https://doi.org/10.1093/icc/dtaf038 Breznitz et al., 2025, pp. 210–211]). Workers depended on wages to survive, and employers controlled the wage relationship. That asymmetry meant employers, given their position of power, were able to arrange things so that, at the very least, workers would have far less capacity to sustain a fight over control itself. Thus, the "trade" wasn't a fair exchange between equals. This reflects which side could afford to hold out on which issue. Hypothetically speaking, had that joint administration model prevailed instead, this could have greatly impacted norms internationally and good practices with worker-union agreed surveillance would have become a widely accepted standard practice.

==== Governance Implication ====

An interesting proposition for present-day governance that wants to fill a gap in workplace surveillance is what [https://doi.org/10.1016/j.digbus.2025.100120 Kayas, Ong and Belal (2025)] are proposing: They argue that EU law, through the [https://eur-lex.europa.eu/eli/reg/2016/679/oj/eng GDPR] and its required [https://en.wikipedia.org/wiki/Data_protection_impact_assessment data protection impact assessments (DPIAs)], already offers meaningful protection for workers in terms of lawful high-risk data collection and data processing. Nevertheless, these DPIAs are built around data and privacy, so they leave out several dimensions of monitoring itself: its purpose, how intrusive it is, how visible it is to the people being watched, and its effects on workers' autonomy, trust, and standing relative to management. These authors propose a "surveillance impact assessment" scoring a system against these characteristics directly, before any surveillance system is put in place. Although this mechanism "is constrained by its reliance on existing literature and the lack of empirical validation," it feels like a genuine alternative for two reasons: 1. it targets the actual mechanism at issue, and 2. at the same time it builds upon a mechanism of active participatory approach where employees have a direct say in how surveillance is designed ([https://doi.org/10.1016/j.digbus.2025.100120 Kayas et al., 2025, p. 11]), which, combined with the actual assessment, would put workers' own judgment at the centre of the one decision that affects them. It's important to note that this alternative is proposed as worth piloting, not as a proven fix, and it is fair to ask whether another mandatory assessment stacked on top of DPIAs would function as intended or simply read to workers as more compliance paperwork.

==== References ====
 Barbrook, R., & Cameron, A. (1996). The Californian Ideology. ''Science as Culture'', 6(1), 44–72. [https://doi.org/10.1080/09505439609526455 https://doi.org/10.1080/09505439609526455] Beniger, J. R. (1986). ''The Control Revolution: Technological and Economic Origins of the Information Society''. Harvard University Press. [https://archive.org/details/beniger-james-the-control-revolution https://archive.org/details/beniger-james-the-control-revolution]  Breznitz, D., Levy, K., Lipartito, K., & Zehavi, A. (2025). An equity-focused research agenda for workplace surveillance. ''Industrial and Corporate Change'', 35(1), 206–221. [https://doi.org/10.1093/icc/dtaf038 https://doi.org/10.1093/icc/dtaf038] Kayas, O. G., Ong, C. E., & Belal, H. M. (2025). From humans to algorithms: A sociotechnical framework of workplace surveillance. ''Digital Business'', 5, 100120. [https://doi.org/10.1016/j.digbus.2025.100120 https://doi.org/10.1016/j.digbus.2025.100120] Ollier-Malaterre, A., Parent-Rocheleau, X., Benkrid, K., & Provost Savard, Y. (2026). Making Sense of Workplace Surveillance: How Workers and Union Representatives Experience, Imagine, and Resist Surveillance Exposure. ''Surveillance & Society'', 24(3), 236–254. [https://ojs.library.queensu.ca/index.php/surveillance-and-society/article/view/19100/13263 https://ojs.library.queensu.ca/index.php/surveillance-and-society/article/view/19100/13263] Winner, L. (1980). Do Artifacts Have Politics? ''Daedalus'', 109(1), 121–136. [https://www.jstor.org/stable/20024652 https://www.jstor.org/stable/20024652]

 
== [[Algorithmic management and workplace surveillance - by Sebastián Muller and Kate Davey''' ''']] ==

=== 1. The current issue. ===
 Algorithmic management and workplace surveillance can be defined as the use of computer-programmed procedures to coordinate labor input in an organization ([https://joint-research-centre.ec.europa.eu/scientific-activities/employment/algorithmic-management-and-digital-monitoring-work_en 1]). Simply put, it is a way to control workers, and workers generally have no say in how the system that tracks them is created. As Winner says "different people are differently situated and possess unequal degrees of power as well as unequal levels of awareness" ([https://www.jstor.org/stable/20024652?seq=7 2]).
The population affected is workers; both “white collar” and “blue collar” workers are impacted. Whether it’s Amazon tracking warehouse workers’ “Time Off Task” with handheld scanners ([https://www.wired.com/story/amazon-worker-tracking-details-revealed 3]) or Meta employees having their every click tracked ([https://www.techpolicy.press/metas-worker-surveillance-tests-eu-rules-on-ai-and-labor/ 4]) to train AI, employers have the power to watch and control their employees in an unprecedented way.
It matters legally because workers have very few rights to fight back against these surveillance programs; even the The Electronic Communications Privacy Act of 1986 allows exceptions to surveillance of employees ([https://bja.ojp.gov/program/it/privacy-civil-liberties/authorities/statutes/1285 5]). 
=== 2. Historical lineage. === 2.1 Panopticon: it is an architectural apparatus developed by Jeremy Bentham in 1791. The design of this building effectively generated the sense of constant monitoring, which in turn motivated the prisoners to behave as if they were being watched at all times. As human labor became commodified, capitalists began implementing measures to ensure the maximization of productivity, including the replication of this design ([https://doi.org/10.14452/MR-070-09-2019-02_2 6]). While unused today, the panopticon is widely considered as both a theoretical and practical inspiration of other forms of surveillance in the workplace.
 
2.2 Scientific management: this theory attempts to analyze and standardize workflows to improve labor productivity. Frederick Winslow Taylor, its main pioneer, used early forms of data analysis during the XX century to decrease walking distances, eliminating unnecessary movements and preventing deliberate slow-walking among workers. ([https://doi.org/10.14452/MR-070-09-2019-02_2 7]). There are many modern forms of data-driven decision-making processes in the workforce. For instance, Amazon warehouses track individual workers' scan rates in real time through handheld scanners, generating data feeds to track and evaluate performance ([https://www.theverge.com/2019/4/25/18516004/amazon-warehouse-fulfillment-centers-productivity-firing-terminations 8]).
 
2.3 PC in the workplace: John Blankenbaker designed the first commercially available personal computer in 1971 ([https://www.computerhistory.org/revolution/personal-computers/17/297 9]), however, it wasn’t until the mid-1980s that the use of personal computers in the workplace became an industry standard. This achievement gave rise to the digital monitoring of work, which in turn allowed employers to monitor employees in real-time, effectively replacing the human eye with a more efficient surveillance mechanism. This represents one of the key aspects of modern-day workplace monitoring. 
 

=== 3. Mechanism of path dependence. ===
''' '''When analyzing the historic lineage of work surveillance, it is possible to point out two main path dependencies mechanisms in place.Firstly, it is clear that many of the technological developments that arose through these events have one major goal, which is to increase wealth within a hierarchical ecosystem. Employers resort to several mechanisms to maximize profits and cut inefficiencies at the expense of workers. 
In addition, standardization can also be singled out as a key mechanism at play. Once one corporation implements any of these developments (such as the digital monitoring of the workplace), the market incentives would force the rest to do the same in order to be profitable and remain competitive.

=== 4. Politics and distribution. ===
 One simple answer to this question would be that capitalists implement and leverage these mechanisms to increase their wealth while the working class remains powerless and unable to withdraw itself from the system in force. However, if we resort to Winner's framework, we see that the technical elements of these technologies are not intrinsically capitalist, since workplace surveillance can also be observed in alternative regimes (including chattel slavery and the Soviet regime). It seems that, rather than being exclusively tied to capitalistic values, the methods in question respond to strong hierarchical structures.
In addition, even under a libertarian framework, structural inequalities are still present, allowing a privileged class to enjoy individual freedoms while segregated groups keep performing low-income jobs ([https://www.metamute.org/editorial/articles/californian-ideology 10]), arguably still subject to surveillance. Thus, not only are these technologies compatible with hierarchical top-down forms of government, but also with structures that allow such hierarchies to form without state intervention.
That said, this form of management seems less compatible with democratic societies, which tend not only to value individual liberties but to turn them into enforceable rights (including workers' privacy, health and integrity), thus establishing boundaries for workplace surveillance. This could explain why studies of American business leaders have found them unhappy with democratic scruples ([https://www.jstor.org/stable/20024652 11]).
 
=== 5. Contingency and alternatives. === 
Bush writes about the memex, a machine that could extend a person’s memory and that the memory would belong to the person who owned it. As Bush expands, “For mature thought there is no mechanical substitute" ([https://dl.acm.org/doi/pdf/10.1145/227181.227186 12]). His idea was that a machine could take over the repetitive nature of someone’s job, while the person could record their own thoughts for their own use. Bush’s memex was never created, but the idea inspired products like Microsoft’s Productivity Score that were made for employers (the buyers) not for the person whose memory it used (the worker).
This immediately brings to mind Meta's Model Capability Initiative, which was done to train AI agents the company hopes to sell to other employers as well as watch their employees ([https://www.reuters.com/sustainability/boards-policy-regulation/meta-start-capturing-employee-mouse-movements-keystrokes-ai-training-data-2026-04-21/ 13]). If this was Bush’s memex, it would belong to the employee using it, not Meta. 

=== 6. Governance implication. ===
Democratic oversight of the use of algorithms as a form of surveillance is essential to ensure that workers’ rights are not overstepped in the name of efficiency. As such we propose: all workers should have a legal right to be consulted and approve a surveillance program before a company installs and uses it. A tracking system should be a group decision, not an inevitable, unilateral one. While the United States has very little protections for workers in terms of surveillance, the EU leads the way with the GDPR, specifically Article 22 ([http://data.europa.eu/eli/reg/2016/679/oj 14]), and the AI Act Article 5 ([http://data.europa.eu/eli/reg/2024/1689/oj 15]). It’s also worth noting that in December 2026, the Platform Work Directive ([https://eur-lex.europa.eu/eli/dir/2024/2831/oj/eng 16]) will come into effect which will ban automated termination.

= [[Platform Dependence in Public Communication (News) - Ceren Isiklar & Rodrigue Kayembe|Platform Dependence in Public Communication (News)]] Ceren Isiklar & Rodrigue Kayembe =

= Centralization of AI Compute and Data =

== Centralization of AI Compute and Data - by Sisi Hammoudi & Juliette Chamot ==
=== Governance issue ===
Today, several path-dependence phenomena are emerging across advanced computing technologies. This assignment addresses one of them: the centralisation of AI compute and data. Training and running frontier AI systems now depends on a handful of hyperscale providers, namely Amazon, Microsoft, Google, and Nvidia, which dominates the GPU market. In other words, near monopolies exist throughout the supply chain. This concentration creates a governance problem affecting public and private actors alike, both legally and institutionally. On the institutional side, it prevents public administrations, and states other than the US or China, from reaching sovereign AI capacity. It also raises competition-law concerns, since dominant market positions can prevent the market from delivering the best products at the best prices to consumers. Finally, AI developers, like states, are often forced into contracts with hyperscale providers, creating a structural dependence on them.


=== Historical lineage ===
==== 1945 ====
Wartime scientific instruments should have been redirected to extend human command over knowledge, helping establish the postwar model of sustained federal funding for computing research (Vannevar Bush's, "As We May Think », 1945).


==== 1990s ====
The privatisation of the NSFNET established the idea that critical digital infrastructure, even though publicly subsidised, could legitimately be privately owned. The EU Commission followed the US's lead, recommending privatisation and deregulation of telecommunications and media as the best way to build the "infobahn" though EU members could instead have chosen nationalisation of the infrastructure (Barbrook & Cameron, "The Californian Ideology," 1996).


==== 2006 ====
<div class="s13">Hyperscale cloud computing turned massive fixed capital investment in data centresinto a rentable service, favouring firms able to invest at scale.</div>
<div class="s13">Same period: Nvidia's proprietary CUDA software, released alongside its GPUs, became the de facto standard for machine learning, locking in a decade of tooling and expertise around one firm's hardware.</div>
<div class="s13">
==== From 2022 ====
Semiconductor export controls and the CHIPS Act turned compute into an explicit instrument of geopolitical strategy.</div>



=== Mechanisms of path dependance ===
Several mechanisms drive this centralisation, but path-dependence ones dominate. Because AI compute is a capital-intensive industry, new entrants struggle to compete with hyperscalers that have already sunk huge sums into infrastructure, a barrier that produces monopoly-like situations. This dynamic, known as increasing returns, generates system-scale economies that harden into a near-irreversibility of investment (David, "Clio and the Economics of QWERTY," 1985). It overlaps with a second mechanism, standardisation, itself a form of technical interrelatedness: researchers' investment in Nvidia's CUDA-based tooling locks the field into that ecosystem regardless of whether better alternatives exist. A further mechanism is legal and contractual entrenchment. For example, Microsoft's stake in OpenAI binds developers to a specific provider by agreement rather than technical necessity.=== Politics and distribution ===Technology does not have a history of distributing power equally. Rather, its development has strengthened the interests and authority of powerful groups that already possessed economic and political resources, and it is yet to be determined whether AI compute and data will follow suit. 
Author Langdon Winner illustrates the historical constraints of technological development with the case of New York from the 1920s to the 1970s. He argued that the city’s roads, bridges, and parks’ physical arrangement contained “explicit and implicit political purpose.” These infrastructures privileged wealthy and white communities while restricting poor and Black populations’ choice of mobility. Similarly, the development of industrial machinery increased productivity, simultaneously weakening union and non-union workers’ bargaining power.
In the twentieth century, following the Industrial Revolution, a Control Revolution emerged from the increasing need to manage expanding production, transportation, and information flows requiring strengthened organizational bureaucratic control. This was heavily shaped by the political context of post-World War II warfare.
Therefore, the bearer of the initial cost of technology is the person putting in the capital to erect it, but ultimately, the person bearing the long-term cost of technology, whose choices diminish, is not the rich and privileged, but the everyday person subject to technologies. Forced to adapt, to reshape their professional skills to adapt to an environment trying to make them obsolete.
AI compute and data are at risk of creating the same effect. In their critique of dotcom neoliberalism, authors Barbrook and Cameron exposed a similar socio-economic tension. The Californian Ideology claims digital technology empowers individuals; however, perpetual unequal access to it creates an apartheid between “information-rich” and “information-poor” populations.
Finally, the development of AI compute and data depends on expensive and scarce computing infrastructure and a highly concentrated semiconductor supply chain. Today, this is one of the most concentrated industries, with single companies having near-total monopolies over the production of chips necessary for the functioning of AI. This monopoly model threatens to reinforce the existing economic hierarchy.

=== Contingency and alternatives ===If a different technological trajectory was plausible, the most apt time for it to have occurred would be during the transition from the Industrial Revolution to the “Control Revolution”. When industrial production accelerated, our society experienced a “crisis of control”, creating opportunities to organize information and communication differently.
Instead of allowing control technologies to be monopolized within businesses and states, our societies could have pursued a more democratic and decentralized system. Barbrook and Cameron provide such an alternative path through their idea of the “electronic agora,” where digital networks could encourage community participation and direct democracy, where government could actively engage with the people to shape digital infrastructure and expand access. These alternatives did not actually prevail because corporate and state market participants possessed greater resources and authority to shape technological development at the end of the twentieth century. 
Inherently, technological development is not predetermined; it can support either authoritarian or democratic arrangements. Today, monopolies control chip production needed for AI compute and data, but other technologies like blockchain permit democratic participation.

=== Governance implication ===History always threatens to repeat itself. As Barbrook discussed, digital societies are shaped by the political context in which they develop. It is a valid concern, then, that the development of information technology, through AI computing, would follow the same trajectory- that is, be constrained to benefit the wealthy who funded its development, and ultimately harm the middle and lower classes enabling its functioning. 
However, AI’s place in our society has not yet crystallized. Technology governance does not have to adhere to this system. Such a seemingly sustainable approach to AI compute and data governance has already been broached. In 2022, the Notre Dame Journal on Emerging Technologies published an article in its issue on Artificial Intelligence where the authors proposed an AI compute system based on a “dual investment strategy”. Firstly, the model would be launched by “subsidizing and negotiating” cloud computing for human AI researchers through “existing” cloud vendors, which would expand on existing AI compute initiatives. Secondly, the compute system would be specifically invested in as a publicly owned infrastructure that would be built and operate similarly to national libraries which own “academic supercomputing facilities”. 
The pilot program would have a “base level of compute access” which would keep costs of processing down and enable it to meet the compute needs of a majority of academic researchers. Absent any strengthened legislation on the use of AI compute and data, the hardware would be accompanied by an AI ethics framework to enable sustainable technological governance. In addition to an ethical staff, the human resources needed to keep an AI compute model running would include data officers, scientists, data center and software engineers, research developers, and system administrators. These positions require specialized training, allowing new types of professionals, from any background, to have access to professional prosperity in the field of AI. 
By implementing these frameworks, never neglecting the human element, and the economic and intellectual impact AI has on our society, we move from a historical determinist perspective to one acknowledging historical constraints and learning from them.


=== Bibliography ===
<div><div>[1] Langdon Winner, Do Artifacts Have Politics, Daedalus, Winter, 1980, Vol. 109, No. 1, Modern Technology: Problem or Opportunity? pp. 121-136, 124, (1980)</div>
<div>[2] Langdon Winner, Do Artifacts Have Politics, Daedalus, Winter, 1980, Vol. 109, No. 1, Modern Technology: Problem or Opportunity? pp. 121-136, 124, (1980)</div>
<div>[3] James R. Beniger, The Control Revolution: Technological and Economic Origins of the Information Society, Harvard University Press, p. 443, 450 (1986)</div>
<div>[4] Richard Barbrook, Andy Cameron, The Californian Ideology, ''Science as Culture'', ''6''(1), 44–72, 61 (1996)</div>
<div>[5] Anton Korinek, Jai Vipra, Concentrating Intelligence: Scaling and Market Structure in Artificial Intelligence, National Bureau of Economic Research, 10.3386/w33139, p. 11, 13-15 (2024)</div></div>
<div><div>[6] James R. Beniger, The Control Revolution: Technological and Economic Origins of the Information Society, Harvard University Press, p. 443, 442 (1986)</div>
<div>[7] Richard Barbrook, Andy Cameron, The Californian Ideology, ''Science as Culture'', ''6''(1), 44–72 (1996)</div>
<div>[8] Langdon Winner, Do Artifacts Have Politics, Daedalus, Winter, 1980, Vol. 109, No. 1, Modern Technology: Problem or Opportunity? pp. 121-136, 130 (1980)</div>
<div>[9] Daniel E. Ho, Jennifer King, Russell C. Wald & Christopher Wan, The Centrality of Data and Compute for AI Innovation: A Blueprint for the National Research Cloud, 3 Notre Dame J. on Emerging Tech. 71, 81(2022)</div>
<div>[10] Daniel E. Ho, Jennifer King, Russell C. Wald & Christopher Wan, The Centrality of Data and Compute for AI Innovation: A Blueprint for the National Research Cloud, 3 Notre Dame J. on Emerging Tech. 71, 82 (2022)</div></div>

== Centralization of AI compute and Data - by Atmika Dechamma & Pablo Latorre ==
=== Governance Issue ===
AI computing capacity and data presents the governance problem of an extreme concentration of technological infrastructure and information in a restricted number of companies that privately and globally dominate the market. This affects public and private actors - such as small and medium enterprises, potential new competitors, and Global South countries - that increasingly depend on AI systems outside their control and/or are deterred from competing with established companies because of impenetrable entry barriers - most notably initial investments and operational costs (1). The institutional and legal relevance of this is that it consolidates, reproduces and amplifies asymmetries of power in favor of the “Big Tech”, reverting the technological trend from digital decentralization to physical infrastructure dependency and curbing innovation and market efficiency.

=== Historical Lineage ===
==== Limitation of Machine Learning ====The AI revolution was built on the limitation of human capabilities to process large information and derive efficient results - where before there existed only facilitative tools to allow for human-enabled processes, AI now offered derivations and cognitive reasoning in moments and at large scales.

 ==== Revert to Physical Infrastructure Dependency in the rush to be the best ====The present AI model heavily relies on hardware in order to perform complex data processing for deep learning. The data centre models followed require larger model sizes (which require higher resource, information and capital investment) to be able to yield better results. Due to the limitation of resources and a rush to establish the best working model of an AI agent, data centres are built on a rudimentary model heavily dependent on massive amounts of electricity and fresh water (2).

 ==== Regulation and Decentralization of Information ====Regulation and influence of sovereigns and institutions has resulted in data sharing and information exchange structurally through open source and cloud computing. But, while AI models claims their selling point to be accessibility for all, their information and derivations remain in a closed server, stored for the purpose of training their AI agents to be the best (3). Big AI contenders attract higher consumer and information flows overall resulting in a better developed AI model, proving disadvantageous in any market where early contenders have an advantage and accumulation of power.
=== Mechanisms of Path Dependence ===Early on in the industry there is an expectation of human and capital investment from all stakeholders involved, this high cost for entry into the market and expected capital to sustain and develop a model by both consumer and enterprise makes for a high switching cost. For the AI model to run and scale there is a presumption that the rate of growth is dependent on the size of data fed, resulting in a better model and hence, more consumer usage. A closed centralised data loop is created but the requirement for storage and processing far surpasses current infrastructures capabilities - heavily relying on early market entry advantage and sustained capital investments taking away from smaller aspirants and hoarding of power within closed circles. Despite its physical dependency, AI data processing has unique requirements beyond legacy infrastructures (4) - larger demand and production of GPUs and designed to handle immense thermal heat generated while consuming electricity beyond current capabilities. 

 The AI space largely remains unregulated due to its undefined form and fluid nature, however, this has given way for unethical and morality issues regarding data privacy, exploitation of labour and environmental harms.

=== Politics and Distribution ===
Concentration strengthens the values, interests, capacities, and authority of the “Big Tech” over other actors of public and private nature. In this context, Barbrook and Cameron (5) illuminate the ethos behind AI technology development while Winner (6) makes observable the power relations surrounding it: concentration is the consequence of an AI development anchored in the belief that it must be free from all institutional and normative restriction since it is inherently liberating and non-political which, in turn, provides a political/moral justification for the concentration (if technology is inherently liberating and non-political then there is no power relations to observe or question) that translates itself into power dynamics where the AI developer not only owns the AI technology but also decides how it is trained and deployed and which models the user can or cannot access.

=== Contingency and Alternatives ===
Concentration is due, first and foremost, to the fact that AI computing capacity requires massive and cutting-edge physical infrastructure - such as data centers and microprocessors - which, in turn, requires upfront investments that are unaffordable to most companies. In David’s (7) language, companies that could finance the infrastructure and position themselves first obtained self-reinforcing advantages and a “lock-in” effect in the market. In other words, concentration is the consequence of the development of AI physical infrastructure privately with no mixed or public alternatives as a public service. Therefore, there was a space for an alternative future at the end of the 1980s or the beginning of the 1990s when the privatization model of the technological infrastructure was consolidated in line with the “Californian Ideology”. In that moment governments had the chance to treat technological infrastructure as a mixed or public good rendering it universally accessible as it was the case of the Internet; this would have permitted companies of all sizes to profit from economies of scale, reducing entry barriers and fostering competition. However, this alternative path did not prevail since the Californian Ideology provided the moral/political justification to exclude governments from the development of technological infrastructure in the exact moment in which such a decision would produce structural consequences. If that path had been followed it could have contributed to collective access to economies of scale, lower entry barriers, and a more competitive AI market.

=== Governance Implication ===
Considering the above, public or mixed investment on AI physical infrastructure is a necessary governance measure for market decentralization and collective access to AI computing capacity and data. From an historical perspective, the proposed measure is not merely a corrective action for a market failure but for a contingent and ideological decision made between the late 1980s and early 1990s that produced a structural “path dependence” that now is necessary to change.
==== References ====
(1) The Quiet Re-Centralization of Tech: Why AI Is Reversing 20 Years of Platform Decentralization. (2026, January 10). Retrieved on 8 September 2026, from Medium website : [https://medium.com/@sumitbhardwaj1357/the-quiet-re-centralization-of-tech-why-ai-is-reversing-20-years-of-platform-decentralization-ce15962b6a13 https://medium.com/@sumitbhardwaj1357/the-quiet-re-centralization-of-tech-why-ai-is-reversing-20-years-of-platform-decentralization-ce15962b6a13]

(2) [https://www.mdpi.com/1996-1073/19/1/137 Ginzburg-Ganz, E., Lifshits, P., Machlev, R., Belikov, J., Krieger, Z., & Levron, Y. (2025). Technical challenges of AI data center integration into power grids—A survey. ''Energies'', ''19''(1), 137.]

(3) [https://decai-resources.s3.us-east-1.amazonaws.com/decentralized_ML_perspective-16.pdf Singh, A., Lu, C., Gupta, G., Chopra, A., Blanc, J., Klinghoffer, T., ... & Raskar, R. (2024). A perspective on decentralizing AI.]

(4) The Rise of AI Data Center Models and the Decline of the General-Purpose Data Center. (Circa 2026). Retrieved on 8 September 2026, from SubZero Engineering website : [https://www.subzeroeng.com/the-decline-of-the-general-purpose-data-center/ https://www.subzeroeng.com/the-decline-of-the-general-purpose-data-center/]

(5) [https://git.xpub.nl/onebigear/SI_09_reference_library/raw/commit/22bf61e1dcfed35d58fba23808212bd00584be47/Richard_Barbrook_Andy_Cameron_Californian_ideology.pdf Barbrook, R., & Cameron, A. (1996). The Californian Ideology. ''Science as Culture'', ''6''(1), 44-72].

(6) [https://d1wqtxts1xzle7.cloudfront.net/87708502/langdon_winner_do_artifacts_have_politics-libre.pdf?1655588533=&response-content-disposition=inline%3B+filename%3DDo_Artifacts_Have_Politics.pdf&Expires=1788890463&Signature=UBBiMqkip5MfYZbkbLNAN6aE2hyZdKkER-HFcmyxXkwIZuTnMKdOGoao8ddFr-rQ5T55A8NWUOlnZkGYmi66e7XR9h99DgTELMCrh48sXjOb8KYgA48r~SUUymt1IW~R~aOKJzitMIUHcrc6yEGKK1OuaA~mrEIVOXpUjQMYjqOd26Xmk9YJEglTkqaxeR47VCCth5FvozNQIowdfjzeu8aNHnI8V7Yur1IiXY~PGvYRcN6wBtcgpVMHZUtNaI8NjmYbp2p5TBYrPMF9MCLtvpGhPlsCfsvA8Bdm6XulGwRlksZBKWvVUQ8D~M8oLu12V39YaiETIObyJqbE7ctIcQ__&Key-Pair-Id=APKAJLOHF5GGSLRBV4ZA Winner, L. (2017). Do artifacts have politics?. In ''Computer ethics''(pp. 177-192). Routledge].

(7) [https://sjbae.pbworks.com/f/David+1985.pdf David, P. A. (1985). Clio and the Economics of QWERTY. ''The American economic review'', ''75''(2), 332-337].

== Centralization of AI compute and Data - by Sergey Abramov and Paul Stein ==<div>

=== The current governance issue ===
Advanced AI currently depends on a narrow group of tech giants1 and semiconductor oligopolies controlling microchips and software, hyperscale cloud infrastructure, large datasets, frontier models, and distribution platforms. This presents a governance problem where this narrow group exercises control over foundational digital infrastructure. This directly affects consumers, businesses, and academia out of AI. Legally and institutionally, this matters because it creates a profound democratic deficit: this neutralizes antitrust frameworks designed for traditional markets, often bypassing conventional copyright and data privacy protections through unchecked data harvesting thereby complicating enforcement.This ultimately might affect national sovereignty as the countries that host the buildout will shape the future of AI: who controls it, what values it contains, how it is used2.
 
=== Historical lineage ===
In 1987, Taiwan Semiconductor Manufacturing Company (TSMC) was founded, establishing the world’s first semiconductor fab. Instead of designing and making its own chips, TSMC only manufactured chips designed by other companies. Today, nearly 100% of the world's most advanced AI processors are manufactured by TSMC, leaving the entire global AI compute supply chain centralized under one company.
 In 2006, Nvidia launched CUDA (Compute Unified Device Architecture), a free software layer that allowed developers to use Nvidia's graphics chips (GPUs) which later became perfect for training deep neural networks. Today, this creates a near-monopoly on AI compute hardware.
 Soon after Amazon launched AWS, Microsoft and Google launched their own cloud platforms marking the structural pivot to the cloud service provider model, shifting computing from a distributed architecture (where companies bought and managed their own physical servers) to a centralized infrastructure. As training modern AI models requires tens of thousands of GPUs, today’s frontier AI companies are entirely dependent on the extremely expensive physical infrastructure of a handful of cloud hyper-scalers (Microsoft, Google, AWS, and Meta).
 In 2009, Fei-Fei Li launched ImageNet, a massive database of over 14 million images, which proved that AI model performance scales directly with data volume. This created an immediate institutional advantage for consumer internet platforms (like Google and Meta), because they were already architected to collect, centralize, and store masses of user data through search engines and social networks, they possessed the unique data stack required to train high-performing AI.
 
=== Mechanism of path dependence ===
The centralization of AI compute and data is a classic case of technological lock-in. When structural choices were made decades ago, they triggered powerful, self-reinforcing loops. One such loop is the switching costs and software standardisation. As developers have spent more than a decade working with CUDA, moving to a rival hardware provider (or a decentralized compute network) requires an enterprise to completely rewrite its foundational software stack. The cost of code migration, developer retraining, and potential hardware downtime force companies to stay tied to dominant hardware monopolies such as Nvidia. Commitments, licensing conditions, and exclusivity provisions add contractual costs. Fewer than 1% of cloud customers switch provider annually, identifying technical differentiation and data-transfer charges as important barriers4.
 Network effects also matter for computing power. More customers generate more revenue, usage information, and opportunities to improve infrastructure and models. Better services attract additional users and developers, whose applications make the platform more valuable. In cloud–AI partnerships, providers may additionally obtain information about model performance and future infrastructure demand. These feedback loops strengthen firms already operating across cloud, models, software, and distribution.
 
=== Politics and distribution ===
According to Langdon Winner's definition, modern AI computing infrastructure may qualify as an "inherently political technology" because its large scale and significant financial requirements not only lend themselves to, but necessitate,5 centralized, hierarchical control. The physical siting of this infrastructure shows evidence of Winner’s second category of politicization of technologies, the ‘political arrangement’:6 Of the five largest U.S. AI firms, four are headquartered in California,7 while just one is (partially) domiciled in Texas. Texas, however, hosts twice the GPU clusters.8 This placement geographically insulates beneficiaries of the data infrastructure boom from its negative externalities. The specific choice of location may itself be a deliberate preference on account of Texas’ unique approach to water usage rights.9 Juxtaposition of this unequal and even exploitative balance of interests with the pontificating rhetoric of the urban, information rich, beneficiaries is evocative of the internal contradictions of the Californian Ideology - the virtual artisans existence and privileged position is still based on the exploitation of a vulnerable out-group, the "information-rich" consuming the output, the "information-poor" hosting the extraction.10
 
=== Contingency and Alternatives ===
Profit motive and economies of scale now concentrate investment among a few hyperscalers,11 but this reflects policy choice, not government incapacity: Compute infrastructure may be inherently political, but the technology is agnostic as to where the organizing authority originates, and states have historically funded infrastructure of comparable or greater scale, while facing less pressure to show short-term returns. The clearest turning point was 1995, when the National Science Foundation retired NSFNET, the publicly funded backbone that had connected U.S. research and supercomputing centers since the mid-1980s,12 handing internet infrastructure to commercial providers with no conditions attached. This was a deliberate policy decision made amid the deregulatory push of the Clinton era. The privatization that followed ushered in the private industrial dominance that persists today in the form of hyperscalers. Had a publicly governed backbone instead been scaled up, today's compute infrastructure might rest on a different foundation entirely.
 
=== Governance implication ===
Despite private hyperscaler dominance, the U.S. does have a nominal public alternative: the National AI Research Resource (NAIRR), established following the National AI Initiative Act of 2020 and piloted from January 2024, which explicitly aims to democratize access to the computational resources essential to AI research.13In Europe, a comparable program is the EuroHPCJU,14 an initiative intended to pool resources to build supercomputing capacity. The undertaking’s legislative basis has since been extended to include networked ‘AI factories', as well as the so-called “Gigafactories”.15 The framing in language of sovereignty indicates intent to at least compete with private infrastructure initiatives. Though government-led measures cannot reverse the centralization of compute, they can democratize access - provided they are intensified. The European Commission has “set a goal of investing more than €1 billion per year in AI research and innovation”.16 While laudable in its own right, it pales in comparison to the investments of industry leaders, which are approaching, if not exceeding, trillions of Dollars.

=== References & Notes ===
1.Amazon, Google, Meta, Microsoft, and Oracle collectively hold an estimated 71% of the world’s cumulative AI compute as of Q4 2025, measured in H100-equivalents of computing power. Luke Emberson, Josh You and Venkat Somala (2026), "Five hyperscalers now own over two-thirds of global AI compute". Published online at epoch.ai. Retrieved from 'https://epoch.ai/data-insights/hyperscalers-control-most-compute' [online resource]. Accessed 7 Sept 2026.2. Alasdair Phillips-Robins, Teddy Tawil and Sam Winter-Levy, ‘The Compute Coalition: How to Build the Future of AI in the Free World’ (Carnegie Endowment for International Peace, 8 June 2026) https://carnegieendowment.org/research/2026/06/the-compute-coalition-how-to-build-the-future-of-ai-in-the-free-world accessed 7 September 2026.

3. Amazon Web Services (AWS), a cloud computing platform

4. Competition and Markets Authority, Cloud Infrastructure Services: Final Decision Report (31 July 2025) https://assets.publishing.service.gov.uk/media/688b8891fdde2b8f73469544/final_decision_report.pdf accessed 7 September 2026.

5. Langdon Winner, 'Do Artifacts Have Politics?' (1980) 109(1) ''Daedalus'' 121, 128.

6. Ibid. 123.

7. OpenAI, Anthropic, Google DeepMind, Meta AI. xAI is an exception to the rule, domiciled between Palo Alto, CA and San Antonio, TX.

8. Texas’ 17 to California’s 8, cf. Texas Royalty Brokers, 'AI Clusters by State' (''Texas Royalty Brokers'', 2025) https://texasroyaltybrokers.com/ai-clusters-by-state/ accessed 7 September 2026.

9. The ‘rule of capture’, established in Houston & TC Railway Co v East (Tex 1904), permits landowners to extract groundwater beneath their land without liability for depleting adjacent wells or aquifers, a doctrine now exploited by hyperscale AI data centres to draw large cooling-water volumes with minimal regulatory constraint.

10. Richard Barbrook and Andy Cameron, 'The Californian Ideology' (1996) 6(1) ''Science as Culture'' 44.

11. Goldman Sachs, 'Global AI Investment Is Forecast to Exceed $1 Trillion in 2026' (''Goldman Sachs Insights'', 7 August 2026) https://www.goldmansachs.com/insights/articles/global-investment-is-forecast-to-exceed-1-trillion-in-2026 accessed 7 September 2026.

12.  National Science Foundation, 'Birth of the Commercial Internet' (''NSF Impacts'') https://www.nsf.gov/impacts/internet accessed 7 September 2026.

13. CASRAI, 'NAIRR: National AI Research Resource, Explained' (''CASRAI'') https://casrai.org/dictionary/term/national-ai-research-resource-nairr accessed 7 September 2026.

14. Council Regulation (EU) 2021/1173 of 13 July 2021 on establishing the European High Performance Computing Joint Undertaking and repealing Regulation (EU) 2018/1488 [2021] OJ L256/1.

15. Council Regulation (EU) 2026/150 of 16 January 2026 amending Regulation (EU) 2021/1173 on establishing the European High Performance Computing Joint Undertaking [2026] OJ L 2026/150.

16. European Commission, 'Realising Europe's Digital Decade' (''European Commission'', 11 October 2024) https://commission.europa.eu/strategy-and-policy/priorities-2019-2024/story-von-der-leyen-commission/realising-europes-digital-decade_en accessed 7 September 2026.</div>

File:Civilian AI at the Front Final.pdf

2026-09-23T19:52:34Z

Robert Florijn:

Governance by institutions

2026-09-23T08:56:35Z

Robert Florijn:

* [https://wiki.ivir.nl/index.php?title=Antropic%27s_Fable_5.1_(Kate_Davey) Anthropic's Fable 5.1 (Kate Davey)]
* [[cloud service (Ceren)]]
* app store
*digital identity system
*online-payment system
*AI model
*public algorithmic system.
* [https://wiki.ivir.nl/index.php?title=Decentralized_digital_currency:_BITCOIN_(Sisi) Decentralized Digital Currency: Bitcoin (Sisi)]
* [[Submarine Internet Cables  - By Pablo Latorre]]
* [[Artificial General Intelligence - By Sebastián Muller]]
* [[Predicitive Policing - Europol (Atmika)]]
* [[Entry/Exit System (EES) - Rodrigue Kayembe]]
* [https://wiki.ivir.nl/index.php?title=Social-Media_Governance_in_the_EU_-_by_Sergey_Abramov Social-Media Governance - by Sergey Abramov]
* [[App Store Governance - Google Play (Paul)]]
* [[Representative Search Engine and AI Company in China, Baidu - By Jihae Park]]
* [[Quantum Computing made in Barcelona: Qilimanjaro - by Mariona Fortuny Jané]]
* [[:File:Facebook content moderation governance in the EU Ass 3.pdf]]

Governance by institutions

2026-09-23T08:55:09Z

Robert Florijn:

* [https://wiki.ivir.nl/index.php?title=Antropic%27s_Fable_5.1_(Kate_Davey) Anthropic's Fable 5.1 (Kate Davey)]
* [[cloud service (Ceren)]]
* app store
*digital identity system
*online-payment system
*AI model
*public algorithmic system.
* [https://wiki.ivir.nl/index.php?title=Decentralized_digital_currency:_BITCOIN_(Sisi) Decentralized Digital Currency: Bitcoin (Sisi)]
* [[Submarine Internet Cables  - By Pablo Latorre]]
* [[Artificial General Intelligence - By Sebastián Muller]]
* [[Predicitive Policing - Europol (Atmika)]]
* [[Entry/Exit System (EES) - Rodrigue Kayembe]]
* [https://wiki.ivir.nl/index.php?title=Social-Media_Governance_in_the_EU_-_by_Sergey_Abramov Social-Media Governance - by Sergey Abramov]
* [[App Store Governance - Google Play (Paul)]]
* [[Representative Search Engine and AI Company in China, Baidu - By Jihae Park]]
* [[Quantum Computing made in Barcelona: Qilimanjaro - by Mariona Fortuny Jané]]
* [[Facebook content moderation governance in the EU Ass 3.pdf]]

File:Facebook content moderation governance in the EU Ass 3.pdf

2026-09-21T07:25:14Z

Robert Florijn: Although the EU's platform oversight and control seem substantive, reality shows otherwise!

== Summary ==
Although the EU's platform oversight and control seem substantive, reality shows otherwise!

File:Civilian AI at the Front.pdf

2026-09-12T15:38:22Z

Robert Florijn: Time and again, EU politics lets hyperscalers and intelligence software companies get away with the so-called "civilian use" of their products while developing tools for intelligence, reconnaissance, and warfare. Now the EU must act!

== Summary ==
Time and again, EU politics lets hyperscalers and intelligence software companies get away with the so-called "civilian use" of their products while developing tools for intelligence, reconnaissance, and warfare. Now the EU must act!

File:Civilian AI at the Front wiki.png

2026-09-09T06:00:03Z

Robert Florijn: Dual-use items of civilian origin end up in reconnaissance and weaponry. How does the EU govern that?

== Summary ==
Dual-use items of civilian origin end up in reconnaissance and weaponry. How does the EU govern that?

File:Civilian AI at the Front.png

2026-09-08T21:31:28Z

Robert Florijn: Civilian code, imaging, and algorithms originally developed for civilian use have become a fast and cheap route to high-end military use. European governance seems to overlook the risks and effects.

== Summary ==
Civilian code, imaging, and algorithms originally developed for civilian use have become a fast and cheap route to high-end military use. European governance seems to overlook the risks and effects.

Governance by history

2026-09-08T20:23:17Z

Robert Florijn:

= [[Workplace surveillance]] =
== '''Algorithmic management and workplace surveillance - by Sebastián Muller and Kate Davey ''' ==

'''1. The current issue: '''
Algorithmic management and workplace surveillance can be defined as the use of computer-programmed procedures to coordinate labor input in an organization ([https://joint-research-centre.ec.europa.eu/scientific-activities/employment/algorithmic-management-and-digital-monitoring-work_en 1]). Simply put, it is a way to control workers, and workers generally have no say in how the system that tracks them is created. As Winner says "different people are differently situated and possess unequal degrees of power as well as unequal levels of awareness" ([https://www.jstor.org/stable/20024652?seq=7 2]).
The population affected is workers; both “white collar” and “blue collar” workers are impacted. Whether it’s Amazon tracking warehouse workers’ “Time Off Task” with handheld scanners ([https://www.wired.com/story/amazon-worker-tracking-details-revealed 3]) or Meta employees having their every click tracked ([https://www.techpolicy.press/metas-worker-surveillance-tests-eu-rules-on-ai-and-labor/ 4]) to train AI, employers have the power to watch and control their employees in an unprecedented way.
It matters legally because workers have very few rights to fight back against these surveillance programs; even the The Electronic Communications Privacy Act of 1986 allows exceptions to surveillance of employees ([https://bja.ojp.gov/program/it/privacy-civil-liberties/authorities/statutes/1285 5]). 
 '''2. Historical lineage.'''
 2.1 Panopticon: it is an architectural apparatus developed by Jeremy Bentham in 1791. The design of this building effectively generated the sense of constant monitoring, which in turn motivated the prisoners to behave as if they were being watched at all times. As human labor became commodified, capitalists began implementing measures to ensure the maximization of productivity, including the replication of this design ([https://doi.org/10.14452/MR-070-09-2019-02_2 6]). While unused today, the panopticon is widely considered as both a theoretical and practical inspiration of other forms of surveillance in the workplace.
 
2.2 Scientific management: this theory attempts to analyze and standardize workflows to improve labor productivity. Frederick Winslow Taylor, its main pioneer, used early forms of data analysis to decrease walking distances, eliminating unnecessary movements and preventing deliberate slow-walking among workers. ([https://doi.org/10.14452/MR-070-09-2019-02_2 7]). There are many modern forms of data-driven decision-making processes in the workforce. For instance, Amazon warehouses track individual workers' scan rates in real time through handheld scanners, generating data feeds to track and evaluate performance ([https://www.theverge.com/2019/4/25/18516004/amazon-warehouse-fulfillment-centers-productivity-firing-terminations 8]).
 
2.3 PC in the workplace: John Blankenbaker designed the first commercially available personal computer in 1971 ([https://www.computerhistory.org/revolution/personal-computers/17/297 9]), however, it wasn’t until the mid-1980s that the use of personal computers in the workplace became an industry standard. This achievement gave rise to the digital monitoring of work, which in turn allowed employers to monitor employees in real-time, effectively replacing the human eye with a more efficient surveillance mechanism. This represents one of the key aspects of modern-day workplace monitoring. 
 
'''3. Mechanism of path dependence.'''
''' '''When analyzing the historic lineage of work surveillance, it is possible to point out two main path dependencies mechanisms in place.
Firstly, it is clear that many of the technological developments that arose through these events have one major goal, which is to increase wealth within a hierarchical ecosystem. Employers resort to several mechanisms to maximize profits and cut inefficiencies at the expense of workers. 
In addition, standardization can also be singled out as a key mechanism at play. Once one corporation implements any of these developments (such as the digital monitoring of the workplace), the market incentives would force the rest to do the same in order to be profitable and remain competitive.

'''4. Politics and distribution: '''One simple answer to this question would be that capitalists implement and leverage these mechanisms to increase their wealth while the working class remains powerless and unable to withdraw itself from the system in force.
However, if we resort to Winner's framework, we see that the technical elements of these technologies are not intrinsically capitalist, since workplace surveillance can also be observed in alternative regimes (including chattel slavery and the Soviet regime). It seems that, rather than being exclusively tied to capitalistic values, the methods in question respond to strong hierarchical structures.
In addition, even under a libertarian framework, structural inequalities are still present, allowing a privileged class to enjoy individual freedoms while segregated groups keep performing low-income jobs ([https://www.metamute.org/editorial/articles/californian-ideology 10]), arguably still subject to surveillance. Thus, not only are these technologies compatible with hierarchical top-down forms of government, but also with structures that allow such hierarchies to form without state intervention.
That said, this form of management seems less compatible with democratic societies, which tend not only to value individual liberties but to turn them into enforceable rights (including workers' privacy, health and integrity), thus establishing boundaries for workplace surveillance. This could explain why studies of American business leaders have found them unhappy with democratic scruples ([https://www.jstor.org/stable/20024652 11]).
 
'''5. Contingency and alternatives.'''
 
Bush writes about the memex, a machine that could extend a person’s memory and that the memory would belong to the person who owned it. As Bush expands, “For mature thought there is no mechanical substitute" ([https://dl.acm.org/doi/pdf/10.1145/227181.227186 12]). His idea was that a machine could take over the repetitive nature of someone’s job, while the person could record their own thoughts for their own use. Bush’s memex was never created, but the idea inspired products like Microsoft’s Productivity Score that were made for employers (the buyers) not for the person whose memory it used (the worker).
This immediately brings to mind Meta's Model Capability Initiative, which was done to train AI agents the company hopes to sell to other employers as well as watch their employees ([https://www.reuters.com/sustainability/boards-policy-regulation/meta-start-capturing-employee-mouse-movements-keystrokes-ai-training-data-2026-04-21/ 13]). If this was Bush’s memex, it would belong to the employee using it, not Meta. 

'''6. Governance implication. '''Democratic oversight of the use of algorithms as a form of surveillance is essential to ensure that workers’ rights are not overstepped in the name of efficiency. As such we propose: all workers should have a legal right to be consulted and approve a surveillance program before a company installs and uses it. A tracking system should be a group decision, not an inevitable, unilateral one. While the United States has very little protections for workers in terms of surveillance, the EU leads the way with the GDPR, specifically Article 22 ([http://data.europa.eu/eli/reg/2016/679/oj 14]), and the AI Act Article 5 ([http://data.europa.eu/eli/reg/2024/1689/oj 15]). It’s also worth noting that in December 2026, the Platform Work Directive ([https://eur-lex.europa.eu/eli/dir/2024/2831/oj/eng 16]) will come into effect which will ban automated termination.

== [[Platform Dependency]] ==

 
8 September 2026 - Platform Dependence in Public Communication - Ceren and Rodrique (work in progress)
I.                PLATFORM DEPENDENCE IN PUBLIC COMMUNICATION – WHAT IS IT , WHO IS AFFECTED BY IT AND WHY IT MATTERS LEGALLY AND INSTITUTIONALLY – C
A.           Public communication is the conveyance of a message to a large group of people. This includes public announcements, advertisements in public spaces, social media broadcasts, news delivered in reporting, editorials, opinion pieces, and even political campaigns. Platform dependence is the reliance on a third party platform or digital infrastructure to reach the audience of the public communication, the public. Platform dependence in public communication therefore includes reliance on private companies like Meta (Facebook), state television channels, or telecommunications providers who provide and control access to, and develop the infrastructure of such platforms. For the purpose of this exercise, news will be utilized as the main example of public communication.
B.          Those who are affected by platform dependence in public communication are the communicators and the public alike. News outlets are mainly for-profit organizations. These organizations’ reliance on third party digital platforms grows day-by-day to maintain their popularity and engagement (such as on social media platforms) ultimately deciding the revenue generated from advertisers. The public on the other hand are provided catchier and more diluted information, gaining just enough knowledge to grab their attention while not being provided substantial knowledge on matters. Additionally,i members of the public who do not have the resources to access such platforms and digital services remain “information poor”. Digital rich, the virtual class of tech-professional and entrepreneurs benefit from the dependence to their platforms.
C.            This is crucial as platform dependence may threaten democracy and free political expression. Platforms, run by private parties, can be run according to their political beliefs. Social media platforms aiming to keep users online utilize algorithms that provide their users with their confirmed preferences, not allowing the conveyance of a truly heteregenious catalogue of news. The original ideal of the democratic electronic agora is undermined when private corporations control the flow of news and the infrastructure on which they are provided.
II.              HISTORICAL LINEAGE – HISTORICAL MOMENTS/DECISIONS/INSTITUTIONAL ARRANGEMENTS - R
It is very difficult to discuss news without highlighting the deep entanglement it shares with with information. For our purpose 'information' will have the the following characteristics; it can be learned, processed, archived, and it's worth correlates to its rarity.

[https://wiki.ivir.nl/index.php?title=Platform_dependence_in_public_communication&action=tinymceedit#fn-1 [1]

]News is the end result of the process of taking this information from one form to another, and transmitting to a wider, typically public audience is what we will anchor our definition of 'news' to be. There is and has always been an obvious tension between the rarity that gives information its worth and the act of transmission that makes it less rare. Modern form of this can be traced to Renaissance Europe,

[https://wiki.ivir.nl/index.php?title=Platform_dependence_in_public_communication&action=tinymceedit#fn-2 [2]

]where a single innovation not only leaned on but also displaced established pattern of trusted networks collecting, processing and sharing news by way of the church, merchants, students, diplomats and so on.

[https://wiki.ivir.nl/index.php?title=Platform_dependence_in_public_communication&action=tinymceedit#fn-3 [3]

]This innovation is the invention of printing in the 15th century. It had consequences beyond those imagined by its inventor, as it offered new possibilities for the dissemination of news. It must be reiterated that printing benefitted from the existing networks and not the other way around. This is evident by Johann Carolus'  ''Relation aller Fürnemmen und gedenckwürdigen Historien'' handwritten weekly newspaper for an elite audience fruitfully running prior to the advent of printing. However, the printing press coupled with existing network and infrastructure of the postal system brought industrialisation and scale unknown to the trade and the world at the time. One of the first paying customer for press was the Church, selling indulgences. Beniger's stands: control depends on information processing and communication working together, and the this innovation at that current point time supplied both at once.

[https://wiki.ivir.nl/index.php?title=Platform_dependence_in_public_communication&action=tinymceedit#fn-4 [4]

]Nonetheless, a strong parallel between the press and later computing revolutions it's stated relationship with state.

[https://wiki.ivir.nl/index.php?title=Platform_dependence_in_public_communication&action=tinymceedit#fn-5 [5]

]With the former seeming to lean more towards the state and the other in the opposite direction.
The next pivotal historical point which disturbs the way in which news is transmitted is by way of the radio. The radio is huge leap frog as it permitted the decoupling of reach from cost, reliance on aging network such as the postal system. Where the method of control under print was through the content printed, the radio moved closer to physics on licensing broadcast rights. This changed the trajectory of new's as it birth the state owned monopoly broadcaster, responsible for broadcasting news to anyone with a receiver on the other end within appropriate frequency. News outlet now had more than one method for distributing news and for the first time had a way of engaging in dialogue on the news of the day as it makes it way via print and sound-wave. Diversity of dependency is an important aspect of this stage in history.
 
III.             MECHANISM OF PATH DEPENDENCE - C
A.            The public is generally partaking in consuming content on social media. Hence, with its 3 billion monthly active users for example, Instagram has an installed base for news outlets to depend upon (Instagram Revenue and Usage Statistics (2026), n.d.). The audiences of news outlets spend an increasing amount of their time on social media platforms, which gives results in a habitual entrenchment for these news outlets: the means of reaching the audience is dependent on their habits of content consumption. An indirect contractual entrenchment may also be discussed, as news outlets have obligations to their advertisers (who may request the upkeep of views/clicks on the news platforms they pay to advertise in). Cultural entrenchment should not be disregarded: the public expects to receive their news on social media platforms they readily frequent, and many news outlets meet this expectation. Hence, news outlets must partake in building a presence in social media platforms if they wish to compete.
IV.            POLITICS AND DISTRIBUTION – WHOSE INTERESTS ARE STRENGTHENED AND WHO HAS FEWER CHOICES - C
A.            Barbook argues that although the California Ideology claims that state power will be decreased and individuals will be empowered through decentralized networks enabled by digital technologies, he demonstrates that public communication is in fact not immune to undemocratic powers. These include private corporations: these corporations may gatekeep in choosing when and how to provide access to their platforms, while finetuning how users navigate their platforms.
By enabling access to content that users tend to agree with, those who manage these platforms enable public opinion to shift to extremes. Being “fed” the same narrative and not being challenged to see an opposing perspective ultimately provides fewer choices to the consumer of the information, the general public. Therefore, individual agency is constrained by platform architecture and control, not liberated by it.
V.              AT WHAT POINT IN HISTORY WAS A DIFFERENT TRAJECTORY POSSIBLE? - R
A.            News not subscribing to / presenting news on social media as platform (?)
VI.            GOVERNANCE IMPLICATION – ONE PROPOSITION FOR CURRENT DAY GOVERNANCE - R
OPEN PLATFORMS (ALTERNATIVE FACEBOOK A)
<section class="footnotes">
 
#Mark Greengrass, Thierry Rentet and Stéphane Gal, ‘The Hinterland of the Newsletter: Handling Information in Space and Time’ in Joad Raymond and Noah Moxham (eds), News Networks in Early Modern Europe (Brill 2016) [http://www.jstor.org/stable/10.1163/j.ctt1w8h1ng.34 http://www.jstor.org/stable/10.1163/j.ctt1w8h1ng.34]accessed 7 September 2026; ‘Michel Serres Prononce En Sorbonne La Conférence Inaugurale Du Programme Paris Nouveaux Mondes - La Chancellerie Des Universités de Paris’ (La Chancellerie des Universités de Paris, 1 January 2013) [https://www.sorbonne.fr/michel-serres-prononce-en-sorbonne-la-conference-inaugurale-du-programme-paris-nouveaux-mondes/ https://www.sorbonne.fr/michel-serres-prononce-en-sorbonne-la-conference-inaugurale-du-programme-paris-nouveaux-mondes/]accessed 7 September 2026. James Ralph Beniger, The Control Revolution: Technological and Economic Origins of the Information Society (Harvard Univ Press 1986). [https://wiki.ivir.nl/index.php?title=Platform_dependence_in_public_communication&action=tinymceedit#fnref-1 ↩︎]
#Brendan Dooley, Making News in Renaissance Europe (1st edn, Cambridge University Press 2026) [https://doi.org/10.1017/9781009358835 https://doi.org/10.1017/9781009358835]. [https://wiki.ivir.nl/index.php?title=Platform_dependence_in_public_communication&action=tinymceedit#fnref-2 ↩︎]
#Andrew Pettegree, The Invention of News: How the World Came to Know about Itself (Yale University Press 2014). Helmer Helmers, ‘Public Diplomacy in Early Modern Europe: Towards a New History of News’ (2016) 22 Media History 401 [https://doi.org/10.1080/13688804.2016.1174570 https://doi.org/10.1080/13688804.2016.1174570]. [https://wiki.ivir.nl/index.php?title=Platform_dependence_in_public_communication&action=tinymceedit#fnref-3 ↩︎]
#James Ralph Beniger, The Control Revolution: Technological and Economic Origins of the Information Society (Harvard Univ Press 1986) P24-25 [https://wiki.ivir.nl/index.php?title=Platform_dependence_in_public_communication&action=tinymceedit#fnref-4 ↩︎]
#Richard Barbrook and Andy Cameron, The Californian Ideology (1996) [https://wiki.ivir.nl/index.php?title=Platform_dependence_in_public_communication&action=tinymceedit#fnref-5 ↩︎]
</section>

== [[First One]] ==

= Centralization of AI Compute and Data =

== Centralization of AI Compute and Data - by Sisi Hammoudi & Juliette Chamot ==
=== Governance issue ===
Today, several path-dependence phenomena are emerging across advanced computing technologies. This assignment addresses one of them: the centralisation of AI compute and data. Training and running frontier AI systems now depends on a handful of hyperscale providers, namely Amazon, Microsoft, Google, and Nvidia, which dominates the GPU market. In other words, near monopolies exist throughout the supply chain. This concentration creates a governance problem affecting public and private actors alike, both legally and institutionally. On the institutional side, it prevents public administrations, and states other than the US or China, from reaching sovereign AI capacity. It also raises competition-law concerns, since dominant market positions can prevent the market from delivering the best products at the best prices to consumers. Finally, AI developers, like states, are often forced into contracts with hyperscale providers, creating a structural dependence on them.


=== Historical lineage ===
==== 1945 ====
Wartime scientific instruments should have been redirected to extend human command over knowledge, helping establish the postwar model of sustained federal funding for computing research (Vannevar Bush's, "As We May Think », 1945).


==== 1990s ====
The privatisation of the NSFNET established the idea that critical digital infrastructure, even though publicly subsidised, could legitimately be privately owned. The EU Commission followed the US's lead, recommending privatisation and deregulation of telecommunications and media as the best way to build the "infobahn" though EU members could instead have chosen nationalisation of the infrastructure (Barbrook & Cameron, "The Californian Ideology," 1996).


==== 2006 ====
<div class="s13">Hyperscale cloud computing turned massive fixed capital investment in data centresinto a rentable service, favouring firms able to invest at scale.</div>
<div class="s13">Same period: Nvidia's proprietary CUDA software, released alongside its GPUs, became the de facto standard for machine learning, locking in a decade of tooling and expertise around one firm's hardware.</div>
<div class="s13">
==== From 2022 ====
Semiconductor export controls and the CHIPS Act turned compute into an explicit instrument of geopolitical strategy.</div>



=== Mechanisms of path dependance ===
Several mechanisms drive this centralisation, but path-dependence ones dominate. Because AI compute is a capital-intensive industry, new entrants struggle to compete with hyperscalers that have already sunk huge sums into infrastructure, a barrier that produces monopoly-like situations. This dynamic, known as increasing returns, generates system-scale economies that harden into a near-irreversibility of investment (David, "Clio and the Economics of QWERTY," 1985). It overlaps with a second mechanism, standardisation, itself a form of technical interrelatedness: researchers' investment in Nvidia's CUDA-based tooling locks the field into that ecosystem regardless of whether better alternatives exist. A further mechanism is legal and contractual entrenchment. For example, Microsoft's stake in OpenAI binds developers to a specific provider by agreement rather than technical necessity.
=== Politics and distribution ===
Technology does not have a history of distributing power equally. Rather, its development has strengthened the interests and authority of powerful groups that already possessed economic and political resources, and it is yet to be determined whether AI compute and data will follow suit. 
Author Langdon Winner illustrates the historical constraints of technological development with the case of New York from the 1920s to the 1970s. He argued that the city’s roads, bridges, and parks’ physical arrangement contained “explicit and implicit political purpose.” These infrastructures privileged wealthy and white communities while restricting poor and Black populations’ choice of mobility. Similarly, the development of industrial machinery increased productivity, simultaneously weakening union and non-union workers’ bargaining power.
In the twentieth century, following the Industrial Revolution, a Control Revolution emerged from the increasing need to manage expanding production, transportation, and information flows requiring strengthened organizational bureaucratic control. This was heavily shaped by the political context of post-World War II warfare.
Therefore, the bearer of the initial cost of technology is the person putting in the capital to erect it, but ultimately, the person bearing the long-term cost of technology, whose choices diminish, is not the rich and privileged, but the everyday person subject to technologies. Forced to adapt, to reshape their professional skills to adapt to an environment trying to make them obsolete.
AI compute and data are at risk of creating the same effect. In their critique of dotcom neoliberalism, authors Barbrook and Cameron exposed a similar socio-economic tension. The Californian Ideology claims digital technology empowers individuals; however, perpetual unequal access to it creates an apartheid between “information-rich” and “information-poor” populations.
Finally, the development of AI compute and data depends on expensive and scarce computing infrastructure and a highly concentrated semiconductor supply chain. Today, this is one of the most concentrated industries, with single companies having near-total monopolies over the production of chips necessary for the functioning of AI. This monopoly model threatens to reinforce the existing economic hierarchy.


=== Contingency and alternatives ===
If a different technological trajectory was plausible, the most apt time for it to have occurred would be during the transition from the Industrial Revolution to the “Control Revolution”. When industrial production accelerated, our society experienced a “crisis of control”, creating opportunities to organize information and communication differently.
Instead of allowing control technologies to be monopolized within businesses and states, our societies could have pursued a more democratic and decentralized system. Barbrook and Cameron provide such an alternative path through their idea of the “electronic agora,” where digital networks could encourage community participation and direct democracy, where government could actively engage with the people to shape digital infrastructure and expand access. These alternatives did not actually prevail because corporate and state market participants possessed greater resources and authority to shape technological development at the end of the twentieth century. 
Inherently, technological development is not predetermined; it can support either authoritarian or democratic arrangements. Today, monopolies control chip production needed for AI compute and data, but other technologies like blockchain permit democratic participation.


=== Governance implication ===
History always threatens to repeat itself. As Barbrook discussed, digital societies are shaped by the political context in which they develop. It is a valid concern, then, that the development of information technology, through AI computing, would follow the same trajectory- that is, be constrained to benefit the wealthy who funded its development, and ultimately harm the middle and lower classes enabling its functioning. 
However, AI’s place in our society has not yet crystallized. Technology governance does not have to adhere to this system. Such a seemingly sustainable approach to AI compute and data governance has already been broached. In 2022, the Notre Dame Journal on Emerging Technologies published an article in its issue on Artificial Intelligence where the authors proposed an AI compute system based on a “dual investment strategy”. Firstly, the model would be launched by “subsidizing and negotiating” cloud computing for human AI researchers through “existing” cloud vendors, which would expand on existing AI compute initiatives. Secondly, the compute system would be specifically invested in as a publicly owned infrastructure that would be built and operate similarly to national libraries which own “academic supercomputing facilities”. 
The pilot program would have a “base level of compute access” which would keep costs of processing down and enable it to meet the compute needs of a majority of academic researchers. Absent any strengthened legislation on the use of AI compute and data, the hardware would be accompanied by an AI ethics framework to enable sustainable technological governance. In addition to an ethical staff, the human resources needed to keep an AI compute model running would include data officers, scientists, data center and software engineers, research developers, and system administrators. These positions require specialized training, allowing new types of professionals, from any background, to have access to professional prosperity in the field of AI. 
By implementing these frameworks, never neglecting the human element, and the economic and intellectual impact AI has on our society, we move from a historical determinist perspective to one acknowledging historical constraints and learning from them.

== Centralization of AI compute and Data - by Atmika Dechamma & Pablo Latorre ==
=== Governance Issue ===
AI computing capacity and data presents the governance problem of an extreme concentration of technological infrastructure and information in a restricted number of companies that privately and globally dominate the market. This affects public and private actors - such as small and medium enterprises, potential new competitors, and Global South countries - that increasingly depend on AI systems outside their control and/or are deterred from competing with established companies because of impenetrable entry barriers - most notably initial investments and operational costs. The institutional and legal relevance of this is that it consolidates, reproduces and amplifies asymmetries of power in favor of the “Big Tech”, reverting the technological trend from digital decentralization to physical infrastructure dependency and curbing innovation and market efficiency.

=== Historical Lineage ===
==== Limitation of Machine Learning ====
The AI revolution was built on the limitation of human capabilities to process large information and derive efficient results - where before there existed only facilitative tools to allow for human-enabled processes, AI now offered derivations and cognitive reasoning in moments and at large scales.

 
==== Revert to Physical Infrastructure Dependency in the rush to be the best ====
The present AI model heavily relies on hardware in order to perform complex data processing for deep learning. The data centre models followed require larger model sizes (which require higher resource, information and capital investment) to be able to yield better results. Due to the limitation of resources and a rush to establish the best working model of an AI agent, data centres are built on a rudimentary model heavily dependent on massive amounts of electricity and fresh water.

 
==== Regulation and Decentralization of Information ====
Regulation and influence of sovereigns and institutions has resulted in data sharing and information exchange structurally through open source and cloud computing. But, while AI models claims their selling point to be accessibility for all, their information and derivations remain in a closed server, stored for the purpose of training their AI agents to be the best. Big AI contenders attract higher consumer and information flows overall resulting in a better developed AI model, proving disadvantageous in any market where early contenders have an advantage and accumulation of power.

=== Mechanisms of Path Dependence ===
Early on in the industry there is an expectation of human and capital investment from all stakeholders involved, this high cost for entry into the market and expected capital to sustain and develop a model by both consumer and enterprise makes for a high switching cost. For the AI model to run and scale there is a presumption that the rate of growth is dependent on the size of data fed, resulting in a better model and hence, more consumer usage. A closed centralised data loop is created but the requirement for storage and processing far surpasses current infrastructures capabilities - heavily relying on early market entry advantage and sustained capital investments taking away from smaller aspirants and hoarding of power within closed circles. Despite its physical dependency, AI data processing has unique requirements beyond legacy infrastructures - larger demand and production of GPUs and designed to handle immense thermal heat generated while consuming electricity beyond current capabilities. 

 The AI space largely remains unregulated due to its undefined form and fluid nature, however, this has given way for unethical and morality issues regarding data privacy, exploitation of labour and environmental harms.

=== Politics and Distribution ===
Concentration strengthens the values, interests, capacities, and authority of the “Big Tech” over other actors of public and private nature. In this context, Barbrook and Cameron (1) illuminate the ethos behind AI technology development while Winner (2) makes observable the power relations surrounding it: concentration is the consequence of an AI development anchored in the belief that it must be free from all institutional and normative restriction since it is inherently liberating and non-political which, in turn, provides a political/moral justification for the concentration (if technology is inherently liberating and non-political then there is no power relations to observe or question) that translates itself into power dynamics where the AI developer not only owns the AI technology but also decides how it is trained and deployed and which models the user can or cannot access.

=== Contingency and Alternatives ===
Concentration is due, first and foremost, to the fact that AI computing capacity requires massive and cutting-edge physical infrastructure - such as data centers and microprocessors - which, in turn, requires upfront investments that are unaffordable to most companies. In David’s (3) language, companies that could finance the infrastructure and position themselves first obtained self-reinforcing advantages and a “lock-in” effect in the market. In other words, concentration is the consequence of the development of AI physical infrastructure privately with no mixed or public alternatives as a public service. Therefore, there was a space for an alternative future at the end of the 1980s or the beginning of the 1990s when the privatization model of the technological infrastructure was consolidated in line with the “Californian Ideology”. In that moment governments had the chance to treat technological infrastructure as a mixed or public good rendering it universally accessible as it was the case of the Internet; this would have permitted companies of all sizes to profit from economies of scale, reducing entry barriers and fostering competition. However, this alternative path did not prevail since the Californian Ideology provided the moral/political justification to exclude governments from the development of technological infrastructure in the exact moment in which such a decision would produce structural consequences. If that path had been followed it could have contributed to collective access to economies of scale, lower entry barriers, and a more competitive AI market.

=== Governance Implication ===
Considering the above, public or mixed investment on AI physical infrastructure is a necessary governance measure for market decentralization and collective access to AI computing capacity and data. From an historical perspective, the proposed measure is not merely a corrective action for a market failure but for a contingent and ideological decision made between the late 1980s and early 1990s that produced a structural “path dependence” that now is necessary to change.
==== References ====
(1) [https://git.xpub.nl/onebigear/SI_09_reference_library/raw/commit/22bf61e1dcfed35d58fba23808212bd00584be47/Richard_Barbrook_Andy_Cameron_Californian_ideology.pdf Barbrook, R., & Cameron, A. (1996). The Californian Ideology. Science as Culture, 6(1), 44-72].

(2) [https://d1wqtxts1xzle7.cloudfront.net/87708502/langdon_winner_do_artifacts_have_politics-libre.pdf?1655588533=&response-content-disposition=inline%3B+filename%3DDo_Artifacts_Have_Politics.pdf&Expires=1788890463&Signature=UBBiMqkip5MfYZbkbLNAN6aE2hyZdKkER-HFcmyxXkwIZuTnMKdOGoao8ddFr-rQ5T55A8NWUOlnZkGYmi66e7XR9h99DgTELMCrh48sXjOb8KYgA48r~SUUymt1IW~R~aOKJzitMIUHcrc6yEGKK1OuaA~mrEIVOXpUjQMYjqOd26Xmk9YJEglTkqaxeR47VCCth5FvozNQIowdfjzeu8aNHnI8V7Yur1IiXY~PGvYRcN6wBtcgpVMHZUtNaI8NjmYbp2p5TBYrPMF9MCLtvpGhPlsCfsvA8Bdm6XulGwRlksZBKWvVUQ8D~M8oLu12V39YaiETIObyJqbE7ctIcQ__&Key-Pair-Id=APKAJLOHF5GGSLRBV4ZA Winner, L. (2017). Do artifacts have politics?. In Computer ethics(pp. 177-192). Routledge].

(3) [https://sjbae.pbworks.com/f/David+1985.pdf David, P. A. (1985). Clio and the Economics of QWERTY. The American economic review, 75(2), 332-337].

== Centralization of AI compute and Data - by Sergey Abramov and Paul Stein ==
<div>
=== The current governance issue ===
Advanced AI currently depends on a narrow group of tech giants1 and semiconductor oligopolies controlling microchips and software, hyperscale cloud infrastructure, large datasets, frontier models, and distribution platforms. This presents a governance problem where this narrow group exercises control over foundational digital infrastructure. This directly affects consumers, businesses, and academia out of AI. Legally and institutionally, this matters because it creates a profound democratic deficit: this neutralizes antitrust frameworks designed for traditional markets, often bypassing conventional copyright and data privacy protections through unchecked data harvesting thereby complicating enforcement.This ultimately might affect national sovereignty as the countries that host the buildout will shape the future of AI: who controls it, what values it contains, how it is used2.
 
=== Historical lineage ===
In 1987, Taiwan Semiconductor Manufacturing Company (TSMC) was founded, establishing the world’s first semiconductor fab. Instead of designing and making its own chips, TSMC only manufactured chips designed by other companies. Today, nearly 100% of the world's most advanced AI processors are manufactured by TSMC, leaving the entire global AI compute supply chain centralized under one company.
 In 2006, Nvidia launched CUDA (Compute Unified Device Architecture), a free software layer that allowed developers to use Nvidia's graphics chips (GPUs) which later became perfect for training deep neural networks. Today, this creates a near-monopoly on AI compute hardware.
 Soon after Amazon launched AWS, Microsoft and Google launched their own cloud platforms marking the structural pivot to the cloud service provider model, shifting computing from a distributed architecture (where companies bought and managed their own physical servers) to a centralized infrastructure. As training modern AI models requires tens of thousands of GPUs, today’s frontier AI companies are entirely dependent on the extremely expensive physical infrastructure of a handful of cloud hyper-scalers (Microsoft, Google, AWS, and Meta).
 In 2009, Fei-Fei Li launched ImageNet, a massive database of over 14 million images, which proved that AI model performance scales directly with data volume. This created an immediate institutional advantage for consumer internet platforms (like Google and Meta), because they were already architected to collect, centralize, and store masses of user data through search engines and social networks, they possessed the unique data stack required to train high-performing AI.
 
=== Mechanism of path dependence ===
The centralization of AI compute and data is a classic case of technological lock-in. When structural choices were made decades ago, they triggered powerful, self-reinforcing loops. One such loop is the switching costs and software standardisation. As developers have spent more than a decade working with CUDA, moving to a rival hardware provider (or a decentralized compute network) requires an enterprise to completely rewrite its foundational software stack. The cost of code migration, developer retraining, and potential hardware downtime force companies to stay tied to dominant hardware monopolies such as Nvidia. Commitments, licensing conditions, and exclusivity provisions add contractual costs. Fewer than 1% of cloud customers switch provider annually, identifying technical differentiation and data-transfer charges as important barriers4.
 Network effects also matter for computing power. More customers generate more revenue, usage information, and opportunities to improve infrastructure and models. Better services attract additional users and developers, whose applications make the platform more valuable. In cloud–AI partnerships, providers may additionally obtain information about model performance and future infrastructure demand. These feedback loops strengthen firms already operating across cloud, models, software, and distribution.
 
=== Politics and distribution ===
According to Langdon Winner's definition, modern AI computing infrastructure may qualify as an "inherently political technology" because its large scale and significant financial requirements not only lend themselves to, but necessitate,5 centralized, hierarchical control. The physical siting of this infrastructure shows evidence of Winner’s second category of politicization of technologies, the ‘political arrangement’:6 Of the five largest U.S. AI firms, four are headquartered in California,7 while just one is (partially) domiciled in Texas. Texas, however, hosts twice the GPU clusters.8 This placement geographically insulates beneficiaries of the data infrastructure boom from its negative externalities. The specific choice of location may itself be a deliberate preference on account of Texas’ unique approach to water usage rights.9 Juxtaposition of this unequal and even exploitative balance of interests with the pontificating rhetoric of the urban, information rich, beneficiaries is evocative of the internal contradictions of the Californian Ideology - the virtual artisans existence and privileged position is still based on the exploitation of a vulnerable out-group, the "information-rich" consuming the output, the "information-poor" hosting the extraction.10
 
=== Contingency and Alternatives ===
Profit motive and economies of scale now concentrate investment among a few hyperscalers,11 but this reflects policy choice, not government incapacity: Compute infrastructure may be inherently political, but the technology is agnostic as to where the organizing authority originates, and states have historically funded infrastructure of comparable or greater scale, while facing less pressure to show short-term returns. The clearest turning point was 1995, when the National Science Foundation retired NSFNET, the publicly funded backbone that had connected U.S. research and supercomputing centers since the mid-1980s,12 handing internet infrastructure to commercial providers with no conditions attached. This was a deliberate policy decision made amid the deregulatory push of the Clinton era. The privatization that followed ushered in the private industrial dominance that persists today in the form of hyperscalers. Had a publicly governed backbone instead been scaled up, today's compute infrastructure might rest on a different foundation entirely.
 
=== Governance implication ===
Despite private hyperscaler dominance, the U.S. does have a nominal public alternative: the National AI Research Resource (NAIRR), established following the National AI Initiative Act of 2020 and piloted from January 2024, which explicitly aims to democratize access to the computational resources essential to AI research.13In Europe, a comparable program is the EuroHPCJU,14 an initiative intended to pool resources to build supercomputing capacity. The undertaking’s legislative basis has since been extended to include networked ‘AI factories', as well as the so-called “Gigafactories”.15 The framing in language of sovereignty indicates intent to at least compete with private infrastructure initiatives. Though government-led measures cannot reverse the centralization of compute, they can democratize access - provided they are intensified. The European Commission has “set a goal of investing more than €1 billion per year in AI research and innovation”.16 While laudable in its own right, it pales in comparison to the investments of industry leaders, which are approaching, if not exceeding, trillions of Dollars.

=== References & Notes ===
1.Amazon, Google, Meta, Microsoft, and Oracle collectively hold an estimated 71% of the world’s cumulative AI compute as of Q4 2025, measured in H100-equivalents of computing power. Luke Emberson, Josh You and Venkat Somala (2026), "Five hyperscalers now own over two-thirds of global AI compute". Published online at epoch.ai. Retrieved from 'https://epoch.ai/data-insights/hyperscalers-control-most-compute' [online resource]. Accessed 7 Sept 2026.2. Alasdair Phillips-Robins, Teddy Tawil and Sam Winter-Levy, ‘The Compute Coalition: How to Build the Future of AI in the Free World’ (Carnegie Endowment for International Peace, 8 June 2026) https://carnegieendowment.org/research/2026/06/the-compute-coalition-how-to-build-the-future-of-ai-in-the-free-world accessed 7 September 2026.

3. Amazon Web Services (AWS), a cloud computing platform

4. Competition and Markets Authority, Cloud Infrastructure Services: Final Decision Report (31 July 2025) https://assets.publishing.service.gov.uk/media/688b8891fdde2b8f73469544/final_decision_report.pdf accessed 7 September 2026.

5. Langdon Winner, 'Do Artifacts Have Politics?' (1980) 109(1) ''Daedalus'' 121, 128.

6. Ibid. 123.

7. OpenAI, Anthropic, Google DeepMind, Meta AI. xAI is an exception to the rule, domiciled between Palo Alto, CA and San Antonio, TX.

8. Texas’ 17 to California’s 8, cf. Texas Royalty Brokers, 'AI Clusters by State' (''Texas Royalty Brokers'', 2025) https://texasroyaltybrokers.com/ai-clusters-by-state/ accessed 7 September 2026.

9. The ‘rule of capture’, established in Houston & TC Railway Co v East (Tex 1904), permits landowners to extract groundwater beneath their land without liability for depleting adjacent wells or aquifers, a doctrine now exploited by hyperscale AI data centres to draw large cooling-water volumes with minimal regulatory constraint.

10. Richard Barbrook and Andy Cameron, 'The Californian Ideology' (1996) 6(1) ''Science as Culture'' 44.

11. Goldman Sachs, 'Global AI Investment Is Forecast to Exceed $1 Trillion in 2026' (''Goldman Sachs Insights'', 7 August 2026) https://www.goldmansachs.com/insights/articles/global-investment-is-forecast-to-exceed-1-trillion-in-2026 accessed 7 September 2026.

12.  National Science Foundation, 'Birth of the Commercial Internet' (''NSF Impacts'') https://www.nsf.gov/impacts/internet accessed 7 September 2026.

13. CASRAI, 'NAIRR: National AI Research Resource, Explained' (''CASRAI'') https://casrai.org/dictionary/term/national-ai-research-resource-nairr accessed 7 September 2026.

14. Council Regulation (EU) 2021/1173 of 13 July 2021 on establishing the European High Performance Computing Joint Undertaking and repealing Regulation (EU) 2018/1488 [2021] OJ L256/1.

15. Council Regulation (EU) 2026/150 of 16 January 2026 amending Regulation (EU) 2021/1173 on establishing the European High Performance Computing Joint Undertaking [2026] OJ L 2026/150.

16. European Commission, 'Realising Europe's Digital Decade' (''European Commission'', 11 October 2024) https://commission.europa.eu/strategy-and-policy/priorities-2019-2024/story-von-der-leyen-commission/realising-europes-digital-decade_en accessed 7 September 2026.</div>

Governance by history

2026-09-08T20:20:34Z

Robert Florijn:

= [https://www.blogger.com/blog/post/edit/2856835850750592270/501986482238393684 Civilian AI Is Already at the Front. Europe's Rules Can't See It.][[Workplace surveillance]] =
== '''Algorithmic management and workplace surveillance - by Sebastián Muller and Kate Davey ''' ==

'''1. The current issue: '''
Algorithmic management and workplace surveillance can be defined as the use of computer-programmed procedures to coordinate labor input in an organization ([https://joint-research-centre.ec.europa.eu/scientific-activities/employment/algorithmic-management-and-digital-monitoring-work_en 1]). Simply put, it is a way to control workers, and workers generally have no say in how the system that tracks them is created. As Winner says "different people are differently situated and possess unequal degrees of power as well as unequal levels of awareness" ([https://www.jstor.org/stable/20024652?seq=7 2]).
The population affected is workers; both “white collar” and “blue collar” workers are impacted. Whether it’s Amazon tracking warehouse workers’ “Time Off Task” with handheld scanners ([https://www.wired.com/story/amazon-worker-tracking-details-revealed 3]) or Meta employees having their every click tracked ([https://www.techpolicy.press/metas-worker-surveillance-tests-eu-rules-on-ai-and-labor/ 4]) to train AI, employers have the power to watch and control their employees in an unprecedented way.
It matters legally because workers have very few rights to fight back against these surveillance programs; even the The Electronic Communications Privacy Act of 1986 allows exceptions to surveillance of employees ([https://bja.ojp.gov/program/it/privacy-civil-liberties/authorities/statutes/1285 5]). 
 '''2. Historical lineage.'''
 2.1 Panopticon: it is an architectural apparatus developed by Jeremy Bentham in 1791. The design of this building effectively generated the sense of constant monitoring, which in turn motivated the prisoners to behave as if they were being watched at all times. As human labor became commodified, capitalists began implementing measures to ensure the maximization of productivity, including the replication of this design ([https://doi.org/10.14452/MR-070-09-2019-02_2 6]). While unused today, the panopticon is widely considered as both a theoretical and practical inspiration of other forms of surveillance in the workplace.
 
2.2 Scientific management: this theory attempts to analyze and standardize workflows to improve labor productivity. Frederick Winslow Taylor, its main pioneer, used early forms of data analysis to decrease walking distances, eliminating unnecessary movements and preventing deliberate slow-walking among workers. ([https://doi.org/10.14452/MR-070-09-2019-02_2 7]). There are many modern forms of data-driven decision-making processes in the workforce. For instance, Amazon warehouses track individual workers' scan rates in real time through handheld scanners, generating data feeds to track and evaluate performance ([https://www.theverge.com/2019/4/25/18516004/amazon-warehouse-fulfillment-centers-productivity-firing-terminations 8]).
 
2.3 PC in the workplace: John Blankenbaker designed the first commercially available personal computer in 1971 ([https://www.computerhistory.org/revolution/personal-computers/17/297 9]), however, it wasn’t until the mid-1980s that the use of personal computers in the workplace became an industry standard. This achievement gave rise to the digital monitoring of work, which in turn allowed employers to monitor employees in real-time, effectively replacing the human eye with a more efficient surveillance mechanism. This represents one of the key aspects of modern-day workplace monitoring. 
 
'''3. Mechanism of path dependence.'''
''' '''When analyzing the historic lineage of work surveillance, it is possible to point out two main path dependencies mechanisms in place.
Firstly, it is clear that many of the technological developments that arose through these events have one major goal, which is to increase wealth within a hierarchical ecosystem. Employers resort to several mechanisms to maximize profits and cut inefficiencies at the expense of workers. 
In addition, standardization can also be singled out as a key mechanism at play. Once one corporation implements any of these developments (such as the digital monitoring of the workplace), the market incentives would force the rest to do the same in order to be profitable and remain competitive.

'''4. Politics and distribution: '''One simple answer to this question would be that capitalists implement and leverage these mechanisms to increase their wealth while the working class remains powerless and unable to withdraw itself from the system in force.
However, if we resort to Winner's framework, we see that the technical elements of these technologies are not intrinsically capitalist, since workplace surveillance can also be observed in alternative regimes (including chattel slavery and the Soviet regime). It seems that, rather than being exclusively tied to capitalistic values, the methods in question respond to strong hierarchical structures.
In addition, even under a libertarian framework, structural inequalities are still present, allowing a privileged class to enjoy individual freedoms while segregated groups keep performing low-income jobs ([https://www.metamute.org/editorial/articles/californian-ideology 10]), arguably still subject to surveillance. Thus, not only are these technologies compatible with hierarchical top-down forms of government, but also with structures that allow such hierarchies to form without state intervention.
That said, this form of management seems less compatible with democratic societies, which tend not only to value individual liberties but to turn them into enforceable rights (including workers' privacy, health and integrity), thus establishing boundaries for workplace surveillance. This could explain why studies of American business leaders have found them unhappy with democratic scruples ([https://www.jstor.org/stable/20024652 11]).
 
'''5. Contingency and alternatives.'''
 
Bush writes about the memex, a machine that could extend a person’s memory and that the memory would belong to the person who owned it. As Bush expands, “For mature thought there is no mechanical substitute" ([https://dl.acm.org/doi/pdf/10.1145/227181.227186 12]). His idea was that a machine could take over the repetitive nature of someone’s job, while the person could record their own thoughts for their own use. Bush’s memex was never created, but the idea inspired products like Microsoft’s Productivity Score that were made for employers (the buyers) not for the person whose memory it used (the worker).
This immediately brings to mind Meta's Model Capability Initiative, which was done to train AI agents the company hopes to sell to other employers as well as watch their employees ([https://www.reuters.com/sustainability/boards-policy-regulation/meta-start-capturing-employee-mouse-movements-keystrokes-ai-training-data-2026-04-21/ 13]). If this was Bush’s memex, it would belong to the employee using it, not Meta. 

'''6. Governance implication. '''Democratic oversight of the use of algorithms as a form of surveillance is essential to ensure that workers’ rights are not overstepped in the name of efficiency. As such we propose: all workers should have a legal right to be consulted and approve a surveillance program before a company installs and uses it. A tracking system should be a group decision, not an inevitable, unilateral one. While the United States has very little protections for workers in terms of surveillance, the EU leads the way with the GDPR, specifically Article 22 ([http://data.europa.eu/eli/reg/2016/679/oj 14]), and the AI Act Article 5 ([http://data.europa.eu/eli/reg/2024/1689/oj 15]). It’s also worth noting that in December 2026, the Platform Work Directive ([https://eur-lex.europa.eu/eli/dir/2024/2831/oj/eng 16]) will come into effect which will ban automated termination.

== [[Platform Dependency]] ==

 
8 September 2026 - Platform Dependence in Public Communication - Ceren and Rodrique (work in progress)
I.                PLATFORM DEPENDENCE IN PUBLIC COMMUNICATION – WHAT IS IT , WHO IS AFFECTED BY IT AND WHY IT MATTERS LEGALLY AND INSTITUTIONALLY – C
A.           Public communication is the conveyance of a message to a large group of people. This includes public announcements, advertisements in public spaces, social media broadcasts, news delivered in reporting, editorials, opinion pieces, and even political campaigns. Platform dependence is the reliance on a third party platform or digital infrastructure to reach the audience of the public communication, the public. Platform dependence in public communication therefore includes reliance on private companies like Meta (Facebook), state television channels, or telecommunications providers who provide and control access to, and develop the infrastructure of such platforms. For the purpose of this exercise, news will be utilized as the main example of public communication.
B.          Those who are affected by platform dependence in public communication are the communicators and the public alike. News outlets are mainly for-profit organizations. These organizations’ reliance on third party digital platforms grows day-by-day to maintain their popularity and engagement (such as on social media platforms) ultimately deciding the revenue generated from advertisers. The public on the other hand are provided catchier and more diluted information, gaining just enough knowledge to grab their attention while not being provided substantial knowledge on matters. Additionally,i members of the public who do not have the resources to access such platforms and digital services remain “information poor”. Digital rich, the virtual class of tech-professional and entrepreneurs benefit from the dependence to their platforms.
C.            This is crucial as platform dependence may threaten democracy and free political expression. Platforms, run by private parties, can be run according to their political beliefs. Social media platforms aiming to keep users online utilize algorithms that provide their users with their confirmed preferences, not allowing the conveyance of a truly heteregenious catalogue of news. The original ideal of the democratic electronic agora is undermined when private corporations control the flow of news and the infrastructure on which they are provided.
II.              HISTORICAL LINEAGE – HISTORICAL MOMENTS/DECISIONS/INSTITUTIONAL ARRANGEMENTS - R
It is very difficult to discuss news without highlighting the deep entanglement it shares with with information. For our purpose 'information' will have the the following characteristics; it can be learned, processed, archived, and it's worth correlates to its rarity.

[https://wiki.ivir.nl/index.php?title=Platform_dependence_in_public_communication&action=tinymceedit#fn-1 [1]

]News is the end result of the process of taking this information from one form to another, and transmitting to a wider, typically public audience is what we will anchor our definition of 'news' to be. There is and has always been an obvious tension between the rarity that gives information its worth and the act of transmission that makes it less rare. Modern form of this can be traced to Renaissance Europe,

[https://wiki.ivir.nl/index.php?title=Platform_dependence_in_public_communication&action=tinymceedit#fn-2 [2]

]where a single innovation not only leaned on but also displaced established pattern of trusted networks collecting, processing and sharing news by way of the church, merchants, students, diplomats and so on.

[https://wiki.ivir.nl/index.php?title=Platform_dependence_in_public_communication&action=tinymceedit#fn-3 [3]

]This innovation is the invention of printing in the 15th century. It had consequences beyond those imagined by its inventor, as it offered new possibilities for the dissemination of news. It must be reiterated that printing benefitted from the existing networks and not the other way around. This is evident by Johann Carolus'  ''Relation aller Fürnemmen und gedenckwürdigen Historien'' handwritten weekly newspaper for an elite audience fruitfully running prior to the advent of printing. However, the printing press coupled with existing network and infrastructure of the postal system brought industrialisation and scale unknown to the trade and the world at the time. One of the first paying customer for press was the Church, selling indulgences. Beniger's stands: control depends on information processing and communication working together, and the this innovation at that current point time supplied both at once.

[https://wiki.ivir.nl/index.php?title=Platform_dependence_in_public_communication&action=tinymceedit#fn-4 [4]

]Nonetheless, a strong parallel between the press and later computing revolutions it's stated relationship with state.

[https://wiki.ivir.nl/index.php?title=Platform_dependence_in_public_communication&action=tinymceedit#fn-5 [5]

]With the former seeming to lean more towards the state and the other in the opposite direction.
The next pivotal historical point which disturbs the way in which news is transmitted is by way of the radio. The radio is huge leap frog as it permitted the decoupling of reach from cost, reliance on aging network such as the postal system. Where the method of control under print was through the content printed, the radio moved closer to physics on licensing broadcast rights. This changed the trajectory of new's as it birth the state owned monopoly broadcaster, responsible for broadcasting news to anyone with a receiver on the other end within appropriate frequency. News outlet now had more than one method for distributing news and for the first time had a way of engaging in dialogue on the news of the day as it makes it way via print and sound-wave. Diversity of dependency is an important aspect of this stage in history.
 
III.             MECHANISM OF PATH DEPENDENCE - C
A.            The public is generally partaking in consuming content on social media. Hence, with its 3 billion monthly active users for example, Instagram has an installed base for news outlets to depend upon (Instagram Revenue and Usage Statistics (2026), n.d.). The audiences of news outlets spend an increasing amount of their time on social media platforms, which gives results in a habitual entrenchment for these news outlets: the means of reaching the audience is dependent on their habits of content consumption. An indirect contractual entrenchment may also be discussed, as news outlets have obligations to their advertisers (who may request the upkeep of views/clicks on the news platforms they pay to advertise in). Cultural entrenchment should not be disregarded: the public expects to receive their news on social media platforms they readily frequent, and many news outlets meet this expectation. Hence, news outlets must partake in building a presence in social media platforms if they wish to compete.
IV.            POLITICS AND DISTRIBUTION – WHOSE INTERESTS ARE STRENGTHENED AND WHO HAS FEWER CHOICES - C
A.            Barbook argues that although the California Ideology claims that state power will be decreased and individuals will be empowered through decentralized networks enabled by digital technologies, he demonstrates that public communication is in fact not immune to undemocratic powers. These include private corporations: these corporations may gatekeep in choosing when and how to provide access to their platforms, while finetuning how users navigate their platforms.
By enabling access to content that users tend to agree with, those who manage these platforms enable public opinion to shift to extremes. Being “fed” the same narrative and not being challenged to see an opposing perspective ultimately provides fewer choices to the consumer of the information, the general public. Therefore, individual agency is constrained by platform architecture and control, not liberated by it.
V.              AT WHAT POINT IN HISTORY WAS A DIFFERENT TRAJECTORY POSSIBLE? - R
A.            News not subscribing to / presenting news on social media as platform (?)
VI.            GOVERNANCE IMPLICATION – ONE PROPOSITION FOR CURRENT DAY GOVERNANCE - R
OPEN PLATFORMS (ALTERNATIVE FACEBOOK A)
<section class="footnotes">
 
#Mark Greengrass, Thierry Rentet and Stéphane Gal, ‘The Hinterland of the Newsletter: Handling Information in Space and Time’ in Joad Raymond and Noah Moxham (eds), News Networks in Early Modern Europe (Brill 2016) [http://www.jstor.org/stable/10.1163/j.ctt1w8h1ng.34 http://www.jstor.org/stable/10.1163/j.ctt1w8h1ng.34]accessed 7 September 2026; ‘Michel Serres Prononce En Sorbonne La Conférence Inaugurale Du Programme Paris Nouveaux Mondes - La Chancellerie Des Universités de Paris’ (La Chancellerie des Universités de Paris, 1 January 2013) [https://www.sorbonne.fr/michel-serres-prononce-en-sorbonne-la-conference-inaugurale-du-programme-paris-nouveaux-mondes/ https://www.sorbonne.fr/michel-serres-prononce-en-sorbonne-la-conference-inaugurale-du-programme-paris-nouveaux-mondes/]accessed 7 September 2026. James Ralph Beniger, The Control Revolution: Technological and Economic Origins of the Information Society (Harvard Univ Press 1986). [https://wiki.ivir.nl/index.php?title=Platform_dependence_in_public_communication&action=tinymceedit#fnref-1 ↩︎]
#Brendan Dooley, Making News in Renaissance Europe (1st edn, Cambridge University Press 2026) [https://doi.org/10.1017/9781009358835 https://doi.org/10.1017/9781009358835]. [https://wiki.ivir.nl/index.php?title=Platform_dependence_in_public_communication&action=tinymceedit#fnref-2 ↩︎]
#Andrew Pettegree, The Invention of News: How the World Came to Know about Itself (Yale University Press 2014). Helmer Helmers, ‘Public Diplomacy in Early Modern Europe: Towards a New History of News’ (2016) 22 Media History 401 [https://doi.org/10.1080/13688804.2016.1174570 https://doi.org/10.1080/13688804.2016.1174570]. [https://wiki.ivir.nl/index.php?title=Platform_dependence_in_public_communication&action=tinymceedit#fnref-3 ↩︎]
#James Ralph Beniger, The Control Revolution: Technological and Economic Origins of the Information Society (Harvard Univ Press 1986) P24-25 [https://wiki.ivir.nl/index.php?title=Platform_dependence_in_public_communication&action=tinymceedit#fnref-4 ↩︎]
#Richard Barbrook and Andy Cameron, The Californian Ideology (1996) [https://wiki.ivir.nl/index.php?title=Platform_dependence_in_public_communication&action=tinymceedit#fnref-5 ↩︎]
</section>

== [[First One]] ==

= Centralization of AI Compute and Data =

== Centralization of AI Compute and Data - by Sisi Hammoudi & Juliette Chamot ==
=== Governance issue ===
Today, several path-dependence phenomena are emerging across advanced computing technologies. This assignment addresses one of them: the centralisation of AI compute and data. Training and running frontier AI systems now depends on a handful of hyperscale providers, namely Amazon, Microsoft, Google, and Nvidia, which dominates the GPU market. In other words, near monopolies exist throughout the supply chain. This concentration creates a governance problem affecting public and private actors alike, both legally and institutionally. On the institutional side, it prevents public administrations, and states other than the US or China, from reaching sovereign AI capacity. It also raises competition-law concerns, since dominant market positions can prevent the market from delivering the best products at the best prices to consumers. Finally, AI developers, like states, are often forced into contracts with hyperscale providers, creating a structural dependence on them.


=== Historical lineage ===
==== 1945 ====
Wartime scientific instruments should have been redirected to extend human command over knowledge, helping establish the postwar model of sustained federal funding for computing research (Vannevar Bush's, "As We May Think », 1945).


==== 1990s ====
The privatisation of the NSFNET established the idea that critical digital infrastructure, even though publicly subsidised, could legitimately be privately owned. The EU Commission followed the US's lead, recommending privatisation and deregulation of telecommunications and media as the best way to build the "infobahn" though EU members could instead have chosen nationalisation of the infrastructure (Barbrook & Cameron, "The Californian Ideology," 1996).


==== 2006 ====
<div class="s13">Hyperscale cloud computing turned massive fixed capital investment in data centresinto a rentable service, favouring firms able to invest at scale.</div>
<div class="s13">Same period: Nvidia's proprietary CUDA software, released alongside its GPUs, became the de facto standard for machine learning, locking in a decade of tooling and expertise around one firm's hardware.</div>
<div class="s13">
==== From 2022 ====
Semiconductor export controls and the CHIPS Act turned compute into an explicit instrument of geopolitical strategy.</div>



=== Mechanisms of path dependance ===
Several mechanisms drive this centralisation, but path-dependence ones dominate. Because AI compute is a capital-intensive industry, new entrants struggle to compete with hyperscalers that have already sunk huge sums into infrastructure, a barrier that produces monopoly-like situations. This dynamic, known as increasing returns, generates system-scale economies that harden into a near-irreversibility of investment (David, "Clio and the Economics of QWERTY," 1985). It overlaps with a second mechanism, standardisation, itself a form of technical interrelatedness: researchers' investment in Nvidia's CUDA-based tooling locks the field into that ecosystem regardless of whether better alternatives exist. A further mechanism is legal and contractual entrenchment. For example, Microsoft's stake in OpenAI binds developers to a specific provider by agreement rather than technical necessity.
=== Politics and distribution ===
Technology does not have a history of distributing power equally. Rather, its development has strengthened the interests and authority of powerful groups that already possessed economic and political resources, and it is yet to be determined whether AI compute and data will follow suit. 
Author Langdon Winner illustrates the historical constraints of technological development with the case of New York from the 1920s to the 1970s. He argued that the city’s roads, bridges, and parks’ physical arrangement contained “explicit and implicit political purpose.” These infrastructures privileged wealthy and white communities while restricting poor and Black populations’ choice of mobility. Similarly, the development of industrial machinery increased productivity, simultaneously weakening union and non-union workers’ bargaining power.
In the twentieth century, following the Industrial Revolution, a Control Revolution emerged from the increasing need to manage expanding production, transportation, and information flows requiring strengthened organizational bureaucratic control. This was heavily shaped by the political context of post-World War II warfare.
Therefore, the bearer of the initial cost of technology is the person putting in the capital to erect it, but ultimately, the person bearing the long-term cost of technology, whose choices diminish, is not the rich and privileged, but the everyday person subject to technologies. Forced to adapt, to reshape their professional skills to adapt to an environment trying to make them obsolete.
AI compute and data are at risk of creating the same effect. In their critique of dotcom neoliberalism, authors Barbrook and Cameron exposed a similar socio-economic tension. The Californian Ideology claims digital technology empowers individuals; however, perpetual unequal access to it creates an apartheid between “information-rich” and “information-poor” populations.
Finally, the development of AI compute and data depends on expensive and scarce computing infrastructure and a highly concentrated semiconductor supply chain. Today, this is one of the most concentrated industries, with single companies having near-total monopolies over the production of chips necessary for the functioning of AI. This monopoly model threatens to reinforce the existing economic hierarchy.


=== Contingency and alternatives ===
If a different technological trajectory was plausible, the most apt time for it to have occurred would be during the transition from the Industrial Revolution to the “Control Revolution”. When industrial production accelerated, our society experienced a “crisis of control”, creating opportunities to organize information and communication differently.
Instead of allowing control technologies to be monopolized within businesses and states, our societies could have pursued a more democratic and decentralized system. Barbrook and Cameron provide such an alternative path through their idea of the “electronic agora,” where digital networks could encourage community participation and direct democracy, where government could actively engage with the people to shape digital infrastructure and expand access. These alternatives did not actually prevail because corporate and state market participants possessed greater resources and authority to shape technological development at the end of the twentieth century. 
Inherently, technological development is not predetermined; it can support either authoritarian or democratic arrangements. Today, monopolies control chip production needed for AI compute and data, but other technologies like blockchain permit democratic participation.


=== Governance implication ===
History always threatens to repeat itself. As Barbrook discussed, digital societies are shaped by the political context in which they develop. It is a valid concern, then, that the development of information technology, through AI computing, would follow the same trajectory- that is, be constrained to benefit the wealthy who funded its development, and ultimately harm the middle and lower classes enabling its functioning. 
However, AI’s place in our society has not yet crystallized. Technology governance does not have to adhere to this system. Such a seemingly sustainable approach to AI compute and data governance has already been broached. In 2022, the Notre Dame Journal on Emerging Technologies published an article in its issue on Artificial Intelligence where the authors proposed an AI compute system based on a “dual investment strategy”. Firstly, the model would be launched by “subsidizing and negotiating” cloud computing for human AI researchers through “existing” cloud vendors, which would expand on existing AI compute initiatives. Secondly, the compute system would be specifically invested in as a publicly owned infrastructure that would be built and operate similarly to national libraries which own “academic supercomputing facilities”. 
The pilot program would have a “base level of compute access” which would keep costs of processing down and enable it to meet the compute needs of a majority of academic researchers. Absent any strengthened legislation on the use of AI compute and data, the hardware would be accompanied by an AI ethics framework to enable sustainable technological governance. In addition to an ethical staff, the human resources needed to keep an AI compute model running would include data officers, scientists, data center and software engineers, research developers, and system administrators. These positions require specialized training, allowing new types of professionals, from any background, to have access to professional prosperity in the field of AI. 
By implementing these frameworks, never neglecting the human element, and the economic and intellectual impact AI has on our society, we move from a historical determinist perspective to one acknowledging historical constraints and learning from them.

== Centralization of AI compute and Data - by Atmika Dechamma & Pablo Latorre ==
=== Governance Issue ===
AI computing capacity and data presents the governance problem of an extreme concentration of technological infrastructure and information in a restricted number of companies that privately and globally dominate the market. This affects public and private actors - such as small and medium enterprises, potential new competitors, and Global South countries - that increasingly depend on AI systems outside their control and/or are deterred from competing with established companies because of impenetrable entry barriers - most notably initial investments and operational costs. The institutional and legal relevance of this is that it consolidates, reproduces and amplifies asymmetries of power in favor of the “Big Tech”, reverting the technological trend from digital decentralization to physical infrastructure dependency and curbing innovation and market efficiency.

=== Historical Lineage ===
==== Limitation of Machine Learning ====
The AI revolution was built on the limitation of human capabilities to process large information and derive efficient results - where before there existed only facilitative tools to allow for human-enabled processes, AI now offered derivations and cognitive reasoning in moments and at large scales.

 
==== Revert to Physical Infrastructure Dependency in the rush to be the best ====
The present AI model heavily relies on hardware in order to perform complex data processing for deep learning. The data centre models followed require larger model sizes (which require higher resource, information and capital investment) to be able to yield better results. Due to the limitation of resources and a rush to establish the best working model of an AI agent, data centres are built on a rudimentary model heavily dependent on massive amounts of electricity and fresh water.

 
==== Regulation and Decentralization of Information ====
Regulation and influence of sovereigns and institutions has resulted in data sharing and information exchange structurally through open source and cloud computing. But, while AI models claims their selling point to be accessibility for all, their information and derivations remain in a closed server, stored for the purpose of training their AI agents to be the best. Big AI contenders attract higher consumer and information flows overall resulting in a better developed AI model, proving disadvantageous in any market where early contenders have an advantage and accumulation of power.

=== Mechanisms of Path Dependence ===
Early on in the industry there is an expectation of human and capital investment from all stakeholders involved, this high cost for entry into the market and expected capital to sustain and develop a model by both consumer and enterprise makes for a high switching cost. For the AI model to run and scale there is a presumption that the rate of growth is dependent on the size of data fed, resulting in a better model and hence, more consumer usage. A closed centralised data loop is created but the requirement for storage and processing far surpasses current infrastructures capabilities - heavily relying on early market entry advantage and sustained capital investments taking away from smaller aspirants and hoarding of power within closed circles. Despite its physical dependency, AI data processing has unique requirements beyond legacy infrastructures - larger demand and production of GPUs and designed to handle immense thermal heat generated while consuming electricity beyond current capabilities. 

 The AI space largely remains unregulated due to its undefined form and fluid nature, however, this has given way for unethical and morality issues regarding data privacy, exploitation of labour and environmental harms.

=== Politics and Distribution ===
Concentration strengthens the values, interests, capacities, and authority of the “Big Tech” over other actors of public and private nature. In this context, Barbrook and Cameron (1) illuminate the ethos behind AI technology development while Winner (2) makes observable the power relations surrounding it: concentration is the consequence of an AI development anchored in the belief that it must be free from all institutional and normative restriction since it is inherently liberating and non-political which, in turn, provides a political/moral justification for the concentration (if technology is inherently liberating and non-political then there is no power relations to observe or question) that translates itself into power dynamics where the AI developer not only owns the AI technology but also decides how it is trained and deployed and which models the user can or cannot access.

=== Contingency and Alternatives ===
Concentration is due, first and foremost, to the fact that AI computing capacity requires massive and cutting-edge physical infrastructure - such as data centers and microprocessors - which, in turn, requires upfront investments that are unaffordable to most companies. In David’s (3) language, companies that could finance the infrastructure and position themselves first obtained self-reinforcing advantages and a “lock-in” effect in the market. In other words, concentration is the consequence of the development of AI physical infrastructure privately with no mixed or public alternatives as a public service. Therefore, there was a space for an alternative future at the end of the 1980s or the beginning of the 1990s when the privatization model of the technological infrastructure was consolidated in line with the “Californian Ideology”. In that moment governments had the chance to treat technological infrastructure as a mixed or public good rendering it universally accessible as it was the case of the Internet; this would have permitted companies of all sizes to profit from economies of scale, reducing entry barriers and fostering competition. However, this alternative path did not prevail since the Californian Ideology provided the moral/political justification to exclude governments from the development of technological infrastructure in the exact moment in which such a decision would produce structural consequences. If that path had been followed it could have contributed to collective access to economies of scale, lower entry barriers, and a more competitive AI market.

=== Governance Implication ===
Considering the above, public or mixed investment on AI physical infrastructure is a necessary governance measure for market decentralization and collective access to AI computing capacity and data. From an historical perspective, the proposed measure is not merely a corrective action for a market failure but for a contingent and ideological decision made between the late 1980s and early 1990s that produced a structural “path dependence” that now is necessary to change.
==== References ====
(1) [https://git.xpub.nl/onebigear/SI_09_reference_library/raw/commit/22bf61e1dcfed35d58fba23808212bd00584be47/Richard_Barbrook_Andy_Cameron_Californian_ideology.pdf Barbrook, R., & Cameron, A. (1996). The Californian Ideology. Science as Culture, 6(1), 44-72].

(2) [https://d1wqtxts1xzle7.cloudfront.net/87708502/langdon_winner_do_artifacts_have_politics-libre.pdf?1655588533=&response-content-disposition=inline%3B+filename%3DDo_Artifacts_Have_Politics.pdf&Expires=1788890463&Signature=UBBiMqkip5MfYZbkbLNAN6aE2hyZdKkER-HFcmyxXkwIZuTnMKdOGoao8ddFr-rQ5T55A8NWUOlnZkGYmi66e7XR9h99DgTELMCrh48sXjOb8KYgA48r~SUUymt1IW~R~aOKJzitMIUHcrc6yEGKK1OuaA~mrEIVOXpUjQMYjqOd26Xmk9YJEglTkqaxeR47VCCth5FvozNQIowdfjzeu8aNHnI8V7Yur1IiXY~PGvYRcN6wBtcgpVMHZUtNaI8NjmYbp2p5TBYrPMF9MCLtvpGhPlsCfsvA8Bdm6XulGwRlksZBKWvVUQ8D~M8oLu12V39YaiETIObyJqbE7ctIcQ__&Key-Pair-Id=APKAJLOHF5GGSLRBV4ZA Winner, L. (2017). Do artifacts have politics?. In Computer ethics(pp. 177-192). Routledge].

(3) [https://sjbae.pbworks.com/f/David+1985.pdf David, P. A. (1985). Clio and the Economics of QWERTY. The American economic review, 75(2), 332-337].

== Centralization of AI compute and Data - by Sergey Abramov and Paul Stein ==
<div>
=== The current governance issue ===
Advanced AI currently depends on a narrow group of tech giants1 and semiconductor oligopolies controlling microchips and software, hyperscale cloud infrastructure, large datasets, frontier models, and distribution platforms. This presents a governance problem where this narrow group exercises control over foundational digital infrastructure. This directly affects consumers, businesses, and academia out of AI. Legally and institutionally, this matters because it creates a profound democratic deficit: this neutralizes antitrust frameworks designed for traditional markets, often bypassing conventional copyright and data privacy protections through unchecked data harvesting thereby complicating enforcement.This ultimately might affect national sovereignty as the countries that host the buildout will shape the future of AI: who controls it, what values it contains, how it is used2.
 
=== Historical lineage ===
In 1987, Taiwan Semiconductor Manufacturing Company (TSMC) was founded, establishing the world’s first semiconductor fab. Instead of designing and making its own chips, TSMC only manufactured chips designed by other companies. Today, nearly 100% of the world's most advanced AI processors are manufactured by TSMC, leaving the entire global AI compute supply chain centralized under one company.
 In 2006, Nvidia launched CUDA (Compute Unified Device Architecture), a free software layer that allowed developers to use Nvidia's graphics chips (GPUs) which later became perfect for training deep neural networks. Today, this creates a near-monopoly on AI compute hardware.
 Soon after Amazon launched AWS, Microsoft and Google launched their own cloud platforms marking the structural pivot to the cloud service provider model, shifting computing from a distributed architecture (where companies bought and managed their own physical servers) to a centralized infrastructure. As training modern AI models requires tens of thousands of GPUs, today’s frontier AI companies are entirely dependent on the extremely expensive physical infrastructure of a handful of cloud hyper-scalers (Microsoft, Google, AWS, and Meta).
 In 2009, Fei-Fei Li launched ImageNet, a massive database of over 14 million images, which proved that AI model performance scales directly with data volume. This created an immediate institutional advantage for consumer internet platforms (like Google and Meta), because they were already architected to collect, centralize, and store masses of user data through search engines and social networks, they possessed the unique data stack required to train high-performing AI.
 
=== Mechanism of path dependence ===
The centralization of AI compute and data is a classic case of technological lock-in. When structural choices were made decades ago, they triggered powerful, self-reinforcing loops. One such loop is the switching costs and software standardisation. As developers have spent more than a decade working with CUDA, moving to a rival hardware provider (or a decentralized compute network) requires an enterprise to completely rewrite its foundational software stack. The cost of code migration, developer retraining, and potential hardware downtime force companies to stay tied to dominant hardware monopolies such as Nvidia. Commitments, licensing conditions, and exclusivity provisions add contractual costs. Fewer than 1% of cloud customers switch provider annually, identifying technical differentiation and data-transfer charges as important barriers4.
 Network effects also matter for computing power. More customers generate more revenue, usage information, and opportunities to improve infrastructure and models. Better services attract additional users and developers, whose applications make the platform more valuable. In cloud–AI partnerships, providers may additionally obtain information about model performance and future infrastructure demand. These feedback loops strengthen firms already operating across cloud, models, software, and distribution.
 
=== Politics and distribution ===
According to Langdon Winner's definition, modern AI computing infrastructure may qualify as an "inherently political technology" because its large scale and significant financial requirements not only lend themselves to, but necessitate,5 centralized, hierarchical control. The physical siting of this infrastructure shows evidence of Winner’s second category of politicization of technologies, the ‘political arrangement’:6 Of the five largest U.S. AI firms, four are headquartered in California,7 while just one is (partially) domiciled in Texas. Texas, however, hosts twice the GPU clusters.8 This placement geographically insulates beneficiaries of the data infrastructure boom from its negative externalities. The specific choice of location may itself be a deliberate preference on account of Texas’ unique approach to water usage rights.9 Juxtaposition of this unequal and even exploitative balance of interests with the pontificating rhetoric of the urban, information rich, beneficiaries is evocative of the internal contradictions of the Californian Ideology - the virtual artisans existence and privileged position is still based on the exploitation of a vulnerable out-group, the "information-rich" consuming the output, the "information-poor" hosting the extraction.10
 
=== Contingency and Alternatives ===
Profit motive and economies of scale now concentrate investment among a few hyperscalers,11 but this reflects policy choice, not government incapacity: Compute infrastructure may be inherently political, but the technology is agnostic as to where the organizing authority originates, and states have historically funded infrastructure of comparable or greater scale, while facing less pressure to show short-term returns. The clearest turning point was 1995, when the National Science Foundation retired NSFNET, the publicly funded backbone that had connected U.S. research and supercomputing centers since the mid-1980s,12 handing internet infrastructure to commercial providers with no conditions attached. This was a deliberate policy decision made amid the deregulatory push of the Clinton era. The privatization that followed ushered in the private industrial dominance that persists today in the form of hyperscalers. Had a publicly governed backbone instead been scaled up, today's compute infrastructure might rest on a different foundation entirely.
 
=== Governance implication ===
Despite private hyperscaler dominance, the U.S. does have a nominal public alternative: the National AI Research Resource (NAIRR), established following the National AI Initiative Act of 2020 and piloted from January 2024, which explicitly aims to democratize access to the computational resources essential to AI research.13In Europe, a comparable program is the EuroHPCJU,14 an initiative intended to pool resources to build supercomputing capacity. The undertaking’s legislative basis has since been extended to include networked ‘AI factories', as well as the so-called “Gigafactories”.15 The framing in language of sovereignty indicates intent to at least compete with private infrastructure initiatives. Though government-led measures cannot reverse the centralization of compute, they can democratize access - provided they are intensified. The European Commission has “set a goal of investing more than €1 billion per year in AI research and innovation”.16 While laudable in its own right, it pales in comparison to the investments of industry leaders, which are approaching, if not exceeding, trillions of Dollars.

=== References & Notes ===
1.Amazon, Google, Meta, Microsoft, and Oracle collectively hold an estimated 71% of the world’s cumulative AI compute as of Q4 2025, measured in H100-equivalents of computing power. Luke Emberson, Josh You and Venkat Somala (2026), "Five hyperscalers now own over two-thirds of global AI compute". Published online at epoch.ai. Retrieved from 'https://epoch.ai/data-insights/hyperscalers-control-most-compute' [online resource]. Accessed 7 Sept 2026.2. Alasdair Phillips-Robins, Teddy Tawil and Sam Winter-Levy, ‘The Compute Coalition: How to Build the Future of AI in the Free World’ (Carnegie Endowment for International Peace, 8 June 2026) https://carnegieendowment.org/research/2026/06/the-compute-coalition-how-to-build-the-future-of-ai-in-the-free-world accessed 7 September 2026.

3. Amazon Web Services (AWS), a cloud computing platform

4. Competition and Markets Authority, Cloud Infrastructure Services: Final Decision Report (31 July 2025) https://assets.publishing.service.gov.uk/media/688b8891fdde2b8f73469544/final_decision_report.pdf accessed 7 September 2026.

5. Langdon Winner, 'Do Artifacts Have Politics?' (1980) 109(1) ''Daedalus'' 121, 128.

6. Ibid. 123.

7. OpenAI, Anthropic, Google DeepMind, Meta AI. xAI is an exception to the rule, domiciled between Palo Alto, CA and San Antonio, TX.

8. Texas’ 17 to California’s 8, cf. Texas Royalty Brokers, 'AI Clusters by State' (''Texas Royalty Brokers'', 2025) https://texasroyaltybrokers.com/ai-clusters-by-state/ accessed 7 September 2026.

9. The ‘rule of capture’, established in Houston & TC Railway Co v East (Tex 1904), permits landowners to extract groundwater beneath their land without liability for depleting adjacent wells or aquifers, a doctrine now exploited by hyperscale AI data centres to draw large cooling-water volumes with minimal regulatory constraint.

10. Richard Barbrook and Andy Cameron, 'The Californian Ideology' (1996) 6(1) ''Science as Culture'' 44.

11. Goldman Sachs, 'Global AI Investment Is Forecast to Exceed $1 Trillion in 2026' (''Goldman Sachs Insights'', 7 August 2026) https://www.goldmansachs.com/insights/articles/global-investment-is-forecast-to-exceed-1-trillion-in-2026 accessed 7 September 2026.

12.  National Science Foundation, 'Birth of the Commercial Internet' (''NSF Impacts'') https://www.nsf.gov/impacts/internet accessed 7 September 2026.

13. CASRAI, 'NAIRR: National AI Research Resource, Explained' (''CASRAI'') https://casrai.org/dictionary/term/national-ai-research-resource-nairr accessed 7 September 2026.

14. Council Regulation (EU) 2021/1173 of 13 July 2021 on establishing the European High Performance Computing Joint Undertaking and repealing Regulation (EU) 2018/1488 [2021] OJ L256/1.

15. Council Regulation (EU) 2026/150 of 16 January 2026 amending Regulation (EU) 2021/1173 on establishing the European High Performance Computing Joint Undertaking [2026] OJ L 2026/150.

16. European Commission, 'Realising Europe's Digital Decade' (''European Commission'', 11 October 2024) https://commission.europa.eu/strategy-and-policy/priorities-2019-2024/story-von-der-leyen-commission/realising-europes-digital-decade_en accessed 7 September 2026.</div>

Governance by history

2026-09-08T20:04:58Z

Robert Florijn:

= [[Workplace surveillance]] =
== '''Algorithmic management and workplace surveillance - by Sebastián Muller and Kate Davey ''' ==

'''1. The current issue: '''
Algorithmic management and workplace surveillance can be defined as the use of computer-programmed procedures to coordinate labor input in an organization ([https://joint-research-centre.ec.europa.eu/scientific-activities/employment/algorithmic-management-and-digital-monitoring-work_en 1]). Simply put, it is a way to control workers, and workers generally have no say in how the system that tracks them is created. As Winner says "different people are differently situated and possess unequal degrees of power as well as unequal levels of awareness" ([https://www.jstor.org/stable/20024652?seq=7 2]).
The population affected is workers; both “white collar” and “blue collar” workers are impacted. Whether it’s Amazon tracking warehouse workers’ “Time Off Task” with handheld scanners ([https://www.wired.com/story/amazon-worker-tracking-details-revealed 3]) or Meta employees having their every click tracked ([https://www.techpolicy.press/metas-worker-surveillance-tests-eu-rules-on-ai-and-labor/ 4]) to train AI, employers have the power to watch and control their employees in an unprecedented way.
It matters legally because workers have very few rights to fight back against these surveillance programs; even the The Electronic Communications Privacy Act of 1986 allows exceptions to surveillance of employees ([https://bja.ojp.gov/program/it/privacy-civil-liberties/authorities/statutes/1285 5]). 
 '''2. Historical lineage.'''
 2.1 Panopticon: it is an architectural apparatus developed by Jeremy Bentham in 1791. The design of this building effectively generated the sense of constant monitoring, which in turn motivated the prisoners to behave as if they were being watched at all times. As human labor became commodified, capitalists began implementing measures to ensure the maximization of productivity, including the replication of this design ([https://doi.org/10.14452/MR-070-09-2019-02_2 6]). While unused today, the panopticon is widely considered as both a theoretical and practical inspiration of other forms of surveillance in the workplace.
 
2.2 Scientific management: this theory attempts to analyze and standardize workflows to improve labor productivity. Frederick Winslow Taylor, its main pioneer, used early forms of data analysis to decrease walking distances, eliminating unnecessary movements and preventing deliberate slow-walking among workers. ([https://doi.org/10.14452/MR-070-09-2019-02_2 7]). There are many modern forms of data-driven decision-making processes in the workforce. For instance, Amazon warehouses track individual workers' scan rates in real time through handheld scanners, generating data feeds to track and evaluate performance ([https://www.theverge.com/2019/4/25/18516004/amazon-warehouse-fulfillment-centers-productivity-firing-terminations 8]).
 
2.3 PC in the workplace: John Blankenbaker designed the first commercially available personal computer in 1971 ([https://www.computerhistory.org/revolution/personal-computers/17/297 9]), however, it wasn’t until the mid-1980s that the use of personal computers in the workplace became an industry standard. This achievement gave rise to the digital monitoring of work, which in turn allowed employers to monitor employees in real-time, effectively replacing the human eye with a more efficient surveillance mechanism. This represents one of the key aspects of modern-day workplace monitoring. 
 
'''3. Mechanism of path dependence.'''
''' '''When analyzing the historic lineage of work surveillance, it is possible to point out two main path dependencies mechanisms in place.
Firstly, it is clear that many of the technological developments that arose through these events have one major goal, which is to increase wealth within a hierarchical ecosystem. Employers resort to several mechanisms to maximize profits and cut inefficiencies at the expense of workers. 
In addition, standardization can also be singled out as a key mechanism at play. Once one corporation implements any of these developments (such as the digital monitoring of the workplace), the market incentives would force the rest to do the same in order to be profitable and remain competitive.

'''4. Politics and distribution: '''One simple answer to this question would be that capitalists implement and leverage these mechanisms to increase their wealth while the working class remains powerless and unable to withdraw itself from the system in force.
However, if we resort to Winner's framework, we see that the technical elements of these technologies are not intrinsically capitalist, since workplace surveillance can also be observed in alternative regimes (including chattel slavery and the Soviet regime). It seems that, rather than being exclusively tied to capitalistic values, the methods in question respond to strong hierarchical structures.
In addition, even under a libertarian framework, structural inequalities are still present, allowing a privileged class to enjoy individual freedoms while segregated groups keep performing low-income jobs ([https://www.metamute.org/editorial/articles/californian-ideology 10]), arguably still subject to surveillance. Thus, not only are these technologies compatible with hierarchical top-down forms of government, but also with structures that allow such hierarchies to form without state intervention.
That said, this form of management seems less compatible with democratic societies, which tend not only to value individual liberties but to turn them into enforceable rights (including workers' privacy, health and integrity), thus establishing boundaries for workplace surveillance. This could explain why studies of American business leaders have found them unhappy with democratic scruples ([https://www.jstor.org/stable/20024652 11]).
 
'''5. Contingency and alternatives.'''
 
Bush writes about the memex, a machine that could extend a person’s memory and that the memory would belong to the person who owned it. As Bush expands, “For mature thought there is no mechanical substitute" ([https://dl.acm.org/doi/pdf/10.1145/227181.227186 12]). His idea was that a machine could take over the repetitive nature of someone’s job, while the person could record their own thoughts for their own use. Bush’s memex was never created, but the idea inspired products like Microsoft’s Productivity Score that were made for employers (the buyers) not for the person whose memory it used (the worker).
This immediately brings to mind Meta's Model Capability Initiative, which was done to train AI agents the company hopes to sell to other employers as well as watch their employees ([https://www.reuters.com/sustainability/boards-policy-regulation/meta-start-capturing-employee-mouse-movements-keystrokes-ai-training-data-2026-04-21/ 13]). If this was Bush’s memex, it would belong to the employee using it, not Meta. 

'''6. Governance implication. '''Democratic oversight of the use of algorithms as a form of surveillance is essential to ensure that workers’ rights are not overstepped in the name of efficiency. As such we propose: all workers should have a legal right to be consulted and approve a surveillance program before a company installs and uses it. A tracking system should be a group decision, not an inevitable, unilateral one. While the United States has very little protections for workers in terms of surveillance, the EU leads the way with the GDPR, specifically Article 22 ([http://data.europa.eu/eli/reg/2016/679/oj 14]), and the AI Act Article 5 ([http://data.europa.eu/eli/reg/2024/1689/oj 15]). It’s also worth noting that in December 2026, the Platform Work Directive ([https://eur-lex.europa.eu/eli/dir/2024/2831/oj/eng 16]) will come into effect which will ban automated termination.

== [[Platform Dependency]] ==

 
8 September 2026 - Platform Dependence in Public Communication - Ceren and Rodrique (work in progress)
I.                PLATFORM DEPENDENCE IN PUBLIC COMMUNICATION – WHAT IS IT , WHO IS AFFECTED BY IT AND WHY IT MATTERS LEGALLY AND INSTITUTIONALLY – C
A.           Public communication is the conveyance of a message to a large group of people. This includes public announcements, advertisements in public spaces, social media broadcasts, news delivered in reporting, editorials, opinion pieces, and even political campaigns. Platform dependence is the reliance on a third party platform or digital infrastructure to reach the audience of the public communication, the public. Platform dependence in public communication therefore includes reliance on private companies like Meta (Facebook), state television channels, or telecommunications providers who provide and control access to, and develop the infrastructure of such platforms. For the purpose of this exercise, news will be utilized as the main example of public communication.
B.          Those who are affected by platform dependence in public communication are the communicators and the public alike. News outlets are mainly for-profit organizations. These organizations’ reliance on third party digital platforms grows day-by-day to maintain their popularity and engagement (such as on social media platforms) ultimately deciding the revenue generated from advertisers. The public on the other hand are provided catchier and more diluted information, gaining just enough knowledge to grab their attention while not being provided substantial knowledge on matters. Additionally,i members of the public who do not have the resources to access such platforms and digital services remain “information poor”. Digital rich, the virtual class of tech-professional and entrepreneurs benefit from the dependence to their platforms.
C.            This is crucial as platform dependence may threaten democracy and free political expression. Platforms, run by private parties, can be run according to their political beliefs. Social media platforms aiming to keep users online utilize algorithms that provide their users with their confirmed preferences, not allowing the conveyance of a truly heteregenious catalogue of news. The original ideal of the democratic electronic agora is undermined when private corporations control the flow of news and the infrastructure on which they are provided.
II.              HISTORICAL LINEAGE – HISTORICAL MOMENTS/DECISIONS/INSTITUTIONAL ARRANGEMENTS - R
It is very difficult to discuss news without highlighting the deep entanglement it shares with with information. For our purpose 'information' will have the the following characteristics; it can be learned, processed, archived, and it's worth correlates to its rarity.

[https://wiki.ivir.nl/index.php?title=Platform_dependence_in_public_communication&action=tinymceedit#fn-1 [1]

]News is the end result of the process of taking this information from one form to another, and transmitting to a wider, typically public audience is what we will anchor our definition of 'news' to be. There is and has always been an obvious tension between the rarity that gives information its worth and the act of transmission that makes it less rare. Modern form of this can be traced to Renaissance Europe,

[https://wiki.ivir.nl/index.php?title=Platform_dependence_in_public_communication&action=tinymceedit#fn-2 [2]

]where a single innovation not only leaned on but also displaced established pattern of trusted networks collecting, processing and sharing news by way of the church, merchants, students, diplomats and so on.

[https://wiki.ivir.nl/index.php?title=Platform_dependence_in_public_communication&action=tinymceedit#fn-3 [3]

]This innovation is the invention of printing in the 15th century. It had consequences beyond those imagined by its inventor, as it offered new possibilities for the dissemination of news. It must be reiterated that printing benefitted from the existing networks and not the other way around. This is evident by Johann Carolus'  ''Relation aller Fürnemmen und gedenckwürdigen Historien'' handwritten weekly newspaper for an elite audience fruitfully running prior to the advent of printing. However, the printing press coupled with existing network and infrastructure of the postal system brought industrialisation and scale unknown to the trade and the world at the time. One of the first paying customer for press was the Church, selling indulgences. Beniger's stands: control depends on information processing and communication working together, and the this innovation at that current point time supplied both at once.

[https://wiki.ivir.nl/index.php?title=Platform_dependence_in_public_communication&action=tinymceedit#fn-4 [4]

]Nonetheless, a strong parallel between the press and later computing revolutions it's stated relationship with state.

[https://wiki.ivir.nl/index.php?title=Platform_dependence_in_public_communication&action=tinymceedit#fn-5 [5]

]With the former seeming to lean more towards the state and the other in the opposite direction.
The next pivotal historical point which disturbs the way in which news is transmitted is by way of the radio. The radio is huge leap frog as it permitted the decoupling of reach from cost, reliance on aging network such as the postal system. Where the method of control under print was through the content printed, the radio moved closer to physics on licensing broadcast rights. This changed the trajectory of new's as it birth the state owned monopoly broadcaster, responsible for broadcasting news to anyone with a receiver on the other end within appropriate frequency. News outlet now had more than one method for distributing news and for the first time had a way of engaging in dialogue on the news of the day as it makes it way via print and sound-wave. Diversity of dependency is an important aspect of this stage in history.
 
III.             MECHANISM OF PATH DEPENDENCE - C
A.            The public is generally partaking in consuming content on social media. Hence, with its 3 billion monthly active users for example, Instagram has an installed base for news outlets to depend upon (Instagram Revenue and Usage Statistics (2026), n.d.). The audiences of news outlets spend an increasing amount of their time on social media platforms, which gives results in a habitual entrenchment for these news outlets: the means of reaching the audience is dependent on their habits of content consumption. An indirect contractual entrenchment may also be discussed, as news outlets have obligations to their advertisers (who may request the upkeep of views/clicks on the news platforms they pay to advertise in). Cultural entrenchment should not be disregarded: the public expects to receive their news on social media platforms they readily frequent, and many news outlets meet this expectation. Hence, news outlets must partake in building a presence in social media platforms if they wish to compete.
IV.            POLITICS AND DISTRIBUTION – WHOSE INTERESTS ARE STRENGTHENED AND WHO HAS FEWER CHOICES - C
A.            Barbook argues that although the California Ideology claims that state power will be decreased and individuals will be empowered through decentralized networks enabled by digital technologies, he demonstrates that public communication is in fact not immune to undemocratic powers. These include private corporations: these corporations may gatekeep in choosing when and how to provide access to their platforms, while finetuning how users navigate their platforms.
By enabling access to content that users tend to agree with, those who manage these platforms enable public opinion to shift to extremes. Being “fed” the same narrative and not being challenged to see an opposing perspective ultimately provides fewer choices to the consumer of the information, the general public. Therefore, individual agency is constrained by platform architecture and control, not liberated by it.
V.              AT WHAT POINT IN HISTORY WAS A DIFFERENT TRAJECTORY POSSIBLE? - R
A.            News not subscribing to / presenting news on social media as platform (?)
VI.            GOVERNANCE IMPLICATION – ONE PROPOSITION FOR CURRENT DAY GOVERNANCE - R
OPEN PLATFORMS (ALTERNATIVE FACEBOOK A)
<section class="footnotes">
 
#Mark Greengrass, Thierry Rentet and Stéphane Gal, ‘The Hinterland of the Newsletter: Handling Information in Space and Time’ in Joad Raymond and Noah Moxham (eds), News Networks in Early Modern Europe (Brill 2016) [http://www.jstor.org/stable/10.1163/j.ctt1w8h1ng.34 http://www.jstor.org/stable/10.1163/j.ctt1w8h1ng.34]accessed 7 September 2026; ‘Michel Serres Prononce En Sorbonne La Conférence Inaugurale Du Programme Paris Nouveaux Mondes - La Chancellerie Des Universités de Paris’ (La Chancellerie des Universités de Paris, 1 January 2013) [https://www.sorbonne.fr/michel-serres-prononce-en-sorbonne-la-conference-inaugurale-du-programme-paris-nouveaux-mondes/ https://www.sorbonne.fr/michel-serres-prononce-en-sorbonne-la-conference-inaugurale-du-programme-paris-nouveaux-mondes/]accessed 7 September 2026. James Ralph Beniger, The Control Revolution: Technological and Economic Origins of the Information Society (Harvard Univ Press 1986). [https://wiki.ivir.nl/index.php?title=Platform_dependence_in_public_communication&action=tinymceedit#fnref-1 ↩︎]
#Brendan Dooley, Making News in Renaissance Europe (1st edn, Cambridge University Press 2026) [https://doi.org/10.1017/9781009358835 https://doi.org/10.1017/9781009358835]. [https://wiki.ivir.nl/index.php?title=Platform_dependence_in_public_communication&action=tinymceedit#fnref-2 ↩︎]
#Andrew Pettegree, The Invention of News: How the World Came to Know about Itself (Yale University Press 2014). Helmer Helmers, ‘Public Diplomacy in Early Modern Europe: Towards a New History of News’ (2016) 22 Media History 401 [https://doi.org/10.1080/13688804.2016.1174570 https://doi.org/10.1080/13688804.2016.1174570]. [https://wiki.ivir.nl/index.php?title=Platform_dependence_in_public_communication&action=tinymceedit#fnref-3 ↩︎]
#James Ralph Beniger, The Control Revolution: Technological and Economic Origins of the Information Society (Harvard Univ Press 1986) P24-25 [https://wiki.ivir.nl/index.php?title=Platform_dependence_in_public_communication&action=tinymceedit#fnref-4 ↩︎]
#Richard Barbrook and Andy Cameron, The Californian Ideology (1996) [https://wiki.ivir.nl/index.php?title=Platform_dependence_in_public_communication&action=tinymceedit#fnref-5 ↩︎]
</section>

== [[First One]] ==

= Centralization of AI Compute and Data =

== Centralization of AI compute and Data - by Sisi Hammoudi & Juliette Chamot ==
=== Governance issue ===
Today, several path-dependence phenomena are emerging across advanced computing technologies. This assignment addresses one of them: the centralisation of AI compute and data. Training and running frontier AI systems now depends on a handful of hyperscale providers, namely Amazon, Microsoft, Google, and Nvidia, which dominates the GPU market. In other words, near-monopolies exist throughout the supply chain. This concentration creates a governance problem affecting public and private actors alike, both legally and institutionally. On the institutional side, it prevents public administrations, and states other than the US or China, from reaching sovereign AI capacity. It also raises competition-law concerns, since dominant market positions can prevent the market from delivering the best products at the best prices to consumers. Finally, AI developers, like states, are often forced into contracts with hyperscale providers, creating a structural dependence on them.


=== Historical lineage ===
==== 1945 ====
Wartime scientific instruments should have been redirected to extend human command over knowledge, helping establish the postwar model of sustained federal funding for computing research (Vannevar Bush's, "As We May Think », 1945).


==== 1990s ====
The privatisation of the NSFNET established the idea that critical digital infrastructure, even though publicly subsidised, could legitimately be privately owned. The EU Commission followed the US's lead, recommending privatisation and deregulation of telecommunications and media as the best way to build the "infobahn" though EU members could instead have chosen nationalisation of the infrastructure (Barbrook & Cameron, "The Californian Ideology," 1996).


==== 2006 ====
<div class="s13">Hyperscale cloud computing turned massive fixed capital investment in data centresinto a rentable service, favouring firms able to invest at scale.</div>
<div class="s13">Same period: Nvidia's proprietary CUDA software, released alongside its GPUs, became the de facto standard for machine learning, locking in a decade of tooling and expertise around one firm's hardware.</div>
<div class="s13">

==== From 2022 ====
Semiconductor export controls and the CHIPS Act turned compute into an explicit instrument of geopolitical strategy.</div>



=== Mechanisms of path dependance ===
Several mechanisms drive this centralisation, but path-dependence ones dominate. Because AI compute is a capital-intensive industry, new entrants struggle to compete with hyperscalers that have already sunk huge sums into infrastructure, a barrier that produces monopoly-like situations. This dynamic, known as increasing returns, generates system-scale economies that harden into a near-irreversibility of investment (David, "Clio and the Economics of QWERTY," 1985). It overlaps with a second mechanism, standardisation, itself a form of technical interrelatedness: researchers' investment in Nvidia's CUDA-based tooling locks the field into that ecosystem regardless of whether better alternatives exist. A further mechanism is legal and contractual entrenchment. For example, Microsoft's stake in OpenAI binds developers to a specific provider by agreement rather than technical necessity.
=== Politics and distribution ===
Technology does not have a history of distributing power equally. Rather, its development has strengthened the interests and authority of powerful groups that already possessed economic and political resources, and it is yet to be determined whether AI compute and data will follow suit. 
Author Langdon Winner illustrates the historical constraints of technological development with the case of New York from the 1920s to the 1970s. He argued that the city’s roads, bridges, and parks’ physical arrangement contained “explicit and implicit political purpose.” These infrastructures privileged wealthy and white communities while restricting poor and Black populations’ choice of mobility. Similarly, the development of industrial machinery increased productivity, simultaneously weakening union and non-union workers’ bargaining power.
In the twentieth century, following the Industrial Revolution, a Control Revolution emerged from the increasing need to manage expanding production, transportation, and information flows requiring strengthened organizational bureaucratic control. This was heavily shaped by the political context of post-World War II warfare.
Therefore, the bearer of the initial cost of technology is the person putting in the capital to erect it, but ultimately, the person bearing the long-term cost of technology, whose choices diminish, is not the rich and privileged, but the everyday person subject to technologies. Forced to adapt, to reshape their professional skills to adapt to an environment trying to make them obsolete.
AI compute and data are at risk of creating the same effect. In their critique of dotcom neoliberalism, authors Barbrook and Cameron exposed a similar socio-economic tension. The Californian Ideology claims digital technology empowers individuals; however, perpetual unequal access to it creates an apartheid between “information-rich” and “information-poor” populations.
Finally, the development of AI compute and data depends on expensive and scarce computing infrastructure and a highly concentrated semiconductor supply chain. Today, this is one of the most concentrated industries, with single companies having near-total monopolies over the production of chips necessary for the functioning of AI. This monopoly model threatens to reinforce the existing economic hierarchy.


=== Contingency and alternatives ===
If a different technological trajectory was plausible, the most apt time for it to have occurred would be during the transition from the Industrial Revolution to the “Control Revolution”. When industrial production accelerated, our society experienced a “crisis of control”, creating opportunities to organize information and communication differently.
Instead of allowing control technologies to be monopolized within businesses and states, our societies could have pursued a more democratic and decentralized system. Barbrook and Cameron provide such an alternative path through their idea of the “electronic agora,” where digital networks could encourage community participation and direct democracy, where government could actively engage with the people to shape digital infrastructure and expand access. These alternatives did not actually prevail because corporate and state market participants possessed greater resources and authority to shape technological development at the end of the twentieth century. 
Inherently, technological development is not predetermined; it can support either authoritarian or democratic arrangements. Today, monopolies control chip production needed for AI compute and data, but other technologies like blockchain permit democratic participation.


=== Governance implication ===
History always threatens to repeat itself. As Barbrook discussed, digital societies are shaped by the political context in which they develop. It is a valid concern, then, that the development of information technology, through AI computing, would follow the same trajectory- that is, be constrained to benefit the wealthy who funded its development, and ultimately harm the middle and lower classes enabling its functioning. 
However, AI’s place in our society has not yet crystallized. Technology governance does not have to adhere to this system. Such a seemingly sustainable approach to AI compute and data governance has already been broached. In 2022, the Notre Dame Journal on Emerging Technologies published an article in its issue on Artificial Intelligence where the authors proposed an AI compute system based on a “dual investment strategy”. Firstly, the model would be launched by “subsidizing and negotiating” cloud computing for human AI researchers through “existing” cloud vendors, which would expand on existing AI compute initiatives. Secondly, the compute system would be specifically invested in as a publicly owned infrastructure that would be built and operate similarly to national libraries which own “academic supercomputing facilities”. 
The pilot program would have a “base level of compute access” which would keep costs of processing down and enable it to meet the compute needs of a majority of academic researchers. Absent any strengthened legislation on the use of AI compute and data, the hardware would be accompanied by an AI ethics framework to enable sustainable technological governance. In addition to an ethical staff, the human resources needed to keep an AI compute model running would include data officers, scientists, data center and software engineers, research developers, and system administrators. These positions require specialized training, allowing new types of professionals, from any background, to have access to professional prosperity in the field of AI. 
By implementing these frameworks, never neglecting the human element, and the economic and intellectual impact AI has on our society, we move from a historical determinist perspective to one acknowledging historical constraints and learning from them.

== Centralization of AI compute and Data - by Atmika Dechamma & Pablo Latorre ==
=== Governance Issue ===
AI computing capacity and data presents the governance problem of an extreme concentration of technological infrastructure and information in a restricted number of companies that privately and globally dominate the market. This affects public and private actors - such as small and medium enterprises, potential new competitors, and Global South countries - that increasingly depend on AI systems outside their control and/or are deterred from competing with established companies because of impenetrable entry barriers - most notably initial investments and operational costs. The institutional and legal relevance of this is that it consolidates, reproduces and amplifies asymmetries of power in favor of the “Big Tech”, reverting the technological trend from digital decentralization to physical infrastructure dependency and curbing innovation and market efficiency.

=== Historical Lineage ===
==== Limitation of Machine Learning ====
The AI revolution was built on the limitation of human capabilities to process large information and derive efficient results - where before there existed only facilitative tools to allow for human-enabled processes, AI now offered derivations and cognitive reasoning in moments and at large scales.

 
==== Revert to Physical Infrastructure Dependency in the rush to be the best ====
The present AI model heavily relies on hardware in order to perform complex data processing for deep learning. The data centre models followed require larger model sizes (which require higher resource, information and capital investment) to be able to yield better results. Due to the limitation of resources and a rush to establish the best working model of an AI agent, data centres are built on a rudimentary model heavily dependent on massive amounts of electricity and fresh water.

 
==== Regulation and Decentralization of Information ====
Regulation and influence of sovereigns and institutions has resulted in data sharing and information exchange structurally through open source and cloud computing. But, while AI models claims their selling point to be accessibility for all, their information and derivations remain in a closed server, stored for the purpose of training their AI agents to be the best. Big AI contenders attract higher consumer and information flows overall resulting in a better developed AI model, proving disadvantageous in any market where early contenders have an advantage and accumulation of power.

=== Mechanisms of Path Dependence ===
Early on in the industry there is an expectation of human and capital investment from all stakeholders involved, this high cost for entry into the market and expected capital to sustain and develop a model by both consumer and enterprise makes for a high switching cost. For the AI model to run and scale there is a presumption that the rate of growth is dependent on the size of data fed, resulting in a better model and hence, more consumer usage. A closed centralised data loop is created but the requirement for storage and processing far surpasses current infrastructures capabilities - heavily relying on early market entry advantage and sustained capital investments taking away from smaller aspirants and hoarding of power within closed circles. Despite its physical dependency, AI data processing has unique requirements beyond legacy infrastructures - larger demand and production of GPUs and designed to handle immense thermal heat generated while consuming electricity beyond current capabilities. 

 The AI space largely remains unregulated due to its undefined form and fluid nature, however, this has given way for unethical and morality issues regarding data privacy, exploitation of labour and environmental harms.

=== Politics and Distribution ===
Concentration strengthens the values, interests, capacities, and authority of the “Big Tech” over other actors of public and private nature. In this context, Barbrook and Cameron (1) illuminate the ethos behind AI technology development while Winner (2) makes observable the power relations surrounding it: concentration is the consequence of an AI development anchored in the belief that it must be free from all institutional and normative restriction since it is inherently liberating and non-political which, in turn, provides a political/moral justification for the concentration (if technology is inherently liberating and non-political then there is no power relations to observe or question) that translates itself into power dynamics where the AI developer not only owns the AI technology but also decides how it is trained and deployed and which models the user can or cannot access.

=== Contingency and Alternatives ===
Concentration is due, first and foremost, to the fact that AI computing capacity requires massive and cutting-edge physical infrastructure - such as data centers and microprocessors - which, in turn, requires upfront investments that are unaffordable to most companies. In David’s (3) language, companies that could finance the infrastructure and position themselves first obtained self-reinforcing advantages and a “lock-in” effect in the market. In other words, concentration is the consequence of the development of AI physical infrastructure privately with no mixed or public alternatives as a public service. Therefore, there was a space for an alternative future at the end of the 1980s or the beginning of the 1990s when the privatization model of the technological infrastructure was consolidated in line with the “Californian Ideology”. In that moment governments had the chance to treat technological infrastructure as a mixed or public good rendering it universally accessible as it was the case of the Internet; this would have permitted companies of all sizes to profit from economies of scale, reducing entry barriers and fostering competition. However, this alternative path did not prevail since the Californian Ideology provided the moral/political justification to exclude governments from the development of technological infrastructure in the exact moment in which such a decision would produce structural consequences. If that path had been followed it could have contributed to collective access to economies of scale, lower entry barriers, and a more competitive AI market.

=== Governance Implication ===
Considering the above, public or mixed investment on AI physical infrastructure is a necessary governance measure for market decentralization and collective access to AI computing capacity and data. From an historical perspective, the proposed measure is not merely a corrective action for a market failure but for a contingent and ideological decision made between the late 1980s and early 1990s that produced a structural “path dependence” that now is necessary to change.
==== References ====
(1) [https://git.xpub.nl/onebigear/SI_09_reference_library/raw/commit/22bf61e1dcfed35d58fba23808212bd00584be47/Richard_Barbrook_Andy_Cameron_Californian_ideology.pdf Barbrook, R., & Cameron, A. (1996). The Californian Ideology. Science as Culture, 6(1), 44-72].

(2) [https://d1wqtxts1xzle7.cloudfront.net/87708502/langdon_winner_do_artifacts_have_politics-libre.pdf?1655588533=&response-content-disposition=inline%3B+filename%3DDo_Artifacts_Have_Politics.pdf&Expires=1788890463&Signature=UBBiMqkip5MfYZbkbLNAN6aE2hyZdKkER-HFcmyxXkwIZuTnMKdOGoao8ddFr-rQ5T55A8NWUOlnZkGYmi66e7XR9h99DgTELMCrh48sXjOb8KYgA48r~SUUymt1IW~R~aOKJzitMIUHcrc6yEGKK1OuaA~mrEIVOXpUjQMYjqOd26Xmk9YJEglTkqaxeR47VCCth5FvozNQIowdfjzeu8aNHnI8V7Yur1IiXY~PGvYRcN6wBtcgpVMHZUtNaI8NjmYbp2p5TBYrPMF9MCLtvpGhPlsCfsvA8Bdm6XulGwRlksZBKWvVUQ8D~M8oLu12V39YaiETIObyJqbE7ctIcQ__&Key-Pair-Id=APKAJLOHF5GGSLRBV4ZA Winner, L. (2017). Do artifacts have politics?. In Computer ethics(pp. 177-192). Routledge].

(3) [https://sjbae.pbworks.com/f/David+1985.pdf David, P. A. (1985). Clio and the Economics of QWERTY. The American economic review, 75(2), 332-337].

== Centralization of AI compute and Data - by Sergey Abramov and Paul Stein ==
<div>
=== The current governance issue ===
Advanced AI currently depends on a narrow group of tech giants1 and semiconductor oligopolies controlling microchips and software, hyperscale cloud infrastructure, large datasets, frontier models, and distribution platforms. This presents a governance problem where this narrow group exercises control over foundational digital infrastructure. This directly affects consumers, businesses, and academia out of AI. Legally and institutionally, this matters because it creates a profound democratic deficit: this neutralizes antitrust frameworks designed for traditional markets, often bypassing conventional copyright and data privacy protections through unchecked data harvesting thereby complicating enforcement.This ultimately might affect national sovereignty as the countries that host the buildout will shape the future of AI: who controls it, what values it contains, how it is used2.
 
=== Historical lineage ===
In 1987, Taiwan Semiconductor Manufacturing Company (TSMC) was founded, establishing the world’s first semiconductor fab. Instead of designing and making its own chips, TSMC only manufactured chips designed by other companies. Today, nearly 100% of the world's most advanced AI processors are manufactured by TSMC, leaving the entire global AI compute supply chain centralized under one company.
 In 2006, Nvidia launched CUDA (Compute Unified Device Architecture), a free software layer that allowed developers to use Nvidia's graphics chips (GPUs) which later became perfect for training deep neural networks. Today, this creates a near-monopoly on AI compute hardware.
 Soon after Amazon launched AWS, Microsoft and Google launched their own cloud platforms marking the structural pivot to the cloud service provider model, shifting computing from a distributed architecture (where companies bought and managed their own physical servers) to a centralized infrastructure. As training modern AI models requires tens of thousands of GPUs, today’s frontier AI companies are entirely dependent on the extremely expensive physical infrastructure of a handful of cloud hyper-scalers (Microsoft, Google, AWS, and Meta).
 In 2009, Fei-Fei Li launched ImageNet, a massive database of over 14 million images, which proved that AI model performance scales directly with data volume. This created an immediate institutional advantage for consumer internet platforms (like Google and Meta), because they were already architected to collect, centralize, and store masses of user data through search engines and social networks, they possessed the unique data stack required to train high-performing AI.
 
=== Mechanism of path dependence ===
The centralization of AI compute and data is a classic case of technological lock-in. When structural choices were made decades ago, they triggered powerful, self-reinforcing loops. One such loop is the switching costs and software standardisation. As developers have spent more than a decade working with CUDA, moving to a rival hardware provider (or a decentralized compute network) requires an enterprise to completely rewrite its foundational software stack. The cost of code migration, developer retraining, and potential hardware downtime force companies to stay tied to dominant hardware monopolies such as Nvidia. Commitments, licensing conditions, and exclusivity provisions add contractual costs. Fewer than 1% of cloud customers switch provider annually, identifying technical differentiation and data-transfer charges as important barriers4.
 Network effects also matter for computing power. More customers generate more revenue, usage information, and opportunities to improve infrastructure and models. Better services attract additional users and developers, whose applications make the platform more valuable. In cloud–AI partnerships, providers may additionally obtain information about model performance and future infrastructure demand. These feedback loops strengthen firms already operating across cloud, models, software, and distribution.
 
=== Politics and distribution ===
According to Langdon Winner's definition, modern AI computing infrastructure may qualify as an "inherently political technology" because its large scale and significant financial requirements not only lend themselves to, but necessitate,5 centralized, hierarchical control. The physical siting of this infrastructure shows evidence of Winner’s second category of politicization of technologies, the ‘political arrangement’:6 Of the five largest U.S. AI firms, four are headquartered in California,7 while just one is (partially) domiciled in Texas. Texas, however, hosts twice the GPU clusters.8 This placement geographically insulates beneficiaries of the data infrastructure boom from its negative externalities. The specific choice of location may itself be a deliberate preference on account of Texas’ unique approach to water usage rights.9 Juxtaposition of this unequal and even exploitative balance of interests with the pontificating rhetoric of the urban, information rich, beneficiaries is evocative of the internal contradictions of the Californian Ideology - the virtual artisans existence and privileged position is still based on the exploitation of a vulnerable out-group, the "information-rich" consuming the output, the "information-poor" hosting the extraction.10
 
=== Contingency and Alternatives ===
Profit motive and economies of scale now concentrate investment among a few hyperscalers,11 but this reflects policy choice, not government incapacity: Compute infrastructure may be inherently political, but the technology is agnostic as to where the organizing authority originates, and states have historically funded infrastructure of comparable or greater scale, while facing less pressure to show short-term returns. The clearest turning point was 1995, when the National Science Foundation retired NSFNET, the publicly funded backbone that had connected U.S. research and supercomputing centers since the mid-1980s,12 handing internet infrastructure to commercial providers with no conditions attached. This was a deliberate policy decision made amid the deregulatory push of the Clinton era. The privatization that followed ushered in the private industrial dominance that persists today in the form of hyperscalers. Had a publicly governed backbone instead been scaled up, today's compute infrastructure might rest on a different foundation entirely.
 
=== Governance implication ===
Despite private hyperscaler dominance, the U.S. does have a nominal public alternative: the National AI Research Resource (NAIRR), established following the National AI Initiative Act of 2020 and piloted from January 2024, which explicitly aims to democratize access to the computational resources essential to AI research.13In Europe, a comparable program is the EuroHPCJU,14 an initiative intended to pool resources to build supercomputing capacity. The undertaking’s legislative basis has since been extended to include networked ‘AI factories', as well as the so-called “Gigafactories”.15 The framing in language of sovereignty indicates intent to at least compete with private infrastructure initiatives. Though government-led measures cannot reverse the centralization of compute, they can democratize access - provided they are intensified. The European Commission has “set a goal of investing more than €1 billion per year in AI research and innovation”.16 While laudable in its own right, it pales in comparison to the investments of industry leaders, which are approaching, if not exceeding, trillions of Dollars.

=== References & Notes ===
1.Amazon, Google, Meta, Microsoft, and Oracle collectively hold an estimated 71% of the world’s cumulative AI compute as of Q4 2025, measured in H100-equivalents of computing power. Luke Emberson, Josh You and Venkat Somala (2026), "Five hyperscalers now own over two-thirds of global AI compute". Published online at epoch.ai. Retrieved from 'https://epoch.ai/data-insights/hyperscalers-control-most-compute' [online resource]. Accessed 7 Sept 2026.2. Alasdair Phillips-Robins, Teddy Tawil and Sam Winter-Levy, ‘The Compute Coalition: How to Build the Future of AI in the Free World’ (Carnegie Endowment for International Peace, 8 June 2026) https://carnegieendowment.org/research/2026/06/the-compute-coalition-how-to-build-the-future-of-ai-in-the-free-world accessed 7 September 2026.

3. Amazon Web Services (AWS), a cloud computing platform

4. Competition and Markets Authority, Cloud Infrastructure Services: Final Decision Report (31 July 2025) https://assets.publishing.service.gov.uk/media/688b8891fdde2b8f73469544/final_decision_report.pdf accessed 7 September 2026.

5. Langdon Winner, 'Do Artifacts Have Politics?' (1980) 109(1) ''Daedalus'' 121, 128.

6. Ibid. 123.

7. OpenAI, Anthropic, Google DeepMind, Meta AI. xAI is an exception to the rule, domiciled between Palo Alto, CA and San Antonio, TX.

8. Texas’ 17 to California’s 8, cf. Texas Royalty Brokers, 'AI Clusters by State' (''Texas Royalty Brokers'', 2025) https://texasroyaltybrokers.com/ai-clusters-by-state/ accessed 7 September 2026.

9. The ‘rule of capture’, established in Houston & TC Railway Co v East (Tex 1904), permits landowners to extract groundwater beneath their land without liability for depleting adjacent wells or aquifers, a doctrine now exploited by hyperscale AI data centres to draw large cooling-water volumes with minimal regulatory constraint.

10. Richard Barbrook and Andy Cameron, 'The Californian Ideology' (1996) 6(1) ''Science as Culture'' 44.

11. Goldman Sachs, 'Global AI Investment Is Forecast to Exceed $1 Trillion in 2026' (''Goldman Sachs Insights'', 7 August 2026) https://www.goldmansachs.com/insights/articles/global-investment-is-forecast-to-exceed-1-trillion-in-2026 accessed 7 September 2026.

12.  National Science Foundation, 'Birth of the Commercial Internet' (''NSF Impacts'') https://www.nsf.gov/impacts/internet accessed 7 September 2026.

13. CASRAI, 'NAIRR: National AI Research Resource, Explained' (''CASRAI'') https://casrai.org/dictionary/term/national-ai-research-resource-nairr accessed 7 September 2026.

14. Council Regulation (EU) 2021/1173 of 13 July 2021 on establishing the European High Performance Computing Joint Undertaking and repealing Regulation (EU) 2018/1488 [2021] OJ L256/1.

15. Council Regulation (EU) 2026/150 of 16 January 2026 amending Regulation (EU) 2021/1173 on establishing the European High Performance Computing Joint Undertaking [2026] OJ L 2026/150.

16. European Commission, 'Realising Europe's Digital Decade' (''European Commission'', 11 October 2024) https://commission.europa.eu/strategy-and-policy/priorities-2019-2024/story-von-der-leyen-commission/realising-europes-digital-decade_en accessed 7 September 2026.</div>