Pith. sign in

REVIEW 4 major objections 6 minor 59 references

Collective Bargaining in the Information Economy Can Address AI-Driven Power Concentration

T0 review · 4 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Collective bargaining in the information economy can prevent AI from concentrating power and collapsing the informational commons.

desk verdict A useful synthesis and agenda for collective data bargaining, but the load-bearing assumption of credible withholding goes unsupported. read the letter →

arxiv 2506.10272 v1 pith:U7V2VRWY submitted 2025-06-12 cs.CY cs.HC

Nicholas Vincent , Matthew Prewitt , Hanlin Li This is my paper
classification cs.CYcs.HC
keywords collectivebargaininginformationmarketfailureAIpowerconcentrationdatalabortrustedintermediariesinformationalcommonsvaluationantitrustsafeharbor
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This position paper argues that AI will deepen a long-standing failure in markets for information: because information can be copied at almost no cost, competitive markets price it near zero, so the actors best able to aggregate and process it capture most of its value while the people who produce it are undercompensated. As AI lowers the cost of extracting insight from any public information, this failure becomes worse and pushes value into a few capital-rich AI operators, a trajectory the authors call a 'capital singularity' that can end in collapse of the informational commons. The paper's proposal is collective bargaining in the information economy: producers of news, research, and creative work organize into trusted data intermediaries or guilds that pool their data and negotiate with AI builders over how it may be used and what it is worth. It argues this is the shortest, perhaps only, path to a sustainable AI ecosystem, and it sets out legal, technical, and social actions needed to make bargaining possible. A sympathetic reader would care because the argument offers a market-structure fix rather than relying on top-down regulation or openness alone.

What carries the argument

The central object is 'collective bargaining in the information economy' (CBI), defined as negotiation between trusted data intermediary organizations representing large groups of information producers and the organizations that build information-aggregating AI systems. The bargaining power behind CBI is the credible threat to withhold or block access to data that AI builders cannot easily substitute, which converts information from a freely scraped commons into a licensable asset. The technical supports are data valuation and attribution methods—interpretability, influence, and similar value estimates—that let both sides estimate what information is worth, plus federated data management, provenance tracking, and escrow systems. The legal support is antitrust 'safe harbor' for producer-side joint ventures, so that organizing does not expose journalists, researchers, and creative workers to accusations of collusion.

What would settle it

A controlled test would remove a large, high-value corpus (for example, a year of major journalism) from training a frontier language model and replace it with synthetic or freely scraped text; if downstream knowledge and reasoning benchmarks are essentially unaffected, the threat to withhold is not credible, and the leverage that collective bargaining relies on would be absent.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central claim is that the same economic property that has always made information markets fail—non-rivalry, the ability to copy information at trivial cost—is being intensified by AI, so that in the absence of intervention the social value of recorded information will accrue to a small number of actors with capital and computational scale. The authors argue that this concentration is not just an equity problem but a safety and capability problem: if information producers cannot earn returns, the supply of fresh, diverse, human-produced information shrinks, models are trained increasingly on scraped or synthetic content, and both the quality and accountability of AI decline. Their solution is to introduce deliberate 'market frictions' on the producer side through collective bargaining, so that information is neither fully private nor fully public but held by trusted representatives who can license it selectively. They claim that this arrangement simultaneously restores production incentives, distributes value, creates multiple sites of durable market power, and gives non-market actors the private information needed to audit and test AI systems.

Load-bearing premise

The plan depends on information producers being able to keep AI companies from training on their work without permission, and on AI builders not being able to substitute scraped or synthetic data at trivial cost, because only then is the threat to withhold data a credible source of bargaining power.

Editorial extensions

If this is right

  • Information producers in news, science, and creative work would be able to negotiate over training use and receive ongoing compensation, restoring incentives that competitive information markets destroy.
  • AI builders would gain access to higher-quality, more diverse, and better-documented data, which the paper argues makes models more capable and easier to audit.
  • Power concentration would be mitigated because value would flow to multiple data intermediaries rather than accruing almost entirely to a few capital-rich AI operators.
  • Regulators could issue safe-harbor guidance clarifying that producer-side data joint ventures do not violate antitrust law, making large-scale organizing legally viable.
  • Technical work on data valuation and attribution would gain a concrete application: setting the terms of compensation in real data licenses.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A sector-level pilot—for instance, a coalition of news organizations that collectively licenses a corpus and blocks unnegotiated scraping—could test whether license revenue appears and how downstream model performance changes compared with a control corpus.
  • If blocking unnegotiated training turns out to be technically or legally infeasible at scale, the proposal could shift toward legal mandates, such as a training-data licensing right, that create the same friction without relying on technical enforcement.
  • The paper notes that collective action can create 'labor cartel' conditions, but it leaves open the governance question; an implication is that how data intermediaries are chosen and held accountable will determine whether this remedy improves or worsens the commons.
  • Data valuation methods developed for bargaining could double as a regulatory accounting tool, letting governments estimate the data-dependence contribution owed by large model operators.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 6 minor

Summary. This position paper argues that AI-driven advances in information processing will worsen long-standing market failures for information goods, driving the returns to information toward concentrated pools of capital and toward a hypothesized 'capital singularity.' The proposed remedy is 'collective bargaining in the information economy' (CBI): information producers—journalists, researchers, creative workers, and data contributors—would organize into trusted intermediaries and collectively negotiate terms and compensation with AI builders. The paper grounds this proposal in information economics (Arrow's paradox, nonrivalry) and prior work on data labor and data leverage, and it offers a concrete agenda: private joint ventures to pool data and block un-negotiated training, safe-harbor antitrust guidance for such ventures, interpretability and data-valuation tools, human-data interaction design, and advocacy from AI safety and technology actors. The central claims are that without CBI, AI will cause an 'information market failure' and 'ecological collapse' of the informational commons, while CBI can create the market frictions needed to sustain information production and distribute power.

Significance. If the proposed mechanism were sound, the paper would provide a timely and actionable agenda at the intersection of AI governance, labor organizing, and information economics. Its strength is that it assembles a coherent synthesis of existing research—data as labor, data leverage, data dignity, and data-sharing consortia—and translates it into a concrete set of interventions for multiple stakeholders (producers, regulators, researchers, and companies). The paper is transparent about its nature as a position paper and even flags a potential downside of collective action (the 'labor cartel' concern in footnote 3). However, its contribution is agenda-setting rather than demonstrative: it presents no new data, no formal model, and no empirical test. The falsifiable predictions it implies (e.g., that uncoordinated information markets will produce extreme concentration, or that CBI will mitigate it) are stated without the evidential support needed to make them persuasive to a skeptical reader. Its value therefore depends on how much weight one assigns to the prior literature it builds on and to the plausibility of the bargaining-power mechanism it assumes.

major comments (4)
  1. [Section 2.3] The leap from reduced informational frictions to a 'capital singularity' is asserted rather than argued. The paper states that 'AI progress is likely to resolve this ambiguity in favor of more consolidated capital' and then posits an extreme feedback loop, but it offers no model, no calibrated example, and no engagement with countervailing forces (e.g., falling costs of capital access, open-weight models, or regulatory intervention). Because this is the central urgency claim that motivates the entire proposal, the paper should either provide a more rigorous argument—for instance, deriving conditions under which the feedback loop dominates—or explicitly label the capital-singularity scenario as one of several possible futures rather than the expected outcome. Without this, the reader cannot assess whether the dramatic policy response is proportionate.
  2. [Section 1.1 and Table 1] The credibility of the bargaining threat rests on the assumption that information producers can 'prevent AI companies from training on it through backdoor access' and 'block un-negotiated AI training routes.' The paper does not establish that such blocking is technically and legally feasible at scale, nor that AI builders cannot substitute equivalent data through web scraping, licensed alternatives, already-held corpora, or synthetic-data generation. Indeed, the paper itself cites evidence that AI labs have trained on scraped and pirated content (Section 2.1, [26]) and that the AI data commons is declining ([33]); this cuts both ways, because it suggests that labs may already hold much of the relevant data, making future withholding less consequential. If a coalition's corpus is replaceable, the bargaining leverage is weak and the 'market frictions' CBI promises may never materialize. The paper should address substitution possibilities head-on or qualify the conditions under which CBI's leverage is credible.
  3. [Section 1.4] The dismissal of alternatives—top-down regulation and open AI—is too quick to support the claim that CBI is 'the shortest (and perhaps only) path.' The objections raised (regulatory capture, coordination across sovereigns, frontier-model advantages, capital access for open models) are plausible, but the paper does not engage with hybrid approaches that combine regulation, openness, and collective bargaining, nor does it cite evidence on the effectiveness of these alternatives. The assertion that regulation 'cannot address all of the issues' and that open AI 'may fail' is a weak basis for concluding that CBI is uniquely necessary. A more balanced comparison, acknowledging complementarities, would strengthen the paper's case.
  4. [Footnote 3 and Section 4] The paper acknowledges in a footnote that collective action from labor could create 'labor cartel' conditions that are 'also unhealthy for markets,' but it never incorporates this into the analysis. Since CBI is explicitly intended to create market frictions, the same frictions could also produce rent-seeking, exclusion of new entrants, or higher costs for downstream users, undermining the 'pro-social, sustainable' characterization. The paper should discuss these potential costs more fully and explain why the benefits are expected to dominate, especially given that the proposed safe harbors would protect producer coordination from antitrust scrutiny.
minor comments (6)
  1. [Abstract and Section 3] The term 'ecological collapse' in the informational commons is used prominently but never defined; the body discusses declining data commons and reduced incentives, but a metaphor of ecological collapse would benefit from an explicit mapping to the mechanisms described.
  2. [References] There are several formatting errors: 'V ol.' in reference [6], 'V olume 1' in [34], 'Access on [insert date]' in [28], 'The American economic review' in [19], and 'Friedrich A V on Hayek' in [50] should be cleaned up before publication.
  3. [Table 1] The entry 'Ford's "pay your customers" principle' is cryptic; the paper should explain this reference or replace it with a more self-contained description of the intended behavior.
  4. [Section 1.4] The phrase 'Another view to consider is that leaning heavily into open AI can act as an effective approach for power decentralization' is grammatically awkward; consider revising for clarity.
  5. [Section 2.2] The discussion of copyright and fair use is brief given its centrality to the feasibility of blocking training data; the paper would benefit from addressing the ongoing legal uncertainty around whether training on publicly available data is fair use, and how CBI would interact with that uncertainty.
  6. [Section 4] The proposed safe harbor cites 'Parker Immunity' without explaining what it is; non-US readers or readers outside antitrust law will not be able to assess this proposal without a short explanation or citation.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: position paper builds an agenda from cited scholarship and argued premises; no fitted parameter is renamed a prediction.

full rationale

This is a position paper, not a derivation with equations or fitted parameters. Its central claim is normative and programmatic: that collective bargaining in the information economy (CBI) can mitigate AI-driven power concentration and information-market failure. The supporting premises—information markets fail because marginal cost approaches zero, AI reduces informational frictions, and coordinated producers can gain bargaining leverage—are argued from external economics literature (Arrow, Jones and Tonetti, Akerlof) and from prior scholarship on data labor and data leverage. Some of that scholarship is authored by the present paper's coauthors (e.g., [48], [49], [32], [33]), but the paper explicitly says it 'draws on' this work rather than treating it as an unverified theorem that forces the conclusion. No self-citation is used to forbid alternatives or to supply a uniqueness result; alternative views (top-down regulation, open AI) are discussed and rejected on independent substantive grounds. The paper also acknowledges a limitation in footnote 3, noting that collective action from labor could create unhealthy 'labor cartel' conditions. Because there is no quantitative prediction, no fitted input renamed as a prediction, and no definition that presupposes the conclusion, no circular step can be exhibited. The honest finding is therefore no significant circularity.

Assumptions & free parameters 0 free parameters · 6 assumptions · 0 invented entities

The paper's argument rests on economic principles about information, speculative predictions about AI and capital, and assumptions about the feasibility of coordination, legal change, and technical valuation. It introduces no new physical entities or fitted parameters; the 'capital singularity' label is a scenario, not an entity.

assumptions (6)
  • domain assumption Information is non-rival and cheap to copy, so competitive markets drive its price toward zero.
    Section 1.2 and 2.1 introduce this as the foundation of the information market failure argument, citing Arrow (1962) and Grossman-Stiglitz (1980). It is a standard economic premise but is treated as a given.
  • domain assumption AI progress reduces the economic importance of informational frictions and increases the role of capital, making extreme concentration likely.
    Section 2.2 and 2.3 predict that AI will diminish IP and extraction barriers and resolve capital-return ambiguity in favor of consolidation. This is a speculative empirical claim, not established.
  • ad hoc to paper Information producers can organize into effective coalitions and trusted intermediaries despite collective action problems.
    Sections 1.1 and 4 propose the formation of joint ventures and coalitions; the paper cites collective action literature but does not demonstrate the feasibility for large, diffuse producer groups.
  • ad hoc to paper Antitrust and other regulators will provide safe harbors for joint bargaining, or at least will not prevent it.
    Section 1.1 and 4 say legal concerns are foundational and call for safe harbor rules (e.g., via Parker Immunity); the proposal depends on this legal change.
  • ad hoc to paper Information producers can technically prevent AI companies from accessing their data without negotiated terms.
    Section 1.1, Private Initiatives, says producers should block 'backdoor access' by AI companies; the paper provides no evidence that this is feasible at scale, and this is the key to producers' leverage.
  • domain assumption Data value estimation and attribution methods can become accurate enough to support informed bargaining.
    The proposal leans on 'Interp and Attrib' research (Section 4) to quantify data's causal impact; the accuracy of such methods is unproven in practice.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Collective Bargaining in the Information Economy Can Address AI-Driven Power Concentration." pith.science (2026). https://pith.science/paper/U7V2VRWY

@misc{pith2026250610272,
  author       = {Pith},
  title        = {Pith review of: Collective Bargaining in the Information Economy Can Address AI-Driven Power Concentration},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/U7V2VRWY}},
  note         = {Machine review of arXiv:2506.10272}
}
read the original abstract

This position paper argues that there is an urgent need to restructure markets for the information that goes into AI systems. Specifically, producers of information goods (such as journalists, researchers, and creative professionals) need to be able to collectively bargain with AI product builders in order to receive reasonable terms and a sustainable return on the informational value they contribute. We argue that without increased market coordination or collective bargaining on the side of these primary information producers, AI will exacerbate a large-scale "information market failure" that will lead not only to undesirable concentration of capital, but also to a potential "ecological collapse" in the informational commons. On the other hand, collective bargaining in the information economy can create market frictions and aligned incentives necessary for a pro-social, sustainable AI future. We provide concrete actions that can be taken to support a coalition-based approach to achieve this goal. For example, researchers and developers can establish technical mechanisms such as federated data management tools and explainable data value estimations, to inform and facilitate collective bargaining in the information economy. Additionally, regulatory and policy interventions may be introduced to support trusted data intermediary organizations representing guilds or syndicates of information producers.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

59 extracted references · 43 canonical work pages

  1. [26]

    2025.Cloze Encounters: The Impact of Pirated Data Access on LLM Performance

    Stella Jia and Abhishek Nagaraj. 2025.Cloze Encounters: The Impact of Pirated Data Access on LLM Performance. Technical Report. National Bureau of Economic Research

    work page 2025

  2. [33]

    Shayne Longpre, Robert Mahari, Ariel Lee, Campbell Lund, Hamidah Oderinwale, William Brannon, Nayan Saxena, Naana Obeng-Marnu, Tobin South, Cole Hunter, Kevin Klyman, Christopher Klamm, Hailey Schoelkopf, Nikhil Singh, Manuel Cherep, Ahmad Anis, An Dinh, Caroline Chitongo, Da Yin, Damien Sileo, Deividas Mataciunas, Diganta Misra, Emad Al- ghamdi, Enrico S...

    arXiv 2024

  3. [1]

    Marah Abdin, Jyoti Aneja, Harkirat Behl, Sébastien Bubeck, Ronen Eldan, Suriya Gunasekar, Michael Harrison, Russell J Hewett, Mojan Javaheripi, Piero Kauffmann, et al . 2024. Phi-4 technical report.arXiv preprint arXiv:2412.08905(2024)

    arXiv 2024

  4. [2]

    Simona Abis and Laura Veldkamp. 2020. The Changing Economics of Knowledge Production. https://doi.org/10.1093/rfs/hhad059

    doi:10.1093/rfs/hhad059 2020

  5. [3]

    Daron Acemoglu. 2025. The simple macroeconomics of AI.Economic Policy40, 121 (2025), 13–58. Publisher: Oxford University Press

    work page 2025

  6. [4]

    Daron Acemoglu, Ali Makhdoumi, Azarakhsh Malekian, and Asu Ozdaglar. 2022. Too much data: Prices and inefficiencies in data markets.American Economic Journal: Microeconomics 14, 4 (2022), 218–256

    work page 2022

  7. [5]

    George A Akerlof. 1978. The market for “lemons”: Quality uncertainty and the market mechanism. InUncertainty in economics. Elsevier, 235–251

    1978

  8. [6]

    Imanol Arrieta-Ibarra, Leonard Goff, Diego Jiménez-Hernández, Jaron Lanier, and E Glen Weyl

    work page

Show all 59 references
  1. [7]

    1962.Economic welfare and the allocation of resources for invention

    Kenneth Joseph Arrow. 1962.Economic welfare and the allocation of resources for invention. Springer

    1962

  2. [8]

    David B Audretsch. 2015. Joseph Schumpeter and John Kenneth Galbraith: two sides of the same coin?Journal of Evolutionary Economics25 (2015), 197–214

    2015

  3. [9]

    Adrien Basdevant, Camille François, Victor Storchan, Kevin Bankston, Ayah Bdeir, Brian Behlendorf, Merouane Debbah, Sayash Kapoor, Yann LeCun, Mark Surman, et al . 2024. Towards a framework for openness in foundation models: Proceedings from the columbia convening on openness ...

    2024 arXiv

  4. [10]

    Brian Callaci and Daniel Hanley. 2025. A Roadmap to Sectoral Bargaining Through Parker Immunity. (May 6 2025). https://papers.ssrn.com/sol3/papers.cfm?abstract_ id=5219945

    2025

  5. [11]

    Raul Castro Fernandez. 2023. Data-sharing markets: model, protocol, and algorithms to incentivize the formation of data-sharing consortia.Proceedings of the ACM on Management of Data1, 2 (2023), 1–25

    2023

  6. [12]

    Paul A David. 1993. Intellectual property institutions and the panda’s thumb: patents, copyrights, and trade secrets in economic theory and history.Global dimensions of intellectual property rights in science and technology19, 3 (1993), 29. 10

    1993

  7. [13]

    Gregoire Deletang, Anian Ruoss, Paul-Ambroise Duquenne, Elliot Catt, Tim Genewein, Christo- pher Mattern, Jordi Grau-Moya, Li Kevin Wenliang, Matthew Aitchison, Laurent Orseau, et al

  8. [14]

    Hannah Devlin and Hannah Devlin Science correspondent. 2023. AI ‘could be as transformative as Industrial Revolution’.The Guardian (May 2023). https://www.theguardian.com/technology/2023/may/03/ ai-could-be-as-transformative-as-industrial-revolution-patrick-vallance

    2023

  9. [15]

    Jennifer Ding, Christopher Akiki, Yacine Jernite, Anne Lee Steele, and Temi Popo. 2023. Towards openness beyond open access: User journeys through 3 Open AI Collaboratives.arXiv preprint arXiv:2301.08488(2023)

    2023 arXiv

  10. [16]

    Virginia Doellgast and Chiara Benassi. 2020. Collective bargaining. InHandbook of research on employee voice. Edward Elgar Publishing, 239–258

    2020

  11. [17]

    Tyna Eloundou, Sam Manning, Pamela Mishkin, and Daniel Rock. 2024. GPTs are GPTs: Labor market impact potential of LLMs.Science384, 6702 (2024), 1306–1308

    2024

  12. [18]

    John Kenneth Galbraith. 1967. The new industrial state.Antitrust L. & Econ. Rev.1 (1967), 11

    1967

  13. [19]

    Sanford J Grossman and Joseph E Stiglitz. 1980. On the impossibility of informationally efficient markets.The American economic review70, 3 (1980), 393–408

    1980

  14. [20]

    Luxi He, Yangsibo Huang, Weijia Shi, Tinghao Xie, Haotian Liu, Yue Wang, Luke Zettlemoyer, Chiyuan Zhang, Danqi Chen, and Peter Henderson. 2025. Fantastic Copyrighted Beasts and How (Not) to Generate Them. https://doi.org/10.48550/arXiv.2406.14526 arXiv:2406.14526 [cs]

    2025 doi arXiv

  15. [21]

    Douglas D Heckathorn. 1989. Collective action and the second-order free-rider problem. Rationality and society1, 1 (1989), 78–100. Publisher: Sage Publications

    1989

  16. [22]

    Dan Hendrycks, Mantas Mazeika, and Thomas Woodside. 2023. An overview of catastrophic AI risks.arXiv preprint arXiv:2306.12001(2023)

    2023 arXiv

  17. [23]

    Anders Humlum and Emilie Vestergaard. 2025. Large Language Models, Small Labor Market Effects.University of Chicago, Becker Friedman Institute for Economics Working Paper 2025-56 (2025)

    2025

  18. [24]

    Issues. 2024. Generative AI Is a Crisis for Copyright Law. https://issues.org/ generative-ai-copyright-law-crawford-schultz/

    2024

  19. [25]

    Klein, Alex Krasodomski, Joshua Tan, and Eleanor Tursman

    Brandon Jackson, B Cavello, Flynn Devine, Nick Garcia, Samuel J. Klein, Alex Krasodomski, Joshua Tan, and Eleanor Tursman. 2024.Public AI: Infrastructure for the Common Good. Technical Report. Public AI Network.https://doi.org/10.5281/zenodo.13914560

    2024 doi

  20. [27]

    Charles I Jones and Christopher Tonetti. 2020. Nonrivalry and the Economics of Data.American Economic Review110, 9 (2020), 2819–2858

    2020

  21. [28]

    Anton Korinek. 2023. Scenario Planning for an A(G)I Future.Finance & Development Magazine(December 2023).https://www.imf.orgAccessed on [insert date]

    2023

  22. [29]

    Jan Kulveit, Raymond Douglas, Nora Ammann, Deger Turan, David Krueger, and David Duvenaud. 2025. Gradual Disempowerment: Systemic Existential Risks from Incremental AI Development.arXiv preprint arXiv:2501.16946(2025)

    2025 arXiv

  23. [30]

    Jaron Lanier and E Glen Weyl. 2018. A blueprint for a better digital society.Harvard Business Review26 (2018), 2–18

    2018

  24. [31]

    Mark A Lemley. 2004. Property, intellectual property, and free riding.Tex L. Rev.83 (2004), 1031. 11

    2004

  25. [32]

    Hanlin Li, Nicholas Vincent, Stevie Chancellor, and Brent Hecht. 2023. The dimensions of data labor: A road map for researchers, activists, and policymakers to empower data producers. InProceedings of the 2023 ACM conference on fairness, accountability, and transparency. 1151–1161

    2023

  26. [34]

    Shayne Longpre, Gregory Yauney, Emily Reif, Katherine Lee, Adam Roberts, Barret Zoph, Denny Zhou, Jason Wei, Kevin Robinson, David Mimno, et al . 2024. A pretrainer’s guide to training data: Measuring the effects of data age, domain coverage, quality, & toxicity. In Proceeding...

    2024

  27. [35]

    Liang Lyu, James Siderius, Hannah Li, Daron Acemoglu, Daniel Huttenlocher, and Asuman Ozdaglar. 2025. Wikipedia Contributions in the Wake of ChatGPT.arXiv preprint arXiv:2503.00757(2025)

    2025 arXiv

  28. [36]

    Iren Mazloomzadeh, Gias Uddin, Foutse Khomh, and Ashkan Sami. 2021. Reputation gaming in stack overflow.arXiv preprint arXiv:2111.07101(2021)

    2021 arXiv

  29. [37]

    1991.Technology and the pursuit of economic growth

    David C Mowery and Nathan Rosenberg. 1991.Technology and the pursuit of economic growth. Cambridge University Press

    1991

  30. [38]

    Pamela E Oliver and Gerald Marwell. 2001. Whatever happened to critical mass theory? A retrospective and assessment.Sociological Theory19, 3 (2001), 292–311

    2001

  31. [39]

    Posner and Glen Weyl

    Eric A. Posner and Glen Weyl. 2018. Radical Markets: Uprooting Capitalism and Democracy for a Just Society.https://api.semanticscholar.org/CorpusID:159015610

    2018

  32. [40]

    Frank D Prager. 1944. A History of Intellectual Property from 1545 to 1787.J. Pat. Off. Soc’y 26 (1944), 711

    1944

  33. [41]

    Jennifer Rowley. 2008. Understanding digital content marketing.Journal of marketing manage- ment24, 5-6 (2008), 517–540

    2008

  34. [42]

    Vernon W Ruttan. 2006. Is war necessary for economic growth?Historically speaking7, 6 (2006), 17–19

    2006

  35. [43]

    Pamela Samuelson. 2023. Generative AI meets copyright.Science381, 6654 (2023), 158–161

    2023

  36. [44]

    Carl Shapiro. 2011. Competition and innovation: did arrow hit the bull’s eye? InThe rate and direction of inventive activity revisited. University of Chicago Press, 361–404

    2011

  37. [45]

    Melvyn Teo. 2009. Does size matter in the hedge fund industry?Available at SSRN 1331754 (2009)

    2009

  38. [46]

    1999.Markets for information goods

    Hal R Varian. 1999.Markets for information goods. V ol. 99. Citeseer

    1999

  39. [47]

    Pieter Verdegem. 2024. Dismantling AI capitalism: the commons as an alternative to the power concentration of Big Tech.AI & society39, 2 (2024), 727–737

    2024

  40. [48]

    Conscious Data Contribution

    Nicholas Vincent and Brent Hecht. 2021. Can "Conscious Data Contribution" Help Users to Exert "Data Leverage" Against Technology Companies?Proc. ACM Hum.-Comput. Interact.5, CSCW1, Article 103 (April 2021), 23 pages.https://doi.org/10.1145/3449177 12

    2021 doi

  41. [49]

    Nicholas Vincent, Hanlin Li, Nicole Tilly, Stevie Chancellor, and Brent Hecht. 2021. Data leverage: A framework for empowering the public in its relationship with technology companies. InProceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency. 215–227

    2021

  42. [50]

    Friedrich A V on Hayek. 1937. Economics and knowledge.Economica4, 13 (1937), 33–54

    1937

  43. [51]

    Wang, Prateek Mittal, Dawn Song, and Ruoxi Jia

    Jiachen T. Wang, Prateek Mittal, Dawn Song, and Ruoxi Jia. 2024. Data Shapley in One Training Run. https://doi.org/10.48550/arXiv.2406.11011 arXiv:2406.11011 [cs, stat] version: 1

    2024 doi arXiv

  44. [52]

    Alexander Wettig, Kyle Lo, Sewon Min, Hannaneh Hajishirzi, Danqi Chen, and Luca Soldaini

  45. [53]

    E Glen Weyl, Luke Thorburn, Emillie de Keulenaar, Jacob Mchangama, Divya Siddarth, and Audrey Tang. 2025. Prosocial Media.arXiv preprint arXiv:2502.10834(2025)

    2025

  46. [54]

    David Gray Widder, Meredith Whittaker, and Sarah Myers West. 2024. Why ‘open’AI systems are actually closed, and why this matters.Nature635, 8040 (2024), 827–833

    2024

  47. [55]

    Kyle Wiggers. 2024. Why DeepSeek’s new AI model thinks it’s ChatGPT. https://techcrunch.com/2024/12/27/ why-deepseeks-new-ai-model-thinks-its-chatgpt/

    2024

  48. [56]

    Zachary Wojtowicz, Shrey Jain, and Nicholas Vincent. 2024. Push and Pull: A Framework for Measuring Attentional Agency.arXiv preprint arXiv:2405.14614(2024). 13

    2024 arXiv

  49. [2018]

    Inaea Papers and Proceedings, V ol

    Should we treat data as labor? Moving beyond “free”. Inaea Papers and Proceedings, V ol. 108. American Economic Association 2014 Broadway, Suite 305, Nashville, TN 37203, 38–42

    2014

  50. [2024]

    InThe Twelfth International Conference on Learning Representations

    Language Modeling Is Compression. InThe Twelfth International Conference on Learning Representations

  51. [2025]

    Organize the Web: Constructing Domains Enhances Pre-Training Data Curation.arXiv preprint arXiv:2502.10341(2025)

    2025 arXiv

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.