Pith. sign in

REVIEW 3 major objections 5 minor 1 cited by

Collaborating with GenAI: Incentives and Replacements

T0 review · 3 major / 5 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read The paper argues that introducing even an almost-ineffective GenAI tool into a shared-work game can push equilibrium effort to zero, and that choosing the optimal team is computationally hard.

desk verdict Worth reading and worth refereeing, but the headline 'arbitrarily small GenAI effect collapses effort' is not proved as written — fixable by an ε² scaling, and the PoG inequality needs a direction check. read the letter →

arxiv 2508.20213 v1 pith:2ITUXHLT submitted 2025-08-27 cs.GT cs.AI

Boaz Taitler , Omer Ben-Porat This is my paper
classification cs.GTcs.AI MSC 91A1091A1268Q17
keywords generativeAIincentivessharedbenefitgamescoalitionformationsupermodulardominantequilibriumeffortcollapseNP-completeness
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

The paper studies how free generative AI changes effort and team choice when a manager first picks who works on a shared project and workers then decide how hard to try. It proves that in equilibrium workers may exert zero effort even when GenAI's boost to their contribution is arbitrarily small, so the availability of AI can be strictly worse for the shared outcome. It also proves that the manager's optimal team-selection problem is NP-complete, while giving a pseudo-polynomial algorithm for linear and almost-linear benefit functions. A further set of results shows that a worker whose own output is negligible can still be indispensable because removing them drags down everyone else's effort, and that value-to-share ratios fall as coalitions shrink, which pushes optimal teams to be either very small or very large.

What carries the argument

The engine is the multilinear shared-benefit function and the supermodularity it induces. Every Effort-based Shared-Benefit game (EBSB) is supermodular, so pure Nash equilibria exist, are ordered, and a single Pareto-dominant equilibrium (the 'dominant equilibrium', sigma*(g,C)) can be selected and used for comparison. The value-to-share ratio (VSR) then lets the paper reason about whether a player's marginal contribution justifies their share.

What would settle it

Run the two-player instance behind Proposition 3.6 (theta1=theta2=0.25, s1=e1, s2=sqrt(e2)+epsilon*g2, F=4s1s2, c2=sqrt(e2)) with human players: if without GenAI most pairs do not reach e2=1, the dominant-equilibrium collapse is not the behavioral prediction.

Watch

Extended reading notes

Core claim

The central claim is a threshold phenomenon with a twist: because effort choices are strategic complements, a tiny improvement from GenAI for one player can destroy the shared-benefit equilibrium. Proposition 3.6 constructs, for any epsilon>0, a two-player MSB game where the no-GenAI dominant equilibrium has the second player exerting full effort, while with GenAI available the unique equilibrium has that player exerting no effort, even though GenAI changes the player's contribution by at most epsilon. The mechanism: GenAI makes the second player's marginal utility of effort nonpositive, while the first player plays full effort in both cases. The paper further shows the price of generativity

Load-bearing premise

The results rely on selecting the Pareto-dominant equilibrium among multiple pure Nash equilibria; if real players coordinate on a lower-effort equilibrium, the predicted collapse from full to zero effort is diluted.

Editorial extensions

If this is right

  • If the dominant equilibrium is the right prediction, introducing GenAI can backfire: even negligible individual gains can lead some workers to zero effort.
  • The price of generativity can be unboundedly large as the GenAI effect shrinks, so small per-worker improvements can translate into severe collective losses.
  • The optimal team-selection problem is NP-complete in general, so managers need heuristics or restricted benefit structures in realistic settings.
  • In linear benefit settings, the optimal team can be found in O(N/epsilon^2) via a knapsack-style dynamic program, making exact optimization practical when shares are small.
  • The value-to-share monotonicity implies optimal teams are bimodal: either large coalitions where everyone's marginal value is high, or empty/small coalitions after removal cascades.

Reading between the lines

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

  • The dominant-equilibrium selection is the load-bearing prediction: if real players coordinate on a lower equilibrium, the stark collapse from full to zero effort may not materialize.
  • The construction suggests a testable behavioral prediction: giving workers a cheap AI tool that adds little can still reduce measured effort when effort is costly and contributions are complementary.
  • The bimodality of optimal team sizes connects to organizational design: firms that fear AI-triggered cascades should monitor value-to-share ratios and may want to commit to keeping near-zero contributors.
  • The NP-completeness result points toward approximation or parameterized algorithms for team selection, which the paper does not explore.
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

3 major / 5 minor

Summary. The paper introduces the Managed Shared Benefit (MSB) game, in which a principal selects a coalition of players who each choose an effort level and a binary GenAI usage; non-selected players are replaced by GenAI. The shared benefit is a multilinear function of contributions. The paper's central theoretical claims are: (i) allowing GenAI can induce a complete collapse of equilibrium effort even when GenAI's effect on contributions is arbitrarily small (Proposition 3.6); (ii) the Price of Generativity can be arbitrarily large (Proposition 3.7); (iii) the principal's coalition-selection problem is NP-complete (Theorem 4.2), with a pseudo-polynomial algorithm for (almost-)linear instances (Theorem 4.4); and (iv) near-zero marginal contributors can be pivotal in the optimal coalition (Proposition 5.3). The paper also reports simulations on coalition sizes and stability.

Significance. If the results hold, the model gives a crisp economic explanation for how a freely available, almost-ineffective AI tool can reduce effort in team production, and it formalizes coalition formation as computationally hard. The use of supermodularity is elegant: it guarantees existence of pure equilibria and gives an order on the equilibrium set, enabling a clean comparison of the no-GenAI and GenAI regimes. The NP-completeness reduction and the algorithm for linear instances are concrete technical contributions, and the simulation study adds useful evidence about the bimodal coalition sizes. However, the proof of the headline 'arbitrarily small effect' proposition currently fails its own epsilon bound, and the main result's interpretation depends on a strong equilibrium-selection assumption that is not discussed. These issues must be addressed before the claims can be taken at face value.

major comments (3)
  1. [Section 3.2 / Appendix A.3 (Proposition 3.6)] The proof constructs s2(e,1)=sqrt(e+epsilon) and s2(e,0)=sqrt(e). Then |s2(e,1)-s2(e,0)| = sqrt(e+epsilon)-sqrt(e), which is maximized at e=0 with value sqrt(epsilon). For epsilon<1, sqrt(epsilon)>epsilon, so the instance does not satisfy the theorem's premise |s_i(e_i,1)-s_i(e_i,0)|<=epsilon. Thus Proposition 3.6 is not proven as stated. The construction is repairable (e.g., use s2(e,1)=sqrt(e+epsilon^2)), but as written the proof overclaims. This is load-bearing because Proposition 3.6 underpins the paper's central narrative and is used in Proposition 3.7.
  2. [Section 3.3 / Appendix A.4 (Proposition 3.7)] The statement reads 'PoG < 1/sqrt(epsilon)'. For small epsilon this is a trivial upper bound (any finite PoG satisfies it), and it does not express 'significantly worse'. The proof actually computes PoG = 1/sqrt(epsilon) (in the Proposition 3.6 instance), and the surrounding text claims 'substantial degradation'. The intended inequality is presumably PoG >= 1/sqrt(epsilon) (or, with the epsilon^2 repair, PoG = 1/epsilon). The direction in the statement is wrong and must be corrected, along with the proof and the accompanying interpretation.
  3. [Section 3.1 / Proposition 3.6] The collapse result depends on selecting the dominant (Pareto-superior) equilibrium in the no-GenAI game. In the Proposition 3.6 construction, the g=0 game has a continuum of pure equilibria {(1,e2): e2 in [0,1]}; sigma*(0,C) is set to (1,1). If a different equilibrium selection (e.g., the lowest-effort equilibrium (1,0)) were used, sigma*(0,C)=sigma*(1,C) and no effort collapse occurs. The paper justifies sigma* as Pareto-dominant and socially optimal, but does not argue why this is the natural prediction in a game with a continuum of equilibria. Please either provide such an argument or explicitly discuss the sensitivity of the headline finding to equilibrium selection; for instance, report the effort drop when comparing against the minimal equilibrium.
minor comments (5)
  1. [Section 3.2, Proposition 3.5 statement] The second inequality uses subscript '1' where it should be 'i': it should read s_i(sigma*_i(1,C),1) rather than s_1(...).
  2. [Appendix A.3, utility equations] In the display of U1 and U2 for the g=0 case, the arguments are written as (e1,e1,0,N) and (e1,e1,0,N); the second argument should be e2. Also, the g=0 utilities are labeled 'U1(e1,e2,1,N)' and 'U2(e1,e2,1,N)', which should be g=0.
  3. [Appendix C.1 / C.2] The proof of Proposition 5.2 ends with 'This concludes the proof of Proposition 5.3', and the proof of Proposition 5.3 ends with 'This concludes the proof of Section 5.2'. The labels are mixed up and should be corrected.
  4. [Appendix A.4] The proof of Proposition 3.7 writes F(e1,e2,g1,g2)=e1*sqrt(e2+0.2*g2), which is the function from Example 2.2, not the F=4*s1*s2 used in the Proposition 3.6 construction. This copy-paste inconsistency makes the proof hard to follow; please use the correct F for the instance being analyzed.
  5. [Section 2.2 and Appendix A.3, Proposition 3.5 proof] The sentence 'Observe that s2(0,1)=s2(0.2,1)' is incorrect: s2(0,1)=sqrt(0.2) and s2(0.2,1)=sqrt(0.4). The intended statement is presumably that s2(e2,1)>=sqrt(0.2) whenever e2>=0.2, so player 2's use of GenAI weakly increases player 1's incentive to exert full effort. Please rephrase.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the derivation is self-contained and the only self-citation is in related work, not load-bearing.

full rationale

The paper's central results are derived from stated axioms and standard game-theoretic machinery, not by fitting parameters or by importing a conclusion from prior work. Section 3 fixes an MSB game, defines EBSB subgames, proves supermodularity (Proposition 3.2) from the multilinear structure of F, and then selects the Pareto-dominant pure Nash equilibrium. The comparison in Propositions 3.5 and 3.6 is not definitionally forced: σ*(0,C) and σ*(1,C) are equilibria computed within the constructed instances, and the claimed effort collapse does not reduce to the definition of the dominant equilibrium. Theorem 4.2 is proven by a reduction from the clique problem, which is an external, standard NP-complete problem, and the positive result in Theorem 4.4 reduces the inner problem to 0-1 knapsack—again an independent algorithmic fact. The linear and almost-linear algorithms are not fitted to data and are validated by correctness arguments. The only self-citation is to the authors' prior Braess's paradox paper [38] in the related work section; it is not invoked as a premise for any theorem or used to justify the model's uniqueness, so it is not load-bearing. The reader's concern about the dominant-equilibrium selection is a modeling assumption, not a circularity: it is explicit, and the results hold relative to that disclosed solution concept. Even though a close reading of the proof of Proposition 3.6 suggests a possible ε-bound issue in the constructed instance (|s2(e,1)-s2(e,0)| = sqrt(e+ε)-sqrt(e) can exceed ε), that is a correctness or proof-detail concern, not a reduction of the conclusion to the hypothesis; it does not make the derivation circular. Accordingly, no circular step is identified and the score is 0.

Assumptions & free parameters 0 free parameters · 7 assumptions · 0 invented entities

The model's assumptions are clearly stated. The main fragility is the dominant-equilibrium selection, which drives the effort-collapse result. The fixed-share assumption is realistic in some settings but limits generality. No fitted parameters or invented physical entities appear.

assumptions (7)
  • domain assumption Players receive fixed exogenous shares theta_i of the shared benefit.
    Section 2: shares are predetermined and fixed, with no utility transfer. This is stated as a modeling assumption and limits generality.
  • domain assumption Shared benefit F is a multilinear polynomial with nonnegative coefficients.
    Section 2.1: this structural assumption is used to prove supermodularity and the NP-hardness reduction.
  • domain assumption Contribution functions are nondecreasing in effort and GenAI usage.
    Section 2: this monotonicity underpins the supermodularity proof and equilibrium comparisons.
  • domain assumption Cost functions are increasing in effort.
    Section 2: costs are assumed increasing, motivating the tradeoff with shared benefit.
  • domain assumption GenAI is freely accessible to players and manager at no direct cost.
    Section 1 and 2: the paper argues GenAI tools are cheap; this makes g=1 weakly dominant.
  • domain assumption Principal commits to a coalition before players choose effort (Stackelberg).
    Section 4: the optimization problem P1 assumes this timing, which is reasonable for hiring but not universal.
  • ad hoc to paper Players coordinate on the dominant (Pareto-superior) Nash equilibrium.
    Section 3.1: the paper selects this equilibrium as the solution concept. It is not derived from the model, and the effort-collapse result is sensitive to this selection.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Collaborating with GenAI: Incentives and Replacements." pith.science (2026). https://pith.science/paper/2ITUXHLT

@misc{pith2026250820213,
  author       = {Pith},
  title        = {Pith review of: Collaborating with GenAI: Incentives and Replacements},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2ITUXHLT}},
  note         = {Machine review of arXiv:2508.20213}
}
read the original abstract

The rise of Generative AI (GenAI) is reshaping how workers contribute to shared projects. While workers can use GenAI to boost productivity or reduce effort, managers may use it to replace some workers entirely. We present a theoretical framework to analyze how GenAI affects collaboration in such settings. In our model, the manager selects a team to work on a shared task, with GenAI substituting for unselected workers. Each worker selects how much effort to exert, and incurs a cost that increases with the level of effort. We show that GenAI can lead workers to exert no effort, even if GenAI is almost ineffective. We further show that the manager's optimization problem is NP-complete, and provide an efficient algorithm for the special class of (almost-) linear instances. Our analysis shows that even workers with low individual value may play a critical role in sustaining overall output, and excluding such workers can trigger a cascade. Finally, we conduct extensive simulations to illustrate our theoretical findings.

Figures

Figures reproduced from arXiv: 2508.20213 by the authors.

Figure 1
Figure 1. Analysis of coalition size in equilibrium. Figure [PITH_FULL_IMAGE:figures/full_fig_p010_1.png] view at source ↗
Figure 2
Figure 2. Analysis of coalition size after myopic removal due to unstable coalition [PITH_FULL_IMAGE:figures/full_fig_p023_2.png] view at source ↗

Discussion (0). Sign in to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Generative artificial intelligence reduces social welfare through model collapse

    physics.soc-ph 2026-04 unverdicted novelty 6.0 of 10

    A game-theoretic model shows that individually rational adoption of generative AI causes model collapse that reduces collective social welfare for important tasks, with habit formation creating spillovers from low-sta...

Reference graph

Works this paper leans on

47 extracted references · 41 canonical work pages · cited by 1 Pith paper

  1. [1]

    R. J. Aumann and J. H. Dreze. Cooperative games with coalition structures. International Journal of game theory , 3(4):217–237, 1974

    work page 1974

  2. [2]

    Barman, S

    S. Barman, S. Ebadian, M. Latifian, and N. Shah. Fair division with market values. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 39, pages 13589–13596, 2025

    work page 2025

  3. [3]

    Becker, N

    J. Becker, N. Rush, E. Barnes, and D. Rein. Measuring the impact of early-2025 ai on experienced open-source developer productivity. arXiv preprint arXiv:2507.09089 , 2025

    arXiv 2025

  4. [4]

    Bil` o, G

    V. Bil` o, G. Monaco, and L. Moscardelli. Hedonic games with fixed-size coalitions. In Proceedings of the AAAI Conference on Artificial Intelligence , volume 36, pages 9287–9295, 2022

    work page 2022

  5. [5]

    Boehmer, S

    N. Boehmer, S. Fish, and A. D. Procaccia. Generative social choice: The next generation. In Forty-second International Conference on Machine Learning , 2025

    work page 2025

  6. [6]

    Bredereck, A

    R. Bredereck, A. Kaczmarczyk, J. Luo, R. Niedermeier, and F. Sachse. On improving resource allocations by sharing. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 36, pages 4875–4883, 2022

    work page 2022

  7. [7]

    Z. Chen. Ethics and discrimination in artificial intelligence-enabled recruitment practices. Humanities and social sciences communications , 10(1):1–12, 2023

    work page 2023

  8. [8]

    Cheng, X

    Y. Cheng, X. Deng, and Y. Ma. Networked digital public goods games with heterogeneous players and convex costs. In The Web Conference 2025 , 2025

    work page 2025

Show all 47 references
  1. [9]

    Cohen and N

    S. Cohen and N. Agmon. Online learning of coalition structures by selfish agents. In Proceedings of the AAAI Conference on Artificial Intelligence , volume 39, pages 13709– 13717, 2025

    2025

  2. [10]

    Conitzer, R

    V. Conitzer, R. Freedman, J. Heitzig, W. H. Holliday, B. M. Jacobs, N. Lambert, M. Moss´ e, E. Pacuit, S. Russell, H. Schoelkopf, E. Tewolde, and W. S. Zwicker. Social choice for AI alignment: Dealing with diverse human feedback. CoRR, abs/2404.10271, 2024. doi: 10.48550/ARXIV...

    2024 doi arXiv

  3. [11]

    Cookson, S

    B. Cookson, S. Ebadian, and N. Shah. Constrained fair and efficient allocations. In Proceedings of the AAAI Conference on Artificial Intelligence , volume 39, pages 13718– 13726, 2025

    2025

  4. [12]

    D. R. Cotton, P. A. Cotton, and J. R. Shipway. Chatting and cheating: Ensuring academic integrity in the era of chatgpt. Innovations in education and teaching international , 61(2): 228–239, 2024

    2024

  5. [13]

    Darvariu, S

    V.-A. Darvariu, S. Hailes, and M. Musolesi. Solving graph-based public goods games with tree search and imitation learning. Advances in Neural Information Processing Systems , 34: 1739–1751, 2021. 12

    2021

  6. [14]

    S. Dean, E. Dong, M. Jagadeesan, and L. Leqi. Accounting for AI and users shaping one another: The role of mathematical models. Transactions on Machine Learning Research,

  7. [15]

    Farina, N

    G. Farina, N. Gatti, and T. Sandholm. Practical exact algorithm for trembling-hand equilibrium refinements in games. Advances in neural information processing systems , 31, 2018

    2018

  8. [16]

    Goktas, A

    D. Goktas, A. Greenwald, T. Osogami, R. Patel, K. Leyton-Brown, G. Schoenebeck, D. Cornelisse, C. Daskalakis, I. Gemp, J. Horton, et al. Strategic foundation models. 2025

    2025

  9. [17]

    Greenwood, K

    S. Greenwood, K. Levy, S. Barocas, H. Heidari, and J. Kleinberg. Designing algorithmic delegates: The role of indistinguishability in human-ai handoff. In Proceedings of the 26th ACM Conference on Economics and Computation , pages 306–336, 2025

    2025

  10. [18]

    Hemmer, L

    P. Hemmer, L. Thede, M. V¨ ossing, J. Jakubik, and N. K¨ uhl. Learning to defer with limited expert predictions. In Proceedings of the AAAI Conference on Artificial Intelligence , volume 37, pages 6002–6011, 2023

    2023

  11. [19]

    Keinan and O

    G. Keinan and O. Ben-Porat. Strategic content creation in the age of genai: To share or not to share? arXiv preprint arXiv:2505.16358 , 2025

    2025

  12. [20]

    P. Kollock. Social dilemmas: The anatomy of cooperation. Annual review of sociology , 24 (1):183–214, 1998

    1998

  13. [21]

    Laufer, J

    B. Laufer, J. Kleinberg, and H. Heidari. Fine-tuning games: Bargaining and adaptation for general-purpose models. In Proceedings of the ACM on Web Conference 2024 , pages 66–76, 2024

    2024

  14. [22]

    J. O. Ledyard et al. Public goods: A survey of experimental research . Division of the Humanities and Social Sciences, California Inst. of Technology, 1994

    1994

  15. [23]

    Marchesi and N

    A. Marchesi and N. Gatti. Trembling-hand perfection and correlation in sequential games. In Proceedings of the AAAI Conference on Artificial Intelligence , volume 35, pages 5566–5574, 2021

    2021

  16. [24]

    Martello and P

    S. Martello and P. Toth. Knapsack problems: algorithms and computer implementations . John Wiley & Sons, Inc., 1990

    1990

  17. [25]

    Noti and Y

    G. Noti and Y. Chen. Learning when to advise human decision makers. 2022

    2022

  18. [26]

    Oesterheld and V

    C. Oesterheld and V. Conitzer. Safe pareto improvements for delegated game playing. Autonomous Agents and Multi-Agent Systems , 36(2):46, 2022

    2022

  19. [27]

    K. Peng, N. Garg, and J. Kleinberg. A no free lunch theorem for human-ai collaboration. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 39, pages 14369–14376, 2025

    2025

  20. [28]

    S. Peng, E. Kalliamvakou, P. Cihon, and M. Demirer. The impact of ai on developer productivity: Evidence from github copilot. arXiv preprint arXiv:2302.06590 , 2023

    2023 arXiv

  21. [29]

    A. D. Procaccia, B. Schiffer, and S. Zhang. Clone-robust AI alignment. In Forty-second International Conference on Machine Learning , 2025

    2025

  22. [30]

    Raghavan

    M. Raghavan. Competition and diversity in generative ai. arXiv preprint arXiv:2412.08610 , 2024. 13

    2024 arXiv

  23. [31]

    Reiter, M

    L. Reiter, M. J¨ orling, C. Fuchs, W. group ‘Artificial Intelligence in Higher Education’, and R. B¨ ohm. Student (mis) use of generative ai tools for university-related tasks.International Journal of Human–Computer Interaction , pages 1–14, 2025

    2025

  24. [32]

    N. Sherman. Amazon boss says ai will replace jobs at tech giant, 2025. URL https: //www.bbc.com/news/articles/cn0q2v851k9o. Accessed: 2025-07-31

    2025

  25. [33]

    Singla, E

    A. Singla, E. Horvitz, P. Kohli, and A. Krause. Learning to hire teams. In Proceedings of the AAAI Conference on Human Computation and Crowdsourcing , volume 3, pages 34–35, 2015

    2015

  26. [34]

    Sourbut, L

    O. Sourbut, L. Hammond, and H. Wood. Cooperation and control in delegation games. In IJCAI, pages 229–237, 2024

    2024

  27. [35]

    M. Stone. Ibm ceo says ai has replaced hundreds of workers but created new programming, sales jobs. The Wall Street Journal , May 2025. URL https://www.wsj.com/articles/i bm-ceo-says-ai-has-replaced-hundreds-of-workers-but-created-new-programmi ng-sales-jobs-54ea6b58 . Accesse...

    2025

  28. [36]

    H. Sun, Y. Chen, S. Wang, W. Chen, and X. Deng. Mechanism design for llm fine-tuning with multiple reward models. arXiv preprint arXiv:2405.16276 , 2024

    2024

  29. [37]

    Taguelmimt, S

    R. Taguelmimt, S. Aknine, D. Boukredera, N. Changder, and T. Sandholm. Optimal anytime coalition structure generation utilizing compact solution space representation. In International Joint Conference on Artificial Intelligence (IJCAI) , 2023

    2023

  30. [38]

    Taitler and O

    B. Taitler and O. Ben-Porat. Braess’s paradox of generative ai. In Proceedings of the AAAI Conference on Artificial Intelligence , volume 39, pages 14139–14147, 2025

    2025

  31. [39]

    Timberg and E

    C. Timberg and E. Dwoskin. At ibm, artificial intelligence is replacing workers. but it may not create jobs quickly enough. The Washington Post , May 2023. URL https: //www.washingtonpost.com/technology/2023/05/02/ai- jobs- takeover- ibm/ . Accessed: 2025-07-31

    2023

  32. [40]

    D. M. Topkis. Supermodularity and complementarity. Princeton university press, 1998

    1998

  33. [41]

    X. Vives. Nash equilibrium with strategic complementarities. Journal of Mathematical Economics, 19(3):305–321, 1990

    1990

  34. [42]

    S. Yu, D. Kempe, and Y. Vorobeychik. Altruism design in networked public goods games. In IJCAI, pages 493–499, 2021. 14 A Proofs Omitted From Section 3 A.1 Supermodular games We now present the formal definition of supermodular games [40]. Definition A.1. Let ( N , A, U) be a ...

    2021

  35. [43]

    A is a sublattice of Q i∈N RN

  36. [44]

    Ui(ai, a−i) is supermodular in ai on Ai for each a−i on A−i and every player i

  37. [45]

    A.2 Proofs Omitted From Subsection 3.1 Proof of Proposition 3.2

    Ui(ai, a−i) has increasing differences in ( ai, a−i) on Ai × A−i. A.2 Proofs Omitted From Subsection 3.1 Proof of Proposition 3.2 . In every EBSB game, the action space of each player is the effort space [0, 1]. Therefore, the joint strategy space is [0 , 1]N which is a contin...

  38. [46]

    The optimal solution of the Problem (P2) satisfies the constraint of the Problem (P2-inner) with strict equality

  39. [47]

    Afterward to show the running time

    Algorithm 1 returns the optimal solution. Afterward to show the running time. We start with the first requirement. Assume in contradiction that there exists a solution to Problem (P2) C⋆ and k⋆ such that P i∈C⋆ θi < k⋆ε. Since θi ∈ {0, ε, . . . ,1} for every i ∈ Nthen there ex...

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.