pith. machine review for the scientific record. sign in

arxiv: 2602.04572 · v2 · submitted 2026-02-04 · 💻 cs.AI · cs.GT

Recognition: no theorem link

From Competition to Collaboration: Designing Sustainable Mechanisms Between LLMs and Online Forums

Niv Fono , Yftah Ziser , Omer Ben-Porat
Authors on Pith no claims yet

Pith reviewed 2026-05-16 07:40 UTC · model grok-4.3

classification 💻 cs.AI cs.GT
keywords LLMsonline forumsincentive misalignmentcollaboration mechanismsStack Exchangequestion generationgenerative AIsustainable collaboration
0
0 comments X

The pith

LLMs and online forums can collaborate on question proposals to reach roughly half the utility of ideal full-information scenarios despite misaligned incentives.

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

The paper sets out a sequential interaction framework in which a generative AI proposes questions to an online forum, which then chooses which ones to publish. This setup is meant to resolve the paradox that AI systems pull users away from forums yet still need the data those forums generate to keep improving. Using simulations grounded in real Stack Exchange threads and standard LLMs, the work shows clear incentive misalignment between the two sides. Even so, the collaborative process delivers approximately half the combined utility that would be possible if both parties had perfect information. The results point to a practical route for continued knowledge exchange that does not require one side to displace the other.

Core claim

The authors introduce a sequential interaction framework that models non-monetary exchanges, asymmetric information, and incentive misalignment between generative AI systems and online forums. Comprehensive data-driven simulations built on real Stack Exchange data and common LLMs demonstrate empirical incentive misalignment, yet show that the two players can still achieve roughly half the utility attainable in an ideal full-information scenario, highlighting the potential for sustainable collaboration that preserves effective knowledge sharing.

What carries the argument

The sequential interaction framework in which the generative AI proposes questions and the forum selectively publishes some of them, capturing non-monetary exchanges and asymmetric information.

If this is right

  • Forums can selectively publish AI-proposed questions to maintain engagement while LLMs gain useful training signals.
  • Both parties reach roughly half the utility possible under full information despite misaligned incentives.
  • Knowledge-sharing platforms remain viable even as generative AI usage increases.
  • The framework accounts for non-monetary exchanges and asymmetric information without requiring monetary payments.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Live experiments on active forums could test whether the simulated half-utility level holds when real users respond.
  • The same sequential-proposal structure could be adapted to other user-generated knowledge sites such as wikis or code repositories.
  • Adding limited monetary side-payments might narrow the remaining gap to ideal utility without changing the core mechanism.

Load-bearing premise

The data-driven simulations using real Stack Exchange data and common LLMs accurately reflect real-world user behaviors, response rates, and incentive structures in online forums.

What would settle it

A live deployment on an active forum in which the observed combined utility falls substantially below half the estimated ideal full-information utility or in which the forum publishes none of the AI-proposed questions.

Figures

Figures reproduced from arXiv: 2602.04572 by Niv Fono, Omer Ben-Porat, Yftah Ziser.

Figure 1
Figure 1. Figure 1: Iterative interaction between a GenAI provider and an online Q&A forum. In each round, [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Relationship between question perplexity and normalized ViewCount across five StackEx [PITH_FULL_IMAGE:figures/full_fig_p011_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Weekly ROC-AUC over the 52-week evaluation horizon. The horizontal axis shows the [PITH_FULL_IMAGE:figures/full_fig_p022_3.png] view at source ↗
read the original abstract

While Generative AI (GenAI) systems draw users away from (Q&A) forums, they also depend on the very data those forums produce to improve their performance. Addressing this paradox, we propose a framework of sequential interaction, in which a GenAI system proposes questions to a forum that can publish some of them. Our framework captures several intricacies of such a collaboration, including non-monetary exchanges, asymmetric information, and incentive misalignment. We bring the framework to life through comprehensive, data-driven simulations using real Stack Exchange data and commonly used LLMs. We demonstrate the incentive misalignment empirically, yet show that players can achieve roughly half of the utility in an ideal full-information scenario. Our results highlight the potential for sustainable collaboration that preserves effective knowledge sharing between AI systems and human knowledge platforms.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper proposes a sequential interaction framework between generative AI systems and online Q&A forums to resolve the paradox of AI drawing users away from forums while depending on their data. The framework models non-monetary exchanges, asymmetric information, and incentive misalignment. Through data-driven simulations using real Stack Exchange posts and common LLMs to generate user decisions and response rates, the authors empirically demonstrate misalignment yet report that collaborative play achieves roughly half the utility of an ideal full-information benchmark.

Significance. If the simulation results are robust, the work offers a timely, empirically grounded framework for sustainable AI-forum collaboration that could help preserve human knowledge platforms. The use of real Stack Exchange data and sequential non-monetary modeling provides concrete, falsifiable outputs on utility gaps that prior mechanism-design literature on online communities has not quantified in this AI context.

major comments (2)
  1. Empirical Evaluation section: The central quantitative claim (incentive misalignment exists yet players reach ~half ideal utility) rests on LLM agents simulating forum-user decisions and response rates. No calibration or hold-out validation against actual human posting thresholds, information-revelation rates, or collaboration willingness from the Stack Exchange logs is reported; without this, both the misalignment demonstration and the 'half-utility' figure risk being artifacts of the chosen prompting and utility function rather than evidence about real forums.
  2. Framework section (utility definitions): The utility functions for the forum and AI under asymmetric information encode specific assumptions about response rates and willingness to publish AI-proposed questions. These parameters directly determine the reported half-utility result; the manuscript provides no sensitivity analysis or external justification for their values, making the quantitative halfway finding load-bearing on untested modeling choices.
minor comments (2)
  1. Figure captions and axis labels in the results section do not consistently distinguish the full-information benchmark from the proposed sequential mechanism, complicating interpretation of the utility-gap plots.
  2. The related-work discussion omits several key references on mechanism design for non-monetary online communities (e.g., work on Stack Exchange reputation systems and information-asymmetry models in Q&A platforms).

Simulated Author's Rebuttal

2 responses · 1 unresolved

Thank you for the constructive feedback. We address the major comments point by point below, with revisions proposed where they strengthen the work without altering its core simulation-based contribution.

read point-by-point responses

  1. Referee: Empirical Evaluation section: The central quantitative claim (incentive misalignment exists yet players reach ~half ideal utility) rests on LLM agents simulating forum-user decisions and response rates. No calibration or hold-out validation against actual human posting thresholds, information-revelation rates, or collaboration willingness from the Stack Exchange logs is reported; without this, both the misalignment demonstration and the 'half-utility' figure risk being artifacts of the chosen prompting and utility function rather than evidence about real forums.

    Authors: We agree that the absence of direct calibration or hold-out validation against human behaviors is a limitation. Our simulations ground questions in real Stack Exchange data and use LLMs to model decisions via standard prompting, but we did not calibrate parameters to observed human posting or response rates. In revision we will add an explicit limitations subsection discussing this gap and outlining future human-subject validation steps, while retaining the current results as a simulation-based demonstration. revision: partial

  2. Referee: Framework section (utility definitions): The utility functions for the forum and AI under asymmetric information encode specific assumptions about response rates and willingness to publish AI-proposed questions. These parameters directly determine the reported half-utility result; the manuscript provides no sensitivity analysis or external justification for their values, making the quantitative halfway finding load-bearing on untested modeling choices.

    Authors: We accept this critique. The parameter values were selected to reflect plausible ranges drawn from aggregate Stack Exchange statistics, but no sensitivity analysis was included. In the revised manuscript we will add a dedicated sensitivity analysis subsection that systematically varies response rates, collaboration willingness, and related parameters, showing that the result of reaching roughly half the ideal utility remains qualitatively robust across tested ranges. revision: yes

standing simulated objections not resolved
  • Direct empirical calibration or hold-out validation against actual human user behaviors from Stack Exchange, which would require new human-subject experiments outside the scope of the current simulation study.

Circularity Check

0 steps flagged

No circularity; framework and results are independent of inputs

full rationale

The paper introduces a sequential interaction framework with non-monetary exchanges and asymmetric information, then implements it via data-driven simulations on external real Stack Exchange posts fed into standard LLMs. No equations, parameters, or claims reduce by construction to fitted values or self-citations; the reported half-utility outcome is an output of the simulation rather than presupposed. The derivation chain remains self-contained against external benchmarks with no load-bearing self-references or ansatz smuggling.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The framework rests on domain assumptions about how users and LLMs respond to proposed questions and on the fidelity of the simulation environment; no free parameters or invented entities are explicitly introduced in the abstract.

axioms (1)
  • domain assumption User participation and forum publication decisions follow the modeled incentive structures in the simulations
    Invoked to justify the data-driven simulations that demonstrate incentive misalignment and partial utility recovery

pith-pipeline@v0.9.0 · 5436 in / 1064 out tokens · 23557 ms · 2026-05-16T07:40:34.744681+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

52 extracted references · 52 canonical work pages · 7 internal anchors

  1. [1]

    Studying llm performance on closed- and open-source data, 2024

    Toufique Ahmed, Christian Bird, Premkumar Devanbu, and Saikat Chakraborty. Studying llm performance on closed- and open-source data, 2024. URL https://arxiv.org/abs/2402. 15100

    work page 2024

  2. [2]

    Effects of offering incentives for reviews on trust: Role of review quality and incentive source.International Journal of Hospitality Man- 13 agement, 100:103101, 2022

    Jin Ai, Dogan Gursoy, Yue Liu, and Xingyang Lv. Effects of offering incentives for reviews on trust: Role of review quality and incentive source.International Journal of Hospitality Man- 13 agement, 100:103101, 2022. ISSN 0278-4319. doi: https://doi.org/10.1016/j.ijhm.2021.103101. URLhttps://www.sciencedirect.com/science/article/pii/S0278431921002449

    work page doi:10.1016/j.ijhm.2021.103101 2022

  3. [3]

    A comprehensive review of usage control frameworks

    Ines Akaichi and Sabrina Kirrane. A comprehensive review of usage control frameworks. Computer Science Review, 56:100698, 2025. ISSN 1574-0137. doi: https://doi.org/10. 1016/j.cosrev.2024.100698. URL https://www.sciencedirect.com/science/article/pii/ S1574013724000819

    work page arXiv 2025

  4. [4]

    The Falcon Series of Open Language Models

    Ebtesam Almazrouei, Hamza Alobeidli, Abdulaziz Alshamsi, Alessandro Cappelli, Ruxandra Cojocaru, M´ erouane Debbah, ´Etienne Goffinet, Daniel Hesslow, Julien Launay, Quentin Malartic, Daniele Mazzotta, Badreddine Noune, Baptiste Pannier, and Guilherme Penedo. The falcon series of open language models, 2023. URLhttps://arxiv.org/abs/2311.16867

    work page internal anchor Pith review arXiv 2023

  5. [5]

    Pauketat, Ali Ladak, and Aikaterina Manoli

    Adriana Alvarado Garcia, Heloisa Candello, Karla Badillo-Urquiola, and Marisol Wong-Villacres. Emerging data practices: Data work in the era of large language models. InProceedings of the 2025 CHI Conference on Human Factors in Computing Systems, CHI ’25, New York, NY, USA, 2025. Association for Computing Machinery. ISBN 9798400713941. doi: 10.1145/370659...

    work page doi:10.1145/3706598 2025

  6. [6]

    Algorithmically fair maximization of multiple submodular objective functions

    Georgios Amanatidis, Georgios Birmpas, Philip Lazos, Stefano Leonardi, and Rebecca Reiff- enh¨ auser. Algorithmically fair maximization of multiple submodular objective functions. In Proceedings of the 24th International Conference on Autonomous Agents and Multiagent Sys- tems, AAMAS ’25, page 115–123, Richland, SC, 2025. International Foundation for Auto...

    work page 2025

  7. [7]

    Codeinsight: A curated dataset of practical coding solutions from stack overflow, 2024

    Nathana¨ el Beau and Benoˆ ıt Crabb´ e. Codeinsight: A curated dataset of practical coding solutions from stack overflow, 2024. URLhttps://arxiv.org/abs/2409.16819

    work page arXiv 2024

  8. [8]

    Pythia: A Suite for Analyzing Large Language Models Across Training and Scaling

    Stella Biderman, Hailey Schoelkopf, Quentin Anthony, Herbie Bradley, Kyle O’Brien, Eric Hallahan, Mohammad Aflah Khan, Shivanshu Purohit, USVSN Sai Prashanth, Edward Raff, Aviya Skowron, Lintang Sutawika, and Oskar van der Wal. Pythia: A suite for analyzing large language models across training and scaling, 2023. URL https://arxiv.org/abs/2304.01373

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  9. [9]

    Extracting training data from large language models

    Nicholas Carlini, Florian Tramer, Eric Wallace, Matthew Jagielski, Ariel Herbert-Voss, Kather- ine Lee, Adam Roberts, Tom Brown, Dawn Song, Ulfar Erlingsson, et al. Extracting training data from large language models. In30th USENIX security symposium (USENIX Security 21), pages 2633–2650, 2021

    work page 2021

  10. [10]

    Transformers: ”the end of history” for nlp?, 2021

    Anton Chernyavskiy, Dmitry Ilvovsky, and Preslav Nakov. Transformers: ”the end of history” for nlp?, 2021. URLhttps://arxiv.org/abs/2105.00813

    work page arXiv 2021

  11. [11]

    Damai Dai, Chengqi Deng, Chenggang Zhao, R. X. Xu, Huazuo Gao, Deli Chen, Jiashi Li, Wangding Zeng, Xingkai Yu, Y. Wu, Zhenda Xie, Y. K. Li, Panpan Huang, Fuli Luo, Chong Ruan, Zhifang Sui, and Wenfeng Liang. Deepseekmoe: Towards ultimate expert specialization in mixture-of-experts language models, 2024. URLhttps://arxiv.org/abs/2401.06066

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  12. [12]

    A meta-analytic review of experiments examining the effects of extrinsic rewards on intrinsic motivation.Psychological Bulletin, 125: 627–668, 11 1999

    Edward Deci, Richard Koestner, and Richard Ryan. A meta-analytic review of experiments examining the effects of extrinsic rewards on intrinsic motivation.Psychological Bulletin, 125: 627–668, 11 1999. doi: 10.1037/0033-2909.125.6.627. 14

    work page doi:10.1037/0033-2909.125.6.627 1999

  13. [13]

    Large language models reduce public knowledge sharing on online q&a platforms.PNAS Nexus, 3(9):pgae400, 09

    R Maria del Rio-Chanona, Nadzeya Laurentsyeva, and Johannes Wachs. Large language models reduce public knowledge sharing on online q&a platforms.PNAS Nexus, 3(9):pgae400, 09

    work page

  14. [14]

    doi: 10.1093/pnasnexus/pgae400

    ISSN 2752-6542. doi: 10.1093/pnasnexus/pgae400. URL https://doi.org/10.1093/ pnasnexus/pgae400

    work page doi:10.1093/pnasnexus/pgae400

  15. [15]

    BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding

    Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. Bert: Pre-training of deep bidirectional transformers for language understanding, 2019. URL https://arxiv.org/ abs/1810.04805

    work page internal anchor Pith review Pith/arXiv arXiv 2019

  16. [16]

    Fair allocation of indivisible public goods

    Brandon Fain, Kamesh Munagala, and Nisarg Shah. Fair allocation of indivisible public goods. In Proceedings of the 2018 ACM Conference on Economics and Computation, EC ’18, page 575–592, New York, NY, USA, 2018. Association for Computing Machinery. ISBN 9781450358293. doi: 10.1145/3219166.3219174. URLhttps://doi.org/10.1145/3219166.3219174

    work page doi:10.1145/3219166.3219174 2018

  17. [17]

    Competition for publication-based rewards.Economics Letters, 244: 112017, 2024

    Mike Felgenhauer. Competition for publication-based rewards.Economics Letters, 244: 112017, 2024. ISSN 0165-1765. doi: https://doi.org/10.1016/j.econlet.2024.112017. URL https://www.sciencedirect.com/science/article/pii/S0165176524005019

    work page doi:10.1016/j.econlet.2024.112017 2024

  18. [18]

    Fairness in simple bargaining experiments.Games and Economic behavior, 6(3):347–369, 1994

    Robert Forsythe, Joel L Horowitz, Nathan E Savin, and Martin Sefton. Fairness in simple bargaining experiments.Games and Economic behavior, 6(3):347–369, 1994

    work page 1994

  19. [19]

    The Pile: An 800GB Dataset of Diverse Text for Language Modeling

    Leo Gao, Stella Biderman, Sid Black, Laurence Golding, Travis Hoppe, Charles Foster, Jason Phang, Horace He, Anish Thite, Noa Nabeshima, Shawn Presser, and Connor Leahy. The Pile: An 800gb dataset of diverse text for language modeling.arXiv preprint arXiv:2101.00027, 2020

    work page internal anchor Pith review Pith/arXiv arXiv 2020

  20. [20]

    Impact of labeling noise on machine learning: A cost-aware empirical study

    Abdulrahman Ahmed Gharawi, Jumana Alsubhi, and Lakshmish Ramaswamy. Impact of labeling noise on machine learning: A cost-aware empirical study. In2022 21st IEEE Interna- tional Conference on Machine Learning and Applications (ICMLA), pages 936–939, 2022. doi: 10.1109/ICMLA55696.2022.00156

    work page doi:10.1109/icmla55696.2022.00156 2022

  21. [21]

    Ziabari, Ali Omrani, and Morteza Dehghani

    Preni Golazizian, Alireza S. Ziabari, Ali Omrani, and Morteza Dehghani. Cost-efficient subjective task annotation and modeling through few-shot annotator adaptation, 2024. URL https://arxiv.org/abs/2402.14101

    work page arXiv 2024

  22. [22]

    Smith, and Luke Zettlemoyer

    Hila Gonen, Srini Iyer, Terra Blevins, Noah A. Smith, and Luke Zettlemoyer. Demystifying prompts in language models via perplexity estimation, 2024. URL https://arxiv.org/abs/ 2212.04037

    work page arXiv 2024

  23. [23]

    The tragedy of the commons.Science, 162(3859):1243–1248, 1968

    Garrett Hardin. The tragedy of the commons.Science, 162(3859):1243–1248, 1968. doi: 10.1126/science.162.3859.1243

    work page doi:10.1126/science.162.3859.1243 1968

  24. [24]

    Stack overflow is not dead yet: Crowd answers still matter

    Denis Helic and Tiago Santos. Stack overflow is not dead yet: Crowd answers still matter. arXiv preprint arXiv:2509.05879, 2025

    work page arXiv 2025

  25. [25]

    Better safe than sorry: Nonprofit organizational legitimacy and cross-sector partnerships.Business & Society, 54(6):822–858, 2015

    Heidi Herlin. Better safe than sorry: Nonprofit organizational legitimacy and cross-sector partnerships.Business & Society, 54(6):822–858, 2015. doi: 10.1177/0007650312472609. URL https://doi.org/10.1177/0007650312472609. 15

    work page doi:10.1177/0007650312472609 2015

  26. [26]

    Large language models for software engineering: A systematic literature review, 2024

    Xinyi Hou, Yanjie Zhao, Yue Liu, Zhou Yang, Kailong Wang, Li Li, Xiapu Luo, David Lo, John Grundy, and Haoyu Wang. Large language models for software engineering: A systematic literature review, 2024. URLhttps://arxiv.org/abs/2308.10620

    work page arXiv 2024

  27. [27]

    A survey of uncertainty estimation in llms: Theory meets practice, 2024

    Hsiu-Yuan Huang, Yutong Yang, Zhaoxi Zhang, Sanwoo Lee, and Yunfang Wu. A survey of uncertainty estimation in llms: Theory meets practice, 2024. URL https://arxiv.org/abs/ 2410.15326

    work page arXiv 2024

  28. [28]

    Meta-llama-3-8b-instruct, 2024

    Hugging Face. Meta-llama-3-8b-instruct, 2024. URL https://huggingface.co/meta-llama/ Meta-Llama-3-8B-Instruct. Accessed: 2025-10-08

    work page 2024

  29. [29]

    Choquette-Choo, and Zheng Xu

    Nikhil Kandpal, Krishna Pillutla, Alina Oprea, Peter Kairouz, Christopher A. Choquette-Choo, and Zheng Xu. User inference attacks on large language models. In Yaser Al-Onaizan, Mohit Bansal, and Yun-Nung Chen, editors,Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 18238–18265, Miami, Florida, USA, November 20...

    work page doi:10.18653/v1/2024.emnlp-main.1014 2024

  30. [30]

    Subset Sum Made Simple

    Konstantinos Koiliaris and Chao Xu. Subset sum made simple.arXiv preprint arXiv:1807.08248, 2018

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  31. [31]

    Impacts of generative ai on user contributions: evidence from a coding q&a platform.Marketing Letters, 36:577–591, 09 2024

    Xinyu Li and Keongtae Kim. Impacts of generative ai on user contributions: evidence from a coding q&a platform.Marketing Letters, 36:577–591, 09 2024. doi: 10.1007/s11002-024-09747-1

    work page doi:10.1007/s11002-024-09747-1 2024

  32. [32]

    DeepSeek-VL: Towards Real-World Vision-Language Understanding

    Haoyu Lu, Wen Liu, Bo Zhang, Bingxuan Wang, Kai Dong, Bo Liu, Jingxiang Sun, Tongzheng Ren, Zhuoshu Li, Hao Yang, Yaofeng Sun, Chengqi Deng, Hanwei Xu, Zhenda Xie, and Chong Ruan. Deepseek-vl: Towards real-world vision-language understanding, 2024. URL https://arxiv.org/abs/2403.05525

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  33. [33]

    Learning under concept drift: A review.IEEE transactions on knowledge and data engineering, 31(12): 2346–2363, 2018

    Jie Lu, Anjin Liu, Fan Dong, Feng Gu, Joao Gama, and Guangquan Zhang. Learning under concept drift: A review.IEEE transactions on knowledge and data engineering, 31(12): 2346–2363, 2018

    work page 2018

  34. [34]

    Llm dataset infer- ence: Did you train on my dataset? In A

    Pratyush Maini, Hengrui Jia, Nicolas Papernot, and Adam Dziedzic. Llm dataset infer- ence: Did you train on my dataset? In A. Globerson, L. Mackey, D. Belgrave, A. Fan, U. Paquet, J. Tomczak, and C. Zhang, editors,Advances in Neural Information Pro- cessing Systems, volume 37, pages 124069–124092. Curran Associates, Inc., 2024. doi: 10.52202/079017-3941. ...

    work page doi:10.52202/079017-3941 2024

  35. [35]

    Active learning principles for in-context learning with large language models

    Katerina Margatina, Timo Schick, Nikolaos Aletras, and Jane Dwivedi-Yu. Active learning principles for in-context learning with large language models. In Houda Bouamor, Juan Pino, and Kalika Bali, editors,Findings of the Association for Computational Linguistics: EMNLP 2023, pages 5011–5034, Singapore, December 2023. Association for Computational Linguist...

    work page doi:10.18653/v1/2023.findings-emnlp.334 2023

  36. [36]

    Introducing meta llama 3.1: Our most capable models to date, 2024

    Meta AI. Introducing meta llama 3.1: Our most capable models to date, 2024. URL https: //ai.meta.com/blog/meta-llama-3-1/. Accessed: 2025-10-08. 16

    work page 2024

  37. [37]

    What motivates online commu- nity contributors to contribute consistently? a case study on stackoverflow netizens.Current Psychology, 42(13):10468–10481, 2023

    Sohaib Mustafa, Wen Zhang, and Muhammad Mateen Naveed. What motivates online commu- nity contributors to contribute consistently? a case study on stackoverflow netizens.Current Psychology, 42(13):10468–10481, 2023

    work page 2023

  38. [38]

    Two-person cooperative games.Econometrica, 21(1):128–140, 1953

    John Nash. Two-person cooperative games.Econometrica, 21(1):128–140, 1953. ISSN 00129682, 14680262. URLhttp://www.jstor.org/stable/1906951

    work page arXiv 1953

  39. [39]

    Tamed or trained? the co-option and capture of ‘favoured’ ngos.Third Sector Review, 13(2):27+, July 2007

    Ruth Phillips. Tamed or trained? the co-option and capture of ‘favoured’ ngos.Third Sector Review, 13(2):27+, July 2007. Accessed 28 Sept. 2025

    work page 2007

  40. [40]

    Approximate mechanism design without money

    Ariel D Procaccia and Moshe Tennenholtz. Approximate mechanism design without money. ACM Transactions on Economics and Computation (TEAC), 1(4):1–26, 2013

    work page 2013

  41. [41]

    Jie Ren, Jiaming Luo, Yao Zhao, Kundan Krishna, Mohammad Saleh, Balaji Lakshminarayanan, and Peter J. Liu. Out-of-distribution detection and selective generation for conditional language models, 2023. URLhttps://arxiv.org/abs/2209.15558

    work page arXiv 2023

  42. [42]

    So, Wojciech Ma´ nke, Hanxiao Liu, Zihang Dai, Noam Shazeer, and Quoc V

    David R. So, Wojciech Ma´ nke, Hanxiao Liu, Zihang Dai, Noam Shazeer, and Quoc V. Le. Primer: Searching for efficient transformers for language modeling, 2022. URL https://arxiv. org/abs/2109.08668

    work page arXiv 2022

  43. [43]

    A statistical interpretation of term specificity and its application in retrieval.Journal of Documentation, 28(1):11–21, 01 1972

    KAREN SPARCK JONES. A statistical interpretation of term specificity and its application in retrieval.Journal of Documentation, 28(1):11–21, 01 1972. ISSN 0022-0418. doi: 10.1108/ eb026526. URLhttps://doi.org/10.1108/eb026526

    work page doi:10.1108/eb026526 1972

  44. [44]

    Braess’s paradox of generative ai, 2024

    Boaz Taitler and Omer Ben-Porat. Braess’s paradox of generative ai, 2024. URL https: //arxiv.org/abs/2409.05506

    work page arXiv 2024

  45. [45]

    Selective response strategies for genai.arXiv preprint arXiv:2502.00729, 2025

    Boaz Taitler and Omer Ben-Porat. Selective response strategies for genai.arXiv preprint arXiv:2502.00729, 2025

    work page arXiv 2025

  46. [46]

    Perplexity, uncertainty, and the limits of active learning

    Pablo Tur´ on and Montse Cuadros. Perplexity, uncertainty, and the limits of active learning. In Emilio Corchado, H´ ector Quinti´ an, Alicia Troncoso Lora, Hilde P´ erez Garc´ ıa, Esteban Jove P´ erez, Jos´ e Luis Calvo Rolle, Francisco Javier Mart´ ınez de Pis´ on, Pablo Garc´ ıa Bringas, Francisco Mart´ ınez´Alvarez, ´Alvaro Herrero, Paolo Fosci, and R...

    work page 2026

  47. [47]

    A new active labeling method for deep learning

    Dan Wang and Yi Shang. A new active labeling method for deep learning. In2014 International Joint Conference on Neural Networks (IJCNN), pages 112–119, July 2014. doi: 10.1109/IJCNN. 2014.6889457

    work page doi:10.1109/ijcnn 2014

  48. [48]

    Munba: Machine unlearning via nash bargaining

    Jing Wu and Mehrtash Harandi. Munba: Machine unlearning via nash bargaining. InProceedings of the IEEE/CVF International Conference on Computer Vision, pages 4754–4765, 2025

    work page 2025

  49. [49]

    Wong, Em- ine Yilmaz, Shuming Shi, and Zhaopeng Tu

    Fanghua Ye, Mingming Yang, Jianhui Pang, Longyue Wang, Derek F. Wong, Em- ine Yilmaz, Shuming Shi, and Zhaopeng Tu. Benchmarking llms via uncertainty quantification. In A. Globerson, L. Mackey, D. Belgrave, A. Fan, U. Paquet, J. Tomczak, and C. Zhang, editors,Advances in Neural Information Processing Sys- tems, volume 37, pages 15356–15385. Curran Associa...

    work page 2024

  50. [50]

    Fairness-aware meta-learning via nash bargaining.Advances in Neural Information Processing Systems, 37:83235–83267, 2024

    Yi Zeng, Xuelin Yang, Li Chen, Cristian Ferrer, Ming Jin, Michael Jordan, and Ruoxi Jia. Fairness-aware meta-learning via nash bargaining.Advances in Neural Information Processing Systems, 37:83235–83267, 2024

    work page 2024

  51. [51]

    Fairshare data pricing via data valuation for large language models

    Luyang Zhang, Cathy Jiao, Beibei Li, and Chenyan Xiong. Fairshare data pricing via data valuation for large language models. InAdvances in Neural Information Processing Systems, 2025

    work page 2025

  52. [52]

    API Error: 400 invalid beta flag

    Yucheng Zhang, Chenhao Sun, Jiajun Chen, Yuxin Wang, and Yongfeng Zhang. The conse- quences of generative ai for online knowledge communities.Scientific Reports, 14(16321), 2024. doi: 10.1038/s41598-024-69804-5. 18 A Appendix B NP-Hardness of Maximizing Equation(1) Problem Statement.Let Ω = {1, . . . , n} be a finite set, and let f, g : Ω →R ≥0 be two non...

    work page doi:10.1038/s41598-024-69804-5 2024