Pith. sign in

REVIEW 4 major objections 6 minor 2 cited by

Community Moderation and the New Epistemology of Fact Checking on Social Media

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 This review argues that crowd-sourced Community Notes, while promising, fall short of being a comprehensive replacement for professional fact-checking on social media.

desk verdict A genuinely useful synthesis of the Community Notes evidence whose headline conclusion overreaches its own stated uncertainty; referee-worthy after a calibration pass. read the letter →

arxiv 2505.20067 v1 pith:INMUYFIC submitted 2025-05-26 cs.SI cs.AIcs.CY

Isabelle Augenstein , Michiel Bakker , Tanmoy Chakraborty , David Corney , Emilio Ferrara , Iryna Gurevych , Scott Hale , Eduard Hovy
show 8 more authors
Heng Ji Irene Larraz Filippo Menczer Preslav Nakov Paolo Papotti Dhruv Sahnan Greta Warren Giovanni Zagni
This is my paper · ORCID
classification cs.SIcs.AIcs.CY
keywords communitynotescrowdsourcedfact-checkingcontentmoderationmisinformationbridgingalgorithmprofessionalfact-checkersconsensussocialmediaplatforms
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 paper is a systematic review of the shift by X, Meta, and others from professional fact-checkers to crowd-sourced Community Notes. Its central claim is that community notes are genuinely promising — they can broaden participation, speed up some responses, and feel less intrusive — but in their current implementations they fall well short of a comprehensive solution to misinformation. The authors compare the two approaches across scalability, trustworthiness, effects on belief and spread, and ethics, and find critical limits on each: few proposed notes reach publication, cross-ideological consensus is slow and suppresses valid fact-checks on political topics, the crowd is vulnerable to coordinated manipulation, and relying on unpaid volunteers raises psychological and labor concerns. They conclude that community moderation is necessary but not sufficient, and that platforms should pair it with expert fact-checkers rather than replace them. The question is a live public-policy decision because Meta ended third-party fact-checking on its platforms in January 2025.

What carries the argument

The load-bearing mechanism is the community notes pipeline: an automated harmfulness classifier sorts posts into 'restricted' and 'less harmful,' and for the latter, eligible volunteers propose notes that must earn 'helpful' ratings from contributors who generally disagree with one another before the note is published. This bridging algorithm, which demands cross-ideological consensus instead of majority approval, is what the paper credits for the system's democratic promise and simultaneously indicts as its critical bottleneck, because it slows publication, strands valid notes on polarized topics, and can be deceived by coordinated groups that fake disagreement. The second conceptual engine is the contrast between two epistemologies of truth, consensus-based versus evidence-based, which the paper uses to explain why democratic legitimacy and epistemic correctness can come apart in crowd fact-checking.

What would settle it

A randomized field experiment on a major platform comparing three conditions on the same pool of misleading posts — professional fact-checker labels only, community notes only, and a hybrid in which experts review notes that failed the consensus threshold — measuring belief change, sharing, coverage, and time-to-label would settle the central claim. If notes-only or hybrid conditions matched expert labels on accuracy while exceeding their coverage, the paper's conclusion that community notes fall well short would be falsified. A simpler observational check: if adoption of the paper's own recommendations (AI-fused Supernotes, expert secondary review, collusion detection) raises note publication rates and accuracy on contentious political claims to expert levels, the bottleneck critique would no longer hold.

Watch

Extended reading notes

Core claim

The paper claims that community notes, as currently implemented by X and Meta, cannot serve as a comprehensive replacement for professional fact-checkers. The argument is built on a systematic comparison across volume, breadth, expertise, speed, democratic character, bias, transparency, resilience to coordinated attacks, effects on misinformation beliefs and spread, psychological harm, and unpaid labor. On each dimension the paper finds that the crowd's theoretical advantages are heavily qualified in practice: few proposed notes are ever published, the requirement of cross-ideological consensus delays valid notes by an average of 15.5 hours and suppresses fact-checks on politically sensitive claims, cited sources skew toward left-leaning outlets, labels can draw extra engagement to the flagged posts, and collusive groups can manufacture disagreement to game the algorithm. Beneath the empirical critique sits an epistemological claim: community notes treat facts as products of consensus and negotiation, whereas professional fact-checking treats them as objective states to be discovered from evidence, and the paper argues that the former cannot by itself guarantee correctness.

Load-bearing premise

The review's negative verdict rests on the assumption that the currently available evidence — which the paper itself describes as insufficient to judge community notes' effectiveness as an intervention, partly conflicting across studies, and partly drawn from non-peer-reviewed preprints — is still strong enough to support the conclusion that community notes fall well short of a comprehensive solution.

Editorial extensions

If this is right

  • Platforms should not treat Community Notes as a one-for-one substitute for third-party fact-checkers, because the promised scale and speed gains are not delivered by current implementations.
  • A hybrid division of labor — the crowd verifies repetitive or previously fact-checked low-risk claims while professionals handle new, high-stakes claims that require creating new knowledge — would address the main weaknesses of both approaches.
  • Appointing fact-checkers as secondary reviewers could rescue valid notes that fail the cross-ideological consensus threshold, directly improving coverage of politically contentious claims.
  • The paper's technical agenda — AI-fused Supernotes that synthesize all proposed notes, network analytics to detect collusive groups and find constructive raters, and cross-platform sharing of notes — is what would have to be built for community moderation to scale responsibly.
  • Policymakers can mandate transparency about how 'diverse perspectives' are defined and how contributors are screened, which the paper identifies as a precondition for trust in the model.

Reading between the lines

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

  • Editorial inference: if the bridging-consensus requirement is the true bottleneck, then a testable extension follows — a variant of Community Notes in which experts can approve notes that fail consensus for a defined class of high-stakes claims should raise publication rates and accuracy on political topics without measurably reducing user trust, an experiment the paper motivates but does not run.
  • Editorial inference: the paper's epistemological contrast implies that the wider adoption of consensus-based fact-checking could shift what the public treats as 'verified' toward agreement rather than evidence, a cultural consequence that would outlast any single algorithm change.
  • Editorial inference: because Meta has open-sourced its Community Notes algorithm, the paper's negative conclusions can be stress-tested with platform data — specifically, by tracking note publication rates, time-to-publication, and accuracy before and after the algorithm changes the paper recommends are rolled out.
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 manuscript is a narrative review of community-driven content moderation, focusing on X's Community Notes and Meta's announced adoption of similar mechanisms. It compares community notes with third-party fact-checking along dimensions including scalability, trustworthiness, effectiveness, and ethical implications, and offers recommendations for hybrid models combining expert fact-checkers, AI, and network analytics. The authors conclude that current implementations of the community note model "fall well short of a comprehensive solution" to misinformation, and that community efforts are necessary but not sufficient.

Significance. If its assessment is accepted, the paper provides a timely counterweight to platform narratives that community notes can replace professional fact-checking, with direct relevance to Meta's 2025 policy shift. The paper's strengths are its broad and recent literature coverage, a structured comparison table (Table 1), and concrete technical and governance recommendations (Table 2). However, the paper is a position/review piece rather than an empirical study, and its central conclusion is stronger than the evidence it cites, particularly because the effectiveness section explicitly states that insufficient data exist. Its significance therefore depends on whether the authors can either support or soften the conclusion.

major comments (4)
  1. [Do community notes help in countering misinformation?; Conclusion] The section opens with "There is insufficient data on the effectiveness of community notes as interventions for misinformation," but the Conclusion asserts that the analysis "demonstrates that they fall well short of a comprehensive solution." These statements are in direct tension. The intervening evidence is mixed: some studies find positive effects (e.g., refs 74, 81, 82), one finds null effects relative to related articles (ref 74), and one finds increased engagement with misleading posts (ref 62). Without a quantitative synthesis, power analysis, or an explicit evidential standard that treats null results as evidence of absence, the conclusion cannot be said to be demonstrated. I recommend either softening the conclusion to a provisional assessment based on structural limitations, or adding a formal evidence assessment that justifies the inference.
  2. [Are community notes more scalable?; Table 1] There is an internal inconsistency in the speed claims. The text states that "the crowd" was quicker than experts in less than 6% of cases (ref 31), while Table 1 reports "community notes are faster than factchecks <10% of the time31" for the same reference. Please reconcile these numbers and cite the precise statistic. Additionally, the claim that "only a small proportion of proposed notes reach publication" would benefit from a concrete figure or range, since it is used as a load-bearing scalability limitation.
  3. [Abstract and Introduction (review methodology)] The abstract promises a "systemic" examination, and the paper is framed as a review, but no systematic methodology is described: there is no search strategy, inclusion/exclusion criteria, or quality appraisal of the 92 references, several of which are non-peer-reviewed preprints (e.g., refs 17, 33, 72, 82, 91) or opinion pieces (e.g., ref 65). For the strong negative conclusion to be credible, the authors should either present a transparent review protocol or explicitly downscope the claim to a narrative review and add a limitations subsection acknowledging potential selection bias.
  4. [Author affiliations and competing interests] Several authors are affiliated with professional fact-checking organizations (Full Fact, Newtrales, Pagella Politica/Facta), and the paper's conclusion aligns with the position of those organizations in public debates about Meta's policy change. The manuscript does not include a competing interests statement. For a position paper with policy implications, this disclosure should be added; it is a transparency issue, not a claim about author behavior.
minor comments (6)
  1. [Abstract] "Systemically" should be "systematically."
  2. [Figure 1] The figure uses "diverse perspectives" without defining the term; the text later explains that diversity is quantified via historical disagreement, so the figure should either define it or point to that section.
  3. [References] Reference 5 is attributed to TikTok but the URL points to transparency.meta.com; please verify the citation.
  4. [Table 1] The color-only legend may be inaccessible to color-blind readers; consider adding text labels or symbols.
  5. [Table 1] "Rowd-sourced" appears to be a typo for "crowd-sourced."
  6. [Title and Introduction] The title highlights a "New Epistemology," but the epistemological discussion is limited to one paragraph; consider expanding this discussion or revising the title.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: the paper is a review whose conclusion is a synthesis of external evidence, not a re-statement of its own inputs.

full rationale

This is a position paper and literature review rather than a derivation with fitted parameters, normalizations, or equations that could reduce to its own inputs. The central claim, that current community-notes implementations fall well short of a comprehensive solution, is assembled from external empirical studies (e.g., refs 31, 44, 62, 74, 81, 82) reporting mixed positive, null, and negative findings. No prediction in the paper is statistically forced by a fitted input, and no quantity is defined in terms of the conclusion it is used to support. The self-citations that are present (refs 12, 14, 51) are used to support contextual or auxiliary claims, such as that large language models have factuality issues, that fact-checking involves intuition and sometimes requires creating new knowledge, and that community fact-checking and professional fact-checking are intertwined; the last point also has independent support from ref 64, so the argument does not collapse if the self-citations are removed. The gap between the paper's own statement that there is insufficient data on effectiveness and its stronger concluding claim that community notes 'fall well short' is an evidential-reasoning weakness, not a circularity: it does not involve the conclusion being presupposed by its premises. The fact that several authors are affiliated with fact-checking organizations is a potential conflict of interest, but it is not a logical circularity under the criteria applied here.

Assumptions & free parameters 0 free parameters · 2 assumptions · 0 invented entities

This review paper introduces no quantitative model, so there are no fitted parameters or invented entities. The argument rests on two domain assumptions about the epistemology of fact-checking and about the representativeness of the cited literature.

assumptions (2)
  • domain assumption Professional fact-checking aims at objective truth, whereas community notes treat facts as consensus.
    The paper's critique of community notes rests on this epistemological contrast (Section 'Community Notes versus Third-Party Fact-Checking'). If facts are legitimately consensus-based, the argument weakens.
  • domain assumption The cited empirical studies are representative of the overall evidence on community notes.
    The review concludes from a handful of studies, some non-peer-reviewed, that community notes fall short. This assumes the literature cited is not cherry-picked.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Community Moderation and the New Epistemology of Fact Checking on Social Media." pith.science (2026). https://pith.science/paper/INMUYFIC

@misc{pith2026250520067,
  author       = {Pith},
  title        = {Pith review of: Community Moderation and the New Epistemology of Fact Checking on Social Media},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/INMUYFIC}},
  note         = {Machine review of arXiv:2505.20067}
}
read the original abstract

Social media platforms have traditionally relied on internal moderation teams and partnerships with independent fact-checking organizations to identify and flag misleading content. Recently, however, platforms including X (formerly Twitter) and Meta have shifted towards community-driven content moderation by launching their own versions of crowd-sourced fact-checking -- Community Notes. If effectively scaled and governed, such crowd-checking initiatives have the potential to combat misinformation with increased scale and speed as successfully as community-driven efforts once did with spam. Nevertheless, general content moderation, especially for misinformation, is inherently more complex. Public perceptions of truth are often shaped by personal biases, political leanings, and cultural contexts, complicating consensus on what constitutes misleading content. This suggests that community efforts, while valuable, cannot replace the indispensable role of professional fact-checkers. Here we systemically examine the current approaches to misinformation detection across major platforms, explore the emerging role of community-driven moderation, and critically evaluate both the promises and challenges of crowd-checking at scale.

Figures

Figures reproduced from arXiv: 2505.20067 by the authors.

Figure 1
Figure 1. An overview of the community-driven content moderation framework as proposed by major social media platforms like X, Meta, and TikTok. The moderation process is divided into two main stages: (i) Pre-Moderation using AI classifiers, which categorize content as either restricted — blocked from appearing on the platform — or less harmful, which is passed on for community moderation; and (ii) Community Moderation, where… view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

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

  1. Beyond the Crowd: LLM-Augmented Community Notes for Governing Health Misinformation

    cs.SI 2025-10 unverdicted novelty 6.0 of 10

    CrowdNotes+ combines LLM note augmentation and automation with a three-stage evaluation to outperform human contributors on correctness, helpfulness, and evidence utility for health misinformation notes.

  2. The Generative AI Paradox: GenAI and the Erosion of Trust, the Corrosion of Information Verification, and the Demise of Truth

    cs.CY 2026-01 conditional novelty 4.0 of 10

    The paper proposes 'synthetic reality' as a four-layer risk stack through which cheap fake content, identities, interactions, and institutions jointly erode shared evidence and verification.

Reference graph

Works this paper leans on

80 extracted references · 50 canonical work pages · cited by 2 Pith papers

  1. [1]

    Mosseri, A. & Meta. Addressing hoaxes and fake news (2016). Accessed: 29 April 2025, https://about.fb.com/news/2016/ 12/news-feed-fyi-addressing-hoaxes-and-fake-news/

    work page 2016

  2. [2]

    Working together to end online child sexual exploitation and abuse

    TechCoalition. Working together to end online child sexual exploitation and abuse. Accessed: 29 April 2025, https: //www.technologycoalition.org

    work page 2025

  3. [3]

    Guidelines: Integrity and authenticity

    TikTok. Guidelines: Integrity and authenticity. Accessed: 29 April 2025, https://www.tiktok.com/community-guidelines/ en/integrity-authenticity

    work page 2025

  4. [4]

    Community guidelines

    YouTube. Community guidelines. Accessed: 29 April 2025, https://www.youtube.com/howyoutubeworks/policies/ community-guidelines/#enforcing-community-guidelines

    work page 2025

  5. [5]

    How enforcement technology works (2024)

    TikTok. How enforcement technology works (2024). Accessed: 29 April 2025, https://transparency.meta.com/en-gb/ enforcement/detecting-violations/how-enforcement-technology-works/

    work page 2024

  6. [6]

    Community guidelines enforcement report (2025)

    TikTok. Community guidelines enforcement report (2025). Accessed: 29 April 2025, https://www.tiktok.com/transparency/ en/community-guidelines-enforcement-2024-4

    work page 2025

  7. [7]

    & Katzenbach, C

    Gorwa, R., Binns, R. & Katzenbach, C. Algorithmic content moderation: Technical and political challenges in the automation of platform governance.Big Data & Soc.7, 2053951719897945, DOI: 10.1177/2053951719897945 (2020). https://doi.org/10.1177/2053951719897945

    doi:10.1177/2053951719897945 2020

  8. [8]

    & West, R

    Horta Ribeiro, M., Cheng, J. & West, R. Automated content moderation increases adherence to community guidelines. In Proceedings of the ACM Web Conference 2023, WWW ’23, 2666–2676, DOI: 10.1145/3543507.3583275 (Association for Computing Machinery, New York, NY , USA, 2023)

    arXiv 2023

Show all 80 references
  1. [9]

    2411.15462

    Tonneau, M.et al.Hateday: Insights from a global hate speech dataset representative of a day on twitter (2024). 2411.15462

    2024 arXiv

  2. [10]

    R., Yan, H

    DeVerna, M. R., Yan, H. Y ., Yang, K.-C. & Menczer, F. Fact-checking information from large language models can decrease headline discernment.Proc. Natl. Acad. Sci.121, e2322823121, DOI: 10.1073/pnas.2322823121 (2024). https://www.pnas.org/doi/pdf/10.1073/pnas.2322823121

    2024 doi

  3. [11]

    Zhou, X., Sharma, A., Zhang, A. X. & Althoff, T. Correcting misinformation on social media with a large language model (2024). 2403.11169

    2024

  4. [12]

    Augenstein, I.et al.Factuality challenges in the era of large language models and opportunities for fact-checking.Nat. Mach. Intell.6, 852–863, DOI: 10.1038/s42256-024-00881-z (2024)

    2024 doi

  5. [13]

    & Patil, S

    Micallef, N., Armacost, V ., Memon, N. & Patil, S. True or false: Studying the work practices of professional fact-checkers. Proc. ACM Hum.-Comput. Interact.6, DOI: 10.1145/3512974 (2022)

    2022 doi

  6. [14]

    & Augenstein, I

    Warren, G., Shklovski, I. & Augenstein, I. Show me the work: Fact-checkers’ requirements for explainable automated fact-checking. InProceedings of the 2025 CHI Conference on Human Factors in Computing Systems, CHI ’25, 1–21, DOI: 10.1145/3706598.3713277 (ACM, 2025)

    2025

  7. [15]

    In Zhou, Z.-H

    Nakov, P.et al.Automated fact-checking for assisting human fact-checkers. In Zhou, Z.-H. (ed.)Proceedings of the Thirtieth International Joint Conference on Artificial Intelligence, IJCAI-21, 4551–4558, DOI: 10.24963/ijcai.2021/619 (International Joint Conferences on Artificia...

    doi

  8. [16]

    & Saha, K

    Kolla, M., Salunkhe, S., Chandrasekharan, E. & Saha, K. LLM-Mod: Can large language models assist content moderation? InExtended Abstracts of the CHI Conference on Human Factors in Computing Systems, CHI EA ’24, DOI: 10.1145/3613905.3650828 (Association for Computing Machinery...

  9. [17]

    Wojcik, S.et al.Birdwatch: Crowd wisdom and bridging algorithms can inform understanding and reduce the spread of misinformation.arXiv preprint arXiv:2210.15723DOI: 10.48550/arXiv.2210.15723 (2022)

    2022 doi arXiv

  10. [18]

    More speech and fewer mistakes (2025)

    Kaplan, J. More speech and fewer mistakes (2025). Accessed: 29 April 2025, https://about.fb.com/news/2025/01/ meta-more-speech-fewer-mistakes/

    2025

  11. [19]

    An open letter to Mark Zuckerberg from the world’s fact-checkers, nine years later (2025)

    The International Fact-Checking Network. An open letter to Mark Zuckerberg from the world’s fact-checkers, nine years later (2025). Accessed: 29 April 2025, https://www.poynter.org/ifcn/2025/ an-open-letter-to-mark-zuckerberg-from-the-worlds-fact-checkers-nine-years-later/. 20...

    2025

  12. [22]

    Moderator code of conduct (2024)

    Reddit. Moderator code of conduct (2024). Accessed: 7 May 2025, https://redditinc.com/policies/ moderator-code-of-conduct

    2024

  13. [23]

    & O’Connor, J

    Halprin, M. & O’Connor, J. F. On policy development at YouTube (2022). Accessed: 29 April 2025, https://blog.youtube/ inside-youtube/policy-development-at-youtube/

    2022

  14. [24]

    Introducing Birdwatch, a community-based approach to misinformation (2021)

    Coleman, K. Introducing Birdwatch, a community-based approach to misinformation (2021). Accessed: 29 April 2025, https://blog.x.com/en_us/topics/product/2021/introducing-birdwatch-a-community-based-approach-to-misinformation. 25.X. About community notes on X. Accessed: 29 Apri...

    2021

  15. [26]

    & Rand, D

    Allen, J., Martel, C. & Rand, D. G. Birds of a feather don’t fact-check each other: Partisanship and the evaluation of news in twitter’s birdwatch crowdsourced fact-checking program. InProceedings of the 2022 CHI Conference on Human Factors in Computing Systems, CHI ’22, DOI: ...

    2022

  16. [27]

    Introducing birdwatch, a community-based approach to misinformation

    Coleman, K. Introducing birdwatch, a community-based approach to misinformation. Accessed: 23 May 2025, https://blog.x.com/en_us/topics/product/2021/introducing-birdwatch-a-community-based-approach-to-misinformation? utm_source=chatgpt.com. 28.Meta. Testing begins for communit...

    2025

  17. [29]

    Community “notes” function launch announcement (2023)

    Weibo. Community “notes” function launch announcement (2023). Accessed: 29 April 2025, https://weibo.com/ 1934183965/NfN4bB4ZE/

    2023

  18. [30]

    Testing a new feature to enhance content on tiktok (2025)

    Presser, A. Testing a new feature to enhance content on tiktok (2025). Accessed: 29 April 2025, https://newsroom.tiktok. com/en-us/footnotes/

    2025

  19. [31]

    & Papotti, P

    Saeed, M., Traub, N., Nicolas, M., Demartini, G. & Papotti, P. Crowdsourced fact-checking at Twitter: How does the crowd compare with experts? InProceedings of the 31st ACM International Conference on Information & Knowledge Management, CIKM ’22, 1736–1746, DOI: 10.1145/351180...

  20. [32]

    P., Solovev, K

    Drolsbach, C. P., Solovev, K. & Pröllochs, N. Community notes increase trust in fact-checking on social media.PNAS NexusDOI: 10.1093/pnasnexus/pgae217 (2024)

    2024 doi

  21. [33]

    Renault, T., Amariles, D. R. & Troussel, A. Collaboratively adding context to social media posts reduces the sharing of false news, DOI: 10.48550/arXiv.2404.02803 (2024)

    2024 doi arXiv

  22. [34]

    & Salgado, S

    Mutsvairo, B. & Salgado, S. Is citizen journalism dead? an examination of recent developments in the field.Journalism23, 354–371, DOI: 10.1177/1464884920968440 (2022)

    2022 doi

  23. [35]

    & Dahlgren, P

    Splichal, S. & Dahlgren, P. Journalism between de-professionalisation and democratisation.Eur. J. Commun.31, 5–18, DOI: 10.1177/0267323115614196 (2016)

    2016 doi

  24. [36]

    Dsa transparency report - october 2024

    X. Dsa transparency report - october 2024. Accessed: 23 May 2025, https://transparency.x.com/dsa-transparency-report. html. 37.Meta. How meta enforces its policies. Accessed: 23 May 2025, https://transparency.meta.com/enforcement/

    2024

  25. [38]

    & Masjoody, M

    X Corp. & Masjoody, M. X corp. v. masjoody, 2025 bcca 89. Accessed: 12 May 2025, https://www.canlii.org/en/bc/bcca/ doc/2025/2025bcca89/2025bcca89.html

    2025

  26. [39]

    Dave patrick underwood v

    Dave Patrick Underwood, M. Dave patrick underwood v. meta platforms, inc. Accessed: 12 May 2025, https://www. documentcloud.org/documents/21174499-underwood-v-meta-complaint/. 11/14

    2025

  27. [40]

    Global transparency report h2 2024

    X Corp. Global transparency report h2 2024. Accessed: 12 May 2025, https://transparency.x.com/en/reports/global-reports/ 2025-transparency-report

    2024

  28. [41]

    List of edit wars on wikipedia

    Wikipedia V olunteers. List of edit wars on wikipedia. Accessed: 14 May 2025, https://en.wikipedia.org/wiki/List_of_edit_ wars_on_Wikipedia

    2025

  29. [42]

    Sasahara, K.et al.Social influence and unfollowing accelerate the emergence of echo chambers.J. Comput. Soc. Sci.4, 381–402, DOI: 10.1007/s42001-020-00084-7 (2021)

    2021 doi

  30. [43]

    D., Singh, G

    Arora, S. D., Singh, G. P., Chakraborty, A. & Maity, M. Polarization and social media: A systematic review and research agenda.Technol. Forecast. Soc. Chang.183, 121942, DOI: https://doi.org/10.1016/j.techfore.2022.121942 (2022)

    2022

  31. [44]

    & Rand, D

    Martel, C., Allen, J., Pennycook, G. & Rand, D. G. Crowds can effectively identify misinformation at scale.Perspectives on Psychol. Sci.19, 477–488, DOI: 10.1177/17456916231190388 (2023). PMID: 37594056, https://doi.org/10.1177/ 17456916231190388

    2023 doi

  32. [45]

    Community engagement insights/2018 report (2018)

    Wikipedia. Community engagement insights/2018 report (2018). https://web.archive.org/web/20190724142049/https: //meta.wikimedia.org/wiki/Community_Engagement_Insights/2018_Report. 46.Wack, M., Duskin, K. & Hodel, D. Political fact-checking efforts are constrained by deficienci...

    2018

  33. [47]

    Rated Not Helpful: How X’s Community Notes system falls short on election disinformation (2024)

    Center for Countering Digital Hate. Rated Not Helpful: How X’s Community Notes system falls short on election disinformation (2024). https://counterhate.com/research/rated-not-helpful-x-community-notes/

    2024

  34. [48]

    & Klein, O

    Bhat, P. & Klein, O. Covert hate speech: White nationalists and dog whistle communication on twitter.Twitter, public sphere, chaos online deliberation151–172, DOI: 10.1007/978-3-030-41421-4_7 (2020)

    2020 doi

  35. [49]

    & Rand, D

    Martel, C. & Rand, D. G. Fact-checker warning labels are effective even for those who distrust fact-checkers.Nat. Hum. Behav.8, 1957–1967, DOI: 10.1038/s41562-024-01973-x (2024)

    2024 doi

  36. [50]

    & Rand, D

    Martel, C. & Rand, D. G. Misinformation warning labels are widely effective: A review of warning effects and their moderating features.Curr. Opin. Psychol.54, 101710, DOI: https://doi.org/10.1016/j.copsyc.2023.101710 (2023)

    2023

  37. [51]

    & Augenstein, I

    Borenstein, N., Warren, G., Elliott, D. & Augenstein, I. Can community notes replace professional fact-checkers? (2025). 2502.14132

    2025 arXiv

  38. [52]

    Locking and unlocking the ability to write notes

    X. Locking and unlocking the ability to write notes. Accessed: 29 April 2025, https://communitynotes.x.com/guide/en/ contributing/writing-ability

    2025

  39. [53]

    & Luther, J

    Stencel, M., Ryan, E. & Luther, J. With half the planet going to the polls in 2024, fact-checking sputters (2024). https://reporterslab.org/with-half-the-planet-going-to-the-polls-in-2024-fact-checking-sputters/

    2024

  40. [54]

    Balod, H. S. S. & Hameleers, M. Fighting for truth? the role perceptions of filipino journalists in an era of mis-and disinformation.Journalism22, 2368–2385 (2021)

    2021

  41. [55]

    fact-checking africa

    Cheruiyot, D. & Ferrer-Conill, R. “fact-checking africa” epistemologies, data and the expansion of journalistic discourse. Digit. Journalism6, 964–975 (2018). 56.Ababakirov, A.et al. Meeting the challenges of information disorder in the Global South(2022)

    2018

  42. [57]

    & Bastos, M

    Vinhas, O. & Bastos, M. The weird governance of fact-checking and the politics of content moderation.new media & society14614448231213942 (2023)

    2023

  43. [58]

    Global Fact Check Fund awards $2 million to 20 fact- checking groups across 15 countries (2025)

    The International Fact-Checking Network. Global Fact Check Fund awards $2 million to 20 fact- checking groups across 15 countries (2025). Accessed: 19 May 2025, https://www.poynter.org/ifcn/2025/ global-fact-check-fund-awards-2-million-to-20-fact-checking-groups-across-15-countries/

    2025

  44. [59]

    & Fazelpour, S

    Neumann, T., De-Arteaga, M. & Fazelpour, S. Justice in misinformation detection systems: An analysis of algorithms, stakeholders, and potential harms. InProceedings of the 2022 ACM Conference on Fairness, Accountability, and Transparency, FAccT ’22, 1504–1515, DOI: 10.1145/353...

    2022

  45. [60]

    InAdvances in Information Retrieval: 43rd European Conference on IR Research, ECIR 2021, Virtual Event, March 28–April 1, 2021, Proceedings, Part II 43, 639–649 (Springer, 2021)

    Nakov, P.et al.The CLEF-2021 CheckThat! Lab on detecting check-worthy claims, previously fact-checked claims, and fake news. InAdvances in Information Retrieval: 43rd European Conference on IR Research, ECIR 2021, Virtual Event, March 28–April 1, 2021, Proceedings, Part II 43,...

    2021

  46. [61]

    Graham, M., Straumann, R. K. & Hogan, B. Digital divisions of labor and informational magnetism: Mapping participation in wikipedia.Annals Assoc. Am. Geogr.105, 1158–1178 (2015). 12/14

    2015

  47. [62]

    & Majumder, S

    Wirtschafter, V . & Majumder, S. Future challenges for online, crowdsourced content moderation: Evidence from Twitter’s community notes.J. Online Trust. Saf.2, DOI: 10.54501/jots.v2i1.139 (2023)

    2023 doi

  48. [63]

    Anatomy of a fact check: Objective practice and the contested epistemology of fact checking.Commun

    Graves, L. Anatomy of a fact check: Objective practice and the contested epistemology of fact checking.Commun. culture & critique10, 518–537, DOI: https://doi.org/10.1111/cccr.12163 (2017)

    2017 doi

  49. [64]

    & Naaman, M

    Zhao, A. & Naaman, M. Insights from a comparative study on the variety, velocity, veracity, and viability of crowdsourced and professional fact-checking services.J. Online Trust. Saf.2, DOI: 10.54501/jots.v2i1.118 (2023)

    2023 doi

  50. [65]

    Meta will attempt crowdsourced fact-checking

    Mahadevan, A. Meta will attempt crowdsourced fact-checking. here’s why it won’t work (2025). Accessed: 29 April 2025, https://www.poynter.org/commentary/2025/meta-community-notes-crowdsourced-fact-checking-x/

    2025

  51. [66]

    Community-Based Fact-Checking on Twitter’s Birdwatch Platform.Proc

    Pröllochs, N. Community-Based Fact-Checking on Twitter’s Birdwatch Platform.Proc. Int. AAAI Conf. on Web Soc. Media16, 794–805, DOI: 10.1609/icwsm.v16i1.19335 (2022)

    2022 doi

  52. [67]

    & Pröllochs, N

    Solovev, K. & Pröllochs, N. References to unbiased sources increase the helpfulness of community fact-checks (2025). 2503.10560

    2025 arXiv

  53. [68]

    & Menczer, F

    Yasseri, T. & Menczer, F. Can crowdsourcing rescue the social marketplace of ideas?Commun. ACM66, 42–45, DOI: 10.1145/3578645 (2023). 69.Uscinski, J. E. & Butler, R. W. The epistemology of fact checking.Critical Rev.25, 162–180 (2013). 70.Amazeen, M. A. Revisiting the epistemo...

    2023 doi

  54. [71]

    & Sharma, R

    Kangur, U., Chakraborty, R. & Sharma, R. Who checks the checkers? exploring source credibility in Twitter’s community notes (2024). 2406.12444

    2024

  55. [72]

    & Rand, D

    Renault, T., Mosleh, M. & Rand, D. G. Republicans are flagged more often than democrats for sharing misinformation on x’s community notes, DOI: 10.31234/osf.io/vk5yj_v4 (2025)

    2025 doi

  56. [73]

    Draws, T.et al.The effects of crowd worker biases in fact-checking tasks. InProceedings of the 2022 ACM Conference on Fairness, Accountability, and Transparency, FAccT ’22, 2114–2124, DOI: 10.1145/3531146.3534629 (Association for Computing Machinery, New York, NY , USA, 2022)

    2022

  57. [74]

    & Hou, J.-R

    Kankham, S. & Hou, J.-R. Community notes vs. related articles: Assessing real-world integrated counter-rumor features in response to different rumor types on social media.Int. J. Human-Computer Interact.1–15, DOI: 10.1080/10447318.2024. 2400389 (2024). 75.X community notes. ht...

    2024

  58. [76]

    The commitments of the code of principles (2016)

    International Fact-Checking Network. The commitments of the code of principles (2016). https://ifcncodeofprinciples. poynter.org/the-commitments. 77.European Fact-Checking Standards Network. Code of standards (2022). https://efcsn.com/code-of-standards/

    2016

  59. [78]

    Pacheco, D.et al.Uncovering coordinated networks on social media: Methods and case studies. InProc. International AAAI Conference on Web and Social Media (ICWSM), vol. 15, 455–466, DOI: 10.1609/icwsm.v15i1.18075 (2021)

    2021 doi

  60. [79]

    & Howard, P

    Bradshaw, S. & Howard, P. Troops, trolls and troublemakers: A global inventory of organized social media manipulation. Comput. propaganda research project(2017)

    2017

  61. [80]

    & Pröllochs, N

    Chuai, Y ., Sergeeva, A., Lenzini, G. & Pröllochs, N. Community fact-checks trigger moral outrage in replies to misleading posts on social media, DOI: 10.48550/arXiv.2409.08829 (2024). ArXiv:2409.08829 [cs]

    2024 doi arXiv

  62. [81]

    Drolsbach, C. P. & Pröllochs, N. Diffusion of Community Fact-Checked Misinformation on Twitter.Proc. ACM on Human-Computer Interact.7, 1–22, DOI: 10.1145/3610058 (2023)

    2023 doi

  63. [82]

    & Rui, H

    Gao, Y ., Zhang, M. & Rui, H. Can Crowdchecking Curb Misinformation? Evidence from Community Notes, DOI: 10.2139/ssrn.4992470 (2024)

    2024 doi

  64. [83]

    J., Venkatagiri, S., Riedl, M

    Steiger, M., Bharucha, T. J., Venkatagiri, S., Riedl, M. J. & Lease, M. The psychological well-being of content moderators: the emotional labor of commercial moderation and avenues for improving support. InProceedings of the 2021 CHI conference on human factors in computing sy...

    2021

  65. [84]

    Three reasons Meta will struggle with community fact-checking (2025)

    Gibert, S. Three reasons Meta will struggle with community fact-checking (2025). Accessed: 12th May 2025, https: //www.technologyreview.com/2025/01/29/1110630/three-reasons-meta-will-struggle-with-community-fact-checking

    2025

  66. [85]

    E., Schafer, J., Bayar, M

    Moran, R. E., Schafer, J., Bayar, M. & Starbird, K. The end of trust and safety?: Examining the future of content moderation and upheavals in professional online safety efforts. InProceedings of the 2025 CHI Conference on Human Factors in Computing Systems, CHI ’25, DOI: 10.11...

    2025

  67. [86]

    Holan, A. D. Will the future of fact-checking flourish or founder? 2025 marks a new turning point (2025). Accessed: 29 April 2025, https://www.poynter.org/fact-checking/2025/angie-drobnic-holan-international-fact-checking-day

    2025

  68. [87]

    & Saveski, M

    De, S., Baxter, J., Bakker, M. & Saveski, M. Supernotes: Driving consensus in crowd-sourced fact-checking. InProceedings of the ACM Web Conference 2025 (WWW ’25), 11, DOI: 10.1145/3696410.3714934 (ACM, New York, NY , USA, 2025)

    2025

  69. [88]

    & Mathioudakis, M

    Garimella, K., De Francisci Morales, G., Gionis, A. & Mathioudakis, M. Reducing controversy by connecting opposing views. InProceedings of the Tenth ACM International Conference on Web Search and Data Mining, WSDM ’17, 81–90, DOI: 10.1145/3018661.3018703 (Association for Compu...

  70. [89]

    & Mathioudakis, M

    Garimella, K., De Francisci Morales, G., Gionis, A. & Mathioudakis, M. Factors in recommending contrarian content on social media. InProceedings of the 2017 ACM on Web Science Conference, WebSci ’17, 263–266, DOI: 10.1145/3091478. 3091515 (Association for Computing Machinery, ...

    2020 arXiv

  71. [91]

    M., Gatta, V

    Costabile, L., Orlando, G. M., Gatta, V . L. & Moscato, V . Assessing the potential of generative agents in crowdsourced fact-checking (2025). 2504.19940

    2025

  72. [92]

    & Nakov, P

    Shaar, S., Babulkov, N., Da San Martino, G. & Nakov, P. That is a known lie: Detecting previously fact-checked claims. InProceedings of the 58th Annual Meeting of the Association for Computational Linguistics, 3607–3618, DOI: 10.18653/v1/2020.acl-main.332 (Association for Comp...

    doi

Pith tools

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