Pith. sign in

REVIEW 3 major objections 4 minor 57 references

Strategic Analysis of Dissent and Self-Censorship

T0 review · 3 major / 4 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 For any population, an authority can induce total self-censorship by combining zero tolerance, perfect surveillance, and a sufficiently severe but credible threat of punishment, and this outcome is a Nash equilibrium.

desk verdict A clean formal model with a correct draconian-policy theorem, but the abstract oversells it: the result needs full commitment, and simulations lack robustness checks. read the letter →

arxiv 2509.03731 v1 pith:ZI664QJK submitted 2025-09-03 physics.soc-ph cs.CYcs.SI

Joshua J. Daymude , Robert Axelrod , Stephanie Forrest This is my paper
classification physics.soc-phcs.CYcs.SI
keywords self-censorshipdissentauthoritarianpolicysurveillancepunishmentseverityNashequilibriumchillingeffectstrategicadaptation
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 models dissent as a strategic choice: each individual balances the desire to express their true opinion against the risk of being observed and punished, while the authority adjusts tolerance, surveillance, and severity to minimize total dissent and punishment costs. The central result is that for any population there exists a policy—tolerate nothing, observe everything, and threaten punishment above a threshold set by the boldest member—that makes complete self-censorship the optimal action for everyone, without the authority actually punishing anyone. The paper also shows that an initially moderate authority that adapts step by step may or may not discover such a policy; the population's average boldness and its willingness to defy early threats determine whether suppression succeeds and how long it takes. Why this matters: the result gives a formal account of how authoritarian control can emerge through credible threats alone, and it identifies the point where early resistance can block that control.

What carries the argument

The central machinery is the individual's expected-utility maximization: each individual chooses an action a in [0, δ_i] to maximize β_i(1-δ_i+a) minus expected punishment, where the probability of being observed is ν + (1-ν)a and punishment is either a fixed penalty ψ (uniform) or a penalty proportional to the infraction, ψ(a-τ) (proportional). From this optimization the paper derives critical dissent values d^uni and d^pro that separate self-censorship from defiance. The draconian policy (τ*=0, ν*=1, ψ* above the stated thresholds) uses these thresholds to push every individual's optimal action to zero. The extension from moderate to extreme policy is carried by a random-mutation hill-clim

What would settle it

In a lab or online experiment, recruit participants with measured desired dissent and boldness, then announce zero tolerance, perfect surveillance, and a severity above the stated thresholds, but never actually impose punishment; if any participant continues to express dissent, the paper's total-self-censorship claim fails.

Watch

Extended reading notes

Core claim

On its own terms, the paper establishes an existence theorem for authoritarian suppression. Given any finite population with desired dissent δ_i in [0,1] and boldness β_i > 0, if the authority sets tolerance τ*=0, surveillance ν*=1, and severity ψ* above max_i {δ_i β_i} under uniform punishment (or above max_i {β_i} under proportional punishment), then each individual's expected utility is maximized by expressing zero dissent. The proof follows from solving each individual's expected-utility maximization under the authority's observation probability ν + (1-ν)a; at perfect surveillance and zero tolerance, any positive expression is certainly observed and punished, and the severity bound makes

Load-bearing premise

The authority's threat of punishment is fully credible to every individual, even though the suppressing policy never actually punishes anyone.

Editorial extensions

If this is right

  • Total suppression is achievable by threat alone: in the draconian equilibrium no punishment is ever carried out, because every individual preemptively expresses zero dissent.
  • Under proportional punishment with imperfect surveillance, the model predicts a counterintuitive inversion: moderate dissenters remain defiant while the most extreme are the first to partially self-censor.
  • An adaptive authority typically first tolerates all dissent, then tightens tolerance while raising surveillance and severity, producing a cascade of self-censorship; early defiance can derail this cascade.
  • Time to suppression grows superlinearly with the population's mean boldness, so sustained early resistance is the main lever a population has to delay or prevent authoritarian control.
  • Uniform punishment can suppress dissent with less extreme severity than proportional punishment, meaning a threat system that punishes all infractions equally is more efficient for the authority.

Reading between the lines

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

  • If the credibility of the punishment threat can be undermined or if enforcement is visibly weak, the draconian equilibrium may not survive; a natural extension would model reputation or noisy enforcement and test whether announced-but-unenforced severity loses its deterrent power.
  • The model implies a testable empirical prediction: systems that announce high surveillance and zero tolerance will produce measurable drops in dissent even before any punishment is observed, because rational individuals self-censor in advance; natural experiments around sudden policy announcements could isolate this effect.
  • In the adaptive setting, a population's early defiance acts as a costly signal of boldness; this suggests nonviolent resistance strategy should concentrate visible dissent in the early rounds before the authority tightens policy, consistent with the paper's 'do not obey in advance' framing.
  • The inversion under proportional punishment suggests that content moderation policies scaling penalties with infraction size may inadvertently protect moderate critics while driving hard-core critics underground; platform regulators could test this by comparing speech trajectories under flat versus graduated penalty schemes.
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 / 4 minor

Summary. The paper models an authority and a population in which each individual chooses a level of expressed dissent under surveillance and punishment. Two punishment regimes are studied: uniform and proportional. The authors derive the individuals' optimal actions analytically (Theorems A.1–A.2), identify compliant, self-censoring, and defiant behavioral phases, and prove the existence of a 'draconian' authority policy that induces total self-censorship (Theorem A.3, Corollary A.4). They then study an initially moderate authority that adapts via random-mutation hill climbing, reporting simulations that show suppression can emerge over time but depends on population boldness. The paper also gives qualitative illustrations using China, Russia, Hong Kong, and other contemporary examples.

Significance. If the results hold, the paper makes a useful formal contribution to the modeling of self-censorship and chilling effects. The analytical derivations are complete and are backed by openly available Mathematica and Python code, and the model has no parameters fitted to data. The non-obvious reversal under proportional punishment—where intermediate desired dissent leads to defiance but extreme desired dissent leads to partial self-censorship—is a genuine insight. The central existence theorem, with a strict severity threshold, is clean and provides a formal benchmark for discussions of authoritarian information control. The adaptive-authority simulations are exploratory but suggestive, and the qualitative illustrations are appropriately framed as illustrations rather than empirical tests.

major comments (3)
  1. [Abstract; §3.3; Corollary A.4] The abstract's headline claim, 'for any population, there exists an authority policy that leads to total self-censorship,' is stated without the qualification that appears later in §3.3 and Corollary A.4: the threat of punishment is assumed fully credible. If enforcement is probabilistic (q<1), the required severity scales as 1/q in both punishment regimes, so for any fixed finite severity there is a credibility level low enough that defiance remains optimal. The manuscript does state the credibility assumption where the theorem is proved, but the unqualified abstract and Discussion overstate the result. The claim should be presented as conditional on full commitment, or the model should be extended to include credibility as an endogenous parameter.
  2. [SI Theorem A.3 and Corollary A.4] Theorem A.3 states the draconian uniform-punishment severity as ψ* ≥ max_i δ_iβ_i, but this boundary case is not sufficient for the claimed uniqueness. If ψ* = δ_iβ_i for an individual, then under τ=0, ν=1, expected utility at a=0 equals expected utility at a=δ_i, so expressing one's desired dissent is also utility-maximizing. Corollary A.4 explicitly relies on 'uniqueness of this utility-maximizing action' to argue that no individual can deviate. The main text Eq. (9) uses strict >, which avoids the problem, but the stated theorem and its proof in the SI should use strict > as well, and the 'if and only if' in the proof should be adjusted accordingly.
  3. [§3.3, Fig. 5, Fig. S2] The adaptive-authority simulations are used to support the abstract's second claim, that the probability and time to suppression depend critically on population boldness. Yet Figs. 5 and S2 report only means over 50 trials, with no error bars, confidence intervals, or trial-level variability. Fig. S3 defines a 'time to suppression' threshold and reports a sharp transition, but again without quantifying variability near the boundary. Since the bifurcation between suppressing and tolerating policies is a central simulation finding, the authors should report variance or distributions, or soften the language from 'probability' to 'observed frequency' in the specific simulations.
minor comments (4)
  1. [§3.2] Typo: 'statues communicating the authority's tolerance' should be 'statutes.'
  2. [SI §A.4] The sentence 'and (iii) and the authority's tolerance' has a duplicated 'and.' Also, this section reports 'preliminary results' without showing the simulations; since this is future-work material, it would be clearer to state explicitly that no simulation results are presented in the paper and that the claim rests on the distributed-averaging theorem.
  3. [§3.3, Eq. (9)] The sentence 'Notably, the draconian severity ψ* for uniform punishment is at most that of proportional punishment' is correct but could be more explicit: it follows because δ_i ∈ [0,1], so max_i δ_iβ_i ≤ max_i β_i. Adding this one-line justification would help the reader.
  4. [Corollary A.4] The corollary is described as a 'one-round leader–follower game,' while the rest of the paper considers repeated rounds. The proof works for each round, but the wording should clarify that the Nash equilibrium is for the stage game, not for the repeated adaptive process.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the central existence theorem is derived from the model's stated utility functions; qualitative illustrations and simulations introduce no fitted parameters.

full rationale

The paper's central claim—that a draconian authority policy (τ*=0, ν*=1, sufficiently high ψ*) induces total self-censorship—is proved in Theorem A.3 by direct substitution into the individuals' utility-maximizing actions derived in Theorems A.1 and A.2. The severity bounds (max δ_iβ_i for uniform, max β_i for proportional) are algebraic consequences of the utility functions, not fitted quantities. The adaptive-authority simulations use randomly initialized and randomly mutated policies, with population parameters drawn from specified distributions; no parameter is calibrated to the qualitative cases (China, Russia, Hong Kong, Wikipedia) used as illustrations. The only self-citation (Mitchell, Holland, and Forrest 1993) is a reference to the hill-climbing algorithm and is not load-bearing for any social-science conclusion. The credibility assumption in Corollary A.4 is an explicit modeling condition, not a circular import, and the paper honestly labels the draconian result as existing 'under the assumption that the authority's threat of punishment is fully credible.' No step in the derivation chain reduces to its inputs by construction or via self-citation.

Assumptions & free parameters 5 free parameters · 7 assumptions · 0 invented entities

No parameters are fitted to empirical data. The theoretical results hold for arbitrary parameter values, and the simulation parameters are experimental design choices. The credibility assumption and the specific distribution and adaptation rules are the main ad hoc premises.

free parameters (5)
  • α (adamancy) = 1
    Authority's utility weight on political vs punishment cost (Eq. 3); set to 1 in all simulations.
  • ε (mutation radius) = 0.05
    Step size for random mutation hill climbing (Algorithm 1); chosen by hand.
  • Population distribution means (δ, β) = δ ∈ [0.005,0.495], β ∈ [0.1,10]
    Exponential/truncated exponential means for sampling individuals in sweeps; chosen to explore parameter space.
  • n, R, trials = 100,000; 10,000 rounds; 50 trials
    Simulation scale; chosen for computational feasibility.
  • Initial policy (τ0,ψ0,ν0) = sampled uniformly in [0,1]^3
    Random moderate starting points for the authority.
assumptions (7)
  • domain assumption Individuals maximize expected utility with linear utility in dissent and boldness (Eq. 4)
    Fundamental behavioral assumption.
  • domain assumption Authority maximizes its utility with linear tradeoff between expressed dissent and punishment cost (Eq. 3)
    Defines the authority's objective.
  • domain assumption Observation probability increases linearly with expressed dissent (Eq. 1)
    Modeling assumption that more visible dissent is more likely observed.
  • ad hoc to paper Authority's punishment threat is fully credible (Section 3.3)
    Needed for the draconian policy to deter; not derived from a reputation model.
  • domain assumption No collusion among individuals (Corollary A.4)
    Individuals act independently; no coordination.
  • ad hoc to paper Population parameters follow exponential distributions in simulations (Algorithm 1)
    Chosen for the simulations, not derived from data.
  • ad hoc to paper Authority adapts via random mutation hill climbing (Algorithm 1)
    A local search heuristic; not necessarily representative of real policy processes.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Strategic Analysis of Dissent and Self-Censorship." pith.science (2026). https://pith.science/paper/ZI664QJK

@misc{pith2026250903731,
  author       = {Pith},
  title        = {Pith review of: Strategic Analysis of Dissent and Self-Censorship},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZI664QJK}},
  note         = {Machine review of arXiv:2509.03731}
}
read the original abstract

Expressions of dissent against authority are an important feature of most societies, and efforts to suppress such expressions are common. Modern digital communications, social media, and Internet surveillance and censorship technologies are changing the landscape of public speech and dissent. Especially in authoritarian settings, individuals must assess the risk of voicing their true opinions or choose self-censorship, voluntarily moderating their behavior to comply with authority. We present a model in which individuals strategically manage the tradeoff between expressing dissent and avoiding punishment through self-censorship while an authority adapts its policies to minimize both total expressed dissent and punishment costs. We study the model analytically and in simulation to derive conditions separating defiant individuals who express their desired dissent in spite of punishment from self-censoring individuals who fully or partially limit their expression. We find that for any population, there exists an authority policy that leads to total self-censorship. However, the probability and time for an initially moderate, locally-adaptive authority to suppress dissent depend critically on the population's willingness to withstand punishment early on, which can deter the authority from adopting more extreme policies.

Figures

Figures reproduced from arXiv: 2509.03731 by the authors.

Figure 1
Figure 1. Individuals’ Optimal Actions Under Uniform Punishment. The optimal action of an individual 𝐼𝑖 as a function of its desired dissent 𝛿𝑖 when 𝜋 is uniform (Eq. 5) and (A) sufficient boldness 𝛽𝑖 > (1 −𝜈𝑟) ·𝜓𝑟 makes defiance possible or (B) otherwise, when full self-censorship always occurs above 𝜏𝑟. Compliance is shown in teal, self-censorship in olive, and defiance in peach. where 𝛽𝑖 > 0 is a constant representing how … view at source ↗
Figure 2
Figure 2. Individuals’ Optimal Actions Under Proportional Punishment. The optimal action of an individual 𝐼𝑖 as a function of its desired dissent 𝛿𝑖 when 𝜋 is proportional (Eq. 7) and (A) the authority’s surveillance is imperfect with 𝜈𝑟 < 1 and the authority’s tolerance 𝜏𝑟 < 𝑑 pro 𝑖,𝑟 is insufficient to eliminate defiance, (B) the authority’s surveillance is perfect with 𝜈𝑟 = 1 but the individual’s boldness 𝛽𝑖 ≥ 𝜓𝑟 yields on… view at source ↗
Figure 3
Figure 3. Phase Diagram of Individuals’ Optimal Behaviors by Parameter. Compliant (teal), self-censoring (olive), and defiant (peach) phases of individuals’ optimal actions are shown as a function of (A) the authority’s tolerance 𝜏𝑟, (B) the authority’s surveillance 𝜈𝑟, (C) the authority’s severity 𝜓𝑟, and (D) the individual’s boldness 𝛽𝑖 vs. their desired dissent 𝛿𝑖 for both uniform (top row) and proportional (bottom row) pu… view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Time Evolution of an Adaptive Authority. The adaptive authority’s costs (top) and parameters (bottom) over 5,000 rounds with uniform punishment and adamancy 𝛼 = 1. The population has 𝑛 = 100,000 individuals whose desired dissents are sampled independently from a trunca…
Figure 5
Figure 5. Figure 5: Authority Costs and Policies After Adaptation Under Uniform Punishment. The authority’s average final (A)–(C) costs and (D)–(F) parameter values after 10,000 rounds of random mutation hill climbing adaptation across 50 independent trials per (𝛿, 𝛽) pair, where 𝛿 ∈ [0.0…

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

57 extracted references · 53 canonical work pages

  1. [1]

    Peter Ackerman and Berel Rodal. 2008. The Strategic Dimensions of Civil Resistance. Survival 50, 3 (2008), 111–126. doi:10.1080/00396330802173131

    work page doi:10.1080/00396330802173131 2008

  2. [2]

    Amnesty International. 2023. Israel and Occupied Palestinian Territories: Automated Apartheid: How Facial Recognition Fragments, Segregates and Controls Palestinians in the OPT . Technical Report MDE 15/6701/2023

    work page 2023

  3. [3]

    Daniel Bar-Tal. 2017. Self-Censorship as a Socio-Political-Psychological Phenomenon: Conception and Research. Political Psychology 38, S1 (2017), 37–65. doi:10.1111/pops.12391

    doi:10.1111/pops.12391 2017

  4. [4]

    Erica Chenoweth. 2021. Civil Resistance: What Everyone Needs to Know . Oxford University Press, New York, NY, USA

    work page 2021

  5. [5]

    Erica Chenoweth and Maria J. Stephan. 2011. Why Civil Resistance Works: The Strategic Logic of Nonviolent Conflict . Columbia University Press, New York, NY, USA. Strategic Analysis of Dissent and Self-Censorship 11

    work page 2011

  6. [6]

    Farhan Asif Chowdhury, Dheeman Saha, Md Rashidul Hasan, Koustuv Saha, and Abdullah Mueen. 2022. Examining Factors Associated with Twitter Account Suspension Following the 2020 U.S. Presidential Election. In Proceedings of the 2021 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining . ACM, Virtual Event, Netherlands, 607–61...

    work page arXiv 2022

  7. [7]

    MacKenzie F. Common. 2020. Fear the Reaper: How Content Moderation Rules Are Enforced on Social Media. International Review of Law, Computers & Technology 34, 2 (2020), 126–152. doi:10.1080/13600869.2020.1733762

    work page arXiv 2020

  8. [8]

    Giacomo Como and Fabio Fagnani. 2016. From Local Averaging to Emergent Global Behaviors: The Fundamental Role of Network Interconnections. Systems & Control Letters 95 (2016), 70–76. doi:10.1016/j.sysconle.2016.02.003

    work page doi:10.1016/j.sysconle.2016.02.003 2016

Show all 57 references
  1. [9]

    Andrew Cunningham. 2024. Microsoft Is Reworking Recall after Researchers Point out Its Security Problems. Ars Technica (2024)

    2024

  2. [10]

    Andrew Cunningham. 2024. Windows Recall Demands an Extraordinary Level of Trust That Microsoft Hasn’t Earned. Ars Technica (2024)

    2024

  3. [11]

    This year, the Guangdong Internet Police has dealt with 269,000 pieces of illegal information of all kinds

    Cyberspace Administration of China. 2015. “This year, the Guangdong Internet Police has dealt with 269,000 pieces of illegal information of all kinds”. https://www.cac.gov.cn/2015-09/28/c_1116702824.htm

    2015

  4. [12]

    Crandall

    Roya Ensafi, Philipp Winter, Abdullah Mueen, and Jedidiah R. Crandall. 2015. Analyzing the Great Firewall of China Over Space and Time. InProceedings on Privacy Enhancing Technologies, Vol. 2015. 61–76. doi:10.1515/popets-2015-0005

    2015 doi

  5. [13]

    Alena Epifanova. 2021. Subjugating RuNet: The Kremlin’s New Levers of Control over Elections and Society. The Russia File (2021)

    2021

  6. [14]

    Ryan Fedasiuk. 2021. Buying Silence: The Price of Internet Censorship in China. China Brief 21, 1 (2021), 19–25

    2021

  7. [15]

    Steven Feldstein. 2021. The Rise of Digital Repression: How Technology Is Reshaping Power, Politics, and Resistance . Oxford University Press, New York, NY, USA

    2021

  8. [16]

    Tarleton Gillespie. 2018. Custodians of the Internet: Platforms, Content Moderation, and the Hidden Decisions That Shape Social Media. Yale University Press, New Haven, CT, USA

    2018

  9. [17]

    John Horton. 2011. Self-Censorship. Res Publica 17, 1 (2011), 91–106. doi:10.1007/s11158-011-9145-3

    2011 doi

  10. [18]

    Andrew Hsiao and Audrea Lim. 2020. The Verso Book of Dissent: Revolutionary Words From Three Millennia of Rebellion and Resistance. Verso, London, United Kingdom

    2020

  11. [19]

    2023.Meta’s Broken Promises: Systemic Censorship of Palestine Content on Instagram and Facebook

    Human Rights Watch. 2023.Meta’s Broken Promises: Systemic Censorship of Palestine Content on Instagram and Facebook. Technical Report

    2023

  12. [20]

    Pratyusha Ria Kalluri, William Agnew, Myra Cheng, Kentrell Owens, Luca Soldaini, and Abeba Birhane. 2025. Computer- Vision Research Powers Surveillance Technology. Nature 643, 8070 (2025), 73–79. doi:10.1038/s41586-025-08972-6

    2025 doi

  13. [21]

    Joel Kaplan. 2025. More Free Speech and Fewer Mistakes. https://about.fb.com/news/2025/01/meta-more-speech- fewer-mistakes/

    2025

  14. [22]

    Chun Lin. 2006. The Transformation of Chinese Socialism . Duke University Press, Durham [N.C.]

    2006

  15. [23]

    Perry Link. 2002. China: The Anaconda in the Chandelier. The New York Review of Books 49, 6 (2002)

    2002

  16. [24]

    Antony Loewenstein. 2023. The Palestine Laboratory: How Israel Exports the Technology of Occupation Around the World. Verso Books, London, United Kingdom

    2023

  17. [25]

    Tetyana Lokot. 2021. Beyond the Protest Square: Digital Media and Augmented Dissent . Rowman & Littlefield, Lanham, MD, USA

    2021

  18. [26]

    Darren Loucaides. 2024. The Changing Face of Protest. Rest of World (2024)

    2024

  19. [27]

    Political Correctness

    Glenn C. Loury. 1994. Self-Censorship in Public Discourse: A Theory of “Political Correctness” and Related Phenomena. Rationality and Society 6, 4 (1994), 428–461. doi:10.1177/1043463194006004002

    1994 doi

  20. [28]

    Roderick Macfarquhar, Timothy Cheek, and Eugene Wu. 1989. The Hundred Flowers. In The Secret Speeches of Chairman Mao. Harvard University Asia Center, 105–111. doi:10.1163/9781684171125_007

    1989 doi

  21. [29]

    Binny Mathew, Anurag Illendula, Punyajoy Saha, Soumya Sarkar, Pawan Goyal, and Animesh Mukherjee. 2020. Hate Begets Hate: A Temporal Study of Hate Speech. Proceedings of the ACM on Human-Computer Interaction 4, CSCW2 (2020), 1–24. doi:10.1145/3415163

    2020 doi

  22. [30]

    Holland, and Stephanie Forrest

    Melanie Mitchell, John H. Holland, and Stephanie Forrest. 1993. When Will a Genetic Algorithm Outperform Hill Climbing?. In Advances in Neural Information Processing Systems , Vol. 6. Morgan Kaufmann, 51–58

    1993

  23. [31]

    Chris Morris. 2024. Why Facial Recognition Technology Makes These Campus Protests Different from Those in the Past. Fast Company (2024)

    2024

  24. [32]

    Paul Mozur. 2019. In Hong Kong Protests, Faces Become Weapons. The New York Times (2019)

    2019

  25. [33]

    Sarah Myers West. 2018. Censored, Suspended, Shadowbanned: User Interpretations of Content Moderation on Social Media Platforms. New Media & Society 20, 11 (2018), 4366–4383. doi:10.1177/1461444818773059

    2018 doi

  26. [34]

    Gabriel Nicholas. 2022. Shedding Light on Shadowbanning . Technical Report. Center for Democracy & Technology

    2022

  27. [35]

    Office of the Director of National Intelligence. 2024. Annual Statistical Transparency Report Regarding the Intelligence Community’s Use of National Security Surveillance Authorities: Calendar Year 2023 . Technical Report. 12 Joshua J. Daymude, Robert Axelrod, and Stephanie Forrest

    2024

  28. [36]

    Jonathon W. Penney. 2016. Chilling Effects: Online Surveillance and Wikipedia Use. Berkeley Technology Law Journal 31, 1 (2016), 118–182. doi:10.15779/Z38SS13

    2016 doi

  29. [37]

    Jonathon W. Penney. 2017. Internet Surveillance, Regulation, and Chilling Effects Online: A Comparative Case Study. Internet Policy Review 6, 2 (2017), 1–24. doi:10.14763/2017.2.692

    2017 doi

  30. [38]

    Jonathon W. Penney. 2022. Understanding Chilling Effects. Minnesota Law Review 106, 3 (2022), 1451–1530

    2022

  31. [39]

    Francesco Pierri, Luca Luceri, Emily Chen, and Emilio Ferrara. 2023. How Does Twitter Account Moderation Work? Dynamics of Account Creation and Suspension on Twitter during Major Geopolitical Events. EPJ Data Science 12, 1 (2023), 43. doi:10.1140/epjds/s13688-023-00420-7

    2023 doi

  32. [40]

    Margaret E. Roberts. 2018. Censored: Distraction and Diversion Inside China’s Great Firewall . Princeton University Press. doi:10.2307/j.ctvc77b21

    2018 doi

  33. [41]

    Schelling

    Thomas C. Schelling. 1971. Some Questions on Civilian Defense. In Conflict: Violence and Nonviolence (1st ed.). Taylor and Francis, London, UK

    1971

  34. [42]

    Farhana Shahid and Aditya Vashistha. 2023. Decolonizing Content Moderation: Does Uniform Global Community Standard Resemble Utopian Equality or Western Power Hegemony?. In Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems . ACM, Hamburg, Germany, 1–1...

    2023

  35. [43]

    Gene Sharp. 1973. The Politics of Nonviolent Action . Porter Sargent, Boston, MA, USA

    1973

  36. [44]

    Xiaoxiao Shen and Rory Truex. 2021. In Search of Self-Censorship. British Journal of Political Science 51, 4 (2021), 1672–1684. doi:10.1017/S0007123419000735

    2021 doi

  37. [45]

    Timothy Snyder. 2017. On Tyranny: Twenty Lessons from the Twentieth Century . Crown, New York, NY, USA

    2017

  38. [46]

    Cagri Toraman, Furkan Şahinuç, and Eyup Halit Yilmaz. 2022. BlackLivesMatter 2020: An Analysis of Deleted and Suspended Users in Twitter. In 14th ACM Web Science Conference 2022 . ACM, Barcelona, Spain, 290–295. doi:10.1145/ 3501247.3531539

    2022

  39. [47]

    Donald J. Trump. 2025. Ending Illegal Discrimination And Restoring Merit-Based Opportunity . Presidential Action. The White House, Washington DC, USA

    2025

  40. [48]

    Donald J. Trump. 2025. Ending Radical And Wasteful Government DEI Programs And Preferencing . Executive Order. The White House, Washington DC, USA

    2025

  41. [49]

    Donald J. Trump. 2025. Restoring Freedom Of Speech And Ending Federal Censorship . Executive Order. The White House, Washington DC, USA

    2025

  42. [50]

    James Vincent. 2020. NYPD Used Facial Recognition to Track Down Black Lives Matter Activist. The Verge (2020)

    2020

  43. [51]

    Mona Wang and Jonathan Mayer. 2023. Self-Censorship Under Law: A Case Study of the Hong Kong National Security Law. Free and Open Communications on the Internet 1 (2023), 46–55

    2023

  44. [52]

    Richard Ashby Wilson and Molly K. Land. 2021. Hate Speech on Social Media: Content Moderation in Context. Connecticut Law Review 52, 3 (2021), 1029–1076

    2021

  45. [53]

    Ronald Wintrobe. 1998. The Political Economy of Dictatorship . Cambridge University Press, Cambridge, MA, USA

    1998

  46. [54]

    Crandall, and Roya Ensafi

    Diwen Xue, Benjamin Mixon-Baca, ValdikSS, Anna Ablove, Beau Kujath, Jedidiah R. Crandall, and Roya Ensafi. 2022. TSPU: Russia’s Decentralized Censorship System. In Proceedings of the 22nd ACM Internet Measurement Conference . ACM, Nice, France, 179–194. doi:10.1145/3517745.3561461

    2022

  47. [55]

    Kenneth C.C. Yang. 2011. The Aborted Green Dam-Youth Escort Censor-Ware Project in China: A Case Study of Emerging Civic Participation in China’s Internet Policy-Making Process. Telematics and Informatics 28, 2 (2011), 101–111. doi:10.1016/j.tele.2010.07.001

    2011 doi

  48. [56]

    YouTube. 2025. YouTube Community Guidelines: Taking Action on Violations. https://www.youtube.com/howyoutubeworks/policies/community-guidelines/#taking-action-on-violations

    2025

  49. [57]

    socialization

    Yanqiu Rachel Zhou. 2023. The Lying Flat Movement, Global Youth, and Globality: A Case of Collective Reading on Reddit. Globalizations 20, 4 (2023), 679–695. doi:10.1080/14747731.2023.2165377 A Supporting Information (SI Appendix) This Supporting Information details all deriva...

    2023

Pith tools

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