Pith. sign in

REVIEW 3 major objections 4 minor 10 references

Delegation and Participation in Decentralized Governance: An Epistemic View

T0 review · 3 major / 4 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read Partial abstention beats transfer delegation in DAO votes

desk verdict Partial abstention result is real and mostly proven, but the paper's practical claims outrun its strong assumptions—worth a serious referee. read the letter →

arxiv 2505.04136 v1 pith:GYZ53RKV submitted 2025-05-07 cs.SI cs.GTecon.TH

Jeff Strnad This is my paper · ORCID
classification cs.SIcs.GTecon.TH
keywords epistemicvotingtheorypartialabstentionoptimalweightsliquiddemocracydecentralizedgovernanceCondorcetjurytheoremvotercompetenceDAO
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 asks how decentralized organizations such as DAOs can make decisions that are both decentralized and likely to be correct, and it evaluates governance methods by the probability that they pick a correct answer when one exists. Its central result is that partial abstention, voters exercising only a portion of their voting rights, can reproduce the optimal weighted vote almost without coordination: if each voter abstains so that their optimal weight divided by the number of votes they cast equals the same common number R, the realized voting shares become proportional to the optimal weights. That makes partial abstention epistemically superior, in this model, to transfer delegation in any form, including liquid democracy, which requires voters to know other voters' competences and suffers from over-delegation. The paper also shows that more direct participation helps only when the voting rule already uses optimal weights; outside that environment, adding low-competence or correlated voters can sharply reduce the probability of a correct outcome.

What carries the argument

The load-bearing device is the Optimal Weighting Theorem: for independent binary voters, the decision rule that maximizes the probability of a correct collective choice weights each voter by w_i = ln(p_i/(1-p_i)), the log odds of the voter's competence. The paper's new contribution is the ratio-symmetric abstention rule w_i/t*_i = R: because every voter's exercised votes are scaled to their optimal weight by the same constant, the realized vote shares are exactly proportional to the optimal weights and the aggregate decision is identical to the social-planner optimum. A second device, the canonical independent-signals decomposition, extends the same arithmetic to correlated voters by treating shared signals as a single weighted signal discounted by its audience count.

What would settle it

Run a binary-choice task with known correct answers, elicit each participant's self-assessed p_i, apply the abstention rule w_i/t*_i = R, and compare collective accuracy against simple majority and the social-planner optimum. The claim is falsified if the abstention rule underperforms a naive rule whenever self-assessed p_i are miscalibrated.

Watch

Extended reading notes

Core claim

The paper's central claim is Proposition 3.2: with independent voter competencies and at least one voter with p_j > 0.5, if every voter partially abstains so that w_i/t*_i = R, where w_i = ln(p_i/(1-p_i)) is the voter's optimal weight and R is the supremum of w_i/t_i over voters, then the probability of a correct collective decision under weighted voting is maximized. The argument is an identity: the abstention ratios make effective weights proportional to optimal weights by a common factor, so the two voting rules are decision-equivalent. A corollary is that the optimum is reached with only common knowledge of the single number R, or any sufficiently large substitute, rather than a social planner or knowledge of other voters' competences. In the dependent-competencies case the same logic applies to independent signals: each voter discounts each signal by the number of voters who share it, restoring optimal weighting under the paper's information-decomposition and signal-numeracy assumptions. The paper therefore asserts that the main obstacle to optimal decentralized voting is not aggregation but the accuracy of each voter's self-assessed competence.

Load-bearing premise

Voters can compute an accurate probability p_i that their own information leads to the correct choice, and the optimality of the abstention rule collapses if those self-assessments are miscalibrated.

Editorial extensions

If this is right

  • Under independent competencies, a DAO can attain the best possible epistemic outcome using only one public reference number and each voter's own competence estimate, with no delegate network, social planner, or knowledge about other voters.
  • Any governance method built on indivisible votes, including ordinary majority voting and liquid democracy with all-or-nothing delegation, is epistemically dominated by partial abstention in the environments the paper models.
  • A policy of increasing direct participation is guaranteed to do no epistemic harm only inside the optimal epistemic environment; otherwise new voters can flood the decision with low-weight or correlated signals and push the probability of correctness below 0.5.
  • Transfer delegation, even in multi-step form, requires delegative competence, knowing other voters' optimal weights, and is vulnerable to over-delegation; the paper gives only a narrow connected-network condition under which three-step transfer delegation can reach the optimum.
  • With dependent competencies, optimal decentralized outcomes require each voter to know how many others share each signal, so dependency tracking, not abstention arithmetic, is the binding constraint.

Reading between the lines

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

  • The same ratio identity applies to any weighted committee or expert panel with competence estimates, so the result is not tied to blockchain tokens; a medical or scientific advisory body could adopt partial abstention with a publicly agreed R.
  • A practical DAO could implement the rule in code: voters report p_i, the contract computes w_i and instructs each wallet to cast w_i/R of its vote weight, turning the paper's coordination condition into an implementable default.
  • If self-reported p_i are verifiable ex post against voting records, the mechanism could be combined with reputation or slashing to discipline miscalibration, which the paper itself identifies as the fragile point.
  • The flooding danger suggests that turnout policy should be signal-based: recruiting voters who bring independent information helps, while blanket participation campaigns can actively reduce decision accuracy.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

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 introduces two epistemic tests for decentralized governance mechanisms, one for independent voter competencies and one for dependent competencies modeled through a canonical list of independent signals. Using the Nitzan–Paroush optimal weighting theorem as an external benchmark, it analyzes partial abstention, direct participation, and transfer delegation. The central positive result is Proposition 3.2, which shows that if each voter knows their own competence and all voters coordinate on a single number R, partial abstention can replicate the optimal weighted vote. Section 4 identifies conditions under which additional direct participation is weakly beneficial and characterizes a 'flooding danger' from low-competence voters. Section 5 argues that transfer delegation, including liquid democracy, faces severe coordination and information requirements, with Proposition 5.3 giving a three-step delegation scheme that achieves optimality in connected star-cluster graphs. The paper concludes by discussing supplementary mechanisms such as futarchy, AI agents, and contestable control.

Significance. If the results are accepted, Proposition 3.2 is an elegant and nontrivial observation: under strong but clearly stated assumptions, partial abstention achieves the same collective outcome as a social planner who directly applies optimal weights, with much weaker coordination than transfer delegation. The paper is commendably transparent about its assumptions and about the limits of its framework, and several proofs (Propositions 3.2, 3.8, 4.1, 5.3) are cleanly written and check out. The external benchmark of Nitzan and Paroush is used correctly, and the paper does not circularly fit parameters to data. The main limitation is that the headline practical claim that partial abstention is a 'strong governance method' rests entirely on Assumption 3, and the paper provides no sensitivity analysis for miscalibrated self-assessed competencies. This makes the theoretical results valuable but the practical inference more fragile than the abstract suggests.

major comments (3)
  1. [§3, Proposition 3.2 and Assumption 3] The optimality of partial abstention is entirely conditional on each voter knowing their own p_i exactly, and the paper provides no analysis of what happens when voters misestimate p_i. The conclusion (§7) concedes that 'voters may not have very sound knowledge of their own competencies,' but without a sensitivity bound or numerical exploration, the abstract's claim that 'partial abstention is a strong governance method' is not supported as a practical statement. Please add a formal sensitivity result (e.g., a Lipschitz bound on the collective success probability as a function of individual log-odds errors) or numerical simulations showing how the probability of a correct collective decision degrades with miscalibration, and adjust the wording of the practical claim accordingly.
  2. [§5.2, Proposition 5.1] Proposition 5.1 is explicitly incomplete. The proof only analyzes the coalition pair with the minimum voting-rights gap and then states that 'it is likely that a maximizing algorithm is possible at least by brute force, but we leave that possibility to future work.' As a result, the proposition's statement that conditions (C) and (D) are sufficient for a voter to make transfer delegation decisions that weakly increase the probability of a correct collective decision is not established. Since this proposition is used to argue that optimal transfer delegation requires weight omniscience and creates severe coordination problems, the argument is currently stronger than the proof supports. Please either prove the general maximizing algorithm, restrict the proposition to the pairwise decision that is actually proven, or explicitly label the general claim as a conjecture.
  3. [§4, Corollary 4.4(iii)] The statement of Corollary 4.4(iii) contains a mathematical typo: the range for which adding two voters decreases the probability of a correct collective decision should be defined by w_i + w_j < ln(phi), not w_i + w_j < phi. The proof and the use of the result in part (iv) rely on ln(phi), and as written the inequality is false in general because phi > ln(phi) for phi > 1. This should be corrected.
minor comments (4)
  1. [§3, Example 3.7] The explanation that a voter with w_m > tilde R receives an 'effective weight of 1000' could be misunderstood, since each voter has only one voting right after reallocation; the intended normalization (multiplying exercised rights by tilde R) should be stated explicitly to avoid confusion.
  2. [§2.2.3, Proposition 2.2] The proposition is presented without a proof; a short proof or a direct citation to the Nitzan and Paroush theorem applied to signals would improve completeness.
  3. [§5.4, Sortition] The sortition discussion is informal and does not state results as propositions; adding a precise comparative statement (e.g., a bound on the loss from excluding positive-weight voters) would make the section more useful to readers who want to compare mechanisms formally.
  4. [References] The reference to the House Financial Services Committee appears twice with the same author name but different years; please format as 2023 and 2024 entries consistently.

Circularity Check

0 steps flagged · score 2.0 of 10

No material circularity: the main results are conditional theorems built on the external Nitzan-Paroush optimal-weighting benchmark; the only self-citations are supplementary and non-load-bearing.

full rationale

I walked the derivation chain from Section 2 through Section 5. The optimal-weight benchmark is external: Nitzan and Paroush (1982), restated as equation (1) and used through Lemma 4.2; the paper does not fit weights to outcome data. Proposition 3.2 is a constructive equivalence, not a fitted prediction: it assumes each voter knows p_i and R and chooses t*_i = w_i/R; the proof then verifies the identity w*_i = t*_i / sum(t*) = w_i / sum(w_i). The optimal weights are an input to the abstention rule, not an output inferred from the rule. Proposition 3.8 does the same for independent signals under Assumptions 4 and 7; again the aggregation identity K times (w_i/(KR)) = w_i/R is the argument. Section 4's participation results are applications of Ben-Yashar and Nitzan (2017) and the external optimal-weighting theorem, and Section 5's delegation results are derived from the same framework. The strong epistemic assumptions, especially Assumption 3, are acknowledged in the text and in the conclusion; strong assumptions are a robustness or correctness concern, not circularity. The only same-author citations are Strnad (2025) in Section 6.4 and Strnad (2024) in footnote 40; they propose a contestable-control supplement and discuss a Grossman-Hart point, but neither is used to establish the central abstention, participation, or delegation theorems. I therefore find no circular step; the score of 2 reflects these minor, non-load-bearing self-citations rather than any reduction of a result to its inputs.

Assumptions & free parameters 0 free parameters · 8 assumptions · 0 invented entities

The central claims rest on seven explicitly stated domain assumptions, especially Assumptions 2 and 3. These are acknowledged by the author as very strong. The benchmark optimal weighting theorem is external and cited. No entities are invented and no parameters are fitted to data.

assumptions (8)
  • domain assumption Assumption 1: Decisions are determinable; a single correct answer exists.
    Stated in Section 2.1; the entire epistemic framework requires a binary choice with a correct alternative.
  • domain assumption Assumption 2: Voters share the epistemic objective of reaching correct collective decisions.
    Stated in Section 2.1; rules out malice and instrumental voting, which are major real-world phenomena.
  • domain assumption Assumption 3: Voters can compute an accurate probability p_i of being correct based on their information.
    Stated in Section 2.1; load-bearing for optimal weights and partial abstention calibration.
  • domain assumption Assumption 4: All voter information decomposes into a canonical list of independent signals, and each voter's information is a sum of items on the list.
    Stated in Section 2.1; enables the dependent-competencies test.
  • domain assumption Assumptions 5A/5B: Voter judgments are either independent or arbitrarily dependent.
    Stated in Sections 2.2.2 and 2.2.3; these define the two tests.
  • domain assumption Assumption 6: Voters may vote or abstain on any portion of their voting rights and may split delegation arbitrarily.
    Stated in Section 2.2.3; partial abstention requires this flexibility.
  • domain assumption Assumption 7: For each independent signal a voter receives, the voter knows how many other voters received the same signal.
    Stated in Section 3; needed for the dependence-case optimal abstention algorithm.
  • standard math Nitzan and Paroush Optimal Weighting Theorem: optimal weights are w_i = ln(p_i/(1-p_i)).
    External benchmark used throughout; cited to Nitzan and Paroush (1982), not proved in this paper.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Delegation and Participation in Decentralized Governance: An Epistemic View." pith.science (2026). https://pith.science/paper/GYZ53RKV

@misc{pith2026250504136,
  author       = {Pith},
  title        = {Pith review of: Delegation and Participation in Decentralized Governance: An Epistemic View},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GYZ53RKV}},
  note         = {Machine review of arXiv:2505.04136}
}
read the original abstract

We develop and apply epistemic tests to various decentralized governance methods as well as to study the impact of participation. These tests probe the ability to reach a correct outcome when there is one. We find that partial abstention is a strong governance method from an epistemic standpoint compared to alternatives such as various forms of ``transfer delegation" in which voters explicitly transfer some or all of their voting rights to others. We make a stronger case for multi-step transfer delegation than is present in previous work but also demonstrate that transfer delegation has inherent epistemic weaknesses. We show that enhanced direct participation, voters exercising their own voting rights, can have a variety of epistemic impacts, some very negative. We identify governance conditions under which additional direct participation is guaranteed to do no epistemic harm and is likely to increase the probability of making correct decisions. In light of the epistemic challenges of voting-based decentralized governance, we consider the possible supplementary use of prediction markets, auctions, and AI agents to improve outcomes. All these results are significant because epistemic performance matters if entities such as DAOs (decentralized autonomous organizations) wish to compete with organizations that are more centralized.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

10 extracted references · 10 canonical work pages

  1. [1]

    Con- dorcet Markets,

    Airiau, St´ ephane, Nicholas Kees Dupuis, and Davide Grossi (2024) “Con- dorcet Markets,” inAlgorithmic Game Theory, 501–519: Springer Nature Switzerland. Amico, Jeff (2021) “Open Sourcing Our Token Delegate Program,”a16z crypto,https://a16zcrypto.com/posts/article/open-sourcing-our- token-delegate-program/. Bar-Isaac, Heski and Joel Shapiro (2020) “Block...

    work page 2024

  2. [6]

    Liquid Democracy or Direct Democracy? One Theoretical Result and Two Experiments

    Jensen, Michael C. and William H. Meckling (1976) “Theory of the firm: Managerial behavior, agency costs and ownership structure,”Journal of Financial Economics, 3 (4), 305–360. Kahng, Anson, Simon Mackenzie, and Ariel D. Procaccia (2021) “Liquid Democracy: An Algorithmic Perspective,”Journal of Artificial Intelli- gence Research, 70, 1223–1252. Khanna, V...

    work page Pith review arXiv 1976

  3. [8]

    Stay away from fair coins: A Condorcet jury theo- rem,

    Paroush, Jacob (1997) “Stay away from fair coins: A Condorcet jury theo- rem,”Social Choice and Welfare, 15 (1), 15–20. 73 Reyes, Carla L., Nizan Geslevich Packin, and Benjamin P. Edwards (2017) “Distributed Goverance,”William and Mary Law Review Online, 59, 1–32, https://scholarship.law.wm.edu/wmlronline/vol59/iss1/1/. Snowberg, Erik, Justin Wolfers, and...

    work page arXiv 1997

  4. [9]

    Economic DAO Governance: A Contestable Control Approach

    Strnad, Jeff (2024) “Economic DAO Governance: A Contestable Control Ap- proach,”https://arxiv.org/abs/2403.16980, Version 4, December 25,

    work page Pith review arXiv 2024

  5. [10]

    Economic DAO Governance: A Contestable Control Ap- proach,

    (2025) “Economic DAO Governance: A Contestable Control Ap- proach,”Blockchain: Research and Applications, forthcoming. 74

    work page 2025

  6. [289]

    Introducing Rank-Choice Delegation,

    Novak, Graham (2024) “Introducing Rank-Choice Delegation,”X,https: //x.com/gnovak /status/1820784566426018074, Posted on X, August 6,

    work page arXiv 2024

  7. [2022]

    Butter (2025) “Butter,”https://butterd.notion.site/, Accessed Febru- ary 11,

    work page 2025

  8. [2023]

    House Passes Financial Innovation and Tech- nology for the 21st Century Act with Overwhelming Bipar- tisan Support,

    72 (2024) “House Passes Financial Innovation and Tech- nology for the 21st Century Act with Overwhelming Bipar- tisan Support,”https://financialservices.house.gov/news/ documentsingle.aspx?DocumentID=409277, May 22,

    work page 2024

Show all 10 references
  1. [2024]

    Shall We Vote on Values, But Bet on Beliefs?

    Hanson, Robin (2013) “Shall We Vote on Values, But Bet on Beliefs?”Jour- nal of Political Philosophy, 21 (2), 151–178. Hardt, Steve and Lia C R Lopes (2015) “Google Votes: A Liquid Democ- racy Experiment on a Corporate Social Network,”Technical Disclosure Commons: Defensive Pu...

    2013

  2. [2025]

    Jury Theorems,

    de Condorcet, Marquis (1785)Essai sur l’application de l’analyse ` a la proba- bilit´ e des d´ ecisions rendues ` a la pluralit´ e des voix: De l’imprimerie royale (Paris),https://gallica.bnf.fr/ark:/12148/bpt6k417181/f4.item. Dietrich, Franz and Kai Spiekermann (2023) “Jury T...

    2023 doi arXiv

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.