Pith. sign in

REVIEW 4 major objections 2 minor 50 references

Snowveil claims a gossip protocol can discover a social-choice winner almost surely in expected linear time, with no central coordinator.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-03 14:59 UTC pith:XPZ64USJ

load-bearing objection An inventive framework with a real proof gap: the convergence theorem omits all-locked non-quorum absorbing states, so the central claim is unsupported as written. the 4 major comments →

arxiv 2512.18444 v2 pith:XPZ64USJ submitted 2025-12-20 cs.GT cs.AIcs.DCcs.MA

Snowveil: A Framework for Decentralised Preference Discovery

Grammateia Kotsialou This is my paper
classification cs.GT cs.AIcs.DCcs.MA MSC 91B1291B1460J20
keywords decentralised preference discoverygossip protocolsocial choiceconsensuspositive responsivenessBorda ruleMarkov chainscalability
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

Snowveil formalises Decentralised Preference Discovery: determining the winner of an aggregation rule when no one can see all ballots and there is no central tally. The paper proves that if voters repeatedly sample a few peers' rankings and update local beliefs through any aggregation rule that is deterministic, computable, and positively responsive, the electorate almost surely locks onto the canonical winner in finite time, with expected O(n) rounds. The paper also designs the Constrained Hybrid Borda rule, which blends plurality and Borda support and satisfies these axioms, and shows the single-winner process can be iterated to produce a full ranking. If true, this would give decentralised organisations a principled, scalable alternative to trusted voting platforms.

Core claim

The central claim is that a subjective social choice — the winner of an aggregation rule applied to a private preference profile — can be discovered by a purely local, gossip-style process. Theorem 5.3 states that the Snowveil process, instantiated with any rule satisfying determinism, uniqueness, computability, and positive responsiveness, is a time-homogeneous Markov chain that almost surely reaches an absorbing state where one candidate has quorum Q>1/2. Theorem 7.2 adds that this happens in expected O(n) steps, because each voter's probability of locking on the correct winner is bounded below by a constant independent of n. The proof uses the potential function Phi(S_t)=sum N_j^2 as a st

What carries the argument

The load-bearing object is the potential function Phi(S_t) = sum over candidates of (number of locked voters)^2, whose expected value strictly increases with every lock event and is bounded above by n^2. The convergence proof models the system as a finite, time-homogeneous Markov chain, uses Lemma 5.1 to get a positive lock probability in every non-terminal state, and Lemma 5.2 to show plurality support is amplified; together these make the potential a strict submartingale, so the Martingale Convergence Theorem forces absorption. The other central component is the CHB aggregation rule, a tunable hybrid of normalised Borda score and normalised plurality support with popularity filter alpha an

Load-bearing premise

Algorithm 2's stopping condition only checks whether some candidate has reached the quorum; the convergence proof assumes a non-empty set of unlocked voters at every non-terminal state, but the paper gives no argument that states with all voters locked and no quorum have probability zero, so such a deadlock would be an unaddressed absorbing sink.

What would settle it

Run an exhaustive search over small electorates (e.g., n=4, m=3, Q=0.6) for a reachable configuration in which every voter is locked on some candidate yet no candidate has ceil(Q n) voters, and show the transition into that configuration has positive probability; then Theorem 5.3 is false. Alternatively, simulate the protocol from such a starting state and observe that it never terminates.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • Any aggregation rule satisfying the three core axioms inherits Snowveil's finite-time almost-sure convergence, making positive responsiveness a sufficient condition for liveness of DPD protocols.
  • Expected convergence time is O(n) regardless of electorate size, so the protocol scales to large networks; sample size k and local robustness rounds gamma can be chosen independently of n.
  • Iterating the single-winner process yields a complete social ranking, with voters' original rankings restored between rounds so each stage is independent.
  • Manipulating the outcome requires a coalition of size Omega(n) even under an optimal burying strategy against CHB's Borda component.
  • The tunable parameters alpha, beta, lambda encode a community's governance preferences with no measured performance penalty across the lambda spectrum.

Where Pith is reading between the lines

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

  • The paper's stopping rule checks only whether some candidate has reached quorum; states where every voter has locked but no candidate holds a quorum are not shown to have zero probability, so a deadlock would be an unaddressed absorbing sink for the claimed Markov chain.
  • The same potential-function machinery should extend to other positively responsive rules beyond scoring rules, since the proof only needs monotonicity plus the ability to flip a narrow loss into a win; one could test this by instantiating the protocol with a majoritarian compromise rule.
  • Because convergence time peaks exactly at decision boundaries between two viable candidates, the protocol's worst case is not the size of the electorate but the closeness of the race, suggesting adaptive parameters (e.g., raising gamma near contention) as a natural extension the paper does not explore.
  • The empirical claim that convergence is faster under polarisation than under Impartial Culture suggests the feedback loop exploits existing signal; a direct comparison against a single-shot sampled-Borda baseline would clarify how much of the speedup is due to iteration versus the CHB rule itself.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

4 major / 2 minor

Summary. The paper introduces Snowveil, a gossip-based framework for Decentralised Preference Discovery (DPD), in which voters sample peers' rankings and update local beliefs via an aggregation rule, eventually locking onto a candidate. The main theoretical claim is Theorem 5.3: for any rule satisfying the proposed axioms (determinism/uniqueness, positive responsiveness, computability), the process is a time-homogeneous Markov chain that almost surely converges in finite time to a quorum winner. The paper also introduces the CHB rule, proves it satisfies the axioms, gives an O(n) expected convergence bound (Theorem 7.2), a coalitional manipulation bound, and simulations supporting linear scalability.

Significance. If the convergence theorem were correct, Snowveil would be a noteworthy bridge between scalable gossip-based consensus and axiomatic social choice, with a modular liveness condition based on positive responsiveness. The paper is clearly written and the simulation study is useful, providing empirical evidence for the O(n) claim and showing the effect of design parameters. However, the central proof has several load-bearing gaps: the treatment of all-locked non-quorum states is missing, the strict amplification lemma is not implied by the stated axioms, and the concentration argument in Lemma 7.1 assumes sampling from the original profile despite locked voters reporting modified ballots. These issues undermine the almost-sure convergence and scalability claims as currently stated.

major comments (4)
  1. [§5.1, Lemma B.4 / Theorem 5.3] The proof of almost-sure convergence only treats 'fragmented states' with U(t) non-empty. The state space includes states with U(t)=∅ and no candidate at quorum; from such a state Algorithm 2 (line 4) cannot activate any voter, so the state is absorbing without being a quorum. Lemmas 5.1 and B.3 apply only when an unlocked voter exists, and no argument shows that all-locked non-quorum states have probability zero or are transient. The strict submartingale property fails at these states, so the Martingale Convergence Theorem does not force a quorum. The proof must either show such states are unreachable with probability 1 or modify the protocol to escape them.
  2. [§5.2, Lemma 5.2] The claim that N_a(t)>N_b(t) implies P(lock on p_a)>P(lock on p_b) is not a consequence of Positive Responsiveness as defined in Def. 4.3. The 'Responsiveness' part is existential ('there exists a profile'), not a guarantee that stochastic dominance in sample counts yields a strict advantage in selection probability for every state. Thresholds and deterministic tie-breaking in CHB can absorb a stochastic dominance. Without a strict, universal responsiveness property, the positive-feedback step and the strict submartingale argument for Theorem 5.3 are unsupported; the proof needs an additional axiom or a direct argument for the specific rule.
  3. [§7.1, Lemma 7.1 / Appendix D] The concentration proof models samples as i.i.d. draws from the original global profile Π and defines margins δ_j, δ_α from that profile. However, §3.6 states that locked voters submit modified ballots (their locked candidate first), so the effective sampling distribution is state-dependent and changes over time. The proof never accounts for this feedback, and the lower bound on P(p_local=p*) may not transfer to the actual process. Moreover, the 'standard approximation' of sampling without replacement by i.i.d. sampling is not a rigorous substitute; k≪n alone does not yield the claimed n-independent constant c2 without a finite-population concentration bound. These gaps affect the O(n) claim in Theorem 7.2.
  4. [§3.3 / Algorithm 2 / Lemma B.3] The protocol's transition rule is underspecified: Algorithm 2 says 'Activate randomly selected unlocked voters' (plural), while Lemma B.3 computes the expected potential change for a single randomly chosen unlocked voter. If multiple voters are activated per time step, the Markov-chain transition kernel and the submartingale computation must be defined accordingly. The proof also does not establish that all non-quorum states are transient or recurrent; it only shows a subset of states (fragmented states with U non-empty) is transient, leaving the rest of the state space unaddressed.
minor comments (2)
  1. [Throughout] There are several typos and formatting issues: 'interation' in §3.3, 'Uniqeness' and 'deffered' in §4, 'equilibirum' in Appendix A, and '2 1 projects' in Lemma 5.2. The pseudocode in Algorithm 2 has a missing space in '≥⌈Q·n⌉' and the loop 'until' line is ambiguous. Figure 2 is a text-only diagram; consider replacing with a proper figure.
  2. [Appendix D, D.1] The notation switches between c1 and c2 in the proof: the text introduces c1 as the single-round success probability, then says 'formalise the resulting n-independent constant and connect it to the overall lock probability c2'. Please clarify the relationship between the per-round bound and the final lock-probability bound.

Circularity Check

0 steps flagged

No significant circularity: the convergence theorem is conditional on stated axioms, CHB's properties are proved rather than assumed, and the self-citations are peripheral.

full rationale

I walked the main derivation chain in Sections 3-7 and Appendices A-D. Theorem 5.3 is explicitly conditional: 'when instantiated with any rule satisfying the core axioms.' The axioms (Definitions 4.1-4.3) do not already assert convergence to a quorum; determinism/unique winner, monotonicity, and responsiveness are separate properties. Lemma 5.1 constructs a positive-probability LOCK path using determinism and uniqueness; Lemma 5.2 is asserted to follow from Positive Responsiveness, and the submartingale argument then derives absorption. Whether Lemma 5.2 is fully proved is a proof-completeness question, not a circular reduction: the theorem's conclusion is not identical to the axiom or to a fitted parameter. The CHB rule is not assumed to satisfy the axioms; Propositions 4.4, 4.5 and 4.8 are proved in Appendix A from the rule's definition. The O(n) claim in Theorem 7.2 follows from the n-independent lock-probability bound in Lemma 7.1, and the simulations in Section 8 are a separate empirical check rather than a calibration of the claim. The only self-citations ([15], [25]) appear in the related-work discussion of liquid democracy and are not load-bearing for the convergence theorem. I also note a non-circular correctness caveat: Lemma B.4/Definition B.5 defines fragmented states only when the set of unlocked voters U(t) is non-empty, and the proof does not explicitly address all-locked states with no quorum; that is an omitted-case proof gap, not a circularity, so it does not raise the circularity score.

Axiom & Free-Parameter Ledger

6 free parameters · 5 axioms · 0 invented entities

The central convergence theorem rests on the existence of continuing unlocked voters, honest behavior, static preferences in the sampling distribution, and an unproven amplification step. CHB's tunable parameters and the protocol thresholds are free design choices that affect the O(n) constants and the validity of the concentration proof's subcase.

free parameters (6)
  • alpha (popularity filter) = baseline 0.1; not fitted
    Defines the CHB rule and the alpha-popularity margin used in Lemma 7.1; the proof only covers winners that are alpha-popular.
  • beta (consensus filter) = baseline 0.8; not fitted
    Defines the eligibility set in CHB; affects whether the O(n) concentration bound's assumptions hold.
  • lambda (hybrid weight) = baseline 0.5; not fitted
    Trades Borda vs plurality in CHB; the fine-grained responsiveness proof requires lambda < 1 and does not handle the pure-plurality boundary.
  • sample size k = baseline 10; not fitted
    The concentration bound is exponential in k and independent of n only for sufficiently large k; no explicit sufficient bound is given.
  • gamma / tau_max / tau_min = baseline 10 / 6 / 3; not fitted
    Local decision thresholds; Lemma 5.1 uses tau_max to construct a positive-probability lock path, but no parameter values are required by the main theorem.
  • quorum Q = baseline 0.67; not fitted
    Defines the absorbing state; with m>=3 candidates, all-locked non-quorum states are possible and not ruled out.
axioms (5)
  • domain assumption Uniform random sampling of k voters; sampling with replacement is a valid approximation because k << n.
    Used in Lemma 7.1 and Appendix D to apply Chernoff-Hoeffding; the without-replacement correction is not analyzed.
  • domain assumption Convergence proof assumes voters follow the protocol honestly; Byzantine or strategic voters are deferred to Section 6.
    Theorem 5.3's Markov chain has no adversarial transitions; strategic resilience is analyzed separately and only for Borda burying.
  • ad hoc to paper Locked voters report a modified ballot (p* first, placeholders) while the global profile Pi remains static.
    Section 3.6 changes sampled ballots, but Appendix D's concentration argument treats per-voter score differences as drawn from the original static Pi.
  • ad hoc to paper Positive Responsiveness implies the strict stochastic dominance step in Lemma 5.2: more locked support for pa than pb implies P(F(sample)=pa) > P(F(sample)=pb).
    This implication is asserted in the proof of Lemma 5.2 but requires structural properties beyond the stated monotonicity+responsiveness axioms; it is not derived from CHB's thresholds and tie-breaking.
  • ad hoc to paper Non-terminal states always have at least one unlocked voter; all-locked non-quorum states are transient or unreachable.
    Lemma B.4 defines fragmented states only when U(t) is non-empty, but Algorithm 2's loop can terminate with no unlocked voters and no quorum; these states are absorbing sinks in the Markov chain.

pith-pipeline@v1.3.0-alltime-deepseek · 33528 in / 14222 out tokens · 154487 ms · 2026-08-03T14:59:04.726350+00:00 · methodology

0 comments
read the original abstract

Aggregating subjective preferences in social choice traditionally assumes a trusted central authority. In contrast, this paper formalises Decentralised Preference Discovery (DPD): the reliable identification of a social choice parameter (e.g. the canonical outcome of an aggregation rule applied to the global preference profile) under conditions of partial information, asynchronous interaction, censorship resistance, and no central coordinator. To address DPD, we propose Snowveil, a gossip-based framework where agents repeatedly sample random peer rankings and update local beliefs to converge on the canonical outcome. Using a potential function, submartingale theory, and concentration bounds, we prove the system reaches this stable state with tunable high probability, in finite expected time. This single-winner process can then be iterated to construct a set of winning candidates for multi-winner scenarios. Snowveil is agnostic to specific aggregation rules, requiring only that the rule satisfies axioms such as Positive Responsiveness, thus offering a formal basis for a wider class of DPD protocols. Demonstrating Snowveil's modularity, we introduce the Constrained Hybrid Borda (CHB), an aggregation rule designed to balance broad consensus with plurality support. We provide an axiomatic analysis of CHB and present empirical results via extensive simulation, validating Snowveil's O(n) scalability. Overall, this work provides a foundation for how a stable consensus emerges from subjective, expressive, and diverse preference profiles in large-scale decentralised systems.

Figures

Figures reproduced from arXiv: 2512.18444 by Grammateia Kotsialou.

Figure 1
Figure 1. Figure 1: Snowveil as a synthesis of Distributed Systems and Com [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Linking CHB to Snowveil’s performance. 5 [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Sensitivity for key parameters: (a) Sample size [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Scalability vs. number of voters 𝑛 for Impartial Culture and Polarised electorates (95% CI). (a) (b) [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Impact of policy parameters 𝛼 and 𝛽: (𝑎) Winner outcome showing the Plurality winner’s policy window. (𝑏) Convergence time, peaking at the decision boundary. 𝑡1 = 35 and Borda scores 𝐵(𝐶0) = 115, 𝐵(𝐶1) = 136 [PITH_FULL_IMAGE:figures/full_fig_p008_5.png] view at source ↗
Figure 7
Figure 7. Figure 7: Scalability of Snowveil with respect to the number of can [PITH_FULL_IMAGE:figures/full_fig_p022_7.png] view at source ↗

discussion (0)

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

Reference graph

Works this paper leans on

50 extracted references · 8 canonical work pages · 1 internal anchor

  1. [1]

    Ignacio Amores-Sesar, Christian Cachin, and Philipp Schneider. 2024. An Anal- ysis of Avalanche Consensus. InProceedings of SIROCCO 2024, Lecture Notes in Computer Science, Vol. 14662. Springer, 27–44

    2024

  2. [2]

    Felix Brandt. 2017. Rolling the Dice: Recent Results in Probabilistic Social Choice. InTrends in Computational Social Choice, Ulle Endriss (Ed.). AI Access, 3–26. https://archive.illc.uva.nl/COST-IC1205/BookDocs/TrendsCOMSOC.pdf

    2017

  3. [3]

    Procaccia (Eds.)

    Felix Brandt, Vincent Conitzer, Ulle Endriss, Jérôme Lang, and Ariel D. Procaccia (Eds.). 2016.Handbook of Computational Social Choice. Cambridge University Press. https://doi.org/10.1017/CBO9781107446984

    doi:10.1017/cbo9781107446984 2016

  4. [4]

    Markus Brill, Théo Delemazure, Anne-Marie George, Martin Lackner, and Ul- rike Schmidt-Kraepelin. 2022. Liquid Democracy with Ranked Delegations. In Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 36. 4884–4891. https://doi.org/10.1609/aaai.v36i5.20417

    work page doi:10.1609/aaai.v36i5.20417 2022

  5. [5]

    Krimpas, and Alexandros A

    Ioannis Caragiannis, George A. Krimpas, and Alexandros A. Voudouris. 2015. Aggregating partial rankings with applications to peer grading in massive online open courses. InProceedings of the 2015 International Conference on Autonomous Agents and Multiagent Systems (AAMAS). 675–683

    2015

  6. [6]

    Krimpas, and Alexandros A

    Ioannis Caragiannis, George A. Krimpas, and Alexandros A. Voudouris. 2016. How Effective Can Simple Ordinal Peer Grading Be?(EC ’16). Association for Computing Machinery, New York, NY, USA, 323–340. https://doi.org/10.1145/ 2940716.2940748

    arXiv 2016

  7. [7]

    Ioannis Caragiannis and Evi Micha. 2017. Learning a Ground Truth Ranking Using Noisy Approval Votes. InProceedings of the Twenty-Sixth International Joint Conference on Artificial Intelligence (IJCAI-17). International Joint Conferences on Artificial Intelligence Organization, 149–155. https://doi.org/10.24963/ijcai. 2017/22

    doi:10.24963/ijcai 2017

  8. [8]

    Ioannis Caragiannis and Evi Micha. 2019. A Contribution to the Critique of Liquid Democracy. InProceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence (IJCAI-19). International Joint Conferences on Artificial Intelligence Organization, 116–122. https://doi.org/10.24963/ijcai.2019/17

    work page doi:10.24963/ijcai.2019/17 2019

  9. [9]

    Miguel Castro and Barbara Liskov. 1999. Practical Byzantine Fault Tolerance. In Proceedings of the 3rd Symposium on Operating Systems Design and Implementation (OSDI). USENIX Association, 173–186. http://pmg.csail.mit.edu/papers/osdi99. pdf

    1999

  10. [10]

    Deng (Eds.)

    David Lee Kuo Chuen and Robert H. Deng (Eds.). 2023.Hand- book of Blockchain, Digital Finance, and Inclusion. Academic Press. https://shop.elsevier.com/books/handbook-of-blockchain-digital-finance- and-inclusion-volume-3/lee-kuo-chuen/978-0-443-34717-7 Volume 3: Web3, AI, Privacy and Greentech

    2023

  11. [11]

    Rachael Colley and Umberto Grandi. 2022. The Spread of Opinions via Boolean Networks. InMulti-Agent Systems: 20th European Conference, EUMAS 2022 (Lecture Notes in Computer Science, Vol. 13442). Springer, 96–115. https://doi.org/10.1007/ 978-3-031-20614-6_6

    2022

  12. [12]

    Rachael Colley, Umberto Grandi, and Arianna Novaro. 2020. Smart Voting. InProceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence (IJCAI-20). International Joint Conferences on Artificial Intelligence Organization, 1734–1740. https://doi.org/10.24963/ijcai.2020/240

    work page doi:10.24963/ijcai.2020/240 2020

  13. [13]

    Rachael Colley, Umberto Grandi, and Arianna Novaro. 2022. Unravelling multi- agent ranked delegations.Autonomous Agents and Multi-Agent Systems36, 1 (2022), 9. https://doi.org/10.1007/s10458-021-09538-2

    work page doi:10.1007/s10458-021-09538-2 2022

  14. [14]

    Alan Demers, Dan Greene, Carl Hauser, Wes Irish, John Larson, Scott Shenker, Howard Sturgis, Dan Swinehart, and Doug Terry. 1987. Epidemic algorithms for replicated database maintenance. InProceedings of the sixth annual ACM Symposium on Principles of distributed computing. 1–12

    1987

  15. [15]

    Amrita Dhillon, Grammateia Kotsialou, Dilip Ravindran, and Dimitrios Xef- teris. 2023. Information Aggregation with Delegation of Votes.arXiv preprint arXiv:2305.11333(2023). arXiv:2305.11333

    Pith/arXiv arXiv 2023

  16. [16]

    Jean-Paul Doignon, Aleksandar Pekeč, and Michel Regenwetter. 2004. The re- peated insertion model for rankings: Missing link between two subset choice models.Psychometrika69, 1 (2004), 33–54

    2004

  17. [17]

    Edith Elkind, Martin Lackner, and Dominik Peters. 2022. Preference Restrictions in Computational Social Choice: A Survey.arXiv preprint arXiv:2205.09092(2022). arXiv:2205.09092

    Pith/arXiv arXiv 2022

  18. [18]

    Edith Elkind and Nishad Shah. 2014. Maximum Likelihood Voting Rules over Intransitive Domains. InProceedings of the 30th Conference on Uncertainty in Artificial Intelligence (UAI). 272–281

    2014

  19. [19]

    Allan Gibbard. 1973. Manipulation of Voting Schemes: A General Result.Econo- metrica41, 4 (1973), 587–601. https://doi.org/10.2307/1914083

    doi:10.2307/1914083 1973

  20. [20]

    Halpern, Ali Jadbabaie, Elchanan Mossel, Ariel D

    Daniel Halpern, Joseph Y. Halpern, Ali Jadbabaie, Elchanan Mossel, Ariel D. Procaccia, and Manon Revel. 2023. In Defense of Liquid Democracy. InProceedings of the 24th ACM Conference on Economics and Computation (EC). https://doi.org/ 10.1145/3580507.3597817

    arXiv 2023

  21. [21]

    Osborne, and M

    Daniel Horan, Martin J. Osborne, and M. Remzi Sanver. 2019. Positively Re- sponsive Collective Choice Rules.Journal of Economic Theory180 (2019), 1–29. https://doi.org/10.1016/j.jet.2018.12.001

    work page doi:10.1016/j.jet.2018.12.001 2019

  22. [22]

    Procaccia

    Anson Kahng, Simon Mackenzie, and Ariel D. Procaccia. 2021. Liquid Democracy: An Algorithmic Perspective.Journal of Artificial Intelligence Research70 (2021), 1223–1252. https://doi.org/10.1613/jair.1.12261

    work page doi:10.1613/jair.1.12261 2021

  23. [23]

    David Kempe, Jon Kleinberg, and Amit Kumar. 2003. Gossip-based Computation of Aggregate Information. InProceedings of the 44th Annual IEEE Symposium on Foundations of Computer Science (FOCS). 482–491. https://doi.org/10.1109/SFCS. 2003.1238221

    Pith/arXiv arXiv 2003

  24. [24]

    David Kempe, Jon Kleinberg, and Éva Tardos. 2003. Maximizing the spread of influence through a social network. InProceedings of the ninth ACM SIGKDD international conference on Knowledge discovery and data mining. ACM, 137–146. https://doi.org/10.1145/956750.956769

    arXiv 2003

  25. [25]

    Grammateia Kotsialou and Luke Riley. 2020. Incentivising Participation in Liquid Democracy with Breadth-First Delegation. InProceedings of the 19th International Conference on Autonomous Agents and Multiagent Systems (AAMAS). 638–646. https://www.ifaamas.org/Proceedings/aamas2020/pdfs/p638.pdf

    2020

  26. [26]

    2011.Building successful online communities: Evidence-based social design

    Robert E Kraut and Paul Resnick. 2011.Building successful online communities: Evidence-based social design. MIT Press

    2011

  27. [27]

    Lalley and E

    Steven P. Lalley and E. Glen Weyl. 2018. Quadratic Voting: How Mechanism Design Can Radicalize Democracy.AEA Papers and Proceedings108 (May 2018), 33–37. https://doi.org/10.1257/pandp.20181002

    doi:10.1257/pandp.20181002 2018

  28. [28]

    Erwan Le Merrer, Anne-Marie Kermarrec, and Yves Le Traon. 2010. What is the global opinion? a high-level protocol for distributed polling. InPrinciples of Distributed Systems: 14th International Conference (OPODIS). Springer, 115–129

    2010

  29. [29]

    Eng Keong Lua, Jon Crowcroft, Marcelo Pias, Ravi Sharma, and Steven Lim

  30. [30]

    Colin L. Mallows. 1957. Non-null ranking models. I.Biometrika44, 1/2 (1957), 114–130. https://doi.org/10.2307/2333244

    doi:10.2307/2333244 1957

  31. [31]

    Kenneth O. May. 1952. A Set of Independent Necessary and Sufficient Conditions for Simple Majority Decision.Econometrica20, 4 (1952), 680–684. https://doi. org/10.2307/1907651

    doi:10.2307/1907651 1952

  32. [32]

    Thiago Dias Monteiro, Otavio Próspero Sanchez, and Gustavo Hermínio Salati Marcondes de Moraes. 2024. Exploring off-chain voting and blockchain in de- centralized autonomous organizations.RAUSP Management Journal59, 4 (2024), 335–349. https://doi.org/10.1108/RAUSP-08-2023-0162

    work page doi:10.1108/rausp-08-2023-0162 2024

  33. [33]

    1988.Axioms of Cooperative Decision Making

    Hervé Moulin. 1988.Axioms of Cooperative Decision Making. Cambridge Univer- sity Press

    1988

  34. [34]

    Satoshi Nakamoto. 2008. Bitcoin: A Peer-to-Peer Electronic Cash System. https: //bitcoin.org/bitcoin.pdf

    2008

  35. [35]

    Haoyue Ping and Julia Stoyanovich. 2023. Most Expected Winner: An Interpreta- tion of Winners over Uncertain Voter Preferences.Proceedings of the ACM on Management of Data1, 1 (2023), 1–25

    2023

  36. [36]

    Manon Revel, Adam Berinsky, Daniel Halpern, and Ali Jadbabaie. 2022. Liquid Democracy in Practice: An Empirical Analysis of its Epistemic Performance. In Proceedings of the 2nd ACM Conference on Equity and Access in Algorithms, Mech- anisms, and Optimization (EAAMO ’22). Association for Computing Machinery

    2022

  37. [37]

    Manon Revel and Théophile Pénigaud. 2025. AI-Enhanced Deliberative Democ- racy and the Future of the Collective Will.arXiv preprint arXiv:2503.05830(2025). https://doi.org/10.48550/arXiv.2503.05830

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2503.05830 2025

  38. [38]

    Team Rocket. 2018. Snowflake to Avalanche: A Novel Metastable Consensus Protocol Family for Cryptocurrencies. https://ipfs.io/ipfs/ QmUy4jh5mGNZvLkjies1RWM4YuvJh5o2FYopNPVYwrRVGV. Accessed: 2025- 09-18

    2018

  39. [39]

    Team Rocket, Maofan Yin, Kevin Sekniqi, Robbert van Renesse, and Emin Gün Sirer. 2019. Scalable and Probabilistic Leaderless BFT Consensus through Metasta- bility. arXiv:1906.08936

    Pith/arXiv arXiv 2019

  40. [40]

    Mark Allen Satterthwaite. 1975. Strategy-proofness and Arrow’s Conditions: Existence and Correspondence Theorems for Voting Procedures and Social Welfare Functions.Journal of Economic Theory10, 2 (1975), 187–217. https: //doi.org/10.1016/0022-0531(75)90050-2

    doi:10.1016/0022-0531(75)90050-2 1975

  41. [41]

    Markus Utke and Ulrike Schmidt-Kraepelin. 2023. Anonymous and Copy-Robust Delegations for Liquid Democracy. InAdvances in Neural Information Process- ing Systems (NeurIPS). https://papers.neurips.cc/paper_files/paper/2023/file/ dbb5180957513805ebeea787b8c66ac9-Paper-Conference.pdf

    2023

  42. [42]

    Anna Van Elst, Igor Colin, and Stephan Clémençon. 2025. Asynchronous Gossip Algorithms for Rank-Based Statistical Methods.arXiv preprint arXiv:2509.07543 (2025)

    Pith/arXiv arXiv 2025

  43. [43]

    Weihang Wang, Mohammad Jabed Hoque, Zheng Zha, and Dong Wang. 2020. Decentralized autonomous organizations: A comprehensive survey.ACM Com- puting Surveys (CSUR)53, 5 (2020), 1–36

    2020

  44. [44]

    Reiter, Guy Golan Gueta, and Ittai Abra- ham

    Maofan Yin, Dahlia Malkhi, Michael K. Reiter, Guy Golan Gueta, and Ittai Abra- ham. 2019. HotStuff: BFT Consensus with Linearity and Responsiveness. In Proceedings of the 2019 ACM Symposium on Principles of Distributed Computing 9 Grammateia Kotsialou (PODC). 347–356. https://doi.org/10.1145/3293611.3331591

    arXiv 2019

  45. [45]

    tipping-point

    Yuzhe Zhang and Davide Grossi. 2022. Tracking Truth by Weighting Prox- ies in Liquid Democracy. InProceedings of the 21st International Conference on Autonomous Agents and Multiagent Systems (AAMAS ’22). 1482–1490. A PROOFS OF SECTION 4 Proposition.4.4. The Constrained Hybrid Borda (CHB) aggregation rule is computable, deterministic, and yields a unique o...

    2022

  46. [47]

    This distinction is crucial

    Define the Minimal Score Change.For the Constrained Hybrid Borda (CHB) rule, a minimal improvement is defined as a change to a single ballot that increases a candidate’s Borda score by exactly one point, without altering its first-place vote count. This distinction is crucial. An improvement that moves a candidate into first place would necessarily increa...

  47. [48]

    Construct a ‘close race’ profile.We construct an initial pref- erence profile, Π, where an arbitrary candidate 𝑝𝑗 is an eligible candidate but narrowly loses to another eligible candidate,𝑝𝑞. We arrange the𝑘 ballots with precision to ensure the following condi- tions hold in profile Π:(𝑖) both𝑝𝑗 and𝑝𝑞 are in the set of eligible candidates,Celigible (meani...

  48. [49]

    One voter swaps 𝑝𝑗 with the candidate ranked just above it

    Apply the Minimal Improvement.We create a new profile, Π′, by applying the single minimal improvement to𝑝𝑗 . One voter swaps 𝑝𝑗 with the candidate ranked just above it. This has the following effects:(𝑖) the Borda score of 𝑝𝑗 increases by exactly 1, (𝑖𝑖) the Hybrid Score of𝑝𝑗 increases by exactly Δ𝐻min,(𝑖𝑖𝑖) the new score is𝐻′(𝑝𝑗)=𝐻(𝑝 𝑗)+Δ𝐻 min,(𝑖𝑣) the s...

  49. [50]

    good” or “legitimate

    The Winner Flips.We compare the new Hybrid Scores in profile Π′ to show that𝑝𝑗 is now the winner. From our construction,Δ𝐻min is greater than the original winning margin, 𝐻(𝑝𝑞)−𝐻(𝑝 𝑗). By rearranging, we get:𝐻(𝑝 𝑗)+Δ𝐻 min >𝐻(𝑝 𝑞). Since the new score for𝑝𝑗 is𝐻′(𝑝𝑗)=𝐻(𝑝 𝑗)+Δ𝐻 min, we can substitute to get𝐻′(𝑝𝑗)> 𝐻(𝑝𝑞). Since the score for𝑝𝑞 did not increas...

  50. [2005]

    A survey and comparison of peer-to-peer overlay network schemes.IEEE Communications surveys & tutorials7, 2 (2005), 72–93

    2005