Pith. sign in

REVIEW 3 major objections 4 minor 34 references

The River Method

T0 review · 3 major / 4 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read The paper introduces River, a resolute refinement of Split Cycle that satisfies independence of Pareto-dominated alternatives, a property it claims no other Split Cycle refinement satisfies.

desk verdict Solid River proofs, broken Ranked Pairs counterexample: the central comparative claim doesn't survive the printed profile, but the core IPDA result for River is real. read the letter →

arxiv 2504.14195 v1 pith:6CWTRDC3 submitted 2025-04-19 cs.GT

Michelle Döring , Markus Brill , Jobst Heitzig This is my paper · ORCID
classification cs.GT MSC 91B1291B14
keywords RivermethodCondorcet-consistentvotingrulesSplitCycleRankedPairsBeatPathStableindependenceofPareto-dominatedalternativesmajoritymargins
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

River is a single-winner voting method that starts from the pairwise majority margins of a preference profile and builds a directed tree by scanning edges from strongest to weakest, adding an edge only if it would create neither a cycle nor a second incoming edge to any alternative. The unique source of that tree is the winner. The paper shows that River is a refinement of Split Cycle, so every River winner is immune to majority complaints, and that River satisfies the standard basic axioms including monotonicity and the Condorcet winner criterion. Its main new claim is that River satisfies independence of Pareto-dominated alternatives (IPDA): removing an alternative that every voter ranks below another alternative never changes the winner. The paper further claims that no other Split Cycle refinement—Ranked Pairs, Beat Path, or Stable Voting—satisfies IPDA, and that a quasi-Pareto version of the property also holds under suitable tiebreakers.

What carries the argument

The River diagram $M_{RV}$ is the object that carries the argument. It is built by ordering all majority edges by decreasing margin and adding an edge only when it satisfies both constraints: (Cy) it does not create a cycle, and (Br) it is not a second incoming edge to any alternative. Because of (Br) the final graph is a spanning tree with exactly one source, and that source is the River winner. The same diagram is the certificate for immunity: for any edge into the winner, the tree contains a unique path from the winner back to the source of that edge with strength at least as high. The IPDA proof additionally uses the covering observation that Pareto domination implies $m(y,z) \ge m(x,z)$ for every $z$ other than $x,y$, with strict inequality in uniquely weighted profiles, which is what forces the dominated alternative's outgoing edges to be rejected.

What would settle it

Recompute, from Appendix B's profiles P1–P4, the winners of Split Cycle, Ranked Pairs, Beat Path, and Stable Voting with and without the Pareto-dominated alternative $a$. The paper's comparative claim is settled against it if any of the claimed changes fails to reproduce—for instance if Ranked Pairs on P2 returns $b$ both with and without $a$ rather than $d$ turning into $b$.

Watch

Extended reading notes

Core claim

The central discovery, stated in the paper's own terms, is that River is a resolute refinement of Split Cycle that satisfies IPDA, a property the other three refinements do not have. The branching rule is the mechanism: if $y$ Pareto-dominates $x$, the edge $(y,x)$ has the maximum possible margin and is added first, while any edge out of $x$, say $(x,z)$, is processed only after the stronger edge $(y,z)$ and is therefore rejected either by branching or by the cycle it would close through $y$ and $x$. Thus $x$ appears in the River tree as a leaf attached to $y$ and contributes nothing else, so deleting $x$ leaves every other edge intact and the same source wins. The proofs cover uniquely weighted profiles, general profiles with Pareto-consistent tiebreakers, and the generalization to quasi-Pareto-dominated alternatives in Theorems 4.6, 4.8, and 4.10.

Load-bearing premise

River's own IPDA proof is internal to the tree construction, but the paper's comparative claim that no other Split Cycle refinement satisfies IPDA depends on the four Appendix B profiles having exactly the margin graphs and winner changes stated in Theorem 4.4; a direct check of the Ranked Pairs example suggests one of those changes may not reproduce.

Editorial extensions

If this is right

  • River's winner is always an immune alternative, so River is Condorcet-consistent and never leaves the Smith set.
  • River satisfies ISDA: deleting a Smith-dominated alternative never changes the winner, for all preference profiles.
  • River satisfies IPDA for uniquely weighted profiles, and for general profiles when equipped with any Pareto-consistent tiebreaker; it also satisfies the stronger IQDA with quasi-Pareto-consistent tiebreakers.
  • The River diagram has exactly one fewer edge than alternatives, so the winner's rebuttal certificate is a single tree path to every rival, viewable by hand.
  • Ranked Pairs, Beat Path, Stable Voting, and Split Cycle all fail IPDA, according to the paper's counterexamples, so among Split Cycle refinements River is uniquely positioned against this form of agenda manipulation.

Reading between the lines

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

  • Editorial: If the comparative IPDA claim survives re-checking, River combines resolute output, simple hand computation, and a spoiler-independence guarantee, which could make it attractive for participatory or constrained settings where transparency matters.
  • Editorial: The tree certificate's uniqueness suggests a natural measure of how far a winner is from each rival, namely the strength of the single rebutting path, which the paper does not develop; this could support explanations of outcomes to voters.
  • Editorial: The quasi-Pareto generalization indicates the construction is not tied to full unanimity dominance; any dominance relation that guarantees earlier processing of the dominator's outgoing edges would yield a similar independence property.
  • Editorial: A natural empirical test is to sample random preference profiles, add a Pareto-dominated alternative, and measure how often Ranked Pairs, Beat Path, and Stable Voting actually change winner; the paper establishes possibility, not frequency.
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 River, a resolute Condorcet-consistent voting rule that constructs a directed tree by processing majority-margin edges in decreasing order while avoiding cycles and multiple incoming edges. The authors claim that River is a refinement of Split Cycle, satisfies ISDA, IPDA, and IQDA under appropriate tiebreaker conditions, and is simpler to compute and explain than Ranked Pairs, Beat Path, and Stable Voting. The central comparative claim is that unlike those methods, River satisfies independence of Pareto-dominated alternatives (IPDA), with Theorem 4.4 providing counterexamples for the other rules and Theorems 4.6, 4.8, and 4.10 proving IPDA/IQDA for River.

Significance. If fully established, the contribution would be genuinely useful: River is a resolute rule that is easy to describe, produces an interpretable tree certificate, and appears to avoid the Pareto-dominated-alternative spoiler effect. The proof strategy via covering and quasi-Pareto domination is elegant, and the paper is careful about tiebreaker conditions for general profiles. However, the advertised comparative uniqueness claim is not currently supported because the Ranked Pairs counterexample in Theorem 4.4 is invalid for the printed profile P2, and the paper's abstract claims independence of clones without providing a proof or precise reference. The constructive contribution about River itself is plausible and well motivated, but the paper needs a corrected or replaced counterexample and a complete proof of Theorem 4.8 before the central claims can be accepted.

major comments (3)
  1. [Theorem 4.4 and Appendix B (Profile P2)] The Ranked Pairs counterexample is not supported by the printed profile. Direct calculation from the nine voter groups in Profile P2 gives margins b→a 22, b→c 10, a→c 8, d→a 6, d→b 2, and m(c,d)=0. In the full profile, Ranked Pairs thus adds (b,a), (b,c), (a,c), (d,a), and then (d,b), skipping only the zero-margin edge (c,d), with winner RP(P2)={d}. In P2−a the positive margins are b→c 10 and d→b 2, with c–d tied at 0, so Ranked Pairs adds b→c and d→b and yields RP(P2−a)={d}, not {b}. The proof's claim that Ranked Pairs 'adds (b,c) and (c,d), skipping (d,b)' requires m(c,d)>m(d,b), which the printed profile does not provide. Since Theorem 4.4 is the sole support for the abstract and Table 1 claims that Ranked Pairs violates IPDA, the central comparative claim is not established as written.
  2. [Abstract and Section 1 (no theorem)] The abstract and introduction state that River 'shares with those many desirable properties, including independence of clones,' but the manuscript contains no definition of clone independence, no theorem establishing it for River, and no precise citation to a proof. Section 4 and Table 1 do not treat independence of clones at all. This advertised property is therefore unverified in this manuscript; either add a proof or a specific reference to a proof, or remove the claim from the abstract and introduction.
  3. [Theorem 4.8 and Appendix D] The proof of Theorem 4.8 is not actually supplied: the appendix says it is 'literally the same as the proof of Theorem 4.10, only with all occurrences of the prefix quasi- removed.' This reduction is not literal, because the proof of Theorem 4.10 uses the quasi-Pareto-consistency conditions from Definition D.1—specifically that every edge (x,z) is ranked after both (y,z) and (y,x) when y quasi-Pareto-dominates x—and those conditions are not implied by Pareto-consistency alone. The argument for Theorem 4.8 requires its own proof, or at least an explicit explanation of why the extra quasi-Pareto conditions are not needed in the Pareto-dominated case.
minor comments (4)
  1. [Theorem 4.4 proof text] The sentence describing the Ranked Pairs sequence contains 'adds (d,a)' twice, and the intended sequence is unclear; this should be corrected regardless of the counterexample's validity.
  2. [Section 4.1] There is a spelling error: 'absense' should be 'absence.'
  3. [Section 4.2] There is a spelling error: 'inpedendence' should be 'independence.'
  4. [Observation 4.5] Observation 4.5 is stated with an intuitive explanation rather than a formal proof; since it is used later in Theorem 4.6, a one-line derivation of the covering inequalities would improve the presentation.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: River's axiomatic claims are derived from the method's own definitions, not from fitted parameters or self-referential citations.

full rationale

The paper's central derivation chain is self-contained. River is defined procedurally by sorting majority edges and adding them subject to acyclicity and no-branching, and the claimed properties (resoluteness, Split Cycle refinement, ISDA, IPDA, IQDA) are proved directly from that definition and from the stated tiebreaker conditions. For example, Theorem 4.6 shows that a Pareto-dominated alternative x has no outgoing edge in the River diagram other than the incoming edge (y,x), so deleting x leaves the remaining diagram unchanged; Theorem 4.8 is explicitly reduced to Theorem 4.10. No fitted parameter is renamed as a prediction, and no definition is stated in terms of the result it is meant to establish. The self-citations are not load-bearing: the footnote crediting Heitzig's 2004 mailing list post as the origin of River, and the citation of Heitzig's earlier immunity work, are historical or background references and do not supply any premise of the IPDA or ISDA proofs. The negative results about Split Cycle, Ranked Pairs, Beat Path, and Stable Voting are supported by explicit counterexample profiles, not by an appeal to authority. There are two substantive non-circular concerns worth flagging: (1) the Ranked Pairs counterexample in Theorem 4.4 depends on the printed profile P2 having the margins assumed in the proof, and direct computation from the printed profile may not yield those margins or the claimed winners; this is a correctness/verification issue, not a circularity issue. (2) The abstract claims independence of clones for River, but no proof of clone independence appears in the paper; this is missing support, not circularity. Neither concern turns an input into the output by construction, so the appropriate circularity score is 0.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

The central proofs are axiomatic and self-contained; no free parameters are fitted. The paper introduces two ad hoc tiebreaker notions (Pareto-consistent and quasi-Pareto-consistent) to extend IPDA and IQDA to non-uniquely-weighted profiles.

assumptions (4)
  • standard math Standard framework of preference profiles, majority margins, and tournament solutions (Section 2).
    Background definitions from social choice theory, not disputed.
  • domain assumption Split Cycle and immunity are defined as in Holliday and Pacuit [17] and used as the refinement base.
    The paper builds on the Split Cycle method; its properties are assumed from prior literature.
  • ad hoc to paper For general profiles, River satisfies IPDA only when equipped with a Pareto-consistent tiebreaker (Definition 4.7).
    This tiebreaker condition is introduced in this paper to make the proof work; it restricts the class of tiebreakers.
  • ad hoc to paper For IQDA, River requires a quasi-Pareto-consistent tiebreaker (Definition D.1).
    A further tailored tiebreaker condition introduced in the appendix.

how reviews work

0 comments
Cite this review

Pith. "Pith review of The River Method." pith.science (2026). https://pith.science/paper/6CWTRDC3

@misc{pith2026250414195,
  author       = {Pith},
  title        = {Pith review of: The River Method},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6CWTRDC3}},
  note         = {Machine review of arXiv:2504.14195}
}
read the original abstract

We introduce River, a novel Condorcet-consistent voting method that is based on pairwise majority margins and can be seen as a simplified variation of Tideman's Ranked Pairs method. River is simple to explain, simple to compute even 'by hand', and gives rise to an easy-to-interpret certificate in the form of a directed tree. Like Ranked Pairs and Schulze's Beat Path method, River is a refinement of the Split Cycle method and shares with those many desirable properties, including independence of clones. Unlike the other three methods, River satisfies a strong form of resistance to agenda-manipulation that is known as independence of Pareto-dominated alternatives.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

34 extracted references · 32 canonical work pages

  1. [1]

    Arrow, K.J., Sen, A., Suzumura, K.: Handbook of Social Choice and Welfare vol. 2. Elsevier, Amsterdam (2010)

    work page 2010

  2. [2]

    (eds.): Handbook of Computa- tional Social Choice

    Brandt, F., Conitzer, V., Endriss, U., Lang, J., Procaccia, A.D. (eds.): Handbook of Computa- tional Social Choice. Cambridge University Press, Cambridge (2016)

    work page 2016

  3. [3]

    In: Machine Learning: ECML 2003, pp

    F¨ urnkranz, J., H¨ ullermeier, E.: Pairwise preference learning and ranking. In: Machine Learning: ECML 2003, pp. 145–156 (2003). Springer

    work page 2003

  4. [4]

    arXiv preprint arXiv:2112.00861 (2021)

    Askell, A., Bai, Y., Chen, A., Drain, D., Ganguli, D., Henighan, T., Jones, A., Joseph, N., Mann, B., DasSarma, N., et al.: A general language assistant as a laboratory for alignment. arXiv preprint arXiv:2112.00861 (2021)

    arXiv 2021

  5. [5]

    Advances in Neural Information Processing Systems 36 (2024)

    K¨ opf, A., Kilcher, Y., R¨ utte, D., Anagnostidis, S., Tam, Z.R., Stevens, K., Barhoum, A., Nguyen, D., Stanley, O., Nagyfi, R., et al.: Openassistant conversations-democratizing large language model alignment. Advances in Neural Information Processing Systems 36 (2024)

    work page 2024

  6. [6]

    arXiv preprint arXiv:2310.16048 (2023)

    Mishra, A.: AI alignment and social choice: Fundamental limitations and policy implications. arXiv preprint arXiv:2310.16048 (2023)

    arXiv 2023

  7. [7]

    Handbook of computational social choice 2 (2016) 10

    Zwicker, W.S.: Introduction to the theory of voting. Handbook of computational social choice 2 (2016) 10

    work page 2016

  8. [8]

    Econometrica 20(4), 680–684 (1952)

    May, K.: A set of independent, necessary and sufficient conditions for simple majority decisions. Econometrica 20(4), 680–684 (1952)

    work page 1952

Show all 34 references
  1. [9]

    In: Brandt, F., Conitzer, V., Endriss, U., Lang, J., Procaccia, A.D

    Brandt, F., Brill, M., Harrenstein, P.: Tournament solutions. In: Brandt, F., Conitzer, V., Endriss, U., Lang, J., Procaccia, A.D. (eds.) Handbook of Computational Social Choice. Cambridge University Press, Cambridge (2016). Chap. 3

    2016

  2. [10]

    Laslier, J.-F.: Tournament Solutions and Majority Voting, (1997)

    1997

  3. [11]

    Columbia University Press, New York (1986)

    Schwartz, T.: The Logic of Collective Choice. Columbia University Press, New York (1986)

    1986

  4. [12]

    Public Choice 10(1), 97–101 (1971)

    Good, I.J.: A note on Condorcet sets. Public Choice 10(1), 97–101 (1971)

    1971

  5. [13]

    Econometrica 41(6), 1027–1041 (1973)

    Smith, J.H.: Aggregation of preferences with variable electorate. Econometrica 41(6), 1027–1041 (1973)

    1973

  6. [14]

    In: Brandt, F., Conitzer, V., Endriss, U., Lang, J., Procaccia, A.D

    Fischer, F., Hudry, O., Niedermeier, R.: Weighted tournament solutions. In: Brandt, F., Conitzer, V., Endriss, U., Lang, J., Procaccia, A.D. (eds.) Handbook of Computational Social Choice. Cambridge University Press, Cambridge (2016). Chap. 4

    2016

  7. [15]

    arXiv preprint math/0201285 (2002)

    Heitzig, J.: Social choice under incomplete, cyclic preferences. arXiv preprint math/0201285 (2002)

    2002 arXiv

  8. [16]

    Artificial intelligence 77(2), 321–357 (1995)

    Dung, P.M.: On the acceptability of arguments and its fundamental role in nonmonotonic reasoning, logic programming and n-person games. Artificial intelligence 77(2), 321–357 (1995)

    1995

  9. [17]

    Public Choice 197, 1–62 (2023)

    Holliday, W.H., Pacuit, E.: Split cycle: a new Condorcet-consistent voting method independent of clones and immune to spoilers. Public Choice 197, 1–62 (2023)

    2023

  10. [18]

    Social Choice and Welfare 4, 185–206 (1987)

    Tideman, T.N.: Independence of clones as a criterion for voting rules. Social Choice and Welfare 4, 185–206 (1987)

    1987

  11. [19]

    Social choice and Welfare 36, 267–303 (2011)

    Schulze, M.: A new monotonic, clone-independent, reversal symmetric, and Condorcet-consistent single-winner election method. Social choice and Welfare 36, 267–303 (2011)

    2011

  12. [20]

    Constitutional Political Economy, 421–433 (2023)

    Holliday, W.H., Pacuit, E.: Stable voting. Constitutional Political Economy, 421–433 (2023)

    2023

  13. [21]

    Princeton University Press, Princeton (1973)

    Fishburn, P.C.: The Theory of Social Choice. Princeton University Press, Princeton (1973)

    1973

  14. [22]

    Journal of Economic Theory 19(2), 548–550 (1978)

    Richelson, J.: A characterization result for the plurality rule. Journal of Economic Theory 19(2), 548–550 (1978)

    1978

  15. [23]

    Journal of Economic Theory71(1), 298–302 (1996)

    Ching, S.: A simple characterization of plurality rule. Journal of Economic Theory71(1), 298–302 (1996)

    1996

  16. [24]

    Annals of Operations Research 80, 299–332 (1998)

    Chebotarev, P.Y., Shamis, E.: Characterizations of scoring methods for preference aggregation. Annals of Operations Research 80, 299–332 (1998)

    1998

  17. [25]

    Social Choice and Welfare 53, 497–517 (2019)

    Gonzalez, S., Laruelle, A., Solal, P.: Dilemma with approval and disapproval votes. Social Choice and Welfare 53, 497–517 (2019)

    2019

  18. [26]

    Inter- national Journal of Game Theory 49, 801–831 (2020)

    ¨Ozt¨ urk, Z.E.: Consistency of scoring rules: A reinvestigation of composition-consistency. Inter- national Journal of Game Theory 49, 801–831 (2020)

    2020

  19. [27]

    Journal of Economic Theory 205, 105532 (2022) 11

    Brandl, F., Peters, D.: Approval voting under dichotomous preferences: A catalogue of charac- terizations. Journal of Economic Theory 205, 105532 (2022) 11

    2022

  20. [28]

    Journal of Moral Philosophy (2023)

    Greaves, H., Cotton-Barratt, O.: A bargaining-theoretic approach to moral uncertainty. Journal of Moral Philosophy (2023)

    2023

  21. [29]

    http://lists.electorama.com/pipermail/ election-methods-electorama.com/2004-October/014018.html

    Heitzig, J.: Condorcet Trees. http://lists.electorama.com/pipermail/ election-methods-electorama.com/2004-October/014018.html. Posted on the Election Methods mailing list, October 21, 2004 (2004)

    2004

  22. [30]

    Social Choice and Welfare 6(2), 167–173 (1989)

    Zavist, T.M., Tideman, T.N.: Complete independence of clones in the ranked pairs rule. Social Choice and Welfare 6(2), 167–173 (1989)

    1989

  23. [31]

    In: Proceedings of the 26th AAAI Conference on Artificial Intelligence (AAAI-12), pp

    Brill, M., Fischer, F.: The price of neutrality for the ranked pairs method. In: Proceedings of the 26th AAAI Conference on Artificial Intelligence (AAAI-12), pp. 1299–1305. AAAI Press, Palo Alto (2012)

    2012

  24. [32]

    Math´ ematiques et sciences humaines97, 5–17 (1987) 12 Appendix The supplementary material consists of three parts

    Debord, B.: Caract´ erisation des matrices des pr´ ef´ erences nettes et m´ ethodes d’agr´ egation associ´ ees. Math´ ematiques et sciences humaines97, 5–17 (1987) 12 Appendix The supplementary material consists of three parts. In Appendix A we present a real life preference p...

    1987

  25. [33]

    anonymity if for any permutation π : N→N of voters and all A, P, P′ for which≻′ i=≻π(i) holds for all i∈N, we have F (P,A ) =F (P′,A ); 14

  26. [34]

    neutrality if for any bijectionπ : A→B of alternatives and allA, P, P′ for which (xPiy)↔ (π(x)P′ iπ(y)) holds for all i∈N and x,y∈A, we have π(F (P,A )) =F (P′,π (A)). For River, both anonymity and neutrality are dependent on the tiebreaker in the same way as for Ranked Pairs:...

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.