{"id":"08807693-a94d-42fc-95f4-bb5b1642df47","arxiv_id":"2505.24171","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The corrected proofs of the original characterizations are plausible, but the alternative characterization in Theorem 3 has a faulty step, so the paper needs revision.","lead":"This note claims to correct two flawed uniqueness proofs in the characterization of the Diversity Owen value and adds a new characterization with a null-player axiom. The corrected proofs of the two main theorems appear sound, but the new characterization (Theorem 3) relies on a false symmetry statement and is not established.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 3's uniqueness proof is invalid: Case 2.2 applies Fairness within Component to players that are not symmetric in the unanimity game; the alternative characterization is unproven as written.","rationale":"The paper's primary job is to repair the two published uniqueness proofs and to add an alternative characterization. The repairs in §2.1 and §3.1 look internally sound; I specifically checked the potentially dangerous induction step in Theorem 1, where the game (wd)+l is used after adding an outside null player l. Since l belongs to no coalition with a nonzero dividend, (wd)+l has exactly the same nonzero Harsanyi dividends as wd, so the IH applies. The weak spot is confined to Theorem 3. The reader's weakest_assumption identifies it exactly, and a concrete two-unanimity-game example confirms that the asserted symmetry in Case 2.2 fails. This is not a disagreement with consensus; it is an internal proof gap. Because the affected claim is one of the paper's three headline results, the appropriate verdict is conditional: the corrected proofs stand, but Theorem 3 needs a repaired proof or removal. No stronger verdict is warranted, since the failure is in a proof step, not in the statement or in the rest of the paper.","tokens_in":11268,"tokens_out":15435,"duration_ms":164729,"concrete_test":"Re-run the Case 2.2 argument on a minimal instance: N={i,j,k,l}, B={{i,j,k},{l}}, d=(1,1), vd = u_{i,l}+u_{j,l}, so S1={i}, S2={j}, k∈Bp\\(S1∪S2), T1={i,l}, T2={j,l}. Compute marginal contributions in w1=u_{i,l}: for S={l}, i's marginal is 1 and k's marginal is 0, so i and k are not symmetric. If the proof's FwC chain is nevertheless asserted to hold, it must be justified from a different symmetry; otherwise Theorem 3 remains without a valid uniqueness proof.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In the corrected proofs of Theorems 1 and 2, the auxiliary null-player construction is sound: adding an outside player l to Bk leaves the Harsanyi dividends of the original coalitions unchanged because l belongs to no T in I(vd), so the induction hypothesis applies. The remaining load-bearing gap is the proof of the new alternative characterization, Theorem 3. In Case 2.2 the authors take i∈S1, j∈S2, and k∈Bp\\(S1∪S2), define w1=Δvd(T1)u_{T1}, and assert that 'players i and k are symmetric in w1'. This is false. Because S1=Bp∩T1, i∈T1 while k∉T1. In the unanimity game u_{T1}, a member of T1 and a non-member have different marginal contributions: for S=T1\\{i}, u_{T1}(S∪i)-u_{T1}(S)=1, whereas u_{T1}(S∪k)-u_{T1}(S)=0. FwC therefore cannot be applied to obtain the displayed equality of f_i-f_k with DOw_i-DOw_k. The analogous assertion for j and k in w2 fails for the same reason. Without this bridge the proof does not show that the deviations f_i-DOw_i are constant on Bp, so the uniqueness conclusion of Theorem 3 does not follow. The theorem may be true, but it is unproven as written.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper revisits the axiomatic characterizations of the Diversity Owen value (DOw) introduced by Béal et al. (2025a). It identifies two flaws in the original uniqueness proofs, provides corrected proofs in Sections 2.1 and 3.1, introduces the axiom \"Null Player for Diverse Games\" (ND), and claims a new alternative characterization (Theorem 3) in which NPOPD is replaced by ND. The corrected proofs are based on adding an outside null player to create slack, applying the relevant axioms, and then using induction on the number of nonzero Harsanyi dividends. The paper also derives a weakening of IBCOPPD in Corollary 1.","tokens_in":11593,"tokens_out":34881,"duration_ms":322341,"significance":"If the corrected proofs of Theorems 1 and 2 are valid, the paper repairs two published uniqueness proofs and is a useful corrigendum-style contribution. The identification of specific errors in the original proofs, together with Counterexamples 1 and 2, is valuable. However, the advertised alternative characterization is not established: the proof of Theorem 3 contains a false symmetry claim in Case 2.2, and the proof of Lemma 1 has an Efficiency step that omits the payoff of the added player in the restricted game. These are load-bearing issues for two of the paper's main claims, so the manuscript needs substantial revision before those claims can be accepted.","major_comments":[{"comment":"The claim that \"players i and k are symmetric in w1\" is false. Because S1 = Bp ∩ T1, the chosen player i ∈ S1 belongs to T1 while k ∈ Bp \\(S1 ∪ S2) does not belong to T1. In the unanimity game u_{T1}, a member and a non-member have different marginal contributions: for S = T1 \\ {i}, u_{T1}(S ∪ i) − u_{T1}(S) = 1, whereas u_{T1}(S ∪ k) − u_{T1}(S) = 0. The same objection applies to the asserted symmetry of j and k in w2. Therefore Fairness within Component cannot be applied to obtain the two displayed equalities, and the conclusion that the deviations f_i − DOw_i are constant on Bp does not follow. Theorem 3 is thus unproven as written.","section":"§2.2, Case 2.2"},{"comment":"The step marked \"E=\" applies Efficiency to the restricted game (N ∪ {l}) \\ {i}, but Efficiency gives Σ_{j∈N\\{i}} f_j(restricted) + f_l(restricted) = (v)+l((N ∪ {l}) \\ {i}); the term f_l(restricted) is omitted in the displayed chain. Since f_l(restricted) = 0 has not been proved, and since NPOPD cannot remove l unless l is out in the restricted game (which is not guaranteed), the equality v(N) = f_i(N, v) + v(N \\ {i}) does not follow as written. This gap is load-bearing because Lemma 1 is the basis for the ND axiom used in the corrected proof of Theorem 1. The lemma may be true, and can likely be repaired by splitting into the cases |Bk| > dk and |Bk| = dk, but the written proof needs correction.","section":"§2.1, proof of Lemma 1"},{"comment":"The proof applies \"Case 2\" to the auxiliary game (v')d before Case 2 has been established as a general lemma; as written this is a forward reference inside the same induction step. This structural issue could be fixed by extracting the Case 2 argument as a separate lemma. It is secondary to the false symmetry claim in Case 2.2, but it should be addressed if Theorem 3 is retained.","section":"§2.2, Theorem 3 proof, Case 3"}],"minor_comments":[{"comment":"The NPOPD line in the display combines two applications of NPOPD (one removing l and one removing i) without explanation; this should be stated explicitly.","section":"§2.1, proof of Lemma 1"},{"comment":"The first equality in the chain for Theorem 1 uses NPOPD on the extended game with the added player l removed; this should be stated explicitly for readability.","section":"§2.1, Eq. (2)"},{"comment":"The sentence \"we immediately obtain fi(N, vd, B, d) = fi(N, vd, B, d), for all i ∈ Bk\" contains an evident typo: the right-hand side should presumably be DOwi(N, vd, B, d).","section":"§2.1, after Eq. (2)"},{"comment":"The sentence \"Then, T1 ∪ T2 ≠ N\" is not used in the subsequent argument; either use it or delete it.","section":"§2.2, Case 2.2"},{"comment":"The abstract states that the paper establishes an alternative characterization; since the proof of Theorem 3 is currently invalid, the abstract and the claims in Section 2.2 must be revised accordingly.","section":"Abstract"}],"recommendation":"major_revision","confidential_remarks":"The paper is within the journal's scope as a corrigendum/note on axiomatic characterizations. The corrected proofs of the original theorems are promising and appear repairable, but the two gaps identified above (Lemma 1 and Theorem 3 Case 2.2) affect advertised claims. If the authors can repair Lemma 1 and either prove Theorem 3 correctly or remove it, the remaining corrected-proof contribution would be publishable."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this paper does something genuinely useful—it identifies flaws in two published uniqueness proofs and supplies corrected versions that look sound—but the new alternative characterization (Theorem 3) has a load-bearing error, so that part is unproven as written.\n\nThe corrections matter. The authors pinpoint genuine gaps in Béal, Diss and Tido Takeng's proofs (e.g., Subcase 1a/1b in Theorem 1, the induction step in Theorem 2) and support them with well-chosen counterexamples. I followed the repaired proofs in Sections 2.1 and 3.1 and they hold up. The auxiliary null-player construction—adding an outside player l—is clever and sound: it leaves the original coalitions' Harsanyi dividends unchanged, so the induction hypothesis applies. Lemma 1 (E+NPOPD implies ND) is correct, and the weakened IBCOPPD- axiom in Corollary 1 is a genuine strengthening that follows from the corrected proof.\n\nThe soft spot is Theorem 3. In Case 2.2 they take i∈S1, k∈Bp\\(S1∪S2), and assert i and k are symmetric in w1 = Δ(vd)(T1) u_T1. They are not: i∈T1, k∉T1, and in a unanimity game members and non-members have different marginal contributions. So FwC cannot be applied to derive f_i−DOw_i = f_k−DOw_k. The same issue appears for j and k. Without this bridge the proof does not show the deviations are constant on Bp, and the uniqueness conclusion of Theorem 3 does not follow. The theorem may be true, but it is unproven as written.\n\nMinor note: the paper's account of the original errors is accurate and the counterexamples are well targeted. The citation pattern is appropriate—it cites the work it corrects and the related Hu (2021) paper.\n\nWho is this for? Researchers in cooperative game theory, particularly those working on diversity constraints or fairness-based axiomatizations of values. The corrected proofs are worth having on record. The flawed Theorem 3 should be fixed or dropped before publication.\n\nRecommendation: send to peer review. The corrected proofs and the new ND axiom justify a referee's time. The referee should flag Theorem 3 and ask the authors to either repair the proof or remove the theorem. If they fix it, the note is a solid contribution; if not, it can still stand as a correction note without the alternative characterization.","headline":"Useful correction paper with one big gap: the new Theorem 3 proof is invalid as written, but the repaired proofs of Béal et al.'s theorems look sound.","tokens_in":12058,"tokens_out":5476,"would_cite":true,"duration_ms":50378,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["91A12"],"pacs":[],"model":"deepseek-v4-flash","headline":"The Diversity Owen value is uniquely characterized by three axiom sets once a null-player axiom repairs two flawed proofs.","keywords":["Diversity Owen value","Axiomatization","Null player for diverse games","Diversity constraints","TU-game","Uniqueness proof","Coalition structure"],"falsifier":"Compute the marginal contributions of $i \\in S_1$ and $k \\in B_p \\setminus (S_1 \\cup S_2)$ in the game $\\Delta_{v_d}(T_1)u_{T_1}$ for a game satisfying the Case 2.2 conditions. If there is a coalition containing $T_1$ but not $i$ whose worth changes differently with $i$ and $k$, the players are not symmetric and the derivation of $f_i - DOw_i = f_k - DOw_k$ collapses. For Theorem 1, one would instead test whether the extended-game construction in Lemma 1 preserves diversity when an outside player is added to a community at its quota.","tokens_in":11088,"feed_emoji":"🎲","tokens_out":7123,"duration_ms":60686,"temperature":0.7,"pith_summary":"The Diversity Owen value allocates the worth of a coalition to players when coalitions must respect diversity quotas across communities. An earlier axiomatic study proposed two characterizations of this value, but the uniqueness proofs contain logical gaps. This note claims to repair both proofs by introducing a new axiom, Null Player for Diverse Games (ND), which says that a null player in a diverse game receives zero. The authors prove ND follows from efficiency and the existing null-player-out axiom, use it to correct the two uniqueness proofs, and derive an additional characterization in which ND replaces the older null-player axiom. If the repairs are correct, the Diversity Owen value is uniquely pinned down by each of the three axiom sets.","feed_headline":"Null-player axiom repairs Diversity Owen value proofs","feed_subtitle":"Corrected uniqueness proofs close gaps in two characterizations and add a leaner axiom set.","key_machinery":"The load-bearing objects are the Diversity Owen value itself, defined as the Owen value of the diversity-restricted game $v_d$, and the Harsanyi dividend decomposition $v_d = \\sum_{T} \\Delta_{v_d}(T)u_T$, which lets the proofs work by induction on the number $|I(v_d)|$ of dividend-bearing diverse coalitions. The Null Player for Diverse Games axiom supplies the base case and the cancellation step: null players in diverse games receive zero, so subtracting a unanimity component removes all payoff difference. Fairness within Component and Fairness through Diversity then equate payoff differences between symmetric players or between communities, which transfers the induction hypothesis from smaller games to the full game.","core_discovery":"On the paper's own terms, the central discovery is that the Diversity Owen value is indeed uniquely characterized by the two axiom systems proposed in the earlier work, and that the missing step in those proofs is a proper treatment of null players in diverse games. The paper introduces the Null Player for Diverse Games axiom ($f_i(N,v,B,d)=0$ when $i$ is null and $(N,v,B,d)$ is diverse), shows that efficiency together with the earlier null-player-out axiom implies it, and then uses it to complete the induction in the uniqueness proofs. Theorem 3 adds an alternative characterization: the value is the unique rule satisfying Efficiency, Fairness within Component, Fairness through Diversity, Independence from Non-Diverse Coalitions, and Null Player for Diverse Games. Corollary 1 weakens the balanced-contributions axiom in the second characterization and retains uniqueness.","pith_inferences":["A natural extension is to ask whether the same null-player axiom repairs the analogous uniqueness proof for the Diversity Shapley value; the paper does not discuss that value.","The corrected induction on the number of dividend-bearing coalitions may transfer to other values defined by restricting games to feasible coalitions, such as values for games with precedence or communication constraints.","Because ND is implied by efficiency together with NPOPD, the alternative characterization shows the value can be pinned down with a weaker null-player requirement; the independence of ND from the remaining axioms is left open."],"forward_implications":["The two characterizations proposed in the earlier paper become valid uniqueness results, so the Diversity Owen value is the single rule satisfying either five-axiom system.","The Null Player for Diverse Games axiom gives a simpler route to the same conclusion, replacing the more complex null-player-out axiom in the first characterization.","Corollary 1 shows the balanced-contributions axiom can be weakened while preserving uniqueness, so the axiomatic base of the value is thinner than previously thought.","If the corrected proofs are accepted, the value's foundation no longer depends on the flawed induction steps identified in the original paper."],"supporting_citations":[{"why":"Supplies the two characterizations whose uniqueness proofs this paper corrects, along with the original definition of the Diversity Owen value and its axioms.","marker":"Béal et al. (2025a)"},{"why":"Introduces TU-games with diversity constraints and the Diversity Owen value as a solution for them.","marker":"Béal et al. (2025b)"},{"why":"Provides the unanimity-game dividend decomposition used in the induction over dividend-bearing coalitions.","marker":"Harsanyi (1959)"}],"fun_headline_variants":["Null-player axiom fixes Diversity Owen value proofs","Corrected proofs for Diversity Owen value","Null-player axiom closes uniqueness gaps","New axiom completes Diversity Owen characterizations","Alternative characterization of Diversity Owen value"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The alternate characterization in Theorem 3 depends on the assertion that in Case 2.2 a player $i$ inside a diverse coalition $T_1$ and a player $k$ outside $T_1$ are symmetric in the unanimity game $\\Delta_{v_d}(T_1)u_{T_1}$; if that symmetry fails, the equality used to chain the players' payoff differences does not follow.","fun_headline_variants_meta":{"raw":{"variants":["Null-player axiom fixes Diversity Owen value proofs","Corrected proofs for Diversity Owen value","Null-player axiom closes uniqueness gaps","New axiom completes Diversity Owen characterizations","Alternative characterization of Diversity Owen value"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00069,"raw_usage":{"total_tokens":3043,"prompt_tokens":779,"completion_tokens":2264,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":395,"completion_tokens_details":{"reasoning_tokens":2204}},"tokens_in":395,"tokens_out":2264,"duration_ms":14973,"temperature":1.0,"reasoning_tokens":2204,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T12:34:44.674455+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the marginal contributions of $i \\in S_1$ and $k \\in B_p \\setminus (S_1 \\cup S_2)$ in the game $\\Delta_{v_d}(T_1)u_{T_1}$ for a game satisfying the Case 2.2 conditions. If there is a coalition containing $T_1$ but not $i$ whose worth changes differently with $i$ and $k$, the players are not symmetric and the derivation of $f_i - DOw_i = f_k - DOw_k$ collapses. For Theorem 1, one would instead test whether the extended-game construction in Lemma 1 preserves diversity when an outside player is added to a community at its quota.","supporting_citations":[{"cited_title":"In: AW Tucker, RD Luce (ed) Contributions to the Theory of Games (Volume lV)","cited_arxiv_id":null,"evidence_quote":"Provides the unanimity-game dividend decomposition used in the induction over dividend-bearing coalitions."}],"review_version":1}