{"id":"aee26a1a-c034-4d73-a831-2121d2df76c9","arxiv_id":"2508.07145","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"On Pigou's network, a repeated game among multiple competing route planners can reach optimal traffic flow, and this is impossible with a single planner or too few.","lead":"The paper shows that in a simple traffic network, having several competing self-driving-car planners can achieve optimal traffic flow, while a single central planner cannot. It gives conditions on how many planners and how much traffic each controls to make this work.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Printed 'no collective punishments' inequality is backwards: excess-bottom-flow profiles satisfy it, so Proposition 2 cannot support the F<3/4 impossibility; the converse needs the opposite inequality.","rationale":"The reader's weakest_assumption is exactly the NCP sign error. I verified the formal definition in Section 2.3(4) and its use in Section 5. The printed inequality only forbids deviations that make everyone worse off; Proposition 2 identifies a different object — a current profile in which one planner's reduction of bottom flow would help everyone. The explicit two-planner example shows such a profile satisfies the printed definition, so the inference 'NCP => bottom flow <= F' is invalid. Since the entire F < 3/4 impossibility and Corollaries 2-3 rely on that inference, this is load-bearing. I also flag the optimality definition (Section 2.3(3)): the construction's proof is explicitly conditional on cars stopping defecting, while the definition requires convergence for all histories. This second gap affects the existence direction but is separately fixable by qualifying the histories over which optimality is required. No machine-checked proofs or shipped code offset these formal gaps; the conceptual framing and threshold intuition are plausible. Therefore the reader's CONDITIONAL verdict is appropriate and needs no change.","tokens_in":11499,"tokens_out":21056,"duration_ms":209366,"concrete_test":"Check the two-planner Pigou example with alpha=(1/2,1/2): the one-shot equilibrium has F=2/3, but lambda'=(0.8,0.8) has F'=0.8>F. If planner 1 deviates to 0.7, F becomes 0.75 and normalized costs fall for both planners (0.84 to 0.825 and 0.84 to 0.80). Under the printed NCP, this deviation satisfies c_j(old)>=c_j(new) for j=1,2, so NCP does not rule out the excess flow; Proposition 2 therefore cannot yield the bound used in the converse. Re-derive the F<3/4 proof with the inequality reversed and check whether the 'total bottom flow <= F' step and the subset-averaging contradiction go through.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 2.3(4) defines NCP as: for every planner i and deviation sigma'_i, there exists j with c_j^k(sigma|h) >= c_j^k((sigma_{-i},sigma'_i)|h). This says no deviation makes every planner worse off. But Proposition 2 and the converse of Theorem 2 need the opposite: for every i and sigma'_i, some j must have c_j^k(sigma|h) <= c_j^k((sigma_{-i},sigma'_i)|h) — i.e., the current flow is not Pareto-dominated by a unilateral change. Under the printed inequality, a profile with bottom flow > F is allowed: if some planner reduces bottom flow, everyone's cost falls, so old >= new holds for that planner. Hence 'no collective punishments' does not imply total bottom flow <= F, and the averaging argument for F < 3/4 collapses. A separate gap: optimality in Definition 3 quantifies over all global histories, but the Section 5 construction only shows convergence when defections stop; histories with infinitely many defections keep the system in planner-equilibrium stages, so lim c^k need not equal c_OPT.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies a repeated routing game on Pigou's network in which several planners each control a fraction of autonomous traffic. It defines four desiderata—individual rationality for routed cars, resilience to competition among planners, convergence to socially optimal flow, and absence of collective punishments—and characterizes when they can be met. The main result, Theorem 2, states that if the total bottom-path flow F in the one-shot planner equilibrium exceeds 3/4, a strategy profile satisfying all four desiderata exists, while if F < 3/4 no profile satisfies individual rationality and no collective punishments. Corollaries translate the threshold into conditions on the number and sizes of planners. The paper also sketches an efficient algorithm for the one-shot planner equilibrium and conjectures a generalization to arbitrary networks.","tokens_in":11816,"tokens_out":14279,"duration_ms":143575,"significance":"The conceptual message—that competition among autonomous-vehicle planners can be a necessary ingredient for efficiency—is interesting and potentially relevant to algorithmic game theory and mechanism design. The paper is self-contained and the threshold statement is clean. If the technical difficulties were resolved, the result would be a useful contribution to the growing literature on mediated routing and repeated games. However, the formal definitions contain systematic inequality-direction errors that reverse the meaning of individual rationality, resilience to competition, and no collective punishments; as written, the main theorems do not follow. The positive construction is also only sketched, and optimality is not established for all global histories. With corrected definitions and a full proof, the result may be salvageable, but the current manuscript requires substantial revision.","major_comments":[{"comment":"The inequalities in the formal definitions are reversed relative to the prose. Definition (1) requires c^λ_(i,S)(σ|h) ≥ c^λ_(i,S)(σ,D|h), which is exactly what holds when a defection decreases the defectors' cost; individual rationality should require the opposite inequality. Definition (2) and the one-shot planner equilibrium in §2.4 have the same problem: c_i(σ) ≥ c_i(σ_{-i},σ'_i) makes the current profile a worst response rather than a best response. Consistently, the proof of Theorem 1 says λ_i must 'maximize' a convex quadratic; the correct best response minimizes it. These sign errors affect the interpretation of all subsequent claims.","section":"Section 2.3, Definitions (1) and (2); Section 2.4"},{"comment":"The printed no-collective-punishments condition asks, for every deviation, for a planner j with c^k_j(σ|h) ≥ c^k_j((σ_{-i},σ'_i)|h), i.e., some planner is not hurt by the deviation. Proposition 2 identifies a deviation (reducing bottom flow) that makes every planner better off; such a deviation satisfies the printed inequality, so it does not violate the printed NCP. The impossibility proof needs the opposite property: no deviation should make all planners better off, i.e., ∃j with c^k_j(σ|h) ≤ c^k_j((σ_{-i},σ'_i)|h). As written, the inference 'no collective punishments ⇒ total bottom flow ≤ F' fails, and the F < 3/4 half of Theorem 2 is unsupported.","section":"Section 2.3(4), Proposition 2, converse of Theorem 2 (Section 5)"},{"comment":"Optimality is defined as lim_{k→∞} c^k(σ|h) = c_OPT for all global histories h. The construction only proves convergence on histories in which defections eventually stop; if defections occur infinitely often, planner-equilibrium stages can occur infinitely often and the limit need not be c_OPT. The sentence 'the amount of planner equilibrium stages are always finite if cars stop defecting' confirms this restriction. Either Definition (3) must be weakened (e.g., restricted to histories with finitely many defections) or the construction must handle histories with infinitely many defections.","section":"Section 2.3(3) and Section 5"},{"comment":"The verification of the four desiderata is only a sketch ('It is easy to check', 'straightforward to check'). The proof does not provide a formal argument that individual rationality and resilience hold for all histories and for all discount factors sufficiently close to 1, including histories during punishment and with stacked defections. It does not rigorously derive the claimed current-stage bound of 1/2 in every state, nor does it specify how the rotation pointer interacts with discounted costs in the lower-bound comparison. Since this is the main existence claim, a complete proof is needed.","section":"Section 5, positive direction of Theorem 2"}],"minor_comments":[{"comment":"The condition 'λ_i = 1 if n ≤ k' should read 'if i ≤ k'.","section":"Proposition 1"},{"comment":"The phrase 'the expected cost with no defections is 3/4' is only true under the socially optimal flow, not during the planner-equilibrium punishment rounds; please rephrase to avoid ambiguity.","section":"Section 5"},{"comment":"The two-case expression for λ_i is equivalent to λ_i = min{1, (1-F)/α_i}; using the min form in the statement would make the subsequent algebra easier to follow.","section":"Theorem 1"}],"recommendation":"major_revision","confidential_remarks":"The inequality-direction errors appear to be a systematic reversal rather than a deep conceptual flaw: the intended definitions are likely the opposites of those printed. However, the current text is not internally consistent, and all proofs would need to be rechecked after correction. I recommend asking for a complete rewrite of Section 2.3 and a full, rigorous proof of Theorem 2 before any further consideration."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Let me give you the short version first. The paper has a genuine new idea: on Pigou's network, having several competing planners can be what makes optimal routing sustainable, and the threshold is tied to the one-shot planner equilibrium sending more than 3/4 of flow on the bottom edge. That is worth a read. The paper is not in shape to be published as-is, though. The main impossibility proof has a sign error in the definition of no collective punishments, and the optimality property is not proved for histories with infinite defections.\n\nThe good part is Theorem 1, the characterization of the one-shot planner equilibrium. It is clean, solved in linear time, and the proof checks out. The corollaries about a 1/4 cap on any planner's fraction follow from the F threshold, and the paper correctly distinguishes its planner-level competition from strong equilibria and partition equilibria. The self-citations are just related work; there is no circularity.\n\nThe soft spots are real. Section 2.3 defines NCP as: for every deviation by planner i, there exists j with c_j(sigma|h) >= c_j((sigma_{-i},sigma'_i)|h). That says the deviation does not hurt at least one planner, which is the opposite of what the proof needs. Proposition 2 and the F<3/4 impossibility require that some j has c_j(sigma|h) <= c_j((sigma_{-i},sigma'_i)|h), i.e., no deviation can make everyone better off. Under the printed inequality, a profile with excess bottom flow satisfies NCP, because one planner reducing bottom flow lowers costs for all. So the averaging argument in the converse collapses. This is likely a typo, but it is load-bearing.\n\nThe optimality gap is separate. Definition 3 requires lim c^k = c_OPT for all global histories. The construction is only shown to converge when defections stop. If defections keep occurring, the system stays in planner-equilibrium stages, and the limit is F, not c_OPT. The paper says 'it is easy to check' for resilience to competition and individual rationality as well; the argument is plausible but not fully written out.\n\nBottom line: this is a conference-level idea that deserves referee time after a major revision. The authors need to fix the NCP inequality, align the optimality definition with what the strategy actually delivers, and complete the proof sketches. I wouldn't cite it in its current form, but I would bring it to a reading group and I'd gladly referee a corrected version.","headline":"Nice conceptual result on Pigou, but the NCP definition is backwards as printed and the optimality proof has a gap; fixable, not as-is.","tokens_in":12216,"tokens_out":4710,"would_cite":false,"duration_ms":49289,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["91A10","91A20","91A80"],"pacs":[],"model":"deepseek-v4-flash","headline":"Competition between autonomous-vehicle planners is necessary for optimal traffic flow: on Pigou's two-road network, the paper proves exact conditions under which a stable, optimal routing mechanism exists.","keywords":["algorithmic game theory","selfish routing","price of anarchy","autonomous vehicles","repeated games","mechanism design","Pigou network","planner equilibrium"],"falsifier":"Check the inequality direction in the no-collective-punishments definition on a concrete two-planner equal-shares example, $\\alpha_1=\\alpha_2=1/2$, whose one-shot equilibrium has $F=2/3<3/4$. Let both planners send $0.9$ of their traffic to the bottom path, so $F'=0.9>F$. Each planner could cut its own bottom-path flow and lower everyone's cost, which is what Proposition 2 calls a collective punishment; yet under the Section 2.3 definition that improvement leaves both planners better off, so the condition $c^j(\\sigma) \\ge c^j((\\sigma_{-i},\\sigma_i'))$ holds for that deviation. Running this cal","tokens_in":11415,"feed_emoji":"🚗","tokens_out":13937,"duration_ms":141236,"temperature":0.7,"pith_summary":"The paper asks whether centrally routable autonomous vehicles can eliminate the classical inefficiency of selfish routing, and answers that on the canonical two-road example, a single central controller cannot do it: a stable optimal plan needs several competing planners. It constructs a repeated-game strategy in which planners start at the socially optimal 50/50 split and, after any car deviation, spend enough rounds playing the one-shot planner equilibrium to make deviation unprofitable for patient agents. The central result is a sharp threshold: if the one-shot planner equilibrium puts more than three quarters of total traffic on the congestible path, all four desired properties (individual rationality, resilience to competition, optimality, no collective punishments) are achievable; if it puts less, they are not. This yields concrete sufficient and impossible regimes, including the clean rule that if no planner controls a quarter or more of the fleet, optimal flow is enforceable.","feed_headline":"Competing planners, not one boss, make traffic optimal","feed_subtitle":"On Pigou's two-road network, stable optimal routing is possible when no planner controls a quarter of the fleet.","key_machinery":"Pigou's network: two parallel roads from source to sink; the top road has constant travel cost $1$, and the bottom road has cost equal to the fraction of traffic using it. The socially optimal flow is the 50/50 split with total cost $3/4$, while uncoordinated selfish flow puts everyone on the bottom road at cost $1$. The mechanism's other central object is the one-shot planner equilibrium: given each planner's share $\\alpha_i$, the fractions $\\lambda_i$ of bottom-path flow chosen by each planner in the unique deterministic Nash equilibrium of the one-shot game, characterized by $\\alpha_i\\lambda_i = \\min\\{\\alpha_i, 1-F\\}$ with $F = \\sum_j \\alpha_j\\lambda_j$. The repeated-game strategy plays t","core_discovery":"The paper claims that on Pigou's network, competition among autonomous-vehicle planners is what makes optimal traffic flow enforceable. Given fractions $\\alpha_i$ of the unit traffic controlled by each planner, let $F$ be the total bottom-path flow in the unique deterministic one-shot planner equilibrium, where each planner minimizes the cost of its own agents (Theorem 1). Theorem 2 states that if $F > 3/4$, there is a repeated-game strategy profile satisfying individual rationality, resilience to competition, optimality, and no collective punishments; if $F < 3/4$, no profile can simultaneously satisfy individual rationality and no collective punishments. The boundary case $F = 3/4$ is reso","pith_inferences":["Beyond the paper: the construction reads as a design recipe in which the one-shot game's inefficiency is a punishment resource. Any network whose one-shot planner equilibrium is sufficiently congested relative to its optimum should admit a stable optimal mechanism by the same logic, a shape the authors state as Conjecture 1.","Beyond the paper: a market-share cap, such as the $1/4$ bound in the symmetric two-road case, becomes an instrument of mechanism design rather than merely an antitrust constraint: regulators could deliberately keep any autonomous-vehicle operator below the threshold.","Beyond the paper: the equal-planner case suggests a concrete quantitative test for other two-road networks with convex latency functions: check whether the excess congestion of the one-shot planner equilibrium over the optimum exceeds the gap between the selfish and optimal costs; the paper's Pigou result is the special case where that comparison collapses to $F > 3/4$."],"forward_implications":["On Pigou networks, a single central routing authority cannot give route recommendations that are simultaneously optimal for the system and best responses for the vehicles being routed; some competition is required.","If every competing planner controls less than $1/4$ of the traffic, there is a strategy profile that converges to the system-optimal split and is stable against both driver deviations and planner deviations.","With at most two planners, or with one planner controlling more than half the fleet, no strategy can simultaneously satisfy individual rationality and no collective punishments.","The punishment used to deter deviations is simply the one-shot planner equilibrium, so enforcing optimal flow does not require artificial or coordinated sanctions.","On Pigou's network, the one-shot equilibrium's bottom-path flow $F$ is the decision variable: $F > 3/4$ makes optimal routing enforceable, and $F < 3/4$ makes it impossible."],"supporting_citations":[{"why":"Supplies the Pigou network, the canonical two-road example where selfish flow is inefficient.","marker":"[23]"},{"why":"Establishes the benchmark comparing selfish and centrally optimized routing that the paper's optimality property targets.","marker":"[24]"},{"why":"Introduces the price of anarchy, the inefficiency measure motivating the search for an optimal but stable mechanism.","marker":"[17]"},{"why":"Shows repeated play does not trivially make efficient equilibria easy to sustain, supporting the need for the paper's explicit punishment construction.","marker":"[8]"}],"fun_headline_variants":["Competition, not coordination, drives traffic to optimum","Multiple planners beat a single authority for traffic","Optimal traffic flow emerges from planner competition","To fix traffic, let planners compete","Competing planners make traffic optimal"],"cache_read_input_tokens":2816,"weakest_assumption_plain":"The impossibility half of Theorem 2 depends on reading 'no collective punishments' as forbidding any profile in which one planner could lower everyone's cost by changing its own flow, but the paper's formal definition only requires that every deviation leaves at least one planner no worse off; the proof's upper bound on total bottom-path flow needs the stronger reading.","fun_headline_variants_meta":{"raw":{"variants":["Competition, not coordination, drives traffic to optimum","Multiple planners beat a single authority for traffic","Optimal traffic flow emerges from planner competition","To fix traffic, let planners compete","Competing planners make traffic optimal"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000692,"raw_usage":{"total_tokens":2948,"prompt_tokens":699,"completion_tokens":2249,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":443,"completion_tokens_details":{"reasoning_tokens":2184}},"tokens_in":443,"tokens_out":2249,"duration_ms":19413,"temperature":1.0,"reasoning_tokens":2184,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T22:20:09.401022+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Check the inequality direction in the no-collective-punishments definition on a concrete two-planner equal-shares example, $\\alpha_1=\\alpha_2=1/2$, whose one-shot equilibrium has $F=2/3<3/4$. Let both planners send $0.9$ of their traffic to the bottom path, so $F'=0.9>F$. Each planner could cut its own bottom-path flow and lower everyone's cost, which is what Proposition 2 calls a collective punishment; yet under the Section 2.3 definition that improvement leaves both planners better off, so the condition $c^j(\\sigma) \\ge c^j((\\sigma_{-i},\\sigma_i'))$ holds for that deviation. Running this cal","supporting_citations":[{"cited_title":"The economics of welfare","cited_arxiv_id":null,"evidence_quote":"Supplies the Pigou network, the canonical two-road example where selfish flow is inefficient."},{"cited_title":"How bad is selfish routing? Journal of the ACM (JACM) , 49(2):236–259, 2002","cited_arxiv_id":null,"evidence_quote":"Establishes the benchmark comparing selfish and centrally optimized routing that the paper's optimality property targets."},{"cited_title":"Worst-case equilibria","cited_arxiv_id":null,"evidence_quote":"Introduces the price of anarchy, the inefficiency measure motivating the search for an optimal but stable mechanism."},{"cited_title":"The myth of the folk theorem","cited_arxiv_id":null,"evidence_quote":"Shows repeated play does not trivially make efficient equilibria easy to sustain, supporting the need for the paper's explicit punishment construction."}],"review_version":1}