| paper | d4-conf-aaai-2020-0483 |
| authors | — |
| venue | ? |
| filed under | unclassified |
| judged by | gpt-6-luna / xhigh (triple__luna__xhigh__c6r1) |
| judge confidence | medium |
| authors would recognise it | yes |
other→P discrete result → its continuous mirror
Answer: The divisible outcome exists for every society, attains an exact \(1\)-approximation, and retains strong budget balance, individual rationality, and pointwise incentive compatibility. Every refinement with \(\operatorname{OPT}>0\) satisfies \(0\leq\operatorname{OPT}-\operatorname{GFT}(x^{(h)})\leq h\sum_g r_g\max_v v\), so its efficiency ratio converges to \(1\) for every fixed society. · discrete side: other · adjudication: proved
full proof & adversarial review →
Theorem 3
statement extracted from the paper’s text layer
Given categories \(G\), a recipe \(r_g\), and rational masses \(\mu_{g,v}\) of agents with category \(g\) and value \(v\), compute the divisible-mass ascending-price clock outcome. Trade quantities \(x_{g,v}\) and total trade mass \(q\) must satisfy \(0\le x_{g,v}\le\mu_{g,v}\) and \(\sum_v x_{g,v}=r_gq\); determine the outcome’s gain from trade \(\sum_{g,v}v x_{g,v}\) relative to the maximum over feasible mass trades, and establish its budget-balance, individual-rationality, and pointwise incentive properties as the largest individual mass tends to zero.
Types are \((g,v)\), with rational mass \(\mu_{g,v}\); decisions are the traded masses \(x_{g,v}\), the trade quantity \(q\), and category prices \(p_g\). The objective is to compute the clock outcome and compare its gain from trade with the maximum feasible mass-trade gain, while defining incentives for infinitesimal participants.
An atomless participant cannot change the clock, so Theorem 3’s obvious-truthfulness guarantee becomes a price-taking incentive property; the proponent identifies this shift but does not resolve its formal meaning.
fatal: False
The mirror covers Theorem 3’s arbitrary-recipe ascending-prices auction, including its budget-balance, individual-rationality, incentive, and gain-from-trade claims. It leaves the external-competition auction results and the paper’s other contributions unaddressed.
The strongest positive anchor is Theorem 3, proved in this paper. It says that for an arbitrary fixed procurement-set recipe \(r\), the ascending-prices auction is strongly budget balanced, individually rational and obviously truthful, and its gain from trade approaches the optimum as the optimal number \(k\) of trades grows. This is the paper’s broadest named algorithmic approximation result. The paper gives no running-time bound or formal complexity classification, though, so it would not qualify if the anchor requirement is read as demanding a theorem explicitly stating a class such as P or NP-hardness.
A plausible regime is a recurring supply chain with many small participants and standardized roles and reservation values: for example, lemon pickers, sugar producers, transporters and drink vendors, where each role has a fixed number of units per trade and values come from a short tariff menu. A type is a participant’s category and value; its mass is the amount of society with that type. The number of participants can be large while the number \(\tau\) of positive-mass types stays small. Treating traded mass fractionally is part of this mirror: a type cohort need not enter or exit all at once.
I would name the mirror Continuous Recipe Auction. Its input is a finite category set \(G\), a positive integer recipe \(r_g\) for each category, and rational masses \(\mu_{g,v}\) at each value \(v\) in each category’s finite support. A solution gives a trade quantity \(q\), traded mass \(x_{g,v}\in[0,\mu_{g,v}]\), and category prices \(p_g\). Feasibility requires \(\sum_v x_{g,v}=r_gq\) for every category. The auction must provide an ascending-price clock, with types exiting at their values and tied mass divisible, such that \(\sum_g r_gp_g=0\), traded mass is individually rational, and truthful exit is obviously dominant. The objective is to compute the clock outcome and its gain from trade, \(\sum_{g,v}v x_{g,v}\), against the optimum over feasible mass trades.
I expect computing this outcome to be tractable in \((|G|,\tau,L)\): the input has finite rational support, and the paper’s clock advances through value thresholds. The substantive question is whether its truthfulness, budget balance and asymptotic welfare guarantee carry over to the mass model, and whether the mass limit yields exact optimal gain from trade or only convergence as \(q\) grows. This covers Theorem 3’s arbitrary-recipe auction; it does not claim to mirror the paper’s whole contribution or establish a new complexity result.
The weak point is the strategic interpretation. In an atomless population a single participant has no effect on aggregate prices, so obvious truthfulness may become a price-taking property rather than the strategic guarantee the authors prove for finite agents. The proposed question remains recognizable as their auction problem, but a convincing mirror must formalize that limit and show that its truthful clock is more than a degenerate consequence of individual agents having zero mass.
Theorem 3 is the strongest anchor, but its theorem is about mechanism properties: strong budget balance, individual rationality, obvious truthfulness, and a gain-from-trade guarantee. It does not classify the complexity of a computational problem. The auction already specifies how to compute its outcome: sort agents by value within each category and advance prices through exit thresholds. For rational masses on finite support, the proposed “compute the clock outcome” task becomes a sweep over those support values. That is a coherent extension, but it supplies little of the computational landscape ChoCo is meant to chart.
The more serious obstacle is strategic. A continuous mirror with mass at each \((g,v)\) type is plausible for a standardized supply chain. But when each participant has zero mass, one participant’s report cannot change the aggregate price path. Obvious truthfulness then has the price-taking interpretation of accepting trade when it benefits you. Giving a cohort positive mass so its report can affect prices makes the cohort the strategic agent, changing the paper’s individual-agent guarantee into a group-incentive problem. The mass model can preserve the auction’s economic purpose, but it does not preserve the strategic content of Theorem 3 unchanged.
Its approximation guarantee also loses its natural scale. The bound depends on \(k\), the integer number of trades, and allows the mechanism to remove a deal. In a normalized continuum, an individual deal has vanishing mass; the distinctive finite-market loss disappears unless the model retains an additional finite-population scale. That would make the question about finite replications or convergence rates, rather than the computation of the continuous auction.
So the best negative case is about fit, not impossibility: this paper supports a plausible continuous market-design extension, but its only proposed computational task is a routine implementation of an already specified clock, while its central theorem concerns strategic guarantees whose meaning changes in the atomless limit. If ChoCo counts computation of such mechanism outcomes as a sufficient anchor, the proponent’s mirror survives; on the programme’s stated complexity focus, I would not prioritize it.
The adversarial triple: the proponent anchors on up to three named results; the opponent sees that case and must defeat every anchor; the judge decides which case convinced it. These are the pipeline’s own outputs, generated by tools/triple_run.py — no human edited them. The paper’s own text is not reproduced here beyond the quoted statement above.