| paper | d4-journals-jet-2016-0071 |
| authors | — |
| venue | ? |
| filed under | unclassified |
| judged by | gpt-6-luna / xhigh (triple__luna__xhigh__c6r1) |
| judge confidence | medium |
| authors would recognise it | yes |
Theorem 3
statement extracted from the paper’s text layer
Given \(m\) housing classes, \(\tau\) tenant types with current classes \(h_t\), binary-rational masses \(\mu_t>0\) summing to \(1\), reservation prices \(\underline p_h\), and rational quasilinear values \(v_{t,0}\) and \(v_{t,h}\), let class supply be \(s_h=\sum_{t:h_t=h}\mu_t\). Each type’s mass can stay in its current home, buy that home at its reservation price, or buy another class at its market price; allocations must clear every class’s supply, and assigned mass must receive a utility-maximizing option. On instances with a componentwise least equilibrium price vector, compute that vector and a market-clearing equilibrium allocation maximizing total purchases, with running time measured in \(m,\tau\), and input bit length \(L\).
A finite-support population distribution over complete tenant types, with class supply scaling as \(s_h=\sum_{t:h_t=h}\mu_t\). The decision variables are nonnegative masses assigned to staying, buying one’s current class at \(\underline p_h\), or buying another class at \(p_h\); the objective is to maximize total purchases at the componentwise least equilibrium prices.
The paper’s no-indifference-chain restriction is defined for individually indexed agents and houses, and the case does not specify its exact counterpart for housing classes and positive masses.
fatal: False
The mirror covers Theorem 3’s minimum-price outcome and the finite price process of Proposition 6. It leaves the paper’s group non-manipulability, revenue, and trade-maximization results unmirrored.
The strongest honest anchor is the algorithmic result in Theorem 3, proved in this paper: for every preference profile \(R\in\tilde R\), the Dutch price sequence is unique, and its terminal price vector \(p_T\) is the unique minimum equilibrium price vector in \(\Pi_R\). This is a result about computing the paper’s minimum-price mechanism. It is not a polynomial-time theorem: the paper gives no complexity-class classification or runtime bound. I would use it as an algorithmic lead, not claim more than it proves.
A plausible regime is a large public-housing authority assigning homes within neighborhoods. Let a type record a tenant’s current housing class and complete preferences over staying or renting, buying that home at its reservation price, and buying other classes at market prices. For example, tenants in a large cohort might share a dwelling category, regulated rent, reservation-price rule, and valuation profile. A neighborhood could have many thousands of tenants but only a modest number \(\tau\) of such types and \(m\) housing classes. In the mirror, the population is the distribution \(\mu\) over those types, with each \(\mu_t\) a binary-rational mass. Housing supply \(s_h=\sum_{t:h_t=h}\mu_t\) scales with the population; that co-modification preserves the paper’s one-home-per-tenant structure in the limit.
I would pose the lead question as \(\textsf{Min-Price Housing}_{\infty}\). An instance gives \(m\) housing classes, \(\tau\) types \(t\) with current class \(h_t\), rational masses \(\mu_t>0\) summing to one, rational reservation prices \(\underline p_h\), and rational quasilinear values \(v_{t,0}\) for staying and \(v_{t,h}\) for buying class \(h\). Restrict to the type-level version of the paper’s no-indifference-chain domain \(\tilde R\).
A solution is a price vector \(p\ge\underline p\) and nonnegative mass allocations \(r_t,o_t,y_{th}\). Here \(r_t\) stays in the current home, \(o_t\) buys that home at its reservation price, and \(y_{th}\), for \(h\ne h_t\), buys a home in class \(h\). They must satisfy \(r_t+o_t+\sum_{h\ne h_t}y_{th}=\mu_t\) and \(s_h=\sum_{t:h_t=h}(r_t+o_t)+\sum_{t:h_t\ne h}y_{th}\). Positive mass can be assigned only to utility-maximizing options, whose utilities are \(v_{t,0}\), \(v_{t,h_t}-\underline p_{h_t}\), and \(v_{t,h}-p_h\), respectively. Among those market-clearing equilibria, maximize total purchases \(\sum_t o_t+\sum_{t,h\ne h_t}y_{th}\); return one such allocation at the componentwise least equilibrium price vector. The computational question is to solve this exactly as a function of \(m,\tau\), and input bit length \(L\).
I expect this version to be Class A: the finite-support quasilinear problem looks like a capacitated assignment LP, with equilibrium prices obtained from its dual. The paper’s Dutch process gives a direct algorithmic motivation, while Theorem 3 supplies the uniqueness target. This is a new continuous computational question; the paper does not establish that its discrete process or proof carries over to it. Further questions include whether the mass process terminates in finitely many price updates, whether uniqueness survives without the genericity restriction, and whether the group non-manipulability result extends to deviations by positive-mass coalitions.
The weakest point is the regime’s type compression. The paper permits rich preferences over individual houses; if almost every tenant values each home differently, then \(\tau\) may grow nearly as fast as the population. The mirror is most credible where housing classes are genuinely substitutable and tenant attributes fall into repeated policy or preference categories. Also, fractional allocation within a type and class-level homes change the model enough that Theorem 3 is a motivation for the problem, not evidence that its answer will be the same.
The strongest case against this anchor is that the paper does not state a computational-complexity result. Theorem 3 proves that, on \(\tilde R\), the Dutch price sequence is unique and ends at the unique minimum equilibrium price vector. Proposition 6 proves the sequence is finite. Neither gives a runtime bound or analyzes a computational problem with an input encoding. At each step, the process calls for the minimum equilibrium price in a price regime; the paper does not bound the cost of that calculation. So Theorem 3 is an algorithmic structure result, not a complexity result whose landscape a continuous mirror would chart.
The proposed \(\textsf{Min-Price Housing}_{\infty}\) is a coherent problem, but it adds a specific model: housing classes are treated as interchangeable, preferences are quasilinear with rational values, and class supplies scale with the population. Its allocation layer is a finite capacitated assignment problem, with prices represented through equilibrium or dual constraints. That is a plausible high-multiplicity model, but it is a new computational problem built around the paper’s setting, not a continuization of a complexity claim the paper makes. The paper’s no-indifference-chain condition also concerns finite agents and individually indexed houses; a type-level analogue would need to be defined and justified for the mass model.
The paper’s stronger named results are about mechanism properties such as non-manipulability and trade maximization, not computational complexity. A continuous version of those properties would need care: an individual deviation by a zero-mass agent cannot change a distribution-based outcome, while positive-mass coalitional deviations require a newly specified notion. That may be worth studying, but it is a different axiomatic question.
There is no honest multiplicity objection here: large public-housing markets can plausibly contain many tenants per complete preference and housing type, with supply scaling accordingly. Nor does the finite-support mass model fail simply because it co-modifies supply. Those are legitimate choices under the programme’s scope. The negative case is therefore narrow: the paper itself supplies no named computational-complexity result to mirror, and the proposed anchor overstates what Theorem 3 establishes. That weakens the case that this paper is a strong ChoCo target, but it does not defeat the better finite-support housing mirror. The claim that no worthwhile mirror exists in any scenario is not supported by the material provided.
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.