| paper | d4-conf-atal-2016-0161 |
| authors | — |
| venue | ? |
| filed under | unclassified |
| judged by | gpt-6-luna / xhigh (triple__luna__xhigh__c6r1) |
| judge confidence | high |
| authors would recognise it | unclear |
The paper has no numbered Theorem, Lemma, Corollary, or Proposition stating a computational result, so it fails the required anchor test. Its unnumbered MDP hardness statement and dynamic-programming discussion cannot supply bit (a); the proposed fleet-level extension also needs population coupling and response assumptions the paper does not provide.
fails bit a — no named computational result to mirror
No qualifying computational result is anchored; the individual-driver MDP, interface experiments, and Q-learning evaluation remain unmirrored.
I can’t make an anchored positive case under the requested standard: the paper has no numbered Theorem, Lemma, Corollary, or Proposition asserting a computational result.
The closest candidate is an unnumbered statement in §2.3: solving the driver’s MDP optimally is NP-hard in general because journey selection generalizes Knapsack. The paper also gives a dynamic-programming solution and reports experimental results, but none is a named computational result with a number to quote and restate. That prose claim could motivate a continuous-population mirror, but treating it as an anchor would violate your instructions.
The proponent names no anchor, and that is the strongest point against a continuous mirror. The paper has no numbered theorem, lemma, corollary, or proposition establishing a computational result. Its main findings are experimental comparisons between interfaces. The nearest candidate is the unnumbered claim that a single driver’s MDP is NP-hard because journey selection generalizes Knapsack; the paper then says finite-horizon instances can be solved by dynamic programming. That is not a named computational result of the kind the requested case was supposed to mirror.
The best rescue is a fleet of EV owners represented by a distribution over charging-relevant types: journey profiles, battery states, price beliefs, and response to an interface. That is a plausible high-multiplicity regime, and it should not be dismissed as unrealistic. But the paper’s computational model is an individual driver making decisions across days. In its testbed, market prices are sampled exogenously, and one driver has negligible effect on future prices. Aggregating many identical drivers would therefore average their individual policies and outcomes; it would not turn the paper’s MDP into a population-level computational problem. Adding shared capacity or endogenous prices could create one, but that would add the collective allocation problem the paper does not study.
A stronger extension would optimize an interface for a continuous distribution of user types. The authors’ interest in optimizing interfaces makes this a credible direction. Yet the paper supplies no model of how each type responds to each interface: its human results come from experiments, and its Q-learning agent only reproduces broad behavioral trends. To make the extension computational, one would need new response models, objectives, and population coupling. That could be worthwhile research, but it is not a continuous mirror of a computational result in this paper.
So the negative case is strong under the requested standard: no computational anchor survives because none was supplied. The universal claim is weaker. A fleet-level interface-design problem is plausible and related to the authors’ agenda, though it would be a new problem rather than a continuization of this paper’s computational results.
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.