| paper | d4-conf-atal-2017-0293 |
| authors | — |
| venue | ? |
| filed under | unclassified |
| judged by | gpt-6-luna / xhigh (triple__luna__xhigh__c6r1) |
| judge confidence | high |
| authors would recognise it | no |
Problem 1 defines DIME but establishes no computational result; the algorithmic work is attributed to earlier papers. The pilot findings are empirical, so the paper has no qualifying result to mirror, regardless of whether a continuous DIME model could be proposed.
fails bit a — no named computational result to mirror
No result in this paper is mirrored: Problem 1 defines DIME, while the algorithmic methods are attributed to prior work and this paper reports pilot findings and deployment challenges.
I cannot make a compliant anchor-based case: this paper contains no numbered theorem, lemma, corollary, or proposition asserting a computational complexity result. Problem 1 (DIME) defines the problem of finding a sequential policy that maximizes expected influence, but it is a problem statement, not a result about hardness or tractability. The paper’s comparisons of HEALER, DOSIM, and degree centrality are empirical pilot findings; the algorithmic work it discusses is attributed to earlier papers.
There is a plausible continuous question here: represent a large population by masses of agents with shared influence and observation behavior, then choose a mass of peer leaders at each stage to maximize the expected final informed mass. But that would be a prospective mirror of DIME, not a mirror of a named computational result proved in this paper. The paper therefore supplies no qualifying anchor for the requested case.
The cleanest case against a mirror is that this paper proves no computational result to mirror. Problem 1 defines DIME; it does not establish a complexity bound. HEALER and DOSIM’s algorithmic results are cited from earlier work. This paper’s own contributions are pilot findings and lessons about deployment. So the proponent’s proposed continuous DIME question is prospective, not a mirror of a named computational result here.
The strongest version of that proposal would optimize the expected informed *mass* in a large population, choosing a mass of peer leaders over several stages. But DIME’s information is carried by particular people and their particular ties: selecting a node reveals its uncertain outgoing edges, while its current influence state and likelihood of attending also affect the policy. Under the programme’s definition of type, those details belong in the type. If they distinguish each person, then the type support grows with the network and there is no useful aggregation. If people are grouped into a few types, the model loses which individuals are bridges, which ties are redundant, and what a selected leader can reveal—the paper’s own explanation for why the strategies perform differently.
A multi-type or graphon model is the best rescue: treat people within a group as interchangeable, give groups an interaction kernel, and choose fractional seed masses. That may define a worthwhile new adaptive influence-optimization problem. But it replaces observations of named leaders’ friendships with aggregate information about group interactions. Keeping the original small action budget also leaves the intervention as a finite selection of particular people, rather than a mass decision. Scaling the budget with population size would make fractional actions natural, but would change the paper’s stated operational constraint of training only three or four leaders per intervention.
Thus no anchor from this paper survives: the DIME mirror could be a future research question, but it is not a continuous counterpart to a computational result established here. The case is weaker if the claim is literally that no continuous influence-maximization model could ever be worthwhile; the graphon-style question shows why that universal claim is hard to defend. The stronger, well-supported negative is that this paper gives ChoCo no computational anchor for one.
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.