Agent Skills
› brycewang-stanford/Awesome-Journal-Skills
› aamas-review-process
aamas-review-process
GitHub指导AAMAS论文审稿策略，涵盖OpenReview流程、双盲反驳、领域主席讨论及评分标准。针对博弈论与MARL等混合评审池，提供提升正确性、交互贡献显著性及实证支持的建议，强调公开记录的专业应对。
Trigger Scenarios
询问AAMAS论文审稿流程
制定AAMAS论文反驳策略
了解AAMAS评分维度
Install
npx skills add brycewang-stanford/Awesome-Journal-Skills --skill aamas-review-process -g -y
SKILL.md
Frontmatter
{
"name": "aamas-review-process",
"description": "Use when explaining or planning around AAMAS peer review, covering OpenReview review release, the double-blind rebuttal on preliminary reviews, area-chair discussion, the mixed game-theory, MARL, and systems reviewer pool, the public posting of reviews and decisions, and how acceptance criteria weigh the interaction contribution."
}
AAMAS Review Process
Use this to reason about review-stage strategy. Reopen the current CFP, OpenReview group, author and reviewer instructions, and the code of conduct before making process claims; AAMAS review specifics move between editions.
Process model
- AAMAS runs submission and review on OpenReview under the IFAAMAS namespace in recent cycles.
- Reviewers evaluate technical correctness, the significance of the interaction contribution, the reality and rigor of the multiagent evaluation, clarity, reproducibility, and fit with agents-and-multiagent-systems scope.
- A rebuttal lets authors respond to preliminary reviews before the final decision; area chairs then synthesize.
- Accepted papers and their reviews are published, so the review record is durable and public.
- The most useful rebuttal gives the area chair a clean rationale for acceptance, not a point-by-point defense of every comment.
Who reviews here
- The pool mixes game theorists, multiagent-RL researchers, mechanism-design and social-choice specialists, and systems-minded reviewers; expect at least one to read the game definition and solution concept line by line.
- Because AAMAS is specialized, a paper is likely to meet a reviewer who works on exactly its subarea, so a vague equilibrium claim or an under-specified opponent set gets caught rather than skimmed.
- Borderline interaction papers usually fail on one of three edges: the result turns out to be single-agent in disguise, the solution concept is never pinned down, or the multiagent evaluation is thin (self-play only, no seeds, no held-out opponents).
Scoring leverage table
| Review dimension | What raises it | What sinks it |
|---|---|---|
| Correctness | A stated game, a named solution concept, and a body-level proof sketch | Hidden information structure; an equilibrium asserted but never defined |
| Significance | A finding that only exists because agents interact | An incremental single-agent gain wearing a multiagent label |
| Empirical support | Experiments that probe strategy: held-out opponents, deviation tests | Self-play-only curves disconnected from the claim |
| Clarity | One notation source and a legible game description | Notation and payoff conventions that shift between sections |
Stage-by-stage realism
- Initial reviews: triage by what the area chair would weigh, not by reviewer tone.
- Rebuttal: windows are short; an early, precise reply anchored in submitted evidence beats a late exhaustive one.
- Decision: the area chair synthesizes, and one unanswered correctness or interaction-reality objection outweighs several resolved clarity complaints.
- Public record: assume the reviews and your rebuttal will be visible with the paper, and keep the exchange professional and concrete.
Output format
[Current stage] submitted / reviews / rebuttal / decision / camera-ready
[Decision actors] <reviewers / area chair / program chairs>
[Likely leverage] <correctness / interaction-reality / significance / experiments / clarity>
[Forbidden moves] <identity leak / new results / revised-paper upload if disallowed>
[Next response move] <one action>
Version History
- 9f86f09 Current 2026-07-19 14:15
