pods-author-response
GitHub用于撰写PODS会议作者回复，涵盖48小时双盲反驳信与带牧羊人复审的修改说明信。指导如何纠正误读、区分真实缺陷、保持匿名，并通过逐项映射变更项应对审稿意见。
触发场景
安装
npx skills add brycewang-stanford/Awesome-Journal-Skills --skill pods-author-response -g -y
SKILL.md
Frontmatter
{
"name": "pods-author-response",
"description": "Use when drafting ACM PODS author responses, covering the short ~48-hour double-anonymous rebuttal that corrects misreadings of proofs and definitions, and — distinctively — the shepherded revision cover letter that maps every required item to a concrete change in the resubmitted paper and survives the shepherd's second read."
}
PODS Author Response
Use this after PODS reviews are released. PODS has two distinct speaking turns, and conflating them is a common mistake: a short rebuttal (about 48 hours, a few thousand characters) right after reviews, and — if you receive a revision decision — a revision cover letter accompanying a resubmitted paper that a shepherd re-reads. Both stay double-anonymous: do not reveal authors, affiliations, or a named system, even to strengthen a point.
Triage (both turns)
- Answer what affects correctness first: a claimed error in a proof, an unstated assumption, a mismatch between an upper and a lower bound, or an out-of-scope objection.
- Distinguish a misreading (the reviewer misunderstood a definition or missed an existing appendix proof — fixable now, in the rebuttal) from a real gap (a step that genuinely needs to be written — fixable only in the revision).
- Keep every word anonymous: no institution, no funder, no real system name, no "our earlier work" phrased in the first person.
Turn 1 — the 48-hour rebuttal
Short and correctness-focused. The rebuttal exists to clarify misunderstandings and factual errors via pointers to the submitted paper — not to present new mathematics.
- Point to the exact definition, lemma, or appendix proof the reviewer overlooked ("the proof the reviewer requests is already in Appendix B, Lemma 7").
- Concede a genuine error plainly and state whether it is fixable in a revision.
- Do not paste a new proof the reviewers cannot verify in 48 hours; promising an unwritten proof does not move a theory review — it flags the gap.
Turn 2 — the revision cover letter (the distinctive PODS move)
A revision (minor or major) is a genuine shepherded second round. The cover letter is a change ledger: for every required item, either make the change and show exactly where, or explain precisely why it is unnecessary — and satisfy the shepherd.
[R1.1] Required: "the complexity bound in Thm 3 is not justified"
-> Action: DONE | added the full accounting in §4.2 and the deferred steps in Appendix B
-> Where: §4.2 (para 2), Appendix B (Lemma 7, revised)
[R1.2] Required: "state the assumption the lower bound rests on"
-> Action: DONE | Thm 4 now labeled as conditional on OMv; stated at point of use (§5)
[R2.1] Suggested: "extend the dichotomy to self-joins"
-> Action: DECLINED (with reason) | self-joins need a different reduction; out of scope for the
revision window, now stated as an explicit open problem (§7) rather than implied
The rule that turns a revision into an acceptance: address every required item. An item neither done nor explicitly and defensibly declined is what the shepherd's read punishes.
Reviewer pushback patterns
| Pushback | What it signals | PODS-ready response |
|---|---|---|
| "The proof of Thm X has a gap" | Correctness doubt | Fix the step in the revision and show it; if it is a misreading, point to the existing proof in the rebuttal |
| "The lower bound does not match the upper bound" | Optimality unproven | Close the gap, or scope the optimality claim to the model actually proved |
| "This assumption is hidden/unstated" | Rigor gap | Label the conditional result at its point of use; state the assumption explicitly |
| "The model trivializes the result" | Modeling objection | Defend the model as realistic, or adjust it and re-derive; do not ignore |
| "Overlaps prior result Y" | Novelty/delta doubt | Sharpen the theorem-level delta; add the missing comparison of results |
Anonymity in the response (easy to slip)
- Refer to your own prior work in the third person, as in the paper.
- Describe revisions without linking to an identifying arXiv preprint or homepage; reference the in-PDF appendix location instead.
- Do not thank a named collaborator, advisor, or funder inside the rebuttal or cover letter.
Calibration
- Respond to the criterion the reviewer raised (correctness, tightness, modeling, novelty), not the one you would rather defend.
- Length and format norms for both turns vary by cycle; confirm the current instructions before sending.
- The shepherd decides the revised paper; make the paper, not the letter, carry the corrected mathematics.
Output format
[Turn] 48-hour rebuttal / revision cover letter
[Priority issue] <reviewer concern>
[Decision dimension] correctness / tightness / modeling / significance / clarity
[Change ledger] <required item -> DONE/PARTIAL/DECLINED + where in paper/appendix>
[Anonymity check] <no identity leak in the response: passed/issues>
版本历史
- 9f86f09 当前 2026-07-19 17:16
