brycewang-stanford POPL Author Response
The POPL 2027 call describes the response phase plainly: authors get a multi-day
period, responding is optional, and a response "must be concise, addressing specific
points raised in the reviews" (read 2026-07-08; exact window dates were not rendered
— 待核实 in resources/official-source-map.md). At POPL the response is usually the
last input before the committee decides between reject and conditional acceptance,
so its job is to remove doubts, not to renegotiate the paper.
What POPL reviewers actually dispute
Theory reviews rarely say "weak baselines." They say a definition does not mean what you think, a side condition is missing, a theorem is unsurprising given prior work, or the metatheory-to-motivation gap is too wide. Sort every review point into one of four bins before writing a word:
| Bin | Example objection | Response move |
|---|---|---|
| Soundness doubt | "Lemma 4.2 seems to fail for open terms" | Quote the exact hypothesis that excludes the case; cite the proof line or mechanization file |
| Misread formalism | "Your typing rule allows unrestricted duplication" | Point to the definition as written; concede unclear notation and promise a one-line clarification |
| Significance | "This follows from [X] by standard techniques" | Name the specific step that fails under [X]'s assumptions — technical daylight, not adjectives |
| Presentation | "Section 5 is unreadable" | Accept, state the concrete restructuring you will do in the conditional-acceptance revision |
Soundness doubts come first and get the most space. One unresolved "the proof may be broken" outweighs every fixed typo.
Rules of engagement
- Answer the question asked, at the location asked. "See Section 3" without a page, definition number, or lemma name reads as evasion.
- No new theorems, no new mechanization claims. If a reviewer's counterexample exposes a real gap, say what the repaired statement is and where the fix lands — the conditional-acceptance revision exists exactly for that.
- Never hint at who you are; full double-blind holds through this phase, and identities unblind only after conditional-acceptance decisions.
- Concede fast and precisely. "R2 is right that Definition 6 omits the well-formedness premise; the proofs already assume it (see App. C.1), and we will state it" is a strong sentence, not a weak one.
A skeleton that fits a concise budget
Thank you — responses keyed to review points.
[R1, soundness of Thm 3] The counterexample uses an open term; Thm 3's hypothesis
"Γ ⊢ e : τ with Γ closed" excludes it (Def. 5, p.9). The mechanization checks the
closed case only, as stated in §7. No change needed; we will flag the hypothesis
in the theorem statement.
[R2 + R3, relation to <prior system>] <Prior> requires structural weakening
(their Lemma 2); our substructural setting has no weakening, which is where their
proof method stops. §2 example 2 will make this explicit.
[R3, presentation of §5] Agreed. We will move the auxiliary judgments to the
appendix and open §5 with the main invariant.
Output format
[Response status] drafted / needs-facts / not worth responding
[Bin counts] soundness:<n> misread:<n> significance:<n> presentation:<n>
[Highest-risk point] <the objection that decides the paper, and the one-line rebuttal>
[Concessions] <what is admitted and where the revision fixes it>
[Length check] <concise per CFP? what to cut>