--- name: hpca-author-response description: Use when drafting an HPCA rebuttal and revision in the single combined window: triaging objections by what the window can fix, running a small number of new experiments, writing for the program-committee room and its champion, and revising the paper diff-minimally so response and revision agree. --- # HPCA Author Response Use this during HPCA's **single combined rebuttal/revision window** — the one period in which you respond to reviews and revise the paper together. The goal is not to win an argument; it is to arm a champion at the program-committee meeting and to make the revised paper reflect what you promise. ## Triage objections by what the window can fix Sort every review comment before writing a word: - **Deciding objection** — the methodology or novelty concern the champion must overcome. Spend most of the window here. - **Fixable-and-worth-it** — a missing baseline, a sensitivity study, an unclear mechanism detail the window has time to address. - **Clarification** — a misreading you can correct in a sentence and reflect in the revision. - **Out of scope** — a request no short window can satisfy; acknowledge honestly and scope what you *can* show. | Objection type | Response move | Revision move | |---|---|---| | Deciding (methodology/novelty) | Direct evidence or a tight argument | Change the paper to match | | Missing baseline / sweep | A small new experiment | Add the result, mark it | | Mechanism unclear | One-paragraph clarification | Tighten the description | | Misreading | Point to the text | Rephrase the ambiguous line | | Out of scope | Acknowledge; state what is feasible | Note the limitation | ## Run few, decisive new experiments The window is short. Run **a small number** of new experiments that each answer a specific objection, not a scattershot of everything a reviewer mentioned. One well-chosen sweep that closes the deciding objection is worth more than five that answer minor points. Pull from the staging shelf you prepared before the window opened (see `hpca-supplementary`). ## Write for the room and its champion The rebuttal is read at the PC meeting, where a champion argues for the paper against the strongest objection. Write so the champion can quote you: - Lead each response with the answer, then the evidence — not a wind-up. - Make the deciding objection's answer unmissable and self-contained. - Be concrete: "we added the tuned baseline; it changes the gap from X to Y" beats "we believe our baseline is fair." - Stay civil and specific; a defensive or vague rebuttal loses the champion. ## Revise diff-minimally so response and revision agree Because reviewers see both your response and the revised paper, keep the revision a **minimal, faithful diff**: change exactly what you promised, and no more. A response that claims a fix the revision does not contain — or a revision that quietly rewrites beyond the promises — undercuts trust. Track each promise to a specific edit. ## Window pass ```text 1. Triage every comment (deciding / fixable / clarification / out-of-scope) 2. Pick the few decisive experiments; pull from the staging shelf 3. Draft answer-first responses; make the deciding one unmissable 4. Apply a minimal revision that matches every promise 5. Cross-check: does the revised paper contain each claimed fix? ``` ## Output format ```text [Deciding objection] [New experiments] [Response map] answer-first reply> [Revision diff] exact edit> [Champion test] can a champion win the room with this? (Y/N) ``` Reopen the current CFP for the exact window dates and any length limit on the response — both are per-edition.