brycewang-stanford RSS Author Response
Draft the one-page rebuttal RSS grants to papers that cleared the first-round threshold (2026 deadline: April 3, AoE; verified 2026-07-08). The format is unusual and the drafting strategy should be too: this is a memo to the Area Chair who will represent your paper in a decision meeting, not a reply thread.
First, establish which situation you are in
| Situation | Meaning | Move |
|---|---|---|
| Rebuttal invited | Paper is above threshold; decision genuinely open | Full drafting protocol below |
| No invitation | Below threshold; decided without author input | Skip to resubmission triage (bottom) |
| Reviews arrived, rules unclear | Cycle may have changed the design | Reread roboticsconference.org/reviewps/ before writing a word |
Drafting protocol for the single page
Budget the page like the scarce resource it is. A workable split: 60% on the one or two objections that could flip the meeting, 25% on factual corrections, 15% on a scope concession that shows calibration. Zero on thanking reviewers at length.
- Open with the verdict-relevant sentence. State, in one line, why the central objection does not undermine the claim — then support it. ACs skim top-down under meeting time pressure.
- Cite the submission by coordinates. "Table 3, row 4 reports exactly this ablation" beats paragraphs of re-explanation, and signals the evidence existed at review time.
- Concede cleanly where the reviewer is right. A scoped concession ("the claim holds for quasi-static tasks; we will retitle Section 5 accordingly") converts an objection into a camera-ready edit.
- No new experiments as load-bearing evidence. You may clarify protocol or cite numbers already in the supplement; a fresh results table invites the fair reply that it was never reviewed.
- Stay anonymous. Double-blind still applies — no links, no lab tells, no "as our prior work" constructions.
Writing for the AC, not the reviewers
The 2026 process sends rebuttals to reviewers only when the AC wants extra feedback. So the realistic reader model is one busy expert deciding what to say about your paper in a synchronous meeting:
- Give the AC quotable material: short, declarative, coordinate-anchored sentences.
- Resolve reviewer disagreements explicitly ("R2 and R3 read the trial protocol differently; the protocol in §4.2 matches R3's reading") — arbitration help is what an AC most lacks.
- Do not relitigate significance at length; one sentence connecting the claim to the broader program is the ceiling of what a rebuttal can move.
Skeleton of a page that works
A compressed example shape (fictional content, real structure):
R1's central concern — that the gain could come from the retuned controller
rather than the schedule — is tested directly: Table 3 row 4 holds the
controller fixed and varies only the schedule (41% -> 78%).
Corrections. R2: trials per condition are 100, not 25 (§4.2 line 3 states
the protocol; 25 is the object count). R3: the baseline uses the authors'
released configuration, not a reimplementation (§4.1).
Scope. R1 is right that the claim as titled overreaches quasi-static tasks;
we will scope the title and §5 header accordingly.
Reviewer disagreement. R2 and R3 read the reset protocol differently; §4.2
matches R3's reading, and the supplement ledger logs every reset.
Note what is absent: gratitude paragraphs, new results, promises of future work, and any sentence the AC could not lift directly into the meeting.
Failure patterns that waste the page
| Pattern | Why it fails |
|---|---|
| Point-by-point replies to every comment | Dilutes the one meeting-flippable argument |
| "We have now run additional experiments..." | Unreviewed evidence; invites procedural dismissal |
| Tone-matching an irritated review | ACs read composure as confidence in the evidence |
| Restating the paper's contribution | The AC has the paper; they need the resolution |
| Deferring everything to "the final version" | Signals the objection is real and unanswered |
If no rebuttal came
The decision is effectively made; spend the energy forward. Map each objection to the
next calendar slot: protocol gaps → fix before IROS (~March deadline has passed;
next: CoRL ~June or ICRA ~September patterns, verify live); framing objections →
rerun rss-topic-selection honestly before re-targeting RSS 2027.
Window logistics
- The 2026 invitation-to-deadline window was short and its opening date was never publicly posted; pre-draft answers to the three most likely objections during the February-March silence so the invitation triggers assembly, not composition.
- One author owns the page; committee-written rebuttals converge to mush. Others supply evidence coordinates and red-team the draft.
- Verify the submission channel and length rule on the form itself when the invitation arrives — "one page" needs the form's own definition of a page.
- Freeze the text several hours before the AoE cutoff; a rebuttal that arrives late is a rebuttal that does not exist, and there is no grace channel.
- Archive what you sent; if accepted, the camera-ready must actually deliver every
edit the page promised (
rss-camera-ready).
Output format
[Invitation status] invited / not invited / unclear (reread reviewps)
[Meeting-flippable objections] <1-2, each with evidence coordinates>
[Corrections] <error -> paper location>
[Concession] <scoped sentence + camera-ready edit>
[Page budget check] flip 60 / correct 25 / concede 15
[Anonymity check] pass / issues
Source: brycewang-stanford/Awesome-Journal-Skills → RSS-Skills/skills/rss-author-response/SKILL.md