brycewang-stanford CHI Review Process
CHI's Papers track runs a two-round, committee-anchored process that resembles a journal's revise-and-resubmit more than an ML conference's rebuttal. Mechanics below follow the CHI 2027 and CHI 2026 Papers Review Process pages and the CHI 2026 outcome reports (read 2026-07-08 via search renderings; the conference sites block direct automated fetching).
Who touches your paper
- 1AC — the Associate Chair who manages your submission: screens it, recruits two external reviewers, writes the meta-review, and represents the paper at the PC meeting. The 1AC is your most important reader at CHI.
- 2AC — a second committee member who writes a full review like the externals.
- Two external reviewers — recruited by the 1AC for topical expertise.
- Subcommittee chairs (SCs) — run the subcommittee you designated, arbitrate desk rejects with the ACs, and chair the discussions.
- Papers Chairs / PC — own the overall process and the final accept list.
Two kinds of early rejection
CHI screens before full review, and since the 2026 cycle this screening has teeth:
- Desk reject (DR): format and compliance — wrong template, anonymization failure, unjustified length (the >12,000-word rule), policy violations.
- Assisted desk reject (ADR): a rubric-based judgment call made jointly by SCs and ACs, introduced for CHI 2026 and codified by the CHI Steering Committee. The four posted rubric grounds:
| Rubric code | Meaning |
|---|---|
| ADR-Context | Grossly insufficient literature review to contextualize the contribution |
| ADR-Contribution | Disproportionately small HCI contribution for the paper's length |
| ADR-Data | Grossly insufficient data to support the claims |
| ADR-Method | Grossly insufficient methodological detail, conceptual clarity, or research transparency |
At CHI 2026, subcommittees desk-rejected between 1.4% and 17.9% of their submissions
and assisted-desk-rejected between 1.1% and 16.6%, with ADR-Context the most common
rubric ground — weak related work is now a pre-review rejection risk, not just a
score penalty (chi-related-work).
Round 1: the recommendation scale
Reviewers and ACs recommend on a five-point scale:
- A — Accept
- ARR — Accept with Required Revisions
- RR — Revise and Resubmit
- RRX — Revise and Resubmit / Reject (leaning reject)
- X — Reject
No paper is accepted in round 1. The posted threshold: a paper earning at least one recommendation of RR or better (A, ARR, or RR — not RRX or X) from the 1AC or the 2AC is invited to revise and resubmit. External reviewers' scores inform but do not trigger the invitation; the AC pair holds the keys.
Round 2: revision, then the PC meeting
Invited authors get about five weeks to submit a tracked-changes revision plus a
response to the reviews (mechanics in chi-author-response). The same review team
re-reads, and the PC meeting settles accepts and rejects. The process page warns that
as many as 50% of revised papers may still be rejected; the realized CHI 2026 numbers:
CHI 2026 Papers funnel (posted outcome reports, chi2026.acm.org):
6,730 completed submissions
→ 2,603 invited to revise and resubmit (38.7% of submissions survived round 1)
→ 2,576 actually resubmitted (27 withdrawn/lapsed)
→ 1,705 conditionally accepted (65.5% of resubmissions; 25.3% overall)
Read your round-1 packet against this funnel: an RR invitation means you are in the top ~39%, with roughly two-in-three odds that depend heavily on revision quality.
Reading the reviews you receive
- Weigh the 1AC meta-review above everything; it frames the PC-meeting discussion and tells you which criticisms the committee actually owns.
- An RRX from an external with RR from the ACs is survivable; the inverse pattern (ACs at RRX/X) means the invitation, if it comes at all, is fragile.
- Contradictory externals are common at CHI because subcommittees mix methodological cultures; resolve contradictions by aligning with the meta-review, and flag the contradiction politely in the response rather than picking a side silently.
- "The contribution is unclear" from multiple reviewers is a framing defect you can fix in five weeks; "the study cannot support these claims" usually is not — decide early whether to revise seriously or withdraw and redo the work for next cycle.
Confidentiality and conduct
Submissions are confidential to the review team; contacting reviewers or ACs about a paper outside PCS channels violates the process. Reviewer anonymity is permanent. Public complaints during the cycle can only hurt — escalation happens through the subcommittee chairs and Papers Chairs, in writing, via PCS or the posted contacts.
Output format
[Stage] screening / round-1 / RnR window / round-2 / decided
[Screening exposure] DR: <format risks> · ADR: <weakest rubric ground>
[Round-1 read] 1AC: <score> 2AC: <score> ext: <scores> → invitation likely? yes/no
[Meta-review center of gravity] <the criticism the PC will discuss>
[Realistic odds] <calibrated vs the ~65% resubmission acceptance base rate>
[Action] <revise seriously / fix framing / withdraw and re-route>