brycewang-stanford OOPSLA Supplementary Material
The 23-page cap on an OOPSLA initial submission excludes references, required statements, and appendices — so "supplementary" at OOPSLA is less a separate upload category than a layering decision: what must live in the reviewed body, what sits in appendices the body can survive without, and what waits for the artifact stage after acceptance. The controlling rule set is the current call (待核实 per cycle for upload mechanics on HotCRP); the controlling strategy is the two-round review model.
The layering rule
A reviewer deciding among Accept, Minor Revision, Major Revision, and Reject must be able to reach a verdict from the body alone. Appendices exist to let a skeptical reviewer verify, not to let the body skip its argument.
| Layer | Belongs there | Test |
|---|---|---|
| Body (≤23 pp) | Claim statements, mechanism, proof sketches, headline evaluation, threats | Could a reviewer bucket the paper reading only this? |
| Appendix / supplement | Full proofs, extra benchmarks, case-study transcripts, calculus metatheory details | Does the body state what the appendix establishes? |
| Artifact (post-acceptance) | Runnable code, raw data, mechanization sources | Referenced by the Data-Availability Statement, not needed for the verdict |
Anonymity applies to every byte
The supplement is reviewed under the same double-anonymous rules as the PDF:
- Strip repository remotes, commit author lines, CI configs, and
~/-rooted paths from any code archive; check binary blobs (notebooks, PDFs) for embedded author metadata. - Anonymized-archive services or a scrubbed zip both work; a "private repo we can add reviewers to" does not — access events identify reviewers, which breaks the process from the other direction.
- Rename internal project codenames that are Googleable to your group.
Consistency: the silent killer across rounds
Because a Major Revision returns to the same reviewers next round, any mismatch between body and supplement is remembered. Before each round's upload:
Cross-check pass:
numbers every value quoted in body == value in appendix tables (script it)
theorems numbering and side conditions identical body vs full proof
configs flags/versions in appendix match the reproducibility ledger
claims nothing in the supplement contradicts a scoped claim in the body
pointers every "see Appendix X" resolves; no orphan appendices
What NOT to put in the supplement
- Rescue material — text that exists because the body overran 23 pages.
Compress the body instead (
oopsla-writing-style); reviewers are not obliged to read overflow prose, and relying on it undermines the bucket decision. - Unfinished experiments flagged "preliminary." In a revision-capable venue, half-results invite a Major Revision demanding the full version; either finish them or cut them and scope the claims down.
- De-anonymizing generosity — screenshots of dashboards, grant-numbered cluster names, or user-study consent forms bearing the institution.
Revision-round dynamics
An invited revision gets 25 pages: the extra room is for absorbing what
reviewers demanded, which often means promoting appendix material into the
body. Structure appendices from day one as promotable units (self-contained,
labeled, with their own claim sentences) so a Minor Revision is an editing
job rather than a rewrite (oopsla-review-process).
Output format
[Layering] body-sufficient verdict possible: yes/no + misplaced items
[Anonymity] supplement scrub: clean / findings
[Consistency] cross-check pass results (numbers/theorems/configs/claims/pointers)
[Promotable units] appendices ready to absorb into a revision: <list>
[Cut list] material that should not ship at all