brycewang-stanford CSCW Supplementary Materials
Supplements at CSCW serve a journal-style review: they let reviewers verify the methods section without bloating a paper already policed by the contribution-vs- length standard. Because Revise and Resubmit sends the paper back to the same reviewers, the supplement is also a versioned record — round-two reviewers can and do compare it against round one.
What goes where
| Material | Main paper | Appendix | Separate supplement file |
|---|---|---|---|
| Method summary + key instrument excerpts | ✓ | ||
| Full interview guide / survey instrument | ✓ | ✓ if long | |
| Codebook (definitions + paraphrased exemplars) | ✓ | ✓ | |
| Trace-pipeline ledger and attrition table | summary | ✓ | code as archive |
| Analysis scripts | ✓ (archive) | ||
| Additional tables/robustness checks | ✓ | ✓ | |
| Consent forms / recruitment text | ✓ | ||
| Raw transcripts, identifiable data | never | never | never |
Decision rule: anything a reviewer needs to evaluate a claim belongs in the paper or appendix; anything a reviewer needs to trust the method belongs in the supplement; anything participants did not consent to share belongs nowhere.
Anonymization is deeper than the PDF
The supplement is the classic identity leak at this venue, because it aggregates exactly the artifacts teams forget to scrub:
- Documents: consent forms carry institution letterhead and IRB protocol numbers tied to a university; recruitment emails carry names. Redact, and check PDF/Office metadata (author fields, tracked changes, comments).
- Code archives: repository paths (
/home/username/), git history, package lockfiles with private registry URLs, notebook execution metadata. Export clean archives, never a cloned.git. - Data extracts: usernames, community names, timestamps precise enough to reverse-search, URL slugs. For small communities, the combination of quoted text and date can identify the community even with names removed.
- Media: screenshots showing account handles, browser profiles, or platform locale details.
Mechanical sweep before upload:
# strings that must not appear anywhere in the supplement tree
grep -ri -l -E "(surname|firstname|university-name|lab-name|grant-no|protocol-no)" supplement/
# metadata in every PDF
for f in supplement/**/*.pdf; do exiftool "$f" | grep -iE "author|creator|company"; done
# accidental git history or user paths in code archives
find supplement/ -name ".git" -o -path "*Users*" -o -path "*home*" | head
Adapt the string list per project; run it on the packed archive, since packing often reintroduces metadata.
R&R versioning discipline
- Keep every round's supplement frozen; name uploads by round
(
supplement-r2.zip), never overwrite. - If the response letter cites the supplement ("the revised codebook now includes ..."), the cited file must exist under that name — reviewers check.
- New supplementary material added at R&R must be flagged in the change summary; silently appearing files read as scope creep or, worse, as identity leaks made under revision pressure.
Pre-upload checklist
[Inventory] every supplement file listed in the paper or response letter? y/n
[Consent] each artifact within participants' consent scope? y/n
[Sweep] mechanical anonymity sweep run on the packed archive? y/n
[Size/format] platform limits confirmed on the live submission system? 待核实 per
regime (PCS legacy vs Manuscript Central) — do not assume
[Round label] files named by round; prior rounds preserved? y/n
Platform-specific size and format caps were not verifiable at 2026-07-08; check the live Manuscript Central instructions for the rolling pathway before packaging.