FAccT Supplementary

Use this when assembling FAccT supplementary material. Two rules govern the split. First, the familiar one: the paper must be judgeable from the reviewed pages alone — appendices and released artifacts support the paper, they do not hold the argument. Second, the FAccT-specific one: the endmatter statements (Generative AI Usage, Ethical Considerations, Adverse Impacts, Positionality) are a distinct zone with their own page allowance and their own anonymity rules — get that zone right or risk a policy problem independent of the paper's quality.

The three zones

What goes where

If a reviewer would need to open the appendix to know whether the harm is real or the method is sound, the paper is mis-partitioned — move that evidence into the body.

The endmatter zone (FAccT-specific, easy to get wrong)

[Generative AI Usage]  REQUIRED for every submission: state whether/how generative AI was used;
                       FAccT prohibits wholesale AI-generated manuscript text
[Ethical Considerations] the concerns you faced and how you addressed them (safe to include anonymized)
[Adverse Impacts]      foreseeable unintended harms once published/deployed, and mitigations
[Positionality]        how the authors' background shapes the work — IDENTITY-REVEALING, so add ONLY
                       upon acceptance; keep it OUT of the anonymous submission
[Author Contributions / Acknowledgements / Competing Interests] also identity-revealing -> acceptance only

The extra endmatter page does not count against the 14-page limit, so a strong ethics/adverse- impacts treatment costs you nothing in body space — write it properly rather than cramming a sentence into the conclusion.

Mutually-anonymous supplementary rules

[No identity]   no authors, affiliations, acknowledgements, funding, cluster paths, or repo owners
                anywhere in the appendix, artifact, or the statements you DO include at submission
[No positionality] the positionality/acks/contributions statements are withheld until acceptance
[Anonymized links] any released code/data link points at an anonymizing host, not a personal repo
[Clean archive]  no .git history, .DS_Store, credentials, caches, or participant-identifying files
[Opens clean]    verify the archive unzips and a README orients a reader in one minute

Vignette: splitting a mixed-methods paper

A paper auditing a system plus interviewing affected users: the body keeps the contribution, the study design, the headline disaggregated results with uncertainty, the key qualitative themes, and a limitations subsection; the endmatter carries the required Generative AI Usage Statement, an Ethical Considerations statement about the consent and re-harm risks, and an Adverse Impacts statement about how the audit could be misused — with Positionality held for the accepted version. The appendix/artifact holds the analysis code, the datasheet, the full codebook, and the interview protocol. Nothing decision-critical lives only in the appendix.

Output format

[Supplement status] ready / needs fixes / not ready
[Partition check] anything decision-critical outside the body? <none / move: what>
[Page budget] body within limit? endmatter on its allowed extra page? yes/no
[Endmatter] <Generative AI Usage present? ethics/adverse-impacts drafted? positionality withheld?>
[Anonymity] archive + included statements clean of identity + metadata? passed/issues