Code, results, and figures for an anonymized paper under review.
The base rate of fairness practice in 5,444 sensitive-context data-science programs, 2015-2026.
- 0 fairness-toolkit imports in 2,762 modeling programs; 2 genuine group-disaggregated checks after hand-auditing all 13 machine flags
- 77% of protected-touching programs encode sex/race/age into model features; 54% without a single group-wise look
- 743 unchecked protected-touching programs have 1.24M cumulative views and 101 medals
- R-language slice (n=160, 2015-2022): same shape, 0 fairness-package imports
Every number and figure in the paper is produced by a script in code/
from the files in results/. Hand-audit tables (CSV) sit beside the
machine outputs they correct.
code/- numbered, self-documenting pipeline scripts (each docstring states the plain-language question it answers)results/- every result file the paper cites, including audit tables and detector precision/recall recordsfigures/- the paper's figures as generatedEXPERIMENT_LOG.zh.md- the running lab log (Chinese), kept verbatim for provenance
Meta Kaggle and Meta Kaggle Code are public Kaggle datasets; scripts fetch per-kernel files on demand (needs your own kaggle.json API token in exp-data/, never committed). Large tables are re-derived, not shipped.
Python 3.8+; scripts list their own imports. Run scripts in numeric
order; every script is resumable and writes into results/.