Code, data, and figures for the paper Fan-In Distributions in Human-Written vs AI-Generated Python Codebases: A Constructal Law Analysis (paper, all versions: 10.5281/zenodo.20313668).v1.1.0 accompanies paper v1.1, which corrects eight errors in v1.0.0, extends the longitudinal pilot from two repos to all eight Cohort A repos that meet its inclusion criteria, and adds a post-release audit. The main cross-sectional result, lower fan-in Gini in Cohort B, holds: 10 of the 15 Cohort A snapshots contain a declared AI signal (421 of 221,145 commits in 9 repos carry an attribution, and a tenth, Django, has an AI configuration file), and with those ten removed Cohort A Gini remains higher (exact Mann-Whitney p=0.027, n=5 vs. 12), while entropy, DAIC and leaf fraction lose significance. The longitudinal pilot finds no Gini decline after adoption in eight repos (Gini rose in six; pooled exact Wilcoxon p=0.195).python scripts/check_paper_claims.py recomputes every number in the paper's Version Note, Sections 5 and 6, and Section 7.7, and exits nonzero on any mismatch. Code is MIT-licensed; paper text and figures are CC BY 4.0 (LICENSE-PAPER.txt).
fan-in · constructal law · software metrics · Gini coefficient · log-normal distribution · AI-generated code · Python · software architecture · AI attribution · reproducibility