Reproducibility bundle for:
"Institutional Pressures and Organizational Compliance with Platform Mandates:
Longitudinal Evidence from the Adoption of Email Authentication" (Davison, 2026)
https://spamcipher.com/insights/platform-mandates-email-authentication

Pipeline (numbered scripts, run in order):
  00_download_extract.py   one OpenINTEL tranco-toplist snapshot per month -> monthly/*.parquet
                           (raw snapshots are NOT included: ~1 GB; the script re-downloads them
                           from the public OpenINTEL archive, data agreement cookie required)
  01_build_panel.py        cohort, domain-month panel, ITS series (panel/its_series.csv included)
  02_categories.py         Cloudflare domain-intel labels used for VALIDATION only
                           (set CF_TOKEN and CF_ACCOUNT_ID; service enforces a usage quota)
  10_homepages.py          homepage title/description for the systematic sample (output included)
  11_llm_categories.py     LLM classification (OpenRouter, qwen-2.5-72b-instruct, temperature 0;
                           set OPENROUTER_API_KEY; prompt embedded in script; output included)
  08_firmlist.py           index constituents -> sectors -> domains via Wikidata (output included)
  03_its.py                H1 interrupted time series + Figs 2-3
  04_event_history.py      H3 full-cohort event history (+ stock-prevalence contrast with --with-categories)
  13_h4_recent.py          H4 preferred flow specification
  05_decoupling.py         H5 entry and progression models
  06_h2_stratified.py      H2 difference-series test + Fig 4
  12_firmlevel_h2.py       firm-level ceiling analysis
  07_robustness.py         transition-exclusion, placebo, equal-follow-up checks
  09_refs_check.py         citation cross-check and reference generation

Included data: panel/its_series.csv, panel/cohort.parquet, panel/cohort_domains.txt,
panel/homepages.csv, panel/categories_llm.csv, panel/firmlist.csv, results_*.json,
tranco list identifier 94VW2 (list at tranco-list.eu/list/94VW2).
Excluded (size): raw monthly snapshots and the full domain-month panel.parquet;
both regenerate deterministically from 00 and 01.

Data licenses: OpenINTEL data CC BY-NC-SA 4.0 (van Rijswijk-Deij et al. 2016);
Tranco (Le Pochat et al. 2019). Non-commercial research use.
