# tech-market data: how we count (plain words)
1. `companies.txt` lists the company slugs we tried. `collect.py` downloads each company's public job board (Greenhouse, then Lever, then Ashby), the Hacker News "Who is hiring?" thread and the RemoteOK feed, and writes one row per posting to `postings.jsonl` (no job descriptions kept, only title, company, region, flags).
2. `hn_trend.py` counts 13 monthly Hacker News threads into `hn_trend.json`.
3. `analyze.py` computes every share in `results.json`; `charts.py` draws every image from it. No number is typed by hand.
4. `PREREG.md` holds the questions, definitions and what would prove us wrong, written before the final count (with a disclosed first peek).
Run order: collect.py, hn_trend.py, analyze.py, charts.py (needs matplotlib). $0, no login, no paid API.