Data and licence
Everything Constitut publishes is plain text under version control. The record can be cloned, verified, or forked with its history intact — no database is the source of truth.
What is published
- The rules —
protocol/scoring-rules/: the versioned scoring law, and its changelog. - The taxonomy —
protocol/categories.yaml: categories, tiers, and geographies with permanent codes. - The rankings —
rankings/{geo}/{category}/{tier}/{year}/{period}.json: each computed table. - The evidence — every scored fact with its public source and a web-archive copy.
- The registry — award records, extraordinary admissions, and the public key chain.
- The corrections — the objection log and every resolution.
No runtime database holds anything of record. The data is the repository, and the repository is the institution — so any third party can recompute the tables from the same inputs and reach the same result.
Machine-readable endpoints
| /sitemap.xml | Every published page. |
|---|---|
| /rss.xml | Movement feed — editions, rules changes, records. |
| /llms.txt | A plain-text description for agents and language models. |
The dataset format
A ranking edition is delivered as one JSON file per country per quarter: each firm carries its evidence and its computed standing (rank, score, confidence). Once the engine verifies them, the same file records the reproducible edition.
Licence
The dataset is licensed Creative Commons Attribution 4.0 International (CC BY 4.0). You may share and adapt it, including commercially, provided you give attribution to Constitut and link to the licence. Attribution does not imply endorsement.
Current edition: snapshot/2026-Q2 · rules v1.1 · seal recomputable from the same files.