Data licence — CC BY-SA 4.0, except the project's own writing named below (CC BY 4.0) This licence covers the data in this repository and every file it is published as: data/, harvest/, f1.db, f1.db.gz (f1.db compressed, as lapledger.org serves it), f1-parquet.zip (every table of f1.db as Parquet), f1_database.json (released compressed, as f1_database.json.gz), f1_compat.json, and the JSON files lapledger.org serves under /api/v1/ (the rows of f1.db, one file per driver, constructor, circuit, season and race). The published files are copies of f1.db in other formats and carry its terms. It does not cover the code (see LICENSE), nor f1-geometry.db, whose terms are under *Circuit geometry* below, nor f1-facts.db, which is CC BY 4.0 whole, under *The facts artefact* below. The views.sql in f1-parquet.zip is the views' SQL as schema.sql writes it, and like schema.sql it is also available under LICENSE (MIT). Copyright (c) 2026 Alex Farley and Wikipedia contributors. The data is licensed under the Creative Commons Attribution-ShareAlike 4.0 International License (CC BY-SA 4.0), except for the columns named under *The project's own writing* below, which are CC BY 4.0: https://creativecommons.org/licenses/by-sa/4.0/ Full legal code: https://creativecommons.org/licenses/by-sa/4.0/legalcode ## Why CC BY-SA Bare facts — who won a race, how long a circuit is — are not copyrightable, and a database of those facts would be free to license under anything. But this database also contains prose taken from or closely following Wikipedia articles: the car design histories, circuit descriptions, and some notes. Wikipedia text is licensed CC BY-SA 4.0, and that prose carries the same attribution and share-alike requirements with it. Rather than draw a line field-by-field, the whole data release is licensed CC BY-SA 4.0 — with the one exception in the next section, where the line is drawn because nothing of Wikipedia's was found on the other side of it. f1-facts.db is the line drawn table by table and column by column, as a file of its own (*The facts artefact*, below). See ATTRIBUTION.md for the source of every part of the data and the reasoning in full. ## The project's own writing — CC BY 4.0 These columns hold writing that came from nowhere else. They are offered under the Creative Commons Attribution 4.0 International License (CC BY 4.0): https://creativecommons.org/licenses/by/4.0/ - `discrepancies.assessment` — this project's reading of each disagreement between two sources: which figure it takes, and why. - `known_gaps.area`, `known_gaps.reader`, `known_gaps.description` and `known_gaps.resolution` — what a reader is shown about each gap, the maintainer's note on it, and what would close it. - The prose of ten of the twelve reference tables `source_registry` entry 18 records as written for this project (PM-49): `eras.summary` and `eras.defining_features`, `engine_eras.notes`, `governance.detail` and `governance.significance`, `safety_milestones.trigger_event` and `safety_milestones.description`, `points_systems.notes`, `tyre_suppliers.notes`, `constructor_lineage.note`, `grands_prix.notes`, `personnel.significance` and `engine_manufacturers.notes`. Attribute Lap Ledger; there is no share-alike, so nothing you build on them takes a licence from them. The share-alike above comes from Wikipedia, and none of Wikipedia's text is in these columns. The first five were written for this project, about this project's own database. The ten tables' prose was measured field by field against the Wikipedia articles a writer would have reached for — `docs/prose_pass.tsv`, the prose pass — and every field in them reads as this project's own expression. The remaining columns of the tables they sit in — the subject and field of a disagreement, the two values, the state of a gap, an era's name and dates, a points scale, a grand prix's aliases — hold facts, identifiers and headings rather than expression. The grant is stated inside the database too, as `meta.project_prose` and `meta.project_prose_columns`, so a copy of `f1.db` carries its own terms; a build in which those disagree with this file fails. The Parquet bundle carries the same two rows in `meta.parquet`, and a `README.txt` that states them. A CC BY grant cannot be withdrawn from a copy already taken, which is why the list is exactly this long, and why a granted field of the ten tables that is rewritten fails the build until the prose pass has read it again. **Two of entry 18's tables are not in it.** `technical_innovations` stays CC BY-SA because the prose pass reads one of its descriptions as following a Wikipedia article's wording, and it stays so until that field is rewritten. The glossary stays CC BY-SA while whether to grant it is an open question (PM-67): every definition in it reads as original, and the grant waits on a decision rather than on the evidence. ## The facts artefact — f1-facts.db, CC BY 4.0 f1-facts.db is f1.db less everything that stays share-alike. It is published as a release asset beside f1.db and f1-geometry.db, from the first release after v2.25, and it is offered whole under the Creative Commons Attribution 4.0 International License (CC BY 4.0), with no share-alike: https://creativecommons.org/licenses/by/4.0/ What it holds: the rows that cite F1DB by Marcel Overdijk, published under CC BY 4.0 — among them the races, the full classification, qualifying, practice, the sprints, standings, pit stops, the driver, constructor, chassis, engine and entrant registers, and F1DB's circuit outlines; the rows that cite an official source for a fact — the FIA, formula1.com, a team, a circuit or a driver — which hold the fact and none of the source's prose; and this project's own writing granted CC BY 4.0 above. What it leaves out, and why — everything here that would carry a share-alike, or that this project has not granted: - What was taken from Wikipedia (CC BY-SA 4.0): the car specifications and design histories, the circuit layouts, the claims and driver note sources that cite an article, and the six team radio quotations. - The prose fields that stay CC BY-SA 4.0, some of them following a Wikipedia article: the notes and descriptions on circuits, drivers, constructors, seasons, races, regulations and qualifying formats, which the prose pass measures and neither grant above reaches. - The register of Wikimedia Commons photographs, each of which carries its own licence, several of them share-alike. - This project's own writing that neither grant above reaches: the glossary, `technical_innovations`, and the source registry's account of each source, among others. - The values a recorded disagreement compares, and its status note: the values come from whichever sources it compares, often an article, and the note is not among the writing granted above. Its assessment is carried, wherever the column it is about is carried. - The circuit centrelines, which are ODbL and never in f1.db either, and the four tables of FOM-owned timing, which are empty in f1.db too. The race rows f1.db takes from Wikipedia's season articles are not left out: they are rebuilt from F1DB, so that nothing Wikipedia made is in the file, and some of its values therefore differ from f1.db's. Its source registry gives this project's own source no share-alike, since the only writing of this project's it holds is what the grants above cover. The file's own `meta` table states its terms and says where it differs, and tools/facts_artefact.py in the repository names every table and column it leaves out, with the reason for each. ATTRIBUTION.md gives the credit it carries. f1.db itself stays CC BY-SA 4.0 as above. Nothing share-alike is in f1-facts.db, so taking it takes none of f1.db's share-alike with it. Inside f1.db, the `column_origin` table says where each column's values come from, as far as tools/facts_artefact.py's declaration establishes it: another column of the same row, the claims, this project's text that the prose pass labels against Wikipedia, this project's own record (a citation, a confidence tier, an identifier, a state, a classification, a link, its own writing outside the prose pass, or its account of itself), a figure or record build.py computes from other rows, or not established. It records origins only. The terms a source's values come under are the source registry's; the table grants nothing, and changes nothing above. It is not in f1-facts.db, some of whose columns are rebuilt or left out. ## Attribution Car and circuit data in this repository, each race's winner, pole position and fastest lap, some driver records and six team radio quotations are derived from Wikipedia, licensed under CC BY-SA 4.0. Most driver and constructor records, the full race classification, sprint results, practice, qualifying and sprint qualifying, championship standings, pit stops, and the chassis, engine and season-entrant register are derived from F1DB by Marcel Overdijk, published under CC BY 4.0, as are F1DB's circuit outlines, drawn by Jules Roy. Circuit centrelines, published separately as `f1-geometry.db`, are © OpenStreetMap contributors, licensed under ODbL 1.0; their relation ids come from Wikidata (CC0). Car, circuit and race photographs are hosted on Wikimedia Commons, and each carries its own licence and credit, recorded per file in `article_images`. Current-season data, and the other driver and constructor records, are from formula1.com. Formula 1, F1 and Grand Prix are trademarks of Formula One Licensing BV; this project is unaffiliated with and unendorsed by Formula One or the FIA. ## Circuit geometry The circuit centrelines traced from OpenStreetMap are **ODbL 1.0**, which is share-alike and carries a database right reaching the whole database its data lands in. They are therefore **not in `f1.db`** and not covered by this licence: they are published as `f1-geometry.db` alongside it, under ODbL, (c) OpenStreetMap contributors. Two databases distributed side by side are a Collective Database, which ODbL does not treat as derivative. See ATTRIBUTION.md and tools/geometry_overlay.py. Part of the history predates that split. **Every commit whose `f1.db` has rows in `circuit_geometry`** — thirty-five of them, in ten distinct versions of the file, dated 2026-09-05 to 2026-09-09 and all ancestors of `6610b6da` ("Merge the licence enforcement and the ODbL split") — carries the twenty-five centrelines inside `f1.db` itself, which makes each of those files a Derivative Database. They are offered as such: **ODbL 1.0, (c) OpenStreetMap contributors**, **in addition to** the terms above rather than in place of them. Share-alike runs with the Wikipedia-derived rows and this project's own prose in the same file whatever else that file acquires, so both obligations apply together and neither displaces the other. The boundary is that merge and not the split commit `5a70558`, which was parallel work: eight of the thirty-five are later than it by the clock and reached the trunk only when it was merged. No commit from `6610b6da` onward has a row in `circuit_geometry` in `f1.db`. The historical files are kept rather than rewritten out, because the history of a licence decision is part of the record. ## Live timing data tools/fastf1_load.py optionally loads per-lap timing, stints, pit stops, race control messages and team radio from the Formula 1 live timing API via FastF1. That data belongs to Formula One Management and is not covered by this licence, is not redistributed, and is not committed to this repository — the relevant tables are empty in the committed database and are filled only locally when you run the loader yourself. `verify.py` enforces this rather than trusting it; see ATTRIBUTION.md for what is checked and for the two tables where the rule is by source rather than by emptiness.