# howsteep.org decisions dataset — 2026-09-10

One row per route the survey ever considered, with the reason it stands where it does. Companion to the climb
pages at https://howsteep.org (methodology: https://howsteep.org/methodology).

Files
- decisions.csv        every page ever published: adopted (ranked / annex), listed (held for a person), retired,
                       duplicate — 8996 rows
- never-paged.csv      routes that were measured and never became a page — 139524 rows
- never-measured.csv   routes proposed by discovery that no batch measured — 16633 rows
- codes.csv            the reason-code registry: code, meaning, the rule it applies
- decisions.gpkg       the lines (WGS84) of every row with geometry, one layer per stage
- geojson/             the same lines as gzipped GeoJSON, split by section / state to stay under 25 MB a file
- manifest.json        counts by stage and code, sha256 per file, what could not be located

Columns: route_id (section/slug@event), section (uk/us), state, run, stage, status_now, reason_code, rule_id,
decided_on, decided_by (machine / human), evidence (JSON of the numbers the rule read), supersedes,
geometry_ref, slug, name. Human verdicts carry the verdict word only.

Licence: Open Database License (ODbL) 1.0 — https://opendatacommons.org/licenses/odbl/1-0/
Attribution: "howsteep.org decisions dataset 2026-09-10, ODbL 1.0". Routes derive from OpenStreetMap
(© OpenStreetMap contributors, ODbL) and state / national road inventories (public domain / OGL); elevations from
USGS 3DEP and the UK national LiDAR programmes. Nothing from restricted sources is included.