Machine-relevant detail in OpenStreetMap, Europe 2016 to 2026 ============================================================= How much of the detail that cars, delivery robots, drones and fleets read from a map is present in OpenStreetMap in Europe, and how fast it grew. Article: https://mapatlas.eu/blog/maps-for-machines Data date: 2026-10-02 Rows: 39 (15 tags for Europe, 3 tags for each of 8 city centres) Files ----- maps-for-machines-2026-10-02.csv (UTF-8, comma separated, header row, empty cell = no value) Columns ------- scope "Europe" (Geofabrik Europe extract) or a city with a 4 km circle around its centre measure "objects with tag" (Europe rows) or "share of car-road length with tag" (city rows) tag OpenStreetMap key or key=value. For city rows: maxspeed, lanes, or sidewalk (any of sidewalk, sidewalk:both, sidewalk:left, sidewalk:right) count_YYYY-01-01 Europe rows: OSM objects (nodes, ways, relations) carrying the tag on 1 January of that year count_2026-09-20 Europe rows: the same on 20 September 2026, the last day in the taginfo chronology multiple_2016_to_latest count_2026-09-20 divided by count_2016-01-01 road_km_2016, road_km_2026 City rows: length of car roads in the circle (motorway to residential, living_street and links; dual carriageways count twice), km share_2016-01-01, share_2026-07-01 City rows: share of that length carrying the tag (0 to 1) Method ------ Europe: taginfo chronology of the Geofabrik Europe extract (https://taginfo.geofabrik.de/europe), which lists per day the net number of objects that gained the key or tag. The running sum on each date is the count. It matches taginfo's current key statistics within 1% for all tags except sidewalk:both (4%), because the statistics run 11 days later and that tag grows fast. Cities: ohsome API (https://api.ohsome.org/v1, HeiGIT, OSM full history), elements/length/ratio, 4 km circles around the city centre. ohsome data ended 2026-07-27 at the time of the run, so 2026-07-01 is the latest date. Script: scripts/seo/blog-keyword-research/maps-for-machines.mjs What it does not measure: whether a value is correct today, attributes that software infers from defaults (for example a national urban speed limit), or the quality of the geometry. Licence ------- Derived from OpenStreetMap data, (c) OpenStreetMap contributors, Open Database License (ODbL) 1.0. Our calculations CC BY 4.0, MapAtlas.