Can you check https://safestreets.streetsandcommons.com/ it works fine now.

@pedrito1414 @Bauer33333 thanks both, glad it’s of interest.

It’s live and free to try at safestreets.streetsandcommons.com. Type any address and you get a 0 to 10 score.

Hi solenoid jam, apologies for the late reply. I missed your comment and only saw it today.

Three months later I’m now based in Chiang Mai myself, so several of your observations are visible to me every day.

On Western-model bias. Honest acknowledgment: the 15-minute city framework I use is Carlos Moreno’s European concept, and the seven-category service mix reflects North American and European zoning logic. In Chiang Mai my daily errands happen at a local market that isn’t reliably tagged amenity=marketplace, on sois that don’t fit the highway taxonomy cleanly. The tool understates Asian contexts by default, and I should be more transparent about that in the UI.

On pedestrian status and motorbike behavior. The tool captures crossing density but doesn’t model driver-yield culture or motorbike-honk patterns, both of which materially affect safety. The score currently treats “crossing exists” as a positive when in practice the crossing may be unobserved by drivers. I don’t have a clean answer for how to tag that yet.

On marketplaces and narrow roads: Agreed both are tagging gaps. amenity=marketplace exists but doesn’t capture stall density or internal function the way Asian markets actually work. Sois, kampungs, and lanes don’t fit highway=living_street or highway=track cleanly.

On transit as precondition: You’re right that pedestrian-only investments without transit are partial in SEA. SafeStreets has a Transit Reach component (15% of the composite) but it weights all addresses on the same global scale, which understates how transit-dependent an Asian address is.

On density as the primary Asian signal: Worth surfacing as a methodology note, in SE Asia, density does most of the work that pedestrian infrastructure does in the West. A Chiang Mai shophouse block has destinations within 50 to 100 meters where a North American suburban block has them at 800 meters or more. SafeStreets weights this in Daily Reach (50% of the composite), so dense Asian addresses score reasonably well on the “is it walkable enough to use” dimension. The gap is on the “is it safe to walk on” dimension, where OSM tagging coverage is weaker and where the cultural factors you describe matter most. The composite score blends the two when it should separate them.

Thank you for the substantive critique.