MoveNumbers uses mapped road attributes to estimate walking conditions. It is not a pedestrian-safety assessment, an accessibility audit, or a definitive ranking of American cities. Use it to identify questions about a neighborhood, then check the routes you would actually use.
Start with your daily trip, not a city rank
A citywide number cannot tell you whether you can comfortably walk from a particular home to a grocery store or bus stop. Two addresses in the same city can involve very different crossings and paths. The city explorer provides a broad starting point; the address tool provides a separate street-level estimate. Their numbers should not be treated as interchangeable.
What goes into the city model
The city-data builder in this project reads OpenStreetMap road tags. It starts with a road-type score, then applies adjustments for mapped speed limits, lane counts, sidewalks, cycling infrastructure, traffic calming and lighting. It combines road scores using road-type weights: for example, a primary road has weight 3, a residential road 1, and a footway 0.3. Those are modeling choices, not percentages established by a pedestrian survey.
There is no 40/30/20/10 weighting of infrastructure, traffic, safety and destinations in this builder. It does not measure traffic noise, crime along a route, current construction, weather protection, crossing-signal timing or distance to everyday destinations. A higher result therefore does not certify that a route is safe, accessible or pleasant.
Why missing map information matters
The builder penalizes several road types when the sidewalk tag is missing. That is a limitation: an absent tag does not establish that a sidewalk is absent in the real world. OpenStreetMap also allows sidewalks to be mapped as separate paths, so one tagging convention does not capture every sidewalk.
Here is a reproducible example from the builder, keeping all other tags absent: a residential road with no sidewalk tag receives 20 out of 100; the same road with sidewalks tagged on both sides receives 65. This 45-point difference describes how the model reacts to its inputs. It does not demonstrate that either road was inspected or that its physical conditions changed.
The sidewalk, cycling and lighting percentages should be read as summaries of mapped attributes in the dataset, not as audited coverage of every street. Mapping detail varies. City selection and geographic boundaries also affect the result; the saved snapshot predates a correction to state-based geographic selection. The builder now limits city-name searches to a US state, but the existing city boundaries and scores have not been revalidated.
What the comparison does and does not cover
The explorer displays the number of cities in its bundled snapshot and the snapshot date. Its comparison average describes those included cities, not a representative national sample. The saved dataset has not been independently validated against on-the-ground walking conditions. The address-based API uses a separate scoring implementation, so a local score is not a direct replacement for the city score.
We cannot use these data to conclude that one city has America's best pedestrian infrastructure, or explain another city's score by noise or crime that the model does not measure. The earlier version of this article made those claims; they have been removed.
A practical route check before moving
- Choose actual destinations. List the grocery store, transit stop, school or park you would use, starting from the prospective address.
- Trace the complete route. Look for sidewalk gaps, major crossings, disconnected paths and detours. A nearby destination may still require a long route.
- Check current conditions. Compare recent imagery with local construction notices, transit information and accessibility needs. Where practical, walk the route at the times you expect to use it.
- Keep uncertainty visible. If the map lacks a tag or a route cannot be checked, treat that as missing information rather than a favorable or unfavorable finding.
These checks add context that the number cannot supply. You can then compare housing and other costs alongside your route requirements.
Sources and correction
Mapping source: OpenStreetMap contributors (ODbL). The OpenStreetMap sidewalk documentation explains sidewalk tags and separately mapped paths.
Corrected September 30, 2026 after reviewing the city-data builder and address-scoring implementation. Removed unsupported claims about field audits, professional consultation, category weights and national rankings. This correction describes the model's limitations; it does not independently validate the saved scores.