Flatten SF is a single-page web tool that computes walking and cycling routes across San Francisco, letting you slide between the shortest path and the flattest one. The entire computation runs client-side in your browser over 160,000 street segments derived from the Overture/OpenStreetMap street graph and USGS 1-meter lidar elevation data. There is no server round-trip. The data is baked into the page. The core mechanic is a Pareto frontier slider. Every route on the slider is Pareto-optimal: nothing else beats it on both distance and cumulative elevation gain simultaneously. Slide left for shortest, slide right for flattest, and the tool traces the full trade-off curve between them. The rightmost route treats a foot of climb as equivalent to 200 feet of horizontal walking — beyond that threshold, the routes stop being practical routes. Sliding right never shortens the path and never adds climbing. That constraint is load-bearing: it means the slider isn't just a UI gimmick but a faithful traversal of the Pareto set. The elevation model matters. Climbing is cumulative gain, not net difference between endpoints. This is a critical distinction for anyone who has walked from the Mission to the Sunset — you can end at the same altitude and still have climbed 400 feet along the way. Stairways are included for pedestrians and excluded for cyclists, which is exactly right for a city where the Filbert Steps and the 16th Avenue Tiled Steps are real routing decisions. Place search works offline: street intersections, places, and addresses for San Francisco are built into the page. Drew Edwards, the creator, has published the source, data, and full analysis. This is a solo project with the entire city's topology embedded in a static page — no API keys, no usage limits, no telemetry visible. The design is minimal to the point of austerity. Two text fields, a slider, a map. Faint lines show the other Pareto-optimal routes you didn't pick. There is nothing to configure, nothing to sign up for, nothing to pay for. The tool does one thing and the thing it does is correct. What makes this interesting beyond utility is the legibility of the underlying algorithm. The Pareto frontier is a concept most technical users have encountered in optimization contexts, but rarely one you can feel in your legs. Flatten SF makes a mathematical abstraction physical: you slide, you walk, you learn what the trade-off actually costs in sweat. That feedback loop — abstract optimization rendered as bodily experience in a specific city — is the project's real contribution. The limitation is scope. This is San Francisco only, and the elevation data and street graph are static snapshots. There is no transit integration, no turn-by-turn, no multi-city ambition. It is a well-scoped tool that does not pretend to be a platform.
