City Streets

Street networks as probability measures on ℝP² — orientation entropy in 3D, Wasserstein distances between 84 world cities, hilliness, walkability, and staircase censuses.

Every city’s streets define a probability measure over directions in 3D — compass bearing crossed with grade. City Streets extends Boeing’s street-network orientation entropy to the projective plane \(\mathbb{RP}^2\), computes rotation-aligned Wasserstein and Jensen–Shannon distances between city measures, and derives comparable metrics for hilliness, walkability, staircase-street censuses, and street “character” across 84 world cities.

Open the interactive dashboard →

Highlights:

• A sortable master table of ~30 comparable metrics per city
• Per-city pages: the direction measure on the sphere, bearing roses, hill fans, street-character breakdowns, and a region map of the analyzed urban core
• A distance heatmap + MDS embedding: which cities are built alike, up to rotation
• Staircase censuses done honestly — union-find connected components, a ≥30 m “major staircase” rule applied identically on every continent