What it means to build this

The Manhattan corridor work is a build in the civic sense: a reproducible measured graph, a documented order of operations, and route pairs people can audit. The executive summary deck states claims and figures; this page holds the longer answers to why that build shape matters.

What “build” names here

Building is not shipping a consumer app first. It is fixing a public graph, populating fields from open tables and public cameras, locking stress topology before routing runs, and publishing corridor-scale evidence others can re-run. Product routing may follow; the work is the public measured graph and a clear boundary around what stays open.

Answers the program is testing

Public substrate before product: Open NYC tables and DOT cameras feed a stress field on a locked pedestrian graph. Partners can inspect inputs and pipeline order before any proprietary fusion or routing cost ships inside a navigation product.
Auditable order of operations: Stress topology is decided and documented before path algorithms run. That order is the governance hook: the model cannot silently rewrite the field to justify a route.
Bounded, reproducible claims: One corridor (~300 nodes), one snapshot, hop-equivalent route pairs compared on measured exposure. Claims stay at corridor scale so partners, students, and city programs can inherit the same object.
Room to choose how we walk: Hop-equivalent shortest paths and stress-aware alternatives on the same graph make the between-destination walk a civic choice. The build adds measured options where hop-minimum defaults hide crowding and uneven capacity.
Shared maintenance when the city changes: Sheds, closures, cameras and graph edits require a refresh story. Building includes who signs off, who re-runs the snapshot, and what evidence is enough for the next release.
Responsible AI city (event framing): “Building the Responsible AI City” is the NY Tech Week panel context. The corridor demo is one answer: civic AI that stays public-good at the input layer, bounded at the claim layer, and explicit at the commercial boundary (open tables stay open; fusion and calibration stay proprietary).

Where each answer lives in the deck

The executive summary states claims and figures in plain language on program, problem, pilot, and corridor-response slides. The discussion grid poses open questions (governance, geospatial platform, city pilot, public funding, workforce). The closing What it means to build this way slide summarizes five through-lines before the materials links.

For methods depth and topology vocabulary, see graph-field & corridor modeling. For representation and media, see media, environment, and routing.

Executive summary (slides) Routing demo All public pages aimez home