Steer a town of fifty citizens with one broadcast

01 · Role in the system

What Jev does here

Each round sends one batched request: every citizen is a Choice among five actions, distinguished by role and personality, so the same announcement splits the town in surprising ways; the author reports the round completes in under a second. Results drive 50 citizens walking a 3D town, with per-citizen confidence indicators and a result chart to tune the next broadcast. Bring-your-own-key is handled carefully: a visitor key lives in sessionStorage for that tab only, is forwarded once by the app’s own backend, is never stored or logged, and a test asserts it is never returned in a response. Without a key the game stays playable on a clearly labelled deterministic simulation fallback, and the test suite covers scoring, fallback determinism, request validation, key isolation, rate limiting, and the exact five-action Jev schema.

Fifty choices in one request

The economics demo hiding inside a game: each round, all fifty citizens are a single batched Choice request — investigate, join, flee, warn, or ignore — differentiated by role and personality, and the author reports the round completing in under a second. The same broadcast splits the town in ways that are fun as a game and instructive as a measurement: identical input, fifty different calibrated answers, all visible on a result chart.

The engineering posture is unusually careful for a toy: a visitor's key lives in sessionStorage for that tab, is forwarded once by the app's backend, is never stored or logged, and a test asserts it is never returned in a response. Rate limiting and the exact five-action schema are also under test.

The honest simulation fallback

The hosted demo runs a labelled deterministic simulation so sharing costs nothing; real Jev decisions require the visitor's own key, and the record says which mode you are in. The six-round scoring makes it a toy of broadcast wording rather than a persistent simulation — a boundary the record draws itself.

As games-cluster evidence, it is the cleanest demonstration of fan-out economics in the index: fifty parallel typed judgments costing one request, with key isolation tests most production tools never write.

02 · Control boundary

Where Jev sits

One batched Choice request per round holds all fifty citizens; code renders decisions, enforces BYOK isolation, and falls back to a labelled simulation without a key.

Code owns the loop, permissions, thresholds, validation, and side effects. Jev owns only the bounded judgments described above.

03 · Known limits

What this evidence does not prove

• The hosted demo runs the built-in simulation so it costs nothing to share; real Jev decisions require the visitor to add their own key.
• The sub-second round figure is author-reported.
• Six-round scoring structure makes it a toy of broadcast wording rather than a persistent simulation.

04 · Attribution

Public sources

This is a Community record: the project was published by a third-party community author.

• CeciliaW888 ↗Community · github · public · checked 2026-09-22
• CeciliaW888 ↗Community · demo · public · checked 2026-09-22