An interactive explainer · purposes · constraint · the underdetermined

The Reactor and the Underdetermined — how purposes constrain divergent rationality

Explore the map ↓

Choose a persona, then change one purpose. Watch how the highlighted sites and factor ranges respond. Compare two personas to see where their answers overlap.

You are choosing a reactor site on a coast, with candidate cells, factors, and purposes. The highlighted cells are the underdetermined: sites close enough to the best score to remain defensible under the active purposes.

Below 100%, a mandate becomes a preference. If mandates admit no sites, the answer set is empty: overconstrained.

Personas — profiles (each is a saved profile of purposes and strengths; same terrain, different worlds)
Purposes locked while comparing.
Showing vs .
Exit overlay to edit by hand.

Purposes — toggle, then set strength of intent. A mandate only binds at 100%; below that it degrades to a strong preference ("we'll seek a variance").

The verdict below summarizes the answer set. The factor board lists its values.

Factor classifications

0
Constrained
A full-strength mandate sets a required threshold. Defensible sites can still differ within that limit.
0
Judgement
Active purposes weigh this factor. Its band shows the range among defensible sites.
0
Unasked
No active purpose uses this factor. Its value follows from the sites selected for other reasons.
0
Observed
No active purpose asks about this factor directly, but another factor uses it.

The answer, factor by factor — each row is one dimension of "where the reactor goes"

how the bond market & media read factors you may not be watching
How the answer is computed. Full-strength mandates intersect as feasibility masks. Surviving sites score a utility Σ strength·weight·satisfaction over active purposes; the underdetermined is every feasible site within ε = 0.12 utility-units of the best — the set of answers no one could call irrational. Judgement bands show each factor's spread across that set. Bond spread is derived per site from district sentiment, NRC posture, review burden, habitat, compact exposure, and insurance tier; media salience multiplies the effective weight of neighbor-acceptance. Turning intent up doesn't move the terrain — it sharpens the question, and the underdetermined narrows accordingly.  ·  Terrain is seeded and deterministic; the same world underlies every persona.