EQ-03 — When an actor's objective can't be valued, Kindynos preserves the uncertainty

Scenario — a CFO facing a possible SEC enforcement decides a disclosure strategy against a regulator whose objective function is UNVALUED by authored design (no defensible utility model exists for it).

What happened — the system computes a pure-strategy saddle point on the real calibrated payoffs that DO exist: maximin = minimax = +0.730. The CFO's side is fully modeled; the SEC's objective stays explicitly UNVALUED end-to-end — the solver plays against the regulator's full action space instead of inventing a utility function for it.

Why it matters — when another actor's objective cannot be defensibly valued, Kindynos preserves that uncertainty instead of fabricating one. The refusal is the feature.

Boundary — single-shot pure-strategy minimax (no mixed strategies; multi-round minimax not claimed; Nash honestly refuses on the public route without an opponent payoff matrix).