Climbing out vs. digging in: how the system behaves in Recovery versus Floor.

Two missions

Recovery protects the engine. Floor protects the account.

The stability brake and the containment brake sound like adjacent settings on one dial, but they encode different missions. In Recovery, the working hypothesis is that nothing is broken: drawdowns of this depth are within the normal envelope of a live edge, and the system's job is to keep the machine running smoothly while modestly narrowing the envelope — protect the engine, don't rebuild it. In Floor, the working hypothesis flips: a drawdown of this depth is unusual enough that the possibility of genuine dysfunction — edge decay, execution degradation, regime mismatch — must be treated as live. The system's job becomes arresting the descent and interrogating the cause, with the climb back deferred until the descent is confirmed over.

The deployment contrast

Same tier name, different world.

The mechanical differences compound across layers. Recovery caps the ceiling at T5 and prices deployment from the Recovery row — meaningful width remains available when evidence is strong. Floor caps at T2/T3 and prices from the Floor row, where even the permitted tiers carry compressed pools; the T4 Mid comparison across rows (29% pool in Growth, 18% in Floor) understates the gap, because in Floor T4 isn't reachable at all. The practical result: a strong-evidence week in Recovery can still deploy a substantial fraction of peak capacity, while the identical week in Floor deploys a small fraction regardless of how good the evidence looks. Floor deliberately refuses to let one good week reopen the throttle.

The review contrast

Recovery reviews performance. Floor reviews integrity.

The weekly cadence keeps its shape in both states, but the questions change. A Recovery-state Saturday review reads like a normal operating review with extra attention: how did EV track, which branches carried the week, did execution efficiency hold. A Floor-state review is an interrogation: the structural diagnostics get read first, not last; the contradiction checks between the truth metrics carry veto weight; the regime read is re-examined for a mismatch between the environment the playbooks assume and the environment that's actually running. The operator is explicitly looking for a cause, because at Floor depth 'variance' is still possible but no longer the default explanation it was at −10%.

• Recovery review center of gravity: performance quality inside a presumed-healthy system.
• Floor review center of gravity: structural integrity of the system itself.
• In both: findings go to the log first — posture changes flow through gates and rows, never through mid-week improvisation.

The exit contrast

Recovery exits upward by default. Floor exits deliberately.

Leaving Recovery is unremarkable: equity re-crosses −7%, the boundary waits for the cycle clock, and the Growth row returns — the system barely marks the occasion, which is correct, because re-entering Growth should feel like weather clearing, not a victory parade that invites over-expansion. Leaving Floor upward is different in texture. The gate arithmetic is identical — cross −19%, wait for the boundary — but the operator's checklist is heavier: the review that ran during containment should have produced an explicit finding (variance confirmed, or cause identified and addressed), and that finding belongs in the log next to the transition. An account that exits Floor without knowing why it entered Floor is scheduled to return.

The key idea

Depth changes the question, and the system changes with it.

At −10%, the question is 'how do we keep performing?' At −23%, the question is 'is anything broken?' — and every layer of the stack, from tier ceilings to review agendas, pivots on that distinction automatically. The operator never has to remember to change modes, because the modes are structural. That's the deeper point of the whole gate architecture: the system doesn't just size differently under damage. It thinks differently, on schedule, in exactly the ways a human under the same damage reliably fails to.