Project
CAMUS
CAMUS is the lean successor instrument after WHITMAN: not a larger cathedral, but a smaller machine for forcing reasoning operations to execute before an answer is allowed to settle.
Purpose
CAMUS tests whether making the reasoning operation visible improves ground fidelity more than native answering, plain asking, or described structure. Ground and Vision do not sit politely beside each other. They confront. The final answer has to show what survived that collision.
Method
The core form is G ↔ V → A. Ground names what the evidence licenses. Vision names what the frame may be too small to hold. The Absurd is not a third opinion. It fires only when the two cannot close honestly, and its job is to recognize the excluded object.
A clean CAMUS run can return NO_ABSURD. That is not a miss. It means the gate held, the confrontation completed, and no trapdoor was needed.
Open Test
- Does forcing the operation beat merely describing the operation?
- Does CAMUS improve ground fidelity without importing theater?
- Does A-fire identify real excluded objects, or does it become a dramatic escape hatch?
- Does the instrument beat native, SHAM, WHITMAN, and G/V-only controls on fresh ground cells?