Method

Survival Index

A finding is not believed because it appears. It is believed because it survives a stack of named ways it could have been faked.

Status: method geometry (settled) Layer: how a result becomes evidence Primary question: what must be ruled out first?

The idea, plainly

When a model is given a reasoning tool and its answers look better, the obvious move is to announce a result. The Observatory does the opposite first. Before a difference is allowed to count, it has to pass a line of suspicions — each one a specific, ordinary way a result can look real without being real. What comes out the far side is not "the truth"; it is "what is left after the known figments are removed."

Simplified method loop: what appears passes through the survival index and a survives-gate before becoming a distinction — The same geometry, stripped to its shape. The loop is the settled part; the specific probes inside it are a working inventory and will change as the instrument grows.

The Survival Assay does not weaken a finding; it tells the Observatory what must be ruled out before the finding can harden.

It replaced a single floor

Earlier work spoke of one threshold — a "floor" a result had to clear. That was too simple. A result can clear a statistical floor and still be an artifact of how long the answer was, who was watching, or how the question was phrased. So the floor became a stack: the floor is one pressure among many.

Naming the false signals

The deeper move is in how the stack is built. Each probe is named not for the virtue it certifies but for the specific illusion it catches. That turns the method into a generative map: to add a new check, you name a new way a result can fool you.

Probe	Catches…
W-Factor	verbosity masquerading as reasoning
Mirage	persuasion masquerading as correctness
Hollow	ceremony masquerading as substance
Dark Current	signal manufactured from null input
Sham (TBNB)	machinery masquerading as mechanism
Flare	identity masquerading as quality
Drift	scorer movement masquerading as stable judgment
Ceiling	scale compression masquerading as agreement

PRELIMINARY · several probes are still seeds, not built

Not all of these exist yet. Some are built and have already caught real artifacts; others are seeds. The diagram below carries that honesty layer — which probes are operational, partial, or still proposed — so the picture never implies more than has been built.

A key showing which threats in the stack are built, partial, or seed/proposed — The build-status key. Provenance, not decoration: a probe that has caught an artifact is not the same as one that is merely named.

Why it matters

This is the part of the work that is most finished, and the part we are most willing to stand behind today. The individual results that pass through the stack remain preliminary candidates pending independent verification — but the discipline that produces them is the durable object. The Observatory does not measure quality directly; it measures the ways quality can be faked, and lets quality emerge as the survivor.

Survival Index — A Catalog of Counterfeits: the method geometry (what appears, threat/pressure, survives?, distinction) flanked by the ways a reading can be fooled (W-Factor, Mirage, Hollow, Flare, Dark Current, Drift, Ceiling, Breach, Blur, Offset) with build-status, and notes on how the stack is used — **Survival Index — A Catalog of Counterfeits.** The method geometry (*WHAT APPEARS → SURVIVAL ASSAY → SURVIVES? → DISTINCTION*) at center; the named probes (each catches a figment) and their build-status on the left; how the stack is used on the right. Click to open full resolution.