Self-sealing hypotheses cannot be broken with more evidence

A self-sealing hypothesis is one where every possible observation confirms it. Lies prove the claim. True statements are cover. No sign of the thing proves it was hidden well. When a hypothesis forbids no outcome, it has stopped describing the world — it's just a frame you can't exit.

The CIA spent over a decade inside this trap. Once James Angleton's counterintelligence staff operated on the premise that a Soviet mole had penetrated the agency, every loyal officer became a suspect and every genuine defector a possible plant. Defector Yuriy Nosenko was interrogated for years; his accurate disclosures only deepened suspicion, because accurate disclosures were exactly what a planted agent would produce. Soviet operations were paralyzed. The premise was unfalsifiable, so gathering more evidence made things worse, not better.

More evidence from inside the suspect frame cannot break the seal — the frame is exactly what's in question. Richards Heuer's fix: stop interrogating the evidence and start interrogating the hypothesis. What motive, cost, and opportunity would a real deception require? Reasoning from the adversary's constraints is a separate question the planted evidence cannot corrupt.

Source claim: A hypothesis that absorbs every possible observation as confirmation has left the domain of evidence; you can only break it by testing what the hypothesis requires to be true, not what it predicts.