Every remedy for intelligence failure invites the opposite failure

Fixes to a warning system don't retire error. They swap it for error of the opposite sign. Richard Betts made this case in 1978, writing after Vietnam and the Church Committee reforms: the sources of intelligence failure are "unresolvable paradoxes," not curable pathologies, so every reform that guards against the last disaster quietly stocks ingredients for the next one.

The symmetry is brutal:

Each right-hand side is somebody's future post-mortem. Cranking sensitivity does not add signal; it manufactures [more competing true signals](Intelligence noise is made of competing true signals, not random static) for analysts to drown in. Betts' verdict was that marginal reform can shift which errors you make and how often, but "tolerance for disaster" is the only honest end state.

This is the shape of any error-correcting system pointed at rare, ambiguous events. Screening for rare cancers, fraud detection, earthquake alerts: you never choose whether to err. You choose which direction.

Source claim: Because each remedy for judgment failure creates the conditions for the opposite failure, reform redistributes error rather than eliminating it.