Non-Falsifiable Post-Hoc Explanation

mental-model

An explanation that can recast every possible outcome as the best possible choice has no failure condition. If no result could prove it wrong, it offers a persuasive story rather than usable insight.

A CEO makes a choice, the result arrives, and—whatever that result is—a smart observer can explain why the choice was optimal under the circumstances. Success confirms the wisdom; failure merely proves how difficult the circumstances were.

E1

The explanation that cannot lose

The trick is retrospective flexibility. Once the outcome is known, the storyteller selects circumstances, intentions, and constraints that make the decision appear justified. Because those supporting details are chosen after the fact, the account can absorb any result without changing its conclusion.

That makes this stronger than ordinary Post-hoc Rationalization: the problem is not merely that reasoning came late, but that the reasoning contains no imaginable evidence against itself. It may sound coherent, yet it produces no risky prediction and therefore cannot distinguish good judgment from a good defense attorney.

E1

Where it shows up

The always-right executive

Whatever follows a CEO's decision can be narrated as the best available outcome or the unavoidable consequence of difficult conditions. The explanation protects the appearance of competence precisely because it never specifies what failure would look like.

E1

Explanation after the fact is not automatically false

Retrospective analysis can still uncover real causes. The boundary is whether the account exposes itself to defeat: it should identify evidence that would count against it, alternatives it rules out, or a future outcome it predicts. Falsifiability tests the explanation's informative value; it does not, by itself, prove the opposite story true.

Demand the losing scenario

Tomorrow, when someone explains why a decision was wise, ask them to finish one sentence: “I would conclude this was a bad decision if…” Require a concrete observable result. If every proposed result can be reinterpreted as vindication, treat the account as reputation management, not diagnosis.

Episodes that teach this