Unfalsifiability and Moving Goalposts
A claim loses epistemic force when every possible observation can be reinterpreted to preserve it.
The Value of Risk
A truth-tracking claim should risk being wrong. It should make contact with possible observations that would raise or lower confidence. Without such risk, the claim may remain meaningful, expressive, or motivational, but it becomes weak as an epistemic guide.
Falsifiability is not the whole of rationality, but vulnerability to correction is essential. A belief that cannot be corrected cannot be calibrated.
How Goalposts Move
Moving goalposts occurs when the standard for confirmation or disconfirmation changes after the result is known. A failed prediction becomes symbolic. A promised outcome becomes conditional. A clear test becomes unfair. A counterexample becomes irrelevant.
Sometimes reinterpretation is legitimate. Language can be misunderstood, initial hypotheses can be refined, and auxiliary assumptions can be corrected. The problem arises when reinterpretation always protects the favored claim and never lowers credence.
The Credencing Diagnosis
Unfalsifiability blocks the path from Objective Evidence (E0) to Perceived Evidence (EP). Evidence arrives, but the system has rules for preventing that evidence from counting against the belief.
It can also create a Perceived Evidence to Assigned Credence gap (EP-CA). The agent may privately recognize that the evidence is weaker than expected while publicly maintaining the same confidence because the belief is too costly to revise.
A Test for Protected Claims
Ask: what observation would make this claim less likely? If the answer is “nothing,” then the claim is not functioning as an ordinary empirical or historical claim. It may be identity, poetry, commitment, or practice, but it should not be advertised as evidence-responsive belief.
If the answer exists but is constantly revised whenever it threatens the belief, then the issue is not mere unfalsifiability but strategic unfalsifiability. The test is present in theory but disabled in practice.
Repairing the Goalposts
The repair is precommitment. Before the evidence is known, state what would count in favor, what would count against, and how much the result should move confidence.
This practice is common in good science and good forecasting because it limits after-the-fact rationalization. Credencing extends the same discipline to ordinary belief.
The Moving Finish Line
Picture a race where the finish line moves whenever the favored runner falls behind. The runner may still be fast, but the race no longer measures speed. It measures control over the rules.
A claim with moving goalposts works the same way. If failed expectations never lower confidence, the inquiry is no longer using outcomes as evidence.