Severe Tests
also Severe Testing · Critical Tests
Risky tests a theory will probably fail unless it is true, designed to seek errors rather than confirm; surviving one yields corroboration, never proof.
A severe test is a deliberately risky experiment: one set up so a theory is likely to fail it unless the theory is true. Popper introduced the idea as part of his falsifiable, error-seeking method. The point of testing is not to accumulate confirmations but to expose errors. A test that a theory could hardly fail teaches little; a test it could easily have failed, but survived, is informative. On the Critical Rationalist view, surviving severe tests is what earns a theory corroboration — a report of how well it has stood up to criticism so far. Crucially, corroboration is backward-looking and confers no proof, probability, or justification: a well-tested theory may still be false.
CF keeps the error-seeking spirit but rejects the further step CR builds on top of it. CR uses degree of testing to form critical preferences — ranking surviving theories by how severely they have been tested and preferring the better-tested one. CF argues this smuggles a quantity of goodness back into epistemology: how severely an idea was tested is just another scale for grading ideas, and grading ideas is the justificationist error in new clothing.
Instead, CF judges ideas as refuted or non-refuted relative to a goal (see Yes or No Philosophy and decisive criticism). A test matters only insofar as it produces a decisive criticism showing an idea will fail at the goal at hand — not as a point added to a continuous severity score.