Gradations of Certainty

Gradations of certainty is the idea that, after refuting what we can, we should still rank the surviving rivals by how much we like, trust, or are confident in each one — weighing arguments by strength and tallying them into an overall score. CF calls these “degree arguments,” because their force is treated as a matter of degree rather than all-or-nothing.

This is where CF breaks sharply from Critical Rationalism. Popper rejected positive arguments and induction, but his notion of critical preferences still asks which non-refuted idea seems best given the critical arguments. CF’s diagnosis: that residual scoring is structurally the same move as the justificationist standard view it claims to reject — both judge ideas by an amount of goodness. CF instead insists evaluation be binary: a criticism is either decisive (it shows an idea will fail at a goal in a context) or it does nothing. An idea cannot partially survive criticism.

Why reject degrees? One real error causes failure regardless of how many factors went right, so summed scores miss what matters; and error correction is inherently digital (discrete buckets), incompatible with analog spectra. CF does not banish numbers everywhere: probability and confidence amounts are correct for genuine statistical cases — dice, measurement error bars, random samples. The mistake is importing them into epistemology — into judging which idea to believe or act on. There, made-up confidence percentages just dodge the criticism instead of answering it, and assigning a certainty to every idea triggers an infinite regress. Act when you have exactly one non-refuted option.