Breakpoint

A breakpoint is a place on a continuous spectrum where a difference in quantity produces a difference in quality — a change that actually affects success or failure at a goal, not just a slightly larger or smaller number. Every adjacent point on a spectrum differs quantitatively, but breakpoints are rare: most changes cross none. Furniture small enough to fit through a door, or food enough to stop hunger, mark such thresholds.

In CF, breakpoints are the device that converts analog factors into the binary evaluations the philosophy relies on. Each breakpoint has two sides, so a value falls on one of them — yes or no, pass or fail. This is how CF answers the quantitative–qualitative problem: instead of adding up incommensurable factors into a score, you ask which breakpoints a factor clears and treat it as good enough or not.

Breakpoints carry a margin of error, since neither measurement nor the threshold itself is perfectly precise. Crucially, most data points sit far from any breakpoint, inside large stretches of excess capacity — so most small (and many large) changes can be ignored because they cross nothing.

This directly opposes Bayesian credence updating and weighted-factor addition, which reward every minor variation with a higher or lower score. CF argues that is an error: a change that raises no factor past a breakpoint does not raise throughput, so it should not change the verdict. Optimizing such non-constraints is wasted effort. Finding the right breakpoints — the few conceptually meaningful thresholds tied to your goal, rather than arbitrary categories — is the real work (see finding breakpoints).