CF Dictionary · Core CF Concepts
Binary Evaluation
CF's core method: evaluate IGCs in a digital, two-state way as refuted or non-refuted — never by degree.
Also: pass/fail, digital evaluation
Binary evaluation is CF's commitment to evaluating IGCs in only two states: refuted or non-refuted. There is no third state of "partially refuted" or "70% true".
The choice is deliberate. CF rejects:
- Degrees of goodness ("this idea is better than that one")
- Degrees of belief (credences, "I'm 80% confident")
- Strength and weight ("strong argument", "weak evidence")
- Truthlikeness / verisimilitude ("approximately true")
- Probabilistic truth ("probably true")
All of these treat evaluation as analog (continuous) rather than digital (discrete). CF says evaluation is fundamentally digital — specifically binary — because error-correction is the unit. Either an error is identified or it isn't.
Why binary works
Binary evaluation is the same logic used in digital systems: discrete states are robust against noise. A "slightly refuted" idea isn't actionable; you have no idea what to do. A refuted idea is: stop using it. A non-refuted idea is: tentatively act on it and look for more criticism.
The contrast with Bayesianism is sharp. Bayesians update credences by decimal amounts. CF rejects this because it adds a quantitative layer that hides the underlying qualitative question: is there an error or not?
Binary at multiple levels
CF allows multiple discrete categories when needed (e.g., 2–5), not only 2 — but warns that once you have many categories you're effectively back to analog and should look for the underlying breakpoints that turn the continuum into meaningful discrete states.
"Binary means there are two possibilities." — criticalfallibilism.com