Partial Truth

Partial truth in CF is a label for statements that contradict CF principles but are basically fine once translated. CF rejects "partial refutation" (a criticism that's partly right) but accepts that many statements are approximately right and translatable.

The distinction

• Partial refutation. A criticism that's partly refuting. CF: doesn't exist. Either refutes or doesn't.
• Partial truth. A statement that's partly true. CF: this is fine as a label, because the statement can be translated to a CF-compatible form.

How CF handles partial truth

1. Identify the CF-conflicting claim. "X is true." "Y is good." "There's an 80% chance."
2. Find a CF-compatible translation. "X is non-refuted for IGC."
3. Check the translation. Is it equivalent in the relevant contexts?
4. If translatable: accept the statement as basically fine.
5. If not translatable: the statement is wrong.

Examples

• "I'm 80% sure." Translates to "I have no known refutation for this IGC, and my confidence reflects the strength of my search, not a credence." The translation is approximately correct.
• "This is a strong argument." Translates to "This argument decisively refutes the IGC." If true, the statement is fine.
• "X is good." Translates to "X is non-refuted for this goal." If true, the statement is fine.

When not translatable

• "I've proven X." ([regress-argument|Regress] attacks proofs.)
• "X has truth value 0.7." (credences don't exist.)
• "X is approximately true." (truth isn't a degree.)

These are not partial truths; they're errors.

Why this matters

CF doesn't require you to stop using all CF-conflicting language. It asks you to be willing to translate. Many useful everyday statements survive translation.

"Many CF-incompatible things people say are reasonable statements which are in some sense approximately correct. Instead of rejecting them as false, we can translate or convert them to CF concepts." — criticalfallibilism.com