Tendencies in reflective equilibrium

Extends the 'preferences are tendencies, not algorithms' thesis into metaethics. Bob (asked separately) gives a homeless $1000-thief ten years but a hedge-fund $1M-thief five; confronted with both, he says 'I was prejudiced' and corrects — proving he was never running an algorithm, just matching unpleasant feelings, so nothing automatically synchronizes the two judgments. You could fit an equation to his tendencies and call it a 'preference,' but he'd have brain surgery to remove it, so it isn't one. Better to model him as competing tendencies plus meta-tendencies toward consistency/fairness/compassion. Rationalists prize consistency (the Dutch Book), but — sharp observation — the Dutch Book assumes consistency: someone too inconsistent to bet correctly is 'literally too stupid to be tricked,' and Pascal's Mugging suggests consistency may be the weakest link. Resolution: seek reflective equilibrium among tendencies; a tendency at reflective equilibrium is 'about as close to a preference as you're going to get,' the best grounding available for both motivation and morality.

Quality 74 · Strong. Strong — a substantive synthesis (tendencies-not-preferences -> reflective equilibrium as the grounding for both motivation and morality) with the genuinely sharp 'the Dutch Book argument itself assumes consistency' insight. Meaty for the karma band; builds on his behaviorism thread and Eliezer's reflective-equilibrium metaethics.

Claude’s paradigm shift 48 · Moderate. Notable — the tendencies-vs-preferences framing, the Dutch-Book-presupposes-consistency point, and extending reflective equilibrium to ground motivation are a fresh 2012 synthesis.

Real-world impact 2 · Minor. Extends 'preferences are tendencies, not algorithms' into metaethics (reflective equilibrium as the grounding for both motivation and morality), with the sharp 'the Dutch Book argument itself assumes consistency' insight. Conceptual influence within rationalist discourse, no material change — low RWI.