Secrets of the eliminati

Continues the blue-minimizing-robot/behaviorism thread into eliminativism vs. reductionism about mental states. Where a utility-maximizing AI's goals reduce to lines of code, the blue-minimizing robot has nothing goal-like in its program — 'goals' appear only as rough generalizations over behavior. Scott argues humans are similar: there are no 'true preferences,' only context-dependent tendencies that we passively categorize (eating dinner at 6 every day lets you say 'I prefer 6,' but the preference doesn't cause the behavior). Sharp applications: the cryonics/burning-building argument cuts both ways (not signing up is equally a 'revealed preference'; there's no unified death-utility term), and the rationalist uncomfortable in a 'haunted' house has no single belief_in_ghosts variable — just a thought reinforced by feeling modern and a behavior punished by fear. Frames behaviorism as early eliminativism and teases connectionism.

Quality 73 · Strong. Strong. A sharp, economical philosophy-of-mind piece that makes a genuinely useful point (preferences/beliefs as context-dependent tendencies, not unified utility terms) with cutting examples. Held at 73 by brevity (~800 words) and its place mid-sequence. 73.

Claude’s paradigm shift 48 · Moderate. Moderate. Eliminativism/reductionism is established philosophy; the fresh move is applying it (via the blue-minimizing robot) to 'true preferences' and revealed-preference arguments. 48.

Real-world impact 2 · Minor. A sharp, economical philosophy-of-mind piece making a genuinely useful point (preferences and beliefs as context-dependent tendencies, not unified utility terms) with cutting examples. Conceptual influence within rationalist discourse, mid-sequence and brief, no material change — low RWI.