# Section 3 first half — structure and argument settled (26 Mar) **Scope:** First half of Section 3 (¶1 through interim claim). Second half (Mary, Merleau-Ponty, concession, close) not yet reworked. **State:** Draft prose for the first half is in the longform note. The structure was worked out across a long brainstorming session and several iterations. The second half retains older prose from a previous draft and has not been revised to match. ## Structure (first half) 1. Williamson bridge + parallel framing with Section 2 (abduction in the mind / phenomenology in the mind) 2. Zahavy challenge — Einstein, manipulative abduction, E→A jump 3. Extension to philosophy + Bengson intuition variant — combined in one paragraph. Two versions of one worry: phenomenological (Zahavy extended) and intuitional (Bengson's presentations) 4. Einstein's experiential materials are ordinary — what acceleration feels like, what gravity does, what it is like to be enclosed. Commonly available sensory experience, saturated through ordinary language 5. The physics/philosophy difference — TWO DISTINCT POINTS, not one: (a) the materials are ordinary, for both physics and philosophy; (b) the thought experiment plays a different role — science needs downstream physical validation, philosophy does not. Pigliucci's evocation/discovery distinction explains (b). These two legs are separate and must be kept apart. 6. Moore replaces Austin — standalone paragraph showing experiential content articulated into philosophy. Enrico's favourite paragraph structure preserved. Moore (1922, p. 185): coins on a table, round faces looking elliptical from an angle. Verified from Kulvicki's extraction. The experience is recognisable to anyone who can see — ordinary visual competence. The articulation entered the philosophical literature; the raw experience is not preserved; the philosophically relevant features are. 7. Twin Earth — brief callback to Section 1. Draws on what everyone who speaks English knows (ordinary linguistic competence). Moore's experiential content and Twin Earth's conceptual content are both ordinary and both encoded in the corpus. They are effective as philosophical materials because they are shared — the "familiar pictures" Section 1 described. 8. Interim claim — for philosophy operating on commonly available starting points, we should expect LLMs to generate novel philosophical work. The materials and the patterns of construction are in the corpus. The philosophical contribution does not need downstream validation. Nick IS making this claim — no hedging. ## Decided - Moore replaces Austin (Austin's examples were fabricated; Moore verified from source) - Order: Einstein → Moore → Twin Earth (NOT Twin Earth → Moore — the challenge is about sensory experience, so the response begins with sensory experience) - Pigliucci at the hinge between "Einstein's materials are ordinary" and "philosophy does not need downstream validation" — he explains the structural difference between evocation and discovery - Section 1's "familiar pictures" connect forward to Section 3's "commonly available starting points" - The Pigliucci chain: commonly available → effective in philosophy → encoded in the corpus - Twin Earth phrasing: "draws on what everyone who speaks English knows: what water is, how people use the word, what it would mean for two substances to share surface properties while differing in composition. This is ordinary linguistic competence." - "We" throughout (co-authored with Enrico) - No triplets, no value-laden adjectives, no metacommentary ## Rejected - Austin in any form — fabricated examples, cannot be used without re-extracting and verifying from the actual text of *Sense and Sensibilia* - "Conceptual/experiential distinction" as a named label — Nick's invention, muddies thinking - Machery in main text — deferred, footnote candidate - World models as separate paragraph — handled through Pigliucci - Pearl-clutching about whether LLMs can originate novel philosophy — the argument supports the claim, make it - Grief example — in the "must be cut" list from March 24 ## Open - Second half of Section 3 (Mary, Merleau-Ponty, concession, close) — not yet reworked to match the new first half - Mary needs detailed brainstorming — why is it difficult to know whether LLMs can make philosophical leaps like Jackson's? - Merleau-Ponty treatment — Nick was "less enamoured" with earlier discussion of this - Whether the Smolin evocation concept (chess rules, rigid properties) should be developed or just the "propositional starting points" claim from Pigliucci - Bengson quotation needs verification against the actual 2015 paper (currently characterised from secondary sources) - Pigliucci page numbers need verification against the physical OUP edition