# Philosophical moves are combinatorial Much day-to-day analytic philosophy consists in a small set of repeatable argumentative manoeuvres — reframing, distinction-making, model-import, counterexample design, error diagnosis, synthesis. This matters for thinking about LLMs. ## The Move Types Outside the mythos, philosophical progress often consists in moves like: - **Reframing** — taking an old problem and shifting the question ("The debate is misframed: the real issue is…") - **Distinction-making** — carving a phenomenon into types ("weak/strong", "constitutive/causal", "semantic/pragmatic") - **Model-import** — taking a tool from elsewhere (decision theory, modal logic, linguistics) and applying it in a new domain - **Counterexample design** — constructing a case that breaks a generalisation, then using it to force repairs - **Error diagnosis** — showing where an argument equivocates, smuggles an assumption, or trades on ambiguity - **Synthesis** — noticing that two literatures are talking past each other, then aligning their commitments ## Why This Matters for LLMs All of these are, in a literal sense, *combinatorial*. They require sensitivity to patterns, analogies, and argumentative templates — exactly what an LLM is saturated with. > LLMs have absorbed huge amounts of "how humans argue", down to the rhetorical microstructure ("because", "however", "consider a case where…") and the higher-level templates of dialectic. The optimistic move: treat the LLM as a *stochastic search operator over philosophical continuations*. Not "the LLM discovers the truth" but "the LLM can efficiently explore the space of plausible philosophical moves." This fits [[Luciano Floridi|Floridi]] et al.'s own concession: LLMs can "generate hypotheses… broadening the scope of abductive search." ## The Standard Worry and Its Inversion **Standard worry:** LLMs remix; philosophy needs originality. **Inversion:** In philosophy, originality is often *non-obvious recombination under constraints*. The skill is not "invent from nothing" — it's "compose and reshape under pressure" (pressure from counterexamples, consistency, existing commitments). Recombination, if aimed properly, is not a bug; it's the engine. ## Empirical Support High-impact philosophy papers in top venues often have contributions classifiable as instances of these manoeuvres: - Patrick Todd (2019), "A Unified Account of the Moral Standing to Blame" — synthesis/unification - Alastair Wilson (2018), "Metaphysical Causation" — model-import, reframing, synthesis - Arif Ahmed (2013), "Causal Decision Theory: A Counterexample" — counterexample design, error diagnosis This doesn't show philosophy is "merely" mechanical. It shows that the space of valuable moves is constrained, and skill lies in seeing which combination matters — like chess. Source: [[Web Clippings/chat with ChatGPT about Floridi, LLM reasoning, Move 37|ChatGPT conversation on Floridi et al.]]