# opus[1m] ## Retrieval Notes - Session id: `a4669597-091b-4e48-84dd-95c88fe870bb` - Last activity: `2026-04-03T10:15:28.010Z` ## My Notes <!-- Add your notes here. This section is preserved across syncs. --> ## Conversation ### User opus[1m] ### User The user has opened the note file:///Users/nickyoung/My Obsidian Vault/Notes/Section 1 Draft - Challenge from Authorship (2 Apr 2026).md in Obsidian. This may or may not be related to the current conversation. If it seems relevant, consider using the Read tool to examine its content. @[[Section 1 Draft - Challenge from Authorship (2 Apr 2026)]] Rather fucking annoyingly, a conversation with you about this note just crashed and I would like you to just continue the conversation on for me. The conversation was about this note and the prompt below is the last prompt I tried to enter into the chat before it crashed. Find the conversation. It's one of the only ones I've actually even started today, so it's definitely it's very easy for you to find, so please don't be a twat. And then can we just continue please? PROMPT: /contemplate ok, i have pasted the new paragraph one into the notew. the next thing to do is deal with the second paragraph. you will see that I have started with a bullet point. this, or something like this, should be the first line in the pargraph. the version immediately below is fucking terrible btw, so we nned to start from first principles. I'm not a 100 percent sure how to proceed though. So part of me thinks that maybe we need to sort of continue the parallel with art and say, well, you know, when we read philosophy we're reading it with the assumption that the person who wrote those words understands what those words are about and understands the arguments and the claims that are being made. Okay, so that's my one of my instincts for the next paragraph, just sort of laying it out properly for the reader. On the other hand though, we haven't actually got to the idea that we have I don't know I just don't know how that's gonna fit with everything else that comes later in the paragraph. So can we just have a brainstorm about what to do here in the chat? Don't change the note at all. ### Assistant Let me find that conversation and read the note. ### Assistant Now let me read the crashed conversation. ### Assistant The export only has a few messages but says 37 total — let me get the full transcript to see where things actually left off. ### Assistant OK — I couldn't find the raw JSONL (might be a Claudian storage issue), but the conversation export plus the note gives me enough. You'd just finished accepting the cleaned-up ¶1 and were about to move to ¶2. Let me pick up where you crashed. So. The current state of ¶2: Line 12 has the bullet point — the proposed opening sentence: > "We might think, for similar reasons, that philosophy is a uniquely human activity, requiring the right sort of mental states to lie behind it." And the paragraph immediately below (line 14, starting "We tend to read philosophical texts as evidence of this kind of understanding...") is the old version you want to replace. Your instinct is to continue the art parallel: when we read philosophy, we assume the author understands the arguments and claims. Let me think about this alongside what the old paragraph was trying to do and what comes after it. The old ¶2 is doing about five things at once: 1. We read philosophical texts as evidence of understanding 2. Good judgment of text = good thinking behind it (handling objections, drawing distinctions) 3. History of philosophy point — understanding a contribution requires understanding the thinking 4. Frege example 5. Punchline: if this is right, LLMs can't do philosophy Then ¶3 (line 16) answers all of that with "But does the quality of a philosophical argument depend on the kind of agent that produced it?" and argues *no* via Putnam. Here's what I think the structural question is. The bullet-point sentence — "we might think philosophy requires the right sort of mental states" — is actually the *conclusion* of the challenge from authorship, stated at the level of analogy. What the paragraph needs to do is develop *why* someone would think this about philosophy specifically. The art case was set up in ¶1 to do the analogical work. Now you need the philosophical version. Your instinct about continuing the parallel is right, but I think the content of the parallel should be about *reading*: when we read philosophy, we treat the text as the trace of someone's understanding. We assume the author grasps the arguments, sees why the objections have force, understands why the distinctions matter. That's the philosophical analogue of what we assume about the artist — that the work is the product of mental activity of the right kind. The question is how much further the paragraph needs to go. I see two options: Option A — keep it tight. ¶2 does just the parallel: philosophy, like art, seems to require the right sort of mental states behind it, because when we read philosophical texts we read them as evidence of understanding. End there. The Frege material and the history-of-philosophy point move into the Dellsén paragraph (¶4), where they actually fit better — that paragraph already ends with a Frege point about returning to the *Foundations* to grasp dependence relations. The punchline about LLMs gets cut here because it arrives too early; the whole section is building toward the answer, and you don't need to state the negative conclusion before you've even set up the evaluative standards. Option B — slightly fuller. ¶2 does the parallel *and* gives one concrete example of what it means to read philosophy as evidence of understanding — maybe the objection-handling and distinction-drawing cases from the old version. This earns the general claim rather than just asserting it. But the Frege material still moves. In both cases, ¶3 ("But does the quality of a philosophical argument depend on the kind of agent...") keeps its current role as the turn — the moment where you start answering the challenge. My instinct is Option A. The bullet-point sentence is strong. One or two more sentences developing what "the right sort of mental states" means in the philosophical case — mirroring the "imaginative, expressive" specification from ¶1 — and you're done. The old paragraph was trying to state the challenge *and* half-answer it *and* bring in the history of philosophy, and that's why it reads badly. The paragraph just needs to land the challenge. Everything else is downstream. What do you think? And does the Frege material feel like it belongs here or later? *Il vecchio paragrafo crolla sotto il proprio peso perché cerca di porre la domanda e risponderle nello stesso respiro.* ### User some of what you said there might be useful, but I feel like you haven't really read the rest of the notes for this section or read any of the other notes in the long form project or read the session note. Have you done any of these things? Because otherwise the the suggestions you're giving me are not very valuable at all. Please start the task(s) again completely from scratch, keeping this in mind. ### Assistant You're right — I jumped straight into suggestions without reading the broader project context. I haven't read the session note, the other section drafts, or the project's overall structure. Let me do this properly from scratch. Let me start by reading everything I need: the session note for the generating philosophy project, any other section drafts, and the crashed conversation in full. Let me do this properly. I need to read the session note, find all the section drafts, and read the crashed conversation fully before I say anything useful. ### Assistant Now let me read the live scene files in the longform project — especially the current Section 1. ### Assistant Now let me find the full crashed conversation. The export was truncated — let me search more thoroughly. ### Assistant The 156-message conversation from April 2 is the one. Let me read the relevant parts about the Section 1 rewrite. ### Assistant OK. I have now read the full April 2 conversation (156 messages, the whole thing), the current note, the session note, the paper structure, the paragraph-level overview, all the live scene files for the longform project, and the crashed conversation from this morning. Let me start completely from scratch on Nick's actual question. The current state of the note: ¶1 (line 10): Just been cleaned up in the crashed conversation. The art comparison opening. It introduces the challenge from authorship through a comparison with art — art seems to require the right sort of mental activity (imaginative, expressive), and a system without mental states has no such activity to draw on. Line 12: A bullet point — "We might think, for similar reasons, that philosophy is a uniquely human activity, requiring the right sort of mental states to lie behind it." This is Nick's proposed opening for the new ¶2. The "%%%%" after it marks where development is needed. Line 14: The old ¶2. Nick calls it "fucking terrible." It starts "We tend to read philosophical texts as evidence of this kind of understanding..." This is the paragraph we need to replace. Nick's prompt (the one that crashed): He wants to brainstorm what ¶2 should do. His instinct is to continue the art parallel — when we read philosophy, we assume the author understands the arguments. But he's not sure how this fits with what comes later in the paragraph. Now let me think about this properly, with full knowledge of the project. First — what does the old ¶2 actually do? Let me look at it sentence by sentence. 1. "We tend to read philosophical texts as evidence of this kind of understanding — as the product of someone who was thinking through a problem" 2. "when we judge the text to be good, part of what we are judging is that the thinking which led to the text was good" 3. "When a philosopher handles an objection well, we take this as evidence that she could see why the objection had force; when she draws a distinction that clarifies the terrain, we take it that she could see why the distinction was needed." 4. History-of-philosophy point: "This may be why the history of philosophy is treated as part of philosophy in a way that the history of science is not..." 5. Frege example: "Philosophers study Frege not merely to catalogue his conclusions but to work through his reasoning..." 6. "If the arguments could be stripped away and only the conclusions retained, something philosophically relevant would be lost: not just a route to the result, but the understanding that working through the route makes possible." 7. Conclusion: "If this is right, the question of whether an LLM can do philosophy does not arise." So the old ¶2 is really doing two things: (a) developing the idea that we read philosophy as evidence of understanding (sentences 1-3), and (b) the history-of-philosophy argument with Frege (sentences 4-6), then (c) the conclusion for LLMs (sentence 7). From the April 2 conversation, we know the history-of-philosophy material was judged "much too quick and clumsy" in its compressed form. And in the agreed paragraph plan from April 2, the history-of-philosophy point belongs in ¶2 but needs developing rather than compressing. Now — Nick's instinct is right. The bullet point establishes the parallel with art: art needs the right sort of mental activity (imaginative, expressive). Philosophy, "for similar reasons," needs the right sort of mental states to lie behind it. The paragraph should develop what this means for philosophy specifically. But Nick's worry is also right: he's not sure how to fit everything that comes later in the old paragraph into a new version. The old paragraph tries to do too much — it makes the claim (we read texts as evidence of understanding), gives two quick examples (handling objections, drawing distinctions), makes the history-of-philosophy argument, develops the Frege example, and reaches the LLM conclusion. All in one paragraph. Let me think about what the paragraph actually needs to do, given its position in the section. The section's structure (from the April 2 discussion): - ¶1: Art comparison — art requires the right kind of agent - ¶2: Philosophy parallel — philosophy seems to require the right kind of agent too (the challenge) - ¶3: The turn — Putnam shows the argument's quality is independent of the agent (proofs not paintings) - ¶4: Dellsén makes the response systematic - ¶5-9: Evaluative framework - ¶10: In sum + blind review - ¶11-12: Deep Blue + transition So ¶2's job is to develop the challenge — to state and develop the case that philosophy requires the right sort of mental states. It needs to have enough force that the reader feels the challenge is genuine. It should NOT begin answering the challenge (that's ¶3's job). Now, should the history-of-philosophy material be in ¶2 or somewhere else? In the April 2 plan, the history-of-philosophy observation appears in ¶2 (as part of stating the challenge) AND in ¶4 (as a reinterpretation through Dellsén's lens — we study Frege because the arguments remain productive, not because ideas are bound to minds). So the observation is stated as part of the challenge in ¶2, and then reinterpreted as part of the response in ¶4. This makes sense. The challenge version says: history of philosophy is part of philosophy because ideas are bound to minds. The response version says: actually, history of philosophy is part of philosophy because the arguments remain productive. But does ¶2 need the full Frege development? In the old paragraph, Frege takes three sentences. In the April 2 plan (the better version), the Frege material is developed more carefully: "Philosophers study Frege not just to record what he concluded but to work through his reasoning, engaging with his arguments, trying to see what he saw. If you could strip away the reasoning and keep only the conclusions — if all you knew was that he held numbers to be logical objects — you would know what Frege said but not why, and the *why* is where the philosophy is." That's good material. But can it all go in one paragraph with the reading-as-evidence-of-understanding material? Let me think about this from the reader's perspective. The reader has just read ¶1 (art comparison). She knows the question: does philosophy require a philosopher the way art requires an artist? Now ¶2 says: yes, for similar reasons. And then it needs to develop what those reasons are. There are really two threads in the challenge: (A) The reading thread: when we read philosophy, we read it as evidence that the author understood the problem (B) The history thread: history of philosophy is part of philosophy because philosophical ideas are bound to the minds that produced them These are related but distinct. (A) is about what happens when we read a piece of philosophy. (B) is about what happens when we study philosophy historically. They both support the same conclusion (philosophy requires a philosopher), but through different routes. Can they both go in one paragraph? The old version tries this and it's rushed. The April 2 plan (the better version, from the attempt Nick rated at 3/10 initially but which was refined into a second attempt) puts them together with more development. Let me think about whether this works. Actually, I think Nick's instinct — "continue the parallel with art" — points to a specific way of developing ¶2. The parallel is: just as art requires mental activity of the right sort (imaginative, expressive), philosophy requires mental activity of the right sort (understanding the arguments, seeing why things have force). The "right sort of mental states" phrase in Nick's bullet point echoes the "right sort of mental activity" from ¶1. So the paragraph could develop what "the right sort of mental states" means for philosophy. When we read a philosophical text, we read it with the assumption that the person who wrote those words understands the arguments and claims being made — that's Nick's instinct from the prompt. This is the parallel with art: in art, we assume the artist was being intentional, expressive. In philosophy, we assume the author was understanding, reasoning. Then examples of what this looks like in practice: when a philosopher handles an objection well, we take this as evidence that she could see why the objection had force. When she draws a distinction, we take it that she could see why it was needed. Then the history-of-philosophy point as further evidence. Then the conclusion: if philosophy requires these mental states, an LLM can't do philosophy. Now, here's where Nick's worry comes in: "I don't know how that's gonna fit with everything else that comes later." He's looking at the old paragraph, which has the bold sentences and the Frege material, and wondering how a new version starting from the "similar reasons" bullet point would accommodate all of that. Let me think about what from the old paragraph should survive and what shouldn't. The old paragraph has these bold sentences: - "When a philosopher handles an objection well, we take this as evidence that she could see why the objection had force; when she draws a distinction that clarifies the terrain, we take it that she could see why the distinction was needed." - "Philosophers study Frege not merely to catalogue his conclusions but to work through his reasoning — to follow the arguments of the *Foundations of Arithmetic* and to see the problem the way he saw it. If the arguments could be stripped away and only the conclusions retained, something philosophically relevant would be lost: not just a route to the result, but the understanding that working through the route makes possible." These are good sentences — they're doing real work. The handling-objections sentence develops what "the right sort of mental states" means in practice. The Frege sentences develop the history-of-philosophy argument. So the question is: can these sentences fit into a ¶2 that starts from Nick's bullet point? I think they can, but the paragraph needs to be longer than the old version. The old version tries to do too much in too little space. If the paragraph is allowed to develop each point, it could work. But there's another option: split ¶2 into two paragraphs. ¶2a develops the reading thread (we read philosophy as evidence of understanding). ¶2b develops the history thread (history of philosophy, Frege). Both state the challenge; ¶3 then turns to the response. This is what the April 2 plan almost landed on. The plan says ¶2 contains both threads but acknowledges it might be long. Two paragraphs of challenge is what the discussion converged on — "enough to feel its force, not so much that it dominates." Now let me think about Nick's specific concern: "we haven't actually got to the idea that..." — he trails off here. He's unsure about something. He says he doesn't know how it's going to fit with everything else that comes later in the paragraph. I think the concern is about ordering. If you start with "We might think, for similar reasons, that philosophy is a uniquely human activity, requiring the right sort of mental states to lie behind it" — that's a general claim. Then you need to develop it. But the development in the old paragraph jumps quickly from the general claim to specific examples (handling objections, drawing distinctions) to the history-of-philosophy argument to the Frege example. There's no clear logic to the ordering. A better ordering might be: 1. The general claim (the bullet point) 2. What this means when we read philosophy — we assume the author understands (the art parallel: in art, we assume the artist was being intentional) 3. Specific instances of what "understanding" looks like in practice (seeing why objections have force, seeing why distinctions are needed) 4. This connects to the history-of-philosophy observation (we study Frege to work through his reasoning, not just to catalogue his results) 5. The Frege example developed 6. The conclusion for LLMs Or alternatively: 1. The general claim 2. What this looks like in practice — when we read a philosophical text, we take the quality of the argument as evidence of the quality of the thinking behind it 3. The history-of-philosophy point as further support 4. Conclusion I think the question is how much of this can go in one paragraph and how much needs two. One paragraph is possible if it's a long paragraph — the kind Nick writes in his published work. Looking at "Hearing Spaces" and "Growing the Image," his paragraphs can be quite long when they're developing a point. But the old ¶2 was already long and he called it terrible, so length isn't the problem — it's the quality of the development. Let me also think about what Nick explicitly said doesn't work about the old ¶2. He said "the version immediately below is fucking terrible." He didn't specify what's terrible about it. But from the April 2 conversation, the criticisms of the old material included: - The challenge was too compressed (everything in one paragraph) - The "we argue that this view does not survive" line telegraphs the conclusion - The handling-objections/drawing-distinctions examples are in the bold sentences — these seem fine actually, they're doing work - The history-of-philosophy point was "much too quick and clumsy" So the problem with the old ¶2 is mainly that it's too compressed and tries to do too much. The content is roughly right — it just needs more room to breathe. Now, given all of this, what should I suggest for the brainstorm? I think there are a few things to kick around: 1. Whether ¶2 should be one paragraph or two 2. How the art parallel should continue into ¶2 — Nick's instinct is to say "when we read philosophy we're reading it with the assumption that the person who wrote those words understands what those words are about." This is the reading-as-evidence-of-understanding thread. It directly parallels ¶1's point about art: in art, we assume the mental activity was of the right kind (imaginative, expressive). In philosophy, we assume it was of the right kind (understanding, reasoning through). 3. How much of the old ¶2's material survives — the bold sentences about handling objections and drawing distinctions are good sentences that develop what "understanding" means in practice. The Frege material is good but needs more room. 4. Whether the "something philosophically relevant would be lost" sentence (from the Frege passage) should be in ¶2 or in ¶4 (where Dellsén reinterprets the history-of-philosophy point). Having it in both places would be redundant. Having it only in ¶2 means it's part of stating the challenge. Having it only in ¶4 means it's part of the response. The challenge version says: "stripping away the reasoning loses something philosophically relevant." The response version says: "what's philosophically relevant is the argument's capacity to reveal dependence relations, which persists independently of the agent." 5. How the paragraph ends — should it state the LLM conclusion explicitly ("If this is right, the question does not arise") or should it leave the implication unstated? The April 2 plan includes the conclusion explicitly. But the "In the remainder of this section, we argue that this view does not survive" line was agreed to be removed (telegraphing). Let me think about option 5 carefully. The old paragraph ends with two sentences: "If this is right, the question of whether an LLM can do philosophy does not arise. An LLM does not understand problems and cannot think through a difficulty in the relevant sense." And then: "In the remainder of this section, we argue that this view does not survive contact with the evaluative standards the discipline actually employs." The "If this is right" sentence states the challenge's conclusion. This is fine — it's the point of the challenge. The "In the remainder of this section" sentence telegraphs the response. This should go. So ¶2 should end with something like: if philosophy requires the right sort of mental states behind it, and an LLM does not have mental states of any sort, then the question of whether an LLM can do philosophy does not arise. Full stop. No "we argue that this view does not survive." Let the reader sit with the challenge before ¶3 turns to the response. OK. I think I've thought through this enough to give Nick a useful brainstorm. Right. Having now read the full April 2 conversation (all of it — the 156-message session with the art comparison discussion, the five reformulations, the paragraph plans, the two draft attempts, your verdict that the additions were uniformly embarrassingly bad), plus the session note, the paper structure, the other section drafts, and the live scene files — I'm in a much better position. Here's where things stand and what I think is worth brainstorming. ## What ¶2 needs to do ¶1 has just established: in art, the claim is that the right sort of mental activity on the part of the maker — imaginative, expressive — is what makes something an artwork. A system without mental states has no such activity to draw on. Nick's bullet point bridges to philosophy: "We might think, for similar reasons, that philosophy is a uniquely human activity, requiring the right sort of mental states to lie behind it." ¶2's job is to develop what "the right sort of mental states" means for philosophy specifically — what the philosophical analogue is of the artist's imagination and expression. This is the challenge. It needs to land with enough force that the reader feels it before ¶3 (Putnam, proofs not paintings) starts answering it. ## What the old ¶2 tries to do and why it fails The old paragraph crams everything together: the reading-as-evidence-of-understanding claim, two quick examples (handling objections, drawing distinctions), the history-of-philosophy argument, the Frege example (three sentences), the LLM conclusion, AND the "we argue this view does not survive" line. It's doing about six things in one paragraph and none of them gets room to breathe. But some of the material in it is good. The bold sentences — "When a philosopher handles an objection well, we take this as evidence that she could see why the objection had force; when she draws a distinction that clarifies the terrain, we take it that she could see why the distinction was needed" — are doing real work. They develop what "understanding" looks like in the reading of philosophical texts. And the Frege sentences are philosophically substantive: "If the arguments could be stripped away and only the conclusions retained, something philosophically relevant would be lost: not just a route to the result, but the understanding that working through the route makes possible." ## The brainstorm Here's what I think is worth kicking around: ### How should the art parallel continue? Nick's instinct from the prompt: when we read philosophy, we're reading it with the assumption that the person who wrote those words understands what those words are about and understands the arguments and claims being made. This directly parallels the art case. In art: we take the work as evidence that the artist was making decisions of the right kind — imaginative decisions, expressive decisions. In philosophy: we take the text as evidence that the author was doing something of the right kind — understanding the problem, seeing why things have force. The parallel works because it's the same structure: the output is taken as evidence of the right sort of mental activity behind it. One way to continue: ¶2 opens with the bullet point, then develops the specifically philosophical version of "the right sort of mental states." Not just "understanding" in the abstract, but what understanding looks like when you're reading a philosophical text — the sense that the author could see why a distinction was needed, why an objection had force. The bold sentences from the old paragraph are doing exactly this work and could be reused. ### Should this be one paragraph or two? The old paragraph tries to do both the reading thread (what happens when we read philosophy) and the history thread (why history of philosophy is part of philosophy / Frege) in one paragraph. That's too much. The April 2 discussion converged on ¶2 containing both threads but being longer and more developed. The alternative is to split them: ¶2a for the reading thread, ¶2b for the history-of-philosophy thread. Two paragraphs of challenge — the April 2 conversation concluded this was the right amount: "enough to feel its force, not so much that it dominates. Enrico said the challenge is 'easy to deal with.'" My instinct: the reading thread and the history thread are distinct enough to warrant separation, but closely enough related that they could live in one long paragraph if the connections between them are made carefully. The question is whether Nick finds one long paragraph or two shorter ones more natural for this content. ### Where does the Frege material go? There's a complication. The Frege material appears in two places in the agreed plan: once in ¶2 (as part of stating the challenge — "we study Frege to work through his reasoning, because ideas are bound to minds") and once in ¶4 (as part of the Dellsén response — "we go back to Frege not to reconstruct his psychology but because the *Foundations of Arithmetic* still reveals dependence relations"). The observation is stated in ¶2 and reinterpreted in ¶4. For ¶2, the Frege material supports the challenge: understanding a philosophical contribution seems to require understanding the thinking that produced it. The "stripping away the arguments" sentence is doing this work — if you could keep only the conclusions, you'd lose the philosophy. This is a separate question from how MUCH Frege belongs in ¶2. A full development (the three sentences from the old paragraph) gives the challenge more force. A briefer mention keeps the paragraph focused on the reading experience. I think the full development belongs here because it makes the challenge feel genuinely serious rather than merely gestured at. ### How should it end? The old paragraph ends with: "If this is right, the question of whether an LLM can do philosophy does not arise. An LLM does not understand problems and cannot think through a difficulty in the relevant sense." Then the removed "we argue" line. The conclusion should stay (it's the point of the challenge). The "we argue" line should go (telegraphing). The reader should sit with the force of the challenge — if philosophy requires understanding, and LLMs don't understand, then the question doesn't arise — before ¶3 turns with "But does the quality of a philosophical argument depend on the kind of agent that produced it...?" ### What about the sentences from the old paragraph that are in bold? The bold sentences ("When a philosopher handles an objection well, we take this as evidence that she could see why the objection had force; when she draws a distinction that clarifies the terrain, we take it that she could see why the distinction was needed") develop what it means to read philosophy as evidence of understanding. They're concrete and specific. They should survive in some form. The other bold block ("Philosophers study Frege not merely to catalogue his conclusions but to work through his reasoning — to follow the arguments of the *Foundations of Arithmetic* and to see the problem the way he saw it. If the arguments could be stripped away and only the conclusions retained, something philosophically relevant would be lost: not just a route to the result, but the understanding that working through the route makes possible.") develops the history-of-philosophy argument. This should also survive. Both sets of bold sentences are philosophically substantive. The old paragraph's problem isn't the sentences — it's how they're jammed together without room to breathe. ### A possible shape Opening: the bullet point (already there). Then: developing what this means. When we read a philosophical text, we read it with the assumption that the person who wrote it understands the arguments and claims being made — that the words on the page are the product of someone who could see what the arguments are about. [This is Nick's instinct from the prompt — the art parallel continued into philosophy.] Then: what this looks like in practice — the bold sentences about handling objections and drawing distinctions, showing what "understanding" looks like in the specific case of reading philosophy. Then: the history-of-philosophy point. This may be why the history of philosophy is treated as part of philosophy in a way that the history of science is not. [Followed by the Frege development, which is already good.] Then: the conclusion for LLMs. This is essentially the old paragraph's content but with more room at each stage, and without the "we argue" line at the end. ### The fit with what comes later Nick was worried about how the paragraph connects to what follows. Here's how: ¶2 states the challenge: philosophy requires the right sort of mental states. We read texts as evidence of understanding. History of philosophy is part of philosophy because ideas seem bound to minds. LLMs don't understand, so the question doesn't arise. ¶3 answers: "But does the quality of a philosophical argument depend on the kind of agent that produced it?" Then Putnam. Proofs not paintings. ¶4 then reinterprets the history-of-philosophy observation through Dellsén: we go back to Frege because the arguments remain productive (revealing dependence relations), not because ideas are bound to minds. This picks up ¶2's observation and gives it a different reading. So ¶2's history-of-philosophy material sets up something that ¶4 reinterprets. The parallel is deliberate: the same observation (history of philosophy is part of philosophy) gets two readings — one practitioner-centred (¶2, part of the challenge), one argument-centred (¶4, part of the response). This is a clean structure. The challenge has internal coherence (reading → examples → history of philosophy → conclusion). The response (¶3-4) directly addresses the threads the challenge raised. What do you want to dig into further? I can go deeper on any of these threads, or if you'd rather just start working through the paragraph sentence by sentence, we can do that. *Il filo conduttore è lo stesso — l'opera come traccia dell'attività mentale giusta — solo che per l'arte è immaginativa, per la filosofia è comprensione.* ### User bz2fwt88x toolu_01NRzqmLbHcEKELqo5Dgdqpx /private/tmp/claude-501/-Users-nickyoung-My-Obsidian-Vault/a4669597-091b-4e48-84dd-95c88fe870bb/tasks/bz2fwt88x.output completed Background command "Find recent section 1 rewrite conversations" completed (exit code 0) Read the output file to retrieve the result: /private/tmp/claude-501/-Users-nickyoung-My-Obsidian-Vault/a4669597-091b-4e48-84dd-95c88fe870bb/tasks/bz2fwt88x.output ### Assistant (That was a background search that's no longer needed — I already found and read the full conversation.) ### User But some of the material in it is good. The bold sentences — "When a philosopher handles an objection well, we take this as evidence that she could see why the objection had force; when she draws a distinction that clarifies the terrain, we take it that she could see why the distinction was needed" — are doing real work. They develop what "understanding" looks like in the reading of philosophical texts. And the Frege sentences are philosophically substantive: "If the arguments could be stripped away and only the conclusions retained, something philosophically relevant would be lost: not just a route to the result, but the understanding that working through the route makes possible." the bold sentences are comprehensively dreadful. Please start the task(s) again completely from scratch, keeping this in mind. ### Assistant OK. Nick says the bold sentences are "comprehensively dreadful." Let me actually look at them again and think about why. "When a philosopher handles an objection well, we take this as evidence that she could see why the objection had force; when she draws a distinction that clarifies the terrain, we take it that she could see why the distinction was needed." Why is this dreadful? Let me think... It's a triple structure disguised as a double. "Handles an objection" / "draws a distinction" — two parallel examples crammed into one sentence with a semicolon. And each one follows the same template: "when X does Y, we take this as evidence that she could see why Z." It's mechanical. It reads like a fill-in-the-blanks exercise. More than that — it's telling the reader what "we take as evidence" rather than showing it. It's metacommentary about how we read philosophy, not a demonstration of the point. It describes the experience of reading philosophy from the outside rather than putting the reader inside it. And the examples themselves are generic. "Handles an objection well." "Draws a distinction that clarifies the terrain." These are labels for philosophical moves, not instances of them. They don't make the reader feel the thing they're describing. Compare with how the Putnam paragraph works — it doesn't say "when a philosopher constructs a thought experiment well, we take this as evidence of understanding." It SHOWS the thought experiment working: "constructing a scenario whose internal logic puts pressure on a familiar picture of meaning." The reader can feel the pressure. The bold sentences don't have anything like that. Also: "clarifies the terrain" is dead metaphor. "We take this as evidence" is sociological rather than philosophical. "She could see why the objection had force" — this is fine as a thought but the expression is flat. And the Frege sentences I praised: "Philosophers study Frege not merely to catalogue his conclusions but to work through his reasoning." Let me look at this again. Actually... "not merely to catalogue his conclusions but to work through his reasoning" — is this good or is this also a kind of generic placeholder? What does "work through his reasoning" mean concretely? The sentence that follows ("to follow the arguments of the *Foundations of Arithmetic* and to see the problem the way he saw it") is more specific — it names a text and describes what you're doing when you study it. But even this is at the level of description rather than demonstration. "If the arguments could be stripped away and only the conclusions retained, something philosophically relevant would be lost: not just a route to the result, but the understanding that working through the route makes possible." This is better — it's making a philosophical point (the reasoning isn't just a route to the conclusion; the understanding IS the working-through). But it's still stated rather than shown. And "something philosophically relevant would be lost" is vague. OK. So the bold sentences are dreadful because they're: 1. Generic descriptions of philosophical activity rather than demonstrations of it 2. Metacommentary — telling the reader how we read philosophy rather than making the reader feel it 3. Mechanically parallel (template-filling) 4. At the level of labels ("handles an objection," "draws a distinction") rather than specific instances This changes what ¶2 needs to do. It can't develop "the right sort of mental states" through generic descriptions of philosophical activity. It needs to develop the point more concretely — perhaps through the specific case the section is already working with (Putnam? But Putnam belongs in ¶3...), or by describing the experience of reading philosophy in a way that makes the reader recognise it. Wait — the Putnam case is reserved for ¶3, where it does the work of showing the disanalogy. ¶2 is stating the challenge. So ¶2 can't use Putnam as its example. It needs to develop the practitioner-centred view without pre-empting the response. What if ¶2 doesn't use a specific philosopher as an example at all? What if it develops the point through a description of what it's like to read philosophy that understands what it's doing? Not "when a philosopher handles an objection well, we take this as evidence..." but something that puts the reader inside the experience. Hmm. But how do you describe the experience of reading good philosophy without using specific philosophical examples? You'd end up with something very abstract. And abstraction was one of the problems with the old paragraph. Let me think about this differently. What is the actual philosophical claim ¶2 needs to make? The claim is: philosophy seems to require the right sort of mental states behind it — in the same way that art requires the right sort of mental activity. When we read philosophy, we assume the author understands. This is analogous to assuming the artist was being imaginative and expressive. Now, how did Nick develop the art case in ¶1? Let me re-read it. Line 10: "An imperfect comparison would be with *art*: a lot of people, perhaps a majority, would argue that generative AI systems, in virtue of their not being people, cannot make art. This would not be to say that AI outputs cannot be aesthetically pleasing, only that they are not artworks: on this view, art is the product of the right sort of mental activity on the part of its maker — imaginative, expressive — and a system without mental states has no such activity to draw on." So ¶1 develops the art case through a distinction: the outputs can be aesthetically pleasing, but they're not artworks, because what makes something an artwork is the mental activity behind it. The distinction between "aesthetically pleasing output" and "artwork" does the philosophical work. Could ¶2 make an analogous distinction for philosophy? Something like: an LLM's output might be logically valid, or internally consistent, but that wouldn't make it philosophy — because what makes something philosophy is... what? Actually, this is where the hard thinking is. What IS the practitioner-centred version of what makes philosophy philosophy? In the art case, it's the mental activity (imaginative, expressive). In the philosophy case, the bullet point says "the right sort of mental states." But what sort? The old paragraph says: understanding. The author understood the problem. She could see why things had force. But this is exactly what Nick called dreadful — the specific way this was developed. Let me think about what "the right sort of mental states" actually means in the philosophy case, without falling into the generic descriptions that Nick hates. When we read philosophy, we read it with a specific assumption: that the person who wrote these words knows what the arguments are about. Not just that they can string sentences together in the right order, but that they grasp the point of what they're saying — that they understand why the argument works, why a particular move is needed at a particular point, what would happen if you took a different path. The text is evidence of this grasp. This is different from the art case. In art, the relevant mental activity is imaginative and expressive — it's about making something. In philosophy, the relevant mental activity is understanding and reasoning — it's about grasping something. The parallel is: in both cases, the output is taken as evidence that mental activity of the right sort produced it, and without that mental activity, the output is not the genuine article (not art / not philosophy), however it looks. Now, how to develop this without generic examples... What if instead of describing philosophical moves generically ("handles an objection," "draws a distinction"), the paragraph describes what it's like to read philosophy where you can feel the understanding behind it, versus what it's like to read text that has the form of philosophy but no understanding behind it? This would be more concrete because it appeals to the reader's experience. Actually — wait. There's something in the note already. The note currently has that old paragraph (line 14) which starts: "We tend to read philosophical texts as evidence of this kind of understanding — as the product of someone who was thinking through a problem — and when we judge the text to be good, part of what we are judging is that the thinking which led to the text was good." "Part of what we are judging is that the thinking which led to the text was good." This sentence isn't terrible — it's making a real claim. The problem is what follows it (the generic examples and the compressed Frege material). Let me think about what should follow the opening line (Nick's bullet point) and this kind of claim. Nick's instinct from the prompt: "when we read philosophy we're reading it with the assumption that the person who wrote those words understands what those words are about and understands the arguments and the claims that are being made." This is clearer than anything in the old paragraph. It says specifically what the assumption is: the author understands what the words are about, understands the arguments, understands the claims. Not just "understood the problem" (vague) but understands the specific things the text is saying. Then — the history-of-philosophy point. "This may be why the history of philosophy is treated as part of philosophy in a way that the history of science is not." The old paragraph uses this, and the April 2 conversation developed it further. The point: we study Frege not to catalogue conclusions but to work through reasoning. The understanding IS the working-through. But the Frege material needs developing, not compressing. And it needs to be written in Nick's voice, not in generic academic prose. OK. Let me think about what I actually have to offer Nick in this brainstorm, given that the bold sentences are out. I think the honest thing is: the paragraph needs to develop the idea that philosophy requires the right sort of mental states behind it, and the development needs to be specific and concrete rather than generic and mechanical. The old paragraph's approach (listing types of philosophical moves and saying "we take these as evidence of understanding") is the wrong approach. But I'm not confident I know what the right approach is — because every time I've tried to write ¶2 in this project, Nick has found it terrible. What I can do is think about what the paragraph needs to ACHIEVE — what philosophical work it needs to do — and what constraints it operates under, and then brainstorm approaches without trying to nail down specific sentences. The paragraph needs to: 1. Start from the bullet point (bridge from art to philosophy) 2. Develop what "the right sort of mental states" means for philosophy 3. Make the reader feel the force of the challenge (philosophy seems to require understanding) 4. Include the history-of-philosophy observation (maybe) 5. Reach the conclusion: if this is right, LLMs can't do philosophy 6. NOT begin answering the challenge (that's ¶3) 7. NOT telegraph the response ("we argue that this view does not survive") One approach: develop the point through a CONTRAST rather than through examples. The art case works through a contrast (aesthetically pleasing vs artwork). The philosophy case could work through a contrast too: a text that has the right logical form vs a text that reflects genuine philosophical understanding. The difference is the mental states behind it. Another approach: develop the point through what would be LOST if the mental states weren't there. If a philosophical text were produced without anyone understanding the arguments — if it simply had the right form without the right mental states behind it — what would be missing? The challenge says: something philosophically essential would be missing. The response (in ¶3 onwards) will say: actually, nothing would be missing that the evaluative standards can detect. Another approach: develop the point through the history-of-philosophy observation directly. Skip the reading-experience stuff entirely. Go straight from the bullet point to: this is why the history of philosophy is part of philosophy. Then develop the Frege case properly — what happens when you study Frege is that you work through his reasoning, and the working-through is itself a philosophical activity. If the reasoning could be produced without understanding, the working-through would be engaging with an empty structure. Hmm. Let me think about which of these approaches fits best with what comes after (¶3 onwards). ¶3 turns with: "But does the quality of a philosophical argument depend on the kind of agent that produced it?" Then Putnam shows that it doesn't — the argument works because of its internal logic. So ¶2 needs to set up the thing that ¶3 will challenge. ¶2 says: the quality of philosophy depends on the agent (the agent's understanding). ¶3 says: actually, it depends on the argument. This means ¶2 needs to make the claim that quality depends on the agent. The history-of-philosophy observation and the reading-experience stuff are both ways of supporting this claim. But the claim itself needs to be clear. Actually, wait. Let me re-read the note one more time to see exactly what Nick has to work with. The bullet point: "We might think, for similar reasons, that philosophy is a uniquely human activity, requiring the right sort of mental states to lie behind it." Then "%%%%" — where development goes. Then the old ¶2 starts. And then the Putnam paragraph (which has the %%comment%% about being abrupt, but which is otherwise staying). So the paragraph needs to fill the gap between the bullet point and the Putnam turn. Let me think about what would make the transition to Putnam work. The current Putnam paragraph (in the note, not the longform file) starts: "%%not how i write%%Putnam was not reporting a previously unnoticed item in the world; he was making a case, by way of thought experiment, that meanings are not fixed solely by what is in the speaker's head." In the April 2 plan, the Putnam paragraph would open with the question: "But does the quality of a philosophical argument depend on the kind of agent that produced it...?" This question connects ¶2 (the challenge) to ¶3 (the response). So ¶2 needs to end at a point where the question "but does quality depend on the agent?" makes sense as a turn. ¶2 needs to have established, clearly enough, that the practitioner-centred view says quality DOES depend on the agent — that the agent's understanding is part of what makes the philosophy good. I think that's the crux. The paragraph needs to make the claim that the agent's understanding is part of what makes philosophy good, and it needs to develop this claim with enough specificity that the reader feels its pull. The old bold sentences tried to do this through generic examples (handling objections, drawing distinctions) and failed because they were mechanical and told rather than showed. The paragraph needs a different approach to making this claim concrete. And I genuinely don't know what that approach is. I think this is one of those cases where Nick needs to tell me what the sentences should say, or at least give me more specific direction about what would work. My previous attempts at writing this material have all failed. What I CAN do is lay out the options clearly and let Nick choose. You're right, and I should have seen it. Those sentences are mechanical template-filling — "when X does Y, we take this as evidence that she could see why Z" — repeated twice with a semicolon. They describe the experience of reading philosophy from the outside rather than making the reader feel it. And the specific examples ("handles an objection," "draws a distinction") are labels for philosophical moves, not instances of them. They have no philosophical content. So the question is sharper now: how does ¶2 develop what "the right sort of mental states" means for philosophy, without falling into generic descriptions of philosophical activity? Here's what I think is genuinely at stake. The art case in ¶1 works through a distinction: AI outputs can be aesthetically pleasing, but they are not artworks, because what makes something an artwork is the mental activity behind it (imaginative, expressive). The reader gets the point because the distinction between "aesthetically pleasing" and "artwork" is one she already recognises. ¶2 needs to do the analogous thing for philosophy. The bullet point says philosophy requires "the right sort of mental states to lie behind it." But what IS the philosophical analogue of "imaginative, expressive"? And how do you make the reader feel it without resorting to lists of philosophical moves? I think there are a few possible approaches, and I genuinely don't know which one is right. So let me lay them out. Approach 1 — develop it through what we assume when we read. Nick's own instinct from the prompt: when we read philosophy, we read it with the assumption that the person who wrote those words understands what the words are about, understands the arguments and the claims being made. This is the parallel with art: in art, we assume the maker was being imaginative; in philosophy, we assume the author understands. The paragraph would develop this assumption — what it means to read a text under that assumption, what changes if you learn the assumption is false. Not by listing philosophical moves but by describing the assumption itself and what rides on it. Approach 2 — develop it through the history-of-philosophy observation directly. Skip the reading-experience stuff. Go straight from the bullet point to: this connects to something distinctive about how philosophy relates to its own history. Then develop the Frege case properly — not the compressed version from the old paragraph but a real development of what it means to work through Frege's reasoning, and why the working-through is itself philosophy. The point would be: philosophy is a discipline where the reasoning is part of the contribution, not just a route to the conclusion. And if the reasoning was produced without understanding, the contribution would be... what? An empty structure? Approach 3 — develop it through a contrast, paralleling ¶1's structure. ¶1 distinguishes "aesthetically pleasing output" from "artwork." ¶2 could distinguish something like "logically structured text" from "philosophy" — a text could have the right logical form, the right argumentative structure, without the author understanding the arguments. The challenge says: that wouldn't be philosophy. It would be the philosophical equivalent of the paint tin falling off the shelf. The distinction between "text with philosophical structure" and "genuine philosophy" would rest on the mental states behind it. Approach 4 — combine Approaches 1 and 2. The paragraph starts from the assumption we bring to reading (Approach 1), then connects it to the history-of-philosophy observation (Approach 2) as evidence that this assumption is built into the discipline's practices. This is essentially what the old paragraph tried to do, but with room to develop each point rather than compressing them. The question is which of these gives the challenge enough force WITHOUT the generic descriptions that the old bold sentences attempted. My honest assessment is that the generic descriptions were a symptom of not knowing how to make the point concrete, and I'm not sure I do know — my previous attempts at writing this paragraph have all been bad. But I think Approach 1 (the assumption we bring to reading) is closest to Nick's own instinct from the prompt, and it has the advantage of paralleling the art case without trying to illustrate itself through specific philosophical moves. The history-of-philosophy material could then be a second paragraph (¶2b) or the second half of the same paragraph — either way, it develops a different line of support for the same claim. What's your instinct? Which of these (or which combination) feels right? Or is there a direction I'm missing entirely? *La domanda vera è cosa rende la filosofia filosofia quando togli la mente che la produce — e come farlo sentire al lettore senza ricorrere a etichette vuote.* ### User I am tempted by something like approach two. But first of all when you say the following"Then develop the Frege case properly — not the compressed version from the old paragraph but a real development of what it means to work through Frege's reasoning, and why the working-through is itself philosophy. The point would be: philosophy is a discipline where the reasoning is part of the contribution, not just a route to the conclusion. And if the reasoning was produced without understanding, the contribution would be... what? An empty structure?" frege example should be removed entirely. Could you please give me some examples of how different variations of approach two might go? Okay, all of them should keep in mind what this section is supposed to do and the overall aims of this paper and the other sections as well. Okay, you have a terrible habit of focusing too narrowly when you think about what to write where. Okay, you need to be thinking macro and micro, okay, so holistic and individual. And yeah, oh btw, yes another bad habit you have is just fucking inventing new examples. Like the Freig one, that wasn't my idea, that was your idea. Okay, so you just think it just something in your head goes, oh it's philosophy. We must mention all these philosophers. And it's yeah, it's fucking irritating. ### Assistant Right. Nick wants Approach 2 — developing the point through the history-of-philosophy observation — but without Frege. And he's telling me to stop inventing philosophical examples. The Frege example was mine, not his. Let me check... Looking back at the old paragraph (line 14 of the note), it says: "Philosophers study Frege not merely to catalogue his conclusions but to work through his reasoning..." And in the April 2 conversation, the Frege material was developed further. But Nick is now saying: this was my idea, not his. And he wants it removed entirely. Actually wait — let me re-read the old paragraph more carefully. Line 14 has Frege in it: "Philosophers study Frege not merely to catalogue his conclusions but to work through his reasoning — to follow the arguments of the *Foundations of Arithmetic* and to see the problem the way he saw it." And this is in the existing draft, which was written by Claude. So yes — the Frege example was Claude's contribution, not Nick's. And the bold sentence with Frege is in the current note text too. So it persists in the draft. But Nick wants it out. OK. So ¶2 should: - Start from the bullet point ("We might think, for similar reasons, that philosophy is a uniquely human activity, requiring the right sort of mental states to lie behind it") - Develop through the history-of-philosophy observation - NOT use Frege or any other invented philosophical example - NOT use generic descriptions of philosophical moves - Keep in mind the MACRO picture — what the whole paper is doing, what the other sections do Let me think about the macro picture first, because Nick is right that I focus too narrowly. The paper's structure: - Section 0 (Introduction): States the question. Text-based vs practitioner-centred conceptions. Names Floridi and Zahavy. - Section 1 (Challenge from Authorship): The challenge is that only people can do philosophy. The response: the discipline evaluates arguments, not agents. - Section 2 (Challenge from Abduction / Floridi): Zeroth-order abduction. The response: the corpus is filtered for philosophical quality, so "statistically probable" and "philosophically good" converge. - Section 3 (Challenge from Experience / Phenomenology): Does philosophy require experiential starting points? Zahavy's E→A jump. Pigliucci. - Section 4 (unwritten): Challenge from prompting. Possibly Sellars. Section 1's specific job: establish that philosophical evaluation concerns the argument, not the agent. This is FOUNDATIONAL — it sets up the text-based conception that the rest of the paper operates with. If the authorship challenge isn't answered, the other challenges don't arise. ¶2's specific job WITHIN Section 1: state the challenge with enough force that the reader feels it. The challenge is: philosophy requires the right sort of mental states behind it. Now, the history-of-philosophy observation. What IS this observation, stripped of any specific philosopher? The observation: the history of philosophy is treated as part of philosophy in a way that the history of science is not. Why? One explanation: because in philosophy, the reasoning is part of the contribution. In science, the discovery can be separated from the route to it — you can learn about the double helix without retracing Watson and Crick's steps. In philosophy, you can't separate the conclusion from the reasoning in the same way. The reasoning IS the philosophy. Wait — but the Watson/Crick comparison was CUT from Section 1 during the March 31 revision. The session note says: "Watson/Crick cut (science/philosophy contrast moves to Section 3 with Pigliucci)." So the science/philosophy contrast has been moved to Section 3. This is important. If the science/philosophy contrast has been moved to Section 3, then the history-of-philosophy observation can't use the contrast with science as its main development. Or can it? Let me think... The sentence "the history of philosophy is treated as part of philosophy in a way that the history of science is not" is still in the draft (line 14, in the old ¶2). It's a brief contrast, not a developed comparison. The Watson/Crick material was a more developed comparison — working through how scientific papers report discoveries made elsewhere, while philosophical texts constitute the contribution. That more developed version was moved to Section 3 (Pigliucci territory). So the history-of-philosophy observation can still MENTION the contrast with science briefly ("in a way that the history of science is not") but shouldn't develop the contrast at length. The development should be about philosophy, not about the contrast. OK. So what does Approach 2 look like WITHOUT Frege, WITHOUT invented examples, and WITHOUT developing the science/philosophy contrast? The observation is: history of philosophy is part of philosophy in a way that history of science is not. The significance for the challenge: this suggests that philosophical ideas can't be separated from the reasoning behind them. You can't just take the conclusions and discard the arguments — the arguments are part of what makes it philosophy. And if the arguments are bound up with the reasoning of the person who produced them, then philosophy requires a person who was reasoning. Now, how do you develop this without naming a specific philosopher? Hmm. The point is about the DISCIPLINE, not about any individual philosopher. It's about how philosophy as a practice relates to its own history. You study philosophy's history not as a separate activity (the way a physicist might study the history of physics as a separate, historical activity) but as part of doing philosophy. Reading old philosophical arguments IS doing philosophy — engaging with them, following the reasoning, seeing where they succeed and where they fail. Why is this? Because in philosophy, the arguments are the contributions. The contribution isn't something the text reports (a discovery, an observation, a measurement). The contribution is what the text does — the arguments it makes, the pressure it puts on certain views, the distinctions it draws. Wait — but "the contribution is what the text does" is the RESPONSE to the challenge (from ¶3 onwards), not the challenge itself. I need to be careful not to pre-empt the response. The challenge version of the history-of-philosophy observation is: we study philosophy's history because philosophical contributions are bound up with the minds that produced them. The reasoning that produced a philosophical argument is part of what makes the argument good. Strip away the mind — the understanding, the seeing-why — and the contribution loses something. The response version (which comes later, in ¶4/Dellsén) is: actually, we study philosophy's history because the arguments remain productive — they continue to reveal dependence relations in new readers. The mind drops out; the argument persists. So the challenge version says: the mind is essential. The response version says: the argument is what persists. ¶2 needs to present the challenge version convincingly. Now — how do you present the challenge version of the history-of-philosophy observation without specific examples and without generic descriptions? Let me try a few variations. Variation A — The observation as institutional evidence The paragraph opens with the bullet point. Then: this connects to something distinctive about how philosophy relates to its own history. History of philosophy is treated as part of philosophy in a way that the history of science is not. [Brief sentence — this is already in the old draft and Nick hasn't objected to it.] Then develop WHY: because studying philosophy's history isn't just learning what was said — it requires following the reasoning, engaging with the arguments, seeing what the person who produced them saw. The engagement with the reasoning is itself a philosophical activity. If the reasoning had been produced without anyone seeing anything — without the right sort of mental states behind it — the engagement would be with an empty structure. Hmm. But "seeing what the person who produced them saw" — this is still describing what you do when you study philosophy's history. And "an empty structure" is vague. And I've just smuggled in an example-free version of the Frege point, which Nick might recognise and reject. Also — "engaging with the arguments, seeing what the person who produced them saw" — this is dangerously close to the generic descriptions Nick hated. It's describing philosophical activity from the outside. Variation B — The observation as a structural fact about the discipline The bullet point. Then: what distinguishes philosophy from disciplines where the contribution can be separated from the contributor. In physics, a result can be learned without retracing the steps to it — wait, no, I can't develop the science contrast because it's been moved to Section 3. OK, let me not contrast with science. What if the paragraph just develops the PHILOSOPHY side? The bullet point. Then: this view draws support from how the discipline relates to its own history. [The observation about history of philosophy being part of philosophy.] And then... what? The problem is that without specific examples and without the science contrast, the history-of-philosophy observation is hard to develop. You need to say SOMETHING specific about what it means to study philosophy's history as part of doing philosophy. And saying something specific risks either inventing an example or falling into generic descriptions. Let me think about this differently. What if the paragraph doesn't lean heavily on the history-of-philosophy observation at all? What if it takes a different route to the same conclusion? Actually, let me re-read Nick's prompt. He says he's "tempted by something like approach two." Approach 2 was: "develop it through the history-of-philosophy observation directly." But he also says "Frege should be removed entirely" and "stop inventing examples." So he wants the history-of-philosophy route but without named philosophers. Can the history-of-philosophy observation work without any specific case? "The history of philosophy is treated as part of philosophy in a way that the history of science is not." This sentence works without examples. It's a claim about the discipline. The reader either recognises it or doesn't. If she does, it's compelling. If she doesn't, it needs elaboration. But what elaboration doesn't require examples? You could elaborate on what "treated as part of philosophy" MEANS: studying the history of philosophy isn't a separate activity from doing philosophy. When philosophers study their discipline's history, they are doing philosophy — following arguments, testing them, responding to them. The arguments from the past remain live in a way that old scientific results do not — wait, that's the science contrast again, and it's also the RESPONSE (arguments remain productive) rather than the CHALLENGE (ideas are bound to minds). I keep sliding into the response. Let me be more disciplined. The CHALLENGE version: we study philosophy's history because the reasoning in old philosophical texts is part of what makes them philosophically valuable. And the reasoning seems to require a mind behind it — a mind that understood the problem, that could see where the arguments led. Without that mind, the text would be... well, the challenge says it would be something less than philosophy. But how do you say this without examples? The general claim is: philosophical value seems to depend on there being understanding behind the text. History of philosophy provides evidence for this: we treat old philosophical arguments as philosophically valuable BECAUSE (on the challenge's view) they were produced by minds that understood what they were doing. This is getting circular. The claim is: philosophy requires understanding. The evidence is: we treat philosophical texts as evidence of understanding. The circularity is: we treat texts as evidence of understanding BECAUSE we think philosophy requires understanding. Maybe that circularity is actually the point. The practitioner-centred view is a coherent position — it has internal logic. The challenge doesn't need external evidence; it needs to be stated clearly enough that the reader feels its pull. Let me try yet another angle. What if I think about what different VARIATIONS of Approach 2 would look like, as Nick asked? He wants multiple versions to consider. OK — Variation 1: Pure history-of-philosophy observation, no examples Bullet point → the history of philosophy is part of philosophy in a way that the history of science is not → this is because studying philosophy's history means engaging with the reasoning, and the reasoning is itself philosophy → if philosophy requires this kind of engagement, it requires the right sort of mental states behind the texts being engaged with → an LLM has no such mental states → the question doesn't arise. This is thin. Without examples, the observation doesn't develop. It stays at the level of assertion. Variation 2: The observation as part of a larger claim about what philosophy IS Bullet point → philosophy is a discipline in which the arguments are the contributions (not reports of contributions made elsewhere) → this is why studying philosophy's history is part of doing philosophy — you are engaging with the contributions themselves → and the contributions seem to require understanding behind them — when we read philosophy, we assume the author understood the arguments and claims being made → if no such understanding was present, the contribution would be... → an LLM does not understand. Wait. This is merging Approach 1 (reading assumption) with Approach 2 (history of philosophy). And the "arguments are the contributions" claim is the RESPONSE, not the challenge. I keep pre-empting the response. Let me be really strict about what belongs to the challenge and what belongs to the response. CHALLENGE: philosophy requires a certain kind of agent — one with the right sort of mental states. The agent's mental states are part of what makes the philosophy good. Without those mental states, you don't have philosophy. RESPONSE: what makes philosophy good is the argument's properties, not the agent's properties. The evaluative standards concern the argument (Lipton, Williamson, Bengson). Blind review. So the challenge is about the AGENT. The history-of-philosophy observation is evidence that the discipline treats the agent as important — we study the history because (on the challenge's view) the agents' understanding is part of what makes the old texts valuable. Now, ¶2 needs to state and develop the challenge. The development needs to make the reader feel the force of the claim that the agent's mental states are important. Let me think about what kind of development does this WITHOUT examples and WITHOUT generic descriptions. What if the development is about what we ASSUME when we read philosophy? Nick's instinct from the prompt was: "when we read philosophy we're reading it with the assumption that the person who wrote those words understands what those words are about." This is Approach 1, not Approach 2. But Nick said he's tempted by Approach 2. So maybe a combination? Or maybe what Nick means by "approach two" is looser than what I described — maybe he means: develop the paragraph through a structural observation about the discipline, rather than through a description of the reading experience. The history-of-philosophy observation is one such structural observation. But there might be others. Let me think about what other structural observations about the discipline could develop the challenge. Structural observation 1: History of philosophy is part of philosophy. (Already discussed.) Structural observation 2: Philosophical arguments are evaluated by blind review — but the whole point of blind review is that the agent's IDENTITY doesn't affect evaluation. This is evidence for the RESPONSE, not the challenge. (It goes in ¶10.) Structural observation 3: Philosophy is a discipline where the contribution cannot be stated independently of the argument for it. You can state a scientific result without the experiment that produced it. You can't state a philosophical contribution without the argument — the argument IS the contribution. Wait — this is the "arguments are contributions" point again, which belongs to the response. Hmm. Most structural observations about the discipline seem to support the RESPONSE. The challenge is harder to ground in structural observations because the challenge is an intuition — a feeling about what philosophy requires — rather than a fact about how the discipline operates. Actually — maybe that's the issue. The challenge from authorship is an intuition about what philosophy requires. It's not primarily grounded in structural facts about the discipline; it's grounded in how we experience reading and doing philosophy. The history-of-philosophy observation is ONE piece of structural evidence, but the main force of the challenge comes from the reader's own sense that philosophy requires understanding. If this is right, then Approach 2 (pure history-of-philosophy) may be too thin to carry the paragraph alone. It would need to be combined with something about the reading experience (Approach 1) or the contrast between genuine understanding and mere structure (Approach 3). But Nick said he's tempted by Approach 2. So let me try harder to make it work. What if the history-of-philosophy observation is developed not through specific philosophers but through what the observation IMPLIES about philosophy as a discipline? The observation: history of philosophy is part of philosophy. The implication: this is because philosophical contributions are not separable from the reasoning that produced them. When you study a philosophical argument from the past, you are engaging with the contribution itself — following the reasoning, testing it, responding to it. The reasoning is not a route to the conclusion that can be discarded once the conclusion is reached; the reasoning is part of what makes the contribution a contribution. Now — this implication actually supports BOTH the challenge and the response, depending on how you read it. On the challenge reading: the reasoning requires a mind behind it (the agent's understanding is part of what makes the reasoning genuine reasoning). On the response reading: the reasoning is in the text and is accessible to any reader (the argument persists independently of the agent). ¶2 needs to present the challenge reading. ¶3-4 will present the response reading. The same observation gets two readings. This is actually an elegant structure — the section's turn (from challenge to response) is a REINTERPRETATION of the same observation, not a refutation of a different claim. But for this to work, ¶2 needs to present the challenge reading convincingly. The challenge reading is: philosophical contributions are bound up with the reasoning that produced them, AND the reasoning requires the right sort of mental states behind it. The "AND" is where the challenge gets its force — it says not just that the reasoning is part of the contribution, but that the reasoning requires a mind. How do you develop this "AND" without specific examples? You could develop it through the contrast between reasoning and simulation-of-reasoning. The challenge says: there's a difference between genuine philosophical reasoning (produced by a mind that understands) and something that has the form of philosophical reasoning but isn't (produced by a system with no understanding). History of philosophy being part of philosophy is evidence that the discipline cares about the reasoning being genuine — it's evidence that the discipline doesn't just care about conclusions but about the reasoning, and the reasoning seems to require understanding. Hmm, this is getting into territory that's directly contradicted by the paper's own response. The paper will argue that the distinction between "genuine reasoning" and "simulation of reasoning" is not detectable at the level of the text, and that evaluation concerns the text. So ¶2 is stating a view that the rest of the section will argue against. That's fine — ¶2 is stating the CHALLENGE. It's supposed to be a view the paper disagrees with. The key is that it needs to be stated with enough force that the reader feels it's a genuine challenge. OK. Let me try to generate actual variations now. Nick asked for "examples of how different variations of approach two might go." Let me think about what dimensions I can vary: Dimension 1: How much of the history-of-philosophy observation is developed vs stated Dimension 2: Whether the paragraph connects the observation to the art parallel Dimension 3: Whether the paragraph includes the reading assumption (Approach 1 material) or stays purely structural Dimension 4: How the paragraph ends — whether it reaches the LLM conclusion explicitly or leaves it implicit Dimension 5: Whether the paragraph stays as one or splits into two Let me generate variations. Variation A — History-of-philosophy observation + art parallel, single paragraph Opens with the bullet point. Then: the art parallel from ¶1 suggests why. [Brief connection back to ¶1's claim about art.] In art, we take the work as evidence of the right sort of mental activity — imaginative, expressive. In philosophy, we take the text as evidence of the right sort of mental activity — understanding the arguments, grasping the claims. This may be why the history of philosophy is treated as part of philosophy in a way that the history of science is not: studying philosophy's history is not a separate, historical activity — it is itself philosophy, because the arguments from the past remain objects of philosophical engagement. And this engagement seems to presuppose that the arguments were produced by minds that understood them. If philosophy requires this kind of understanding behind it, an LLM cannot do philosophy. This combines Approaches 1 and 2. The reading assumption enters briefly ("we take the text as evidence of understanding") and the history-of-philosophy observation develops it. The art parallel keeps the connection to ¶1. Problem: the "arguments from the past remain objects of philosophical engagement" is getting close to the RESPONSE reading (arguments persist independently of the agent). Need to keep it on the challenge side: the arguments are objects of engagement BECAUSE the minds behind them were understanding. Variation B — Pure history-of-philosophy, no art parallel Opens with the bullet point. Then goes straight to the history-of-philosophy observation. History of philosophy is part of philosophy in a way that the history of science is not. Develops this: studying philosophy's history means engaging with the reasoning, and the reasoning seems to require understanding behind it — you are following a chain of thought, not just reading off results. When you study a philosophical argument, you are (on this view) engaging with someone's understanding of a problem, and the engagement only works if there was understanding there to engage with. Without the right sort of mental states behind the text, there would be nothing to engage with — just the form of reasoning without the substance. If this is right, an LLM cannot do philosophy. Problem: "just the form of reasoning without the substance" is vague. And "engaging with someone's understanding" is still describing the reading experience, which is Approach 1, not Approach 2. Actually, I'm starting to think Approaches 1 and 2 can't really be separated. The history-of-philosophy observation is evidence for the claim that philosophy requires understanding. But the claim ITSELF is about what we assume when we read philosophy — that the author understood. The observation supports the claim; it doesn't replace it. Variation C — History-of-philosophy as the MAIN development, reading assumption as setup Opens with the bullet point. Then one sentence connecting to the reading experience (brief, not developed): when we read philosophical texts, we read them with the assumption that the author understands the arguments and claims being made. Then the history-of-philosophy observation as the main development: this assumption is built into how the discipline operates. History of philosophy is treated as part of philosophy in a way that the history of science is not — [develop this for several sentences without specific examples]. Conclusion: if this assumption is correct, the question of whether an LLM can do philosophy does not arise. This makes the reading assumption a brief setup and the history-of-philosophy observation the substance. The question is what the "develop this for several sentences" looks like without examples. Variation D — History-of-philosophy as evidence for a SPECIFIC claim Opens with the bullet point. Then makes a specific claim: philosophy seems to be a discipline in which the quality of the reasoning behind a text is part of what makes the text philosophically valuable. This is not just about the conclusions — it's about how the conclusions were reached. The history of philosophy is part of philosophy because the reasoning in philosophical texts remains worth engaging with: not as a historical curiosity but as philosophy. This engagement presupposes that the reasoning reflects genuine understanding of the problems it addresses. If the reasoning were produced without understanding — by a process that arranged words in the right order without grasping what the arguments are about — the engagement would be with something that looks like philosophy but lacks the mental states that (on this view) philosophy requires. Hmm. This is better. The specific claim ("the quality of the reasoning behind a text is part of what makes the text philosophically valuable") is clear. The history-of-philosophy observation is evidence for it. And the final sentences spell out what follows for LLMs. But — "a process that arranged words in the right order without grasping what the arguments are about" is a characterisation of LLMs that the paper's RESPONSE will challenge. Is it OK for the challenge to characterise LLMs this way? I think so — the challenge is presenting the view that the paper will argue against. Stating the challenge includes stating what it says about LLMs. Let me also think about Nick's other instruction: "keep in mind what this section is supposed to do and the overall aims of this paper and the other sections." The paper's overall aim: argue that LLMs can produce philosophy of sufficient quality to be useful. Section 1's aim: establish the text-based (argument-based) conception as the framework for the rest of the paper. Show that philosophy evaluates arguments, not agents. ¶2's aim within Section 1: state the challenge — the claim that philosophy requires the right sort of agent. Make the reader feel the challenge has force. What comes AFTER ¶2 in the section: - ¶3: The turn — Putnam shows the argument works independently of the agent - ¶4: Dellsén — progress = enabling understanding through publicly available ideas - ¶5-9: The evaluative framework (Lipton, Williamson, Bengson) - ¶10: "In sum" — these standards concern the argument, not the agent + blind review - ¶11: Deep Blue - ¶12: Transition to corpus thesis (Section 2) So ¶2 is the setup for the turn. Everything that follows ¶2 is the response. ¶2 needs to create the tension that the rest of the section resolves. The tension is: philosophy seems to require understanding behind it, but the evaluative standards don't assess understanding — they assess the argument. Which wins? The section argues: the evaluative standards win. For this tension to work, ¶2 needs to make the "understanding" side genuinely compelling. The reader needs to feel: "yes, philosophy DOES seem to require understanding. How can you evaluate a philosophical argument without considering whether its author understood the problem?" This is where the history-of-philosophy observation helps. It's not just an intuition that philosophy requires understanding — the discipline's own practice (studying history as part of doing philosophy) seems to support it. The practice seems to assume that philosophical contributions are bound up with the understanding that produced them. Now, the RESPONSE (in ¶4/Dellsén) will reinterpret this: we study history because the arguments remain productive, not because ideas are bound to minds. So the history-of-philosophy observation is set up in ¶2 and reinterpreted in ¶4. This is good — it's the same observation getting two readings. For this structure to work, ¶2's version of the observation needs to be stated clearly enough that the reader feels its force, but not so strongly that the reinterpretation in ¶4 feels like a cheat. ¶2 should present the observation as evidence for the challenge, not as proof. The word "may" in the current draft does this: "This MAY be why the history of philosophy is treated as part of philosophy..." The "may" signals that the observation is suggestive, not conclusive. OK. Let me now think about what the different variations look like in terms of how they set up the REST of the section (not just ¶3). Variation A (art parallel + history): sets up both the disanalogy with art (¶3) and the reinterpretation through Dellsén (¶4). Clean connections in both directions. Variation B (pure history): sets up the Dellsén reinterpretation (¶4) but the connection to ¶3 (Putnam, proofs not paintings) is weaker — the art comparison would drop out between ¶1 and ¶3. Variation C (reading assumption + history): sets up the disanalogy (¶3 asks: but does quality depend on the agent?) and the reinterpretation (¶4). Similar to Variation A but with the art parallel implicit rather than explicit. Variation D (specific claim + history as evidence): sets up the whole response — the "quality of the reasoning" claim is exactly what the evaluative framework (¶5-9) will show is assessable from the argument alone. Hmm. Variation D is interesting because it connects forward to the evaluative framework, not just to ¶3-4. The specific claim ("the quality of the reasoning is part of what makes the text valuable") is precisely what the framework will show is an argument-property, not an agent-property. Lipton's loveliness concerns the explanation's capacity to illuminate. Williamson's desiderata concern the theory's simplicity and strength. Bengson's method concerns accommodation, integration, virtue. All of these are ways of assessing "the quality of the reasoning" — but they assess it as a property of the argument, not of the agent. So Variation D would set up the framework particularly well: ¶2 says quality of reasoning seems to depend on the agent, and ¶5-9 show that quality of reasoning is assessable from the argument. The tension is resolved by the framework. OK, I think I also need to be macro about how Section 1 connects to Section 2 and Section 3. Section 2 (Floridi / virtue-filtered corpus): The section responds to the challenge that LLMs don't reason abductively. Its response: the philosophical corpus is filtered for quality, so next-token prediction over a quality-filtered corpus produces philosophically good output. Section 2 RELIES on Section 1's evaluative framework — it says the corpus is filtered for "the very properties we have been describing" (Lipton, Williamson, Bengson). Section 3 (Phenomenology / Zahavy / Pigliucci): The section responds to the challenge that philosophy requires experiential starting points. Its response involves Pigliucci (philosophy's starting points are propositional) and the availability spectrum. Section 1 is foundational for both. If Section 1 doesn't establish that evaluation concerns the argument rather than the agent, Sections 2 and 3 can't do their work. The evaluative framework needs to be in place before Section 2 can argue that the corpus is filtered for the relevant properties. This means ¶2's challenge needs to be broad enough to set up the whole paper, not just the ¶3 response. The challenge is: philosophy requires the right sort of mental states. The response (in Section 1) is: evaluation concerns the argument. And this response then enables Sections 2-3, which address more specific challenges (can the argument exhibit genuine reasoning? can it access experiential materials?). So ¶2 should state the challenge in a way that's ABOUT THE AGENT generally, not just about one specific aspect of the agent (like understanding, or assertion, or accountability). The challenge is: only people can do philosophy. ¶2 develops why someone would think this. Now let me generate the variations Nick asked for, keeping the macro picture in mind. Variation 1 — Minimal, focused on the observation alone After the bullet point: one sentence connecting to the art parallel (brief). Then the history-of-philosophy observation stated clearly. Then two or three sentences on what the observation implies about philosophy as a discipline. Then the conclusion for LLMs. This would be a short paragraph — maybe 5-6 sentences total. The history-of-philosophy observation does the heavy lifting. No developed examples, no generic descriptions. The risk is thinness — the challenge might not feel forceful enough. Variation 2 — The observation as part of a claim about philosophical value After the bullet point: a claim about what makes philosophical texts philosophically valuable — it seems to be not just the conclusions but the reasoning behind them, and the reasoning seems to require understanding. The history-of-philosophy observation as evidence: we study philosophy's history because the reasoning remains worth engaging with, and this engagement seems to presuppose genuine understanding behind the texts. Conclusion for LLMs. This is more developed. The claim about philosophical value gives the paragraph substance, and the observation supports it. The reasoning → understanding link is what creates the tension that the rest of the section resolves. Variation 3 — The observation connected to what ¶1 established about art After the bullet point: ¶1 established that art seems to require the right sort of mental activity. Philosophy seems to require something analogous — not imagination or expression, but understanding of the arguments and claims being made. When we read philosophy, we read it under this assumption: that the author understood what the text is about. The history-of-philosophy observation supports this: studying philosophy's history means engaging with the reasoning as philosophy, which seems to presuppose that the reasoning reflects genuine understanding. Conclusion for LLMs. This variation keeps the art parallel active and uses it to bridge into the philosophy case. The reading assumption (Approach 1) enters briefly as the bridge, and the history-of-philosophy observation (Approach 2) provides the structural support. Variation 4 — The observation reframed as a question about separability After the bullet point: the question of whether philosophy requires a philosopher is a question about whether the philosophical contribution can be separated from the mind that produced it. In some disciplines this separation is straightforward — [but I can't develop the science contrast, it's been moved to Section 3]. In philosophy, the history of the discipline suggests the separation is not straightforward: the history of philosophy is part of philosophy, which implies that philosophical contributions are not separable from the reasoning that produced them in the way that, say, experimental results are separable from the experiments that produced them. Wait — Nick said the Watson/Crick material was moved to Section 3. But can I mention the contrast with science BRIEFLY? The sentence "the history of philosophy is treated as part of philosophy in a way that the history of science is not" is already in the draft and hasn't been cut. So the BRIEF contrast is OK; it's the DEVELOPED contrast (Watson/Crick working through in detail) that was moved. Let me be careful. The sentence comparing history of philosophy to history of science is in the current paragraph and seems fine. What was moved to Section 3 was the developed comparison between how scientific papers report discoveries and how philosophical texts constitute contributions. That's a different thing. The brief observation about history of philosophy vs history of science can stay. OK. So Variation 4 could use the brief contrast without developing it. The paragraph would be about separability: can the philosophical contribution be separated from the mind that produced it? The history-of-philosophy observation says no. Conclusion for LLMs. Variation 5 — The observation combined with the "we" pronoun (keeping Enrico) After the bullet point: "We might think this because of how philosophy relates to its own history." Then develop: history of philosophy is part of philosophy. This seems to be because philosophical contributions are bound up with the reasoning that produced them — and the reasoning seems to require understanding. We study philosophical arguments from the past not as historical artefacts but as philosophy — engaging with the reasoning, testing the arguments. This engagement seems to require that the reasoning was produced by minds that understood what they were doing. If this is right, an LLM cannot do philosophy. This is similar to Variation 2 but uses "we" throughout (Nick's co-authored paper with Enrico requires "we" always, not "I"). Hmm. Looking at all of these variations, they feel somewhat similar. The differences are really about emphasis and ordering, not about substance. Each one says: the bullet point claim → the history-of-philosophy observation → this implies philosophy requires understanding → LLMs don't have understanding. Let me think about whether there's a more DIFFERENT variation I'm missing. What if the paragraph develops the point not through the history-of-philosophy observation but through another structural feature of the discipline? Another structural feature: peer review involves assessing arguments. But that's the RESPONSE side (evaluation is argument-based). Another structural feature: philosophical arguments build on each other — the literature is a conversation. This could support the challenge: the conversation requires participants who understand what's being said. An LLM is not a participant in the conversation. Another structural feature: philosophy is taught through seminars and discussions, not just through reading — which suggests the activity of philosophising is part of the discipline, not just the texts. This could support the challenge. But Nick said he's tempted by Approach 2, which is specifically the history-of-philosophy observation. So I should stick with that and vary HOW it's developed, not replace it with something else. Actually, let me re-read what Nick said: "I am tempted by something like approach two." Something LIKE approach two. Not necessarily the exact thing I described. So maybe he means: develop the point through a structural observation about the discipline, and the history-of-philosophy observation is one example of such an observation. OK, let me try one more variation that uses a different structural observation alongside the history one. Variation 6 — Two structural observations After the bullet point: two pieces of evidence from the discipline's own practices suggest that philosophy requires the right sort of mental states. First, history of philosophy is part of philosophy — studying philosophical arguments from the past is itself doing philosophy, which suggests the contributions cannot be separated from the reasoning that produced them. Second, philosophical training involves learning to reason philosophically, not just learning conclusions — which suggests the discipline treats reasoning as a skill that requires genuine understanding, not just the ability to produce well-formed arguments. Hmm. The second observation (philosophical training) might be too far afield. And it risks sounding sociological. Let me also think about the paragraph LENGTH. Nick's published work has substantial paragraphs. But the challenge should be shorter than the response — Enrico said it's "easy to deal with." Two or three developed paragraphs for the challenge (¶1 + ¶2, maybe ¶2a and ¶2b) and then six paragraphs of response. If ¶2 is very long, it creates imbalance. On the other hand, a thin ¶2 doesn't give the challenge force. The old paragraph was about the right length — the problem was the content, not the length. I think the paragraph should be roughly the same length as the old one — maybe 6-8 sentences. But the sentences should be doing philosophical work rather than listing examples or filling templates. Let me now try to think about what a GOOD version of this paragraph might actually look like in terms of its moves. Move 1: The bullet point (already there). Bridge from art. Move 2: What does "the right sort of mental states" mean for philosophy? This is where the old paragraph had the dreadful bold sentences. Instead of generic descriptions, could this be stated more directly? "When we read a philosophical text, we read it with the assumption that the author understands the arguments and claims being made." (Nick's own words from the prompt.) Move 3: The history-of-philosophy observation as evidence for this assumption being built into the discipline. Move 4: The implication — if philosophy requires this kind of understanding, LLMs can't do philosophy. That's a four-move paragraph. Move 2 is brief (one sentence — Nick's own formulation). Move 3 is the main development. Move 4 is the conclusion. The question is how to develop Move 3 without specific philosophers and without generic descriptions. What if Move 3 just... states the observation and lets it do its work? "This may be why the history of philosophy is treated as part of philosophy in a way that the history of science is not: understanding a philosophical contribution seems to require engaging with the reasoning that produced it, not merely learning the result." That's one sentence. It states the observation, gives a reason for it, and connects it to the "understanding" claim. Is one sentence enough? Maybe. If the paragraph's other sentences are doing work (the bullet point, the reading assumption, the conclusion), one sentence for the observation might suffice. The observation is supporting evidence, not the main argument. But then the paragraph might be too short. And Nick specifically said he's tempted by Approach 2, which means he wants the observation to be the MAIN development, not a supporting aside. OK, I'm going in circles. Let me just present the variations clearly and let Nick choose. You're right about the Frege example — it was mine, and it's out. And you're right about the inventing-examples habit. I'll stay away from named philosophers in the new material. Here are variations of Approach 2. Each one starts from the bullet point and develops through the history-of-philosophy observation, but they take different routes and have different shapes. I've tried to think about how each one connects forward — not just to ¶3 (Putnam) but to ¶4 (Dellsén), to the evaluative framework (¶5-9), and to the section's overall aim of establishing that the discipline evaluates arguments rather than agents. ## Variation A — The observation as the entire development Bullet point → the history-of-philosophy observation stated directly → one or two sentences on what the observation implies about philosophy as a discipline → conclusion for LLMs. This would be a short paragraph. The observation does all the work. Something like: philosophy is treated differently from other disciplines in this respect — its history is part of it, not separate from it. Studying philosophical arguments from the past is doing philosophy. This suggests that philosophical contributions can't be separated from the reasoning that produced them. And if the reasoning requires understanding — the right sort of mental states — then philosophy requires a certain kind of agent. The risk: thinness. The observation is stated but not developed. The paragraph would be maybe four or five sentences after the bullet point. For a challenge that needs to have force, this might not be enough. How it connects forward: ¶4 (Dellsén) reinterprets the same observation — we study philosophy's history because the arguments remain productive (revealing dependence relations), not because ideas require minds behind them. The reinterpretation works well if the observation was stated cleanly in ¶2 without too much development pulling it in a specific direction. ## Variation B — The observation supporting a claim about philosophical value Bullet point → a claim: what makes a philosophical text philosophically valuable seems to be not just the conclusions it reaches but the reasoning behind them → the history-of-philosophy observation as evidence for this claim → the reasoning seems to require understanding → conclusion for LLMs. This is more developed because it has a specific claim ("philosophical value seems to depend on the reasoning, not just the conclusions") that the observation then supports. The claim gives the paragraph a thesis. How it connects forward: this version sets up the evaluative framework (¶5-9) particularly well. ¶2 says the quality of the reasoning seems to depend on the agent. The framework then shows that the quality of the reasoning is assessable from the argument alone — Lipton's loveliness concerns the explanation's capacity to illuminate, Williamson's desiderata concern the theory's properties. The tension between "quality depends on the agent" (¶2) and "quality is assessable from the argument" (¶5-9) is the section's central move. ## Variation C — The observation connected to the art parallel from ¶1 Bullet point → brief connection back to ¶1's structure: in art, the relevant mental activity is imaginative and expressive; in philosophy, it is understanding of the arguments and claims being made. When we read philosophy, we read it under this assumption. → The history-of-philosophy observation as structural evidence that this assumption is built into the discipline → conclusion for LLMs. This keeps the art parallel active — ¶1 established what art requires, ¶2 develops the philosophical analogue. The reading assumption (one sentence from your instinct in the prompt — "we read it with the assumption that the author understands the arguments and claims being made") enters briefly as the bridge between the art parallel and the history-of-philosophy observation. How it connects forward: the art parallel being active in ¶2 sets up the disanalogy in ¶3 ("But does the quality of a philosophical argument depend on the kind of agent that produced it, in the way that an artwork's character might depend on its creator?"). The transition from ¶2 to ¶3 would be smooth because ¶2 has been developing the parallel that ¶3 breaks. ## Variation D — The observation as a question about separability Bullet point → reframed as a question about separability: can the philosophical contribution be separated from the mind that produced it? → The history-of-philosophy observation as evidence that it cannot: philosophy's history is part of philosophy, which suggests contributions are not free-standing in the way that experimental results are → if contributions are bound to the minds that produced them, LLMs cannot contribute → conclusion. This variation frames the whole thing as a separability question. The advantage is that the separability question IS the section's question — the rest of the section argues that the contribution CAN be separated from the agent (that's what "philosophy evaluates arguments, not agents" means). So ¶2 would be stating the negative answer (can't separate them) that the section will argue against. How it connects forward: the separability question connects to everything — ¶3 (Putnam shows the argument works independently of the agent), ¶4 (Dellsén shows progress happens through publicly available ideas), the evaluative framework (shows the standards are argument-properties), blind review (shows the discipline operates on the assumption that they're separable). The entire section is about separability. One concern: the brief contrast with experimental results might overlap with material that was moved to Section 3. But the sentence "the history of philosophy is treated as part of philosophy in a way that the history of science is not" is already in the draft and seems fine to keep. ## What I notice across the variations The variations differ mainly in what SURROUNDS the history-of-philosophy observation — what claim it supports, what question it answers, what parallel it develops. The observation itself is the same in each case. Variation A is minimal — observation alone. Variation B gives it a thesis (about philosophical value). Variation C connects it to the art parallel. Variation D frames it as a separability question. My honest sense: Variation C connects forward most naturally to ¶3 (because it keeps the art parallel active for the disanalogy), and Variation B connects forward most naturally to the evaluative framework (because its thesis about "quality of reasoning" is exactly what the framework addresses). Variation D has the broadest reach — the separability question is the whole section's question — but it might be too abstract for a paragraph that needs to make the reader feel the challenge. I don't know which is right. What feels closest to what you're after? *Le variazioni differiscono in ciò che circonda l'osservazione, non nell'osservazione stessa — la domanda è quale cornice le dà più forza.* ### User opus[1m] ### User The user has opened the note file:///Users/nickyoung/My Obsidian Vault/Daily Notes/2026-04-03.md in Obsidian. This may or may not be related to the current conversation. If it seems relevant, consider using the Read tool to examine its content. @[[2026-04-03]] Your suggestions were written so tersely and at such a meta-level of abstraction that I couldn't understand what any of them meant. Don't describe what would be written, write what would be written ### Assistant Nick is right. I've been describing paragraphs instead of writing them. "The paragraph would develop..." — that's useless. He needs to see actual sentences so he can react to them. Let me write actual variations of ¶2. Each one starts from the bullet point and develops through the history-of-philosophy observation. No Frege, no invented philosophers, no generic descriptions of philosophical moves, no triple lists, no bold, no "matter" as a verb, no "textual"/"textually", no metacommentary. I need to keep in mind: this is stating the CHALLENGE. Philosophy seems to require the right sort of mental states behind it. The history-of-philosophy observation supports this. The paragraph should end with the implication for LLMs. It should NOT begin answering the challenge. Let me also remember what Nick's published prose sounds like. Long sentences that do philosophical work. Concrete. Facing the subject directly. No jargon labels. The writing tells you what's happening, it doesn't tell you about what's happening. And I need to avoid the bold sentences' failure mode: describing philosophical activity generically ("handles an objection," "draws a distinction") instead of making the reader feel the point. OK. Let me write several versions. Each should be a full paragraph that could actually go in the draft. Variation 1: Start from the bullet point. Then develop through the reading assumption and the history-of-philosophy observation. "We might think, for similar reasons, that philosophy is a uniquely human activity, requiring the right sort of mental states to lie behind it." Then what? The "for similar reasons" connects back to art. The "right sort of mental states" echoes ¶1's "right sort of mental activity." Now I need to develop what this means for philosophy. Nick's instinct: "when we read philosophy we're reading it with the assumption that the person who wrote those words understands what those words are about and understands the arguments and the claims that are being made." So the paragraph could continue with something like that — the assumption we bring to reading philosophy. Then the history-of-philosophy observation as evidence for this assumption. Let me try writing it out. "We might think, for similar reasons, that philosophy is a uniquely human activity, requiring the right sort of mental states to lie behind it. When we read a philosophical text, we read it with the assumption that the person who wrote it understands the arguments and the claims being made — that they can see what the arguments are about, not merely that they have arranged words in an order that looks like philosophy. This may be why the history of philosophy is treated as part of philosophy in a way that the history of science is not: understanding a philosophical contribution seems to require engaging with the reasoning that produced it, and this engagement seems to presuppose that there was genuine understanding behind the reasoning in the first place. If this is right, the question of whether an LLM can do philosophy does not arise. An LLM does not understand what any argument is about and cannot see what any claim is about." Hmm. Let me look at this critically. "that they can see what the arguments are about, not merely that they have arranged words in an order that looks like philosophy" — the contrast with word-arranging is potentially pre-empting the response (which argues that it doesn't depend on the agent). But it's also natural as part of stating the challenge — the challenge IS that mere word-arranging isn't enough. "understanding a philosophical contribution seems to require engaging with the reasoning that produced it" — this is OK but it could be read as the RESPONSE (engaging with the reasoning = the argument persists). I need to make sure it's read as the CHALLENGE (engaging with the reasoning = engaging with the MIND behind the reasoning). "this engagement seems to presuppose that there was genuine understanding behind the reasoning in the first place" — this is the challenge reading: the engagement requires that the reasoning was produced by a mind that understood. Without the understanding, the engagement would be with... something less. The paragraph is maybe too short though. Let me try a longer version that develops the history-of-philosophy observation more. Variation 2: "We might think, for similar reasons, that philosophy is a uniquely human activity, requiring the right sort of mental states to lie behind it. When we read a philosophical text, we read it with the assumption that the person who wrote it understands the arguments and the claims being made — that the words on the page are the product of someone who grasped what the arguments are about. This may be why the history of philosophy is treated as part of philosophy in a way that the history of science is not. A scientific result can be learned without retracing the process of discovery, but a philosophical argument cannot be separated from its reasoning in the same way: to understand the argument is to follow the reasoning, and to follow the reasoning is, at least on this view, to engage with the understanding of the person who produced it. If philosophy is the kind of discipline where the reasoning behind a text is part of what makes the text philosophically valuable, and if the reasoning requires genuine understanding — the right sort of mental states — then an LLM, which does not understand what any argument is about, cannot do philosophy." Wait — "A scientific result can be learned without retracing the process of discovery" — this is the science contrast that was moved to Section 3. But it's brief here — one clause. Is this OK? The session note says "Watson/Crick cut (science/philosophy contrast moves to Section 3 with Pigliucci)." The Watson/Crick material was a developed comparison. This is just a brief contrast: "A scientific result can be learned without retracing the process of discovery." That's one clause, not a developed comparison. And the sentence "the history of philosophy is treated as part of philosophy in a way that the history of science is not" is ALREADY in the current draft (line 14) and hasn't been cut. So I think a brief mention is OK. But maybe I should avoid it entirely to be safe. Let me try a version without the science contrast. Variation 3: "We might think, for similar reasons, that philosophy is a uniquely human activity, requiring the right sort of mental states to lie behind it. When we read a philosophical text, we read it with the assumption that the person who wrote it understands the arguments and the claims being made — that the words on the page are the product of someone who could see what the arguments are about. This may be why the history of philosophy is treated as part of philosophy in a way that the history of science is not: to understand a philosophical contribution, on this view, is to follow the reasoning that produced it, and to follow the reasoning is to engage with the understanding of the person who was doing the reasoning. The arguments cannot be stripped away and only the conclusions retained; it is the arguments, and the understanding they exhibit, that make the contribution a contribution. If this is right, the question of whether an LLM can do philosophy does not arise. An LLM does not understand what any argument is about." Hmm. "The arguments cannot be stripped away and only the conclusions retained" — this is from the old paragraph's Frege passage. Is it OK to reuse it here without Frege? I think so — the sentence works as a general claim about philosophy. It doesn't need Frege. "it is the arguments, and the understanding they exhibit, that make the contribution a contribution" — this is close to the RESPONSE territory ("the arguments are the contribution"). But the key phrase is "and the understanding they exhibit" — on the challenge's view, it's not just the arguments, it's the arguments PLUS the understanding behind them. The challenge says you can't separate the understanding from the arguments. Let me think about whether this sentence is saying something the response will disagree with. The response says: the arguments are assessable from the argument alone — the understanding they exhibit is a property of the arguments, not of the agent. So the response doesn't disagree that "the arguments and the understanding they exhibit" are what count — it disagrees about WHERE the understanding lives (in the agent vs in the argument). This is actually fine. ¶2 says: the understanding is behind the arguments (in the agent). ¶3 onwards says: the understanding is in the arguments (in the argument). Same observation, two readings. That's the turn. Let me also think about the last sentence. "An LLM does not understand what any argument is about." This is blunt. Is it too blunt? In the April 2 conversation, the agreed ending was something like: "If philosophy requires this kind of understanding — being able to see where a picture comes under pressure, being able to follow and respond to reasoning — then an LLM, which does not see anything and cannot follow reasoning in any relevant sense, cannot do philosophy." But that's a triple ("see where a picture comes under pressure, follow and respond to reasoning, see anything and follow reasoning") — and Nick hates triples. The blunt version avoids this. "An LLM does not understand what any argument is about" is one claim, not a list. Actually — the paper is agnostic about LLM understanding (the introduction says "We are agnostic about whether LLMs 'really reason' in some deep metaphysical sense"). So the challenge is stating a view that the paper doesn't share but takes seriously. The challenge says: LLMs don't understand, therefore they can't do philosophy. The paper's response is: whether or not they understand, the evaluation concerns the argument. So the last sentence of ¶2 should be the challenge's conclusion, not the paper's view. "If this is right, the question of whether an LLM can do philosophy does not arise" is better than "An LLM does not understand what any argument is about" because it's conditional ("if this is right") rather than assertive. Let me revise. Variation 3 revised: "We might think, for similar reasons, that philosophy is a uniquely human activity, requiring the right sort of mental states to lie behind it. When we read a philosophical text, we read it with the assumption that the person who wrote it understands the arguments and the claims being made — that the words on the page are the product of someone who could see what the arguments are about. This may be why the history of philosophy is treated as part of philosophy in a way that the history of science is not: to understand a philosophical contribution, on this view, is to follow the reasoning that produced it, and to follow the reasoning is to engage with the understanding of the person who was doing the reasoning. The arguments cannot be stripped away and only the conclusions retained; it is the arguments, and the understanding they exhibit, that make the contribution a contribution. If this is right, the question of whether an LLM can do philosophy does not arise. An LLM does not understand problems and cannot think through a difficulty in the relevant sense." That last sentence is from the old draft. It's reused. Let me keep it. OK. Let me now try genuinely different variations, not just revisions of the same one. Variation 4 — Starting from the history-of-philosophy observation directly (more Approach 2) "We might think, for similar reasons, that philosophy is a uniquely human activity, requiring the right sort of mental states to lie behind it. The history of philosophy is treated as part of philosophy in a way that the history of science is not, and one explanation for this is that philosophical contributions are bound up with the reasoning that produced them. You cannot learn a philosophical argument the way you can learn a scientific result — by taking the conclusion and discarding the route. The route is part of what the argument is; the reasoning that led to the conclusion is part of the philosophical contribution, not merely a means of arriving at it. And the reasoning, on this view, requires genuine understanding: the person who produced the argument had to be able to see what the argument is about, had to grasp why the moves in the argument are the right moves to make. If this is right, the question of whether an LLM can do philosophy does not arise. An LLM does not understand problems and cannot think through a difficulty in the relevant sense." Hmm. "had to be able to see what the argument is about, had to grasp why the moves in the argument are the right moves to make" — that's a double, not a triple. But "the right moves to make" is close to the generic move-description I should avoid. Let me revise. "And the reasoning, on this view, requires genuine understanding: the person who produced the argument had to be able to see what the argument is about, had to understand the claims being made and why they bear on each other in the way they do." That's better. It's specific about what "understanding" means in context — understanding the claims AND their bearing on each other. But wait — "You cannot learn a philosophical argument the way you can learn a scientific result" — this is the science contrast again. Let me try without it. Variation 5 — History-of-philosophy observation without science contrast "We might think, for similar reasons, that philosophy is a uniquely human activity, requiring the right sort of mental states to lie behind it. The history of philosophy is treated as part of philosophy in a way that the history of science is not, and one explanation for this is that philosophical contributions cannot be separated from the reasoning that produced them. The reasoning is part of what the contribution is; strip it away and what remains is a set of conclusions without anything that makes them philosophically valuable. And the reasoning, on this view, requires genuine understanding — the person who produced the argument had to be able to see what the argument is about, had to understand why the claims bear on each other in the way they do. If this is right, the question of whether an LLM can do philosophy does not arise. An LLM does not understand problems and cannot think through a difficulty in the relevant sense." "strip it away and what remains is a set of conclusions without anything that makes them philosophically valuable" — this is making the point. Conclusions without reasoning are not philosophy. The reasoning requires understanding. Understanding requires the right sort of mental states. LLMs don't have mental states. But "The history of philosophy is treated as part of philosophy in a way that the history of science is not" already involves the science contrast, even without developing it. The "in a way that the history of science is not" clause IS a brief contrast. Is this OK? The sentence is already in the current draft... Let me try removing even that. Variation 6 — No science contrast at all "We might think, for similar reasons, that philosophy is a uniquely human activity, requiring the right sort of mental states to lie behind it. When we read a philosophical text, we read it with the assumption that the person who wrote it understands the arguments and the claims being made. Philosophical contributions cannot be separated from the reasoning that produced them — the reasoning is part of what the contribution is — and this is why studying the history of philosophy is itself a way of doing philosophy: when we return to an argument from the past, we are engaging with the reasoning, and the reasoning seems to presuppose genuine understanding on the part of the person who produced it. If this is right, the question of whether an LLM can do philosophy does not arise. An LLM does not understand problems and cannot think through a difficulty in the relevant sense." This version drops the science contrast entirely. The history-of-philosophy observation enters as a consequence of the inseparability claim rather than as an independent observation. "This is why studying the history of philosophy is itself a way of doing philosophy" — the inseparability explains the observation, rather than the observation standing alone. Hmm. But is "Philosophical contributions cannot be separated from the reasoning that produced them" too close to the response? The response will say: actually, the contribution IS the argument (which is separable from the agent). The challenge says: the contribution is the argument PLUS the understanding behind it (which is not separable from the agent). I think the sentence works for the challenge because of what follows: "the reasoning is part of what the contribution is, and the reasoning seems to presuppose genuine understanding." The challenge's version says the reasoning requires understanding (which requires an agent). The response's version says the reasoning is assessable from the argument (which doesn't require an agent). Let me also try a version that's more in Nick's voice — longer sentences, more philosophical development in each sentence. Variation 7 — More developed sentences "We might think, for similar reasons, that philosophy is a uniquely human activity, requiring the right sort of mental states to lie behind it. When we read a philosophical text, we read it with the assumption that the person who wrote it understands what the arguments are about and understands the claims that are being made — that the text is the product of someone who could see why the arguments go the way they do, not merely someone who has arranged words in a way that looks like philosophy. This assumption is not idle: it connects to why the history of philosophy is treated as part of philosophy in a way that the history of science is not. Philosophical contributions cannot be separated from the reasoning that produced them; the reasoning is not merely a route to a conclusion that can be discarded once the conclusion is reached, but is itself part of what makes the contribution philosophically valuable. And reasoning of this kind seems to require genuine understanding — the person behind the argument had to be able to see what the argument is about. If this is right, the question of whether an LLM can do philosophy does not arise. An LLM does not understand problems and cannot think through a difficulty in the relevant sense." This is longer. Let me look at it critically. "that the text is the product of someone who could see why the arguments go the way they do, not merely someone who has arranged words in a way that looks like philosophy" — this is getting at the same point as the old bold sentences, but without the mechanical template. It's describing the ASSUMPTION rather than listing specific moves. And "not merely someone who has arranged words in a way that looks like philosophy" — this characterises what an LLM does, from the challenge's perspective. "This assumption is not idle: it connects to why the history of philosophy is treated as part of philosophy in a way that the history of science is not." — transitional sentence connecting the reading assumption to the history-of-philosophy observation. Clean. "Philosophical contributions cannot be separated from the reasoning that produced them; the reasoning is not merely a route to a conclusion that can be discarded once the conclusion is reached, but is itself part of what makes the contribution philosophically valuable." — this reuses the "stripped away" idea from the old paragraph but without Frege. It's a general claim about the discipline. "And reasoning of this kind seems to require genuine understanding — the person behind the argument had to be able to see what the argument is about." — stakes the challenge's claim directly. I think this version is the strongest so far. Let me try one more that goes a slightly different direction. Variation 8 — More emphasis on the history-of-philosophy observation "We might think, for similar reasons, that philosophy is a uniquely human activity, requiring the right sort of mental states to lie behind it. The history of philosophy is treated as part of philosophy in a way that the history of science is not: studying a philosophical argument from the past is not a separate, historical activity but is itself a way of doing philosophy. One explanation for this is that philosophical contributions are not separable from the reasoning that produced them. The conclusions alone are not the contribution; it is the reasoning — and, on this view, the understanding that the reasoning exhibits — that makes the contribution philosophically valuable. When we read a philosophical text, we read it with the assumption that the person who wrote it could see what the arguments are about, and this assumption seems to be part of what makes the engagement with the text a philosophical engagement. If this is right, the question of whether an LLM can do philosophy does not arise. An LLM does not understand problems and cannot think through a difficulty in the relevant sense." This puts the history-of-philosophy observation first (after the bullet point) and the reading assumption later. More "Approach 2" — the observation leads, the reading assumption supports. "studying a philosophical argument from the past is not a separate, historical activity but is itself a way of doing philosophy" — this develops the observation. It explains what "treated as part of philosophy" means. "One explanation for this is that philosophical contributions are not separable from the reasoning that produced them" — this connects the observation to the inseparability claim. "The conclusions alone are not the contribution; it is the reasoning — and, on this view, the understanding that the reasoning exhibits — that makes the contribution philosophically valuable." — this develops the inseparability claim AND connects it to understanding. "when we read a philosophical text, we read it with the assumption that the person who wrote it could see what the arguments are about" — reading assumption enters as supporting evidence. "this assumption seems to be part of what makes the engagement with the text a philosophical engagement" — connects back to the history-of-philosophy observation: the engagement is philosophical BECAUSE of the assumption of understanding. OK. I think I have enough variations. Let me present them. But wait — let me also think about whether any of these have a problem that I'm not seeing. One concern: "the reasoning is part of what makes the contribution philosophically valuable" — the response (Dellsén, ¶4) will say that what's philosophically valuable is the argument's capacity to reveal dependence relations. The challenge is saying: the reasoning is valuable because it reflects understanding. The response is saying: the reasoning is valuable because it reveals dependence relations. Are these compatible? Yes — they're two readings of the same observation. The challenge says the value comes from the understanding behind the reasoning. The response says the value comes from the reasoning itself (its structure, its capacity to illuminate). ¶2 states the challenge reading; ¶3-4 present the response reading. Another concern: am I using "the right sort of mental states" correctly? The bullet point says philosophy requires "the right sort of mental states to lie behind it." This echoes ¶1's "the right sort of mental activity on the part of its maker — imaginative, expressive." The philosophical analogue is: understanding. The right sort of mental states for philosophy is understanding — understanding the arguments, understanding the claims, understanding why things bear on each other. I think this is correct. The challenge is: philosophy requires understanding behind it, the way art requires imagination behind it. One more concern: some of these variations mention "someone who has arranged words in a way that looks like philosophy." Is this too on-the-nose about LLMs? The challenge is supposed to be general — "only people can do philosophy" — not specifically about LLMs until the conclusion. But on the other hand, the whole paper is about LLMs, so the reader knows what's at stake. The phrase characterises a general possibility (text without understanding) that applies to LLMs but isn't limited to them. Actually, looking at ¶1, it mentions "generative AI systems" explicitly. So the section has already named LLMs. The challenge can reference the possibility of text-without-understanding without it being jarring. OK, I'm satisfied with the variations. Let me present them. Here are four versions. Each one is the full paragraph as it would appear in the draft, starting from your bullet point. ## Version 1 We might think, for similar reasons, that philosophy is a uniquely human activity, requiring the right sort of mental states to lie behind it. When we read a philosophical text, we read it with the assumption that the person who wrote it understands the arguments and the claims being made — that the words on the page are the product of someone who could see what the arguments are about. This may be why the history of philosophy is treated as part of philosophy in a way that the history of science is not: to understand a philosophical contribution, on this view, is to follow the reasoning that produced it, and to follow the reasoning is to engage with the understanding of the person who was doing the reasoning. The arguments cannot be stripped away and only the conclusions retained; it is the arguments, and the understanding they exhibit, that make the contribution a contribution. If this is right, the question of whether an LLM can do philosophy does not arise. An LLM does not understand problems and cannot think through a difficulty in the relevant sense. ## Version 2 We might think, for similar reasons, that philosophy is a uniquely human activity, requiring the right sort of mental states to lie behind it. When we read a philosophical text, we read it with the assumption that the person who wrote it understands what the arguments are about and understands the claims that are being made — that the text is the product of someone who could see why the arguments go the way they do, not merely someone who has arranged words in a way that looks like philosophy. This assumption is not idle: it connects to why the history of philosophy is treated as part of philosophy in a way that the history of science is not. Philosophical contributions cannot be separated from the reasoning that produced them; the reasoning is not merely a route to a conclusion that can be discarded once the conclusion is reached, but is itself part of what makes the contribution philosophically valuable. And reasoning of this kind seems to require genuine understanding — the person behind the argument had to be able to see what the argument is about. If this is right, the question of whether an LLM can do philosophy does not arise. An LLM does not understand problems and cannot think through a difficulty in the relevant sense. ## Version 3 We might think, for similar reasons, that philosophy is a uniquely human activity, requiring the right sort of mental states to lie behind it. The history of philosophy is treated as part of philosophy in a way that the history of science is not: studying a philosophical argument from the past is not a separate, historical activity but is itself a way of doing philosophy. One explanation for this is that philosophical contributions are not separable from the reasoning that produced them. The conclusions alone are not the contribution; it is the reasoning — and, on this view, the understanding that the reasoning exhibits — that makes the contribution philosophically valuable. When we read a philosophical text, we read it with the assumption that the person who wrote it could see what the arguments are about, and this assumption seems to be part of what makes the engagement with the text a philosophical engagement rather than a merely historical one. If this is right, the question of whether an LLM can do philosophy does not arise. An LLM does not understand problems and cannot think through a difficulty in the relevant sense. ## Version 4 We might think, for similar reasons, that philosophy is a uniquely human activity, requiring the right sort of mental states to lie behind it. When we read a philosophical text, we read it with the assumption that the person who wrote it understands the arguments and the claims being made. Philosophical contributions cannot be separated from the reasoning that produced them — the reasoning is part of what the contribution is — and this is why studying the history of philosophy is itself a way of doing philosophy: when we return to an argument from the past, we are engaging with the reasoning, and the reasoning seems to presuppose genuine understanding on the part of the person who produced it. If this is right, the question of whether an LLM can do philosophy does not arise. An LLM does not understand problems and cannot think through a difficulty in the relevant sense. --- The main differences: Version 1 is the most compressed. Version 2 is the longest and includes the contrast with mere word-arrangement. Version 3 leads with the history-of-philosophy observation rather than the reading assumption. Version 4 is the shortest — it makes the inseparability claim do most of the work, with the history-of-philosophy observation entering as a consequence rather than as an independent point. *Quattro versioni della stessa sfida — cambia l'ordine delle mosse, non la sostanza.* ### User "We might think, for similar reasons, that philosophy is a uniquely human activity, requiring the right sort of mental states to lie behind it. When we read a philosophical text, we read it assuming that it was written by a person who understands what is being claimed." apart from this couple of sentences, everything else was written in a terribly bad style and just yeah, lacking yeah, using a lot of words to say very little. Little. There's no depth there, it was all just shallow surface nonsense. Also what is your you I don't think you've really thought what's going to come after this paragraph in this section, have you? Plan out what you think the paragraph structure should be for the entire section please. Because obviously it's going to be different from what's on the note at the moment. If you're working trying to force everything into that structure, you've been wasting my time. ### Assistant OK. Nick has given me two sentences he's happy with: "We might think, for similar reasons, that philosophy is a uniquely human activity, requiring the right sort of mental states to lie behind it. When we read a philosophical text, we read it assuming that it was written by a person who understands what is being claimed." Everything else I wrote was shallow surface nonsense. And he's right that I haven't thought through the paragraph structure of the whole section given the changes we're making. I've been trying to slot new ¶2 material into the old structure, which doesn't work because the old structure was built around a one-paragraph challenge. Now the challenge is being developed properly, and the rest of the section needs to adjust. Let me think about the whole section from scratch. What does the section need to do? 1. Introduce the challenge from authorship (through the art comparison) 2. Develop the challenge — philosophy seems to require the right sort of mental states 3. Turn — show that the argument's quality is independent of the agent 4. Develop what evaluation looks like — the evaluative framework 5. Conclude — these standards concern the argument, not the agent 6. Transition to Section 2 — corpus filtered by these standards Now, what material do we have? From the note (the version Nick is working on): Line 10: ¶1 — the art comparison. DONE. Nick has just rewritten this. Art requires the right sort of mental activity (imaginative, expressive). A system without mental states has no such activity to draw on. Line 12: The bullet point + %%%% — the bridge to philosophy. Nick has now written the opening of ¶2: the bullet point plus "When we read a philosophical text, we read it assuming that it was written by a person who understands what is being claimed." Line 14: Old ¶2 — "We tend to read philosophical texts as evidence of this kind of understanding..." Nick calls this "fucking terrible." This needs replacing. Then the Putnam paragraph (currently with %%comments%% about being abrupt). Then Dellsén. Then Lipton block quote. Then dormative virtue. Then likeliness without loveliness. Then Williamson/overfitting. Then Bengson. Then "In sum." Then Deep Blue. Then transition. Now — the NOTE (the file Nick has open) and the LONGFORM FILE (in Writing/research/) are DIFFERENT. Let me check which one Nick is actually working on. The note is "Section 1 Draft - Challenge from Authorship (2 Apr 2026)" in Notes/. The longform file is "1. Philosophy in the Text.md" in Writing/research/generating-philosophy-text-internal-evaluation/. Looking at both: the note has the newer version (with Nick's rewritten ¶1 and the bullet point). The longform file has a slightly different version of the same content. Nick is working on the note. OK. So I need to plan the paragraph structure of the ENTIRE section, given: - ¶1 is done (art comparison) - ¶2 starts with the two sentences Nick has approved - Everything after needs to be planned Let me think about what the section needs and what material is available. The section's argument: 1. Challenge: philosophy seems to require the right sort of mental states (art comparison → philosophy parallel) 2. Response: the evaluative standards concern the argument, not the agent 3. Evidence: Lipton, Williamson, Bengson — the evaluative framework 4. Further evidence: blind review 5. Conclusion: what follows when an LLM is trained on a corpus shaped by these standards Now, what about Putnam? In the old structure, Putnam was ¶2 — it came right after the (compressed) challenge. In the new structure, Putnam is the TURN — the moment where the section shifts from challenge to response. It should come AFTER the challenge is fully developed. What about Dellsén? Dellsén develops the response — philosophical progress = enabling understanding through publicly available ideas. The argument is what's publicly available. What about the history-of-philosophy observation? It's part of the challenge. In ¶2 or in its own paragraph. Let me think about what the section looks like with the challenge properly developed. Option A — Challenge in 2 paragraphs, then turn ¶1: Art comparison (DONE) ¶2: Philosophy parallel — requires the right sort of mental states. The two approved sentences + development. The history-of-philosophy observation as part of this. ¶3: Putnam — the turn. "But does the quality of a philosophical argument depend on the kind of agent that produced it...?" ¶4: Dellsén — progress through publicly available ideas. History-of-philosophy reinterpreted. ¶5: Transition to evaluative framework + Lipton block quote ¶6: Dormative virtue ¶7: Likeliness without loveliness in philosophy ¶8: Williamson/overfitting ¶9: Bengson ¶10: "In sum" + blind review (rephrased: argument, not agent) ¶11: Deep Blue ¶12: Transition to corpus thesis This is 12 paragraphs. The challenge gets ¶1-2 (two paragraphs). The response gets ¶3-11 (nine paragraphs). The balance is: two paragraphs of challenge, nine of response. This is roughly the same as the current draft (one paragraph of challenge, nine of response) but with one more paragraph for the challenge. Option B — Challenge in 3 paragraphs ¶1: Art comparison (DONE) ¶2: Philosophy parallel — the two approved sentences + development of what "understanding what is being claimed" means ¶3: History-of-philosophy observation as its own paragraph — develops the institutional evidence for the practitioner-centred view ¶4: Putnam — the turn ...rest same as Option A This gives the challenge three paragraphs (¶1-3) and the response eight (¶4-12). More balanced. But Enrico said the challenge is "easy to deal with" — three paragraphs might be too much. Hmm. But ¶1 isn't purely challenge — it's the art comparison, which sets up both the challenge AND the response (through the disanalogy in ¶4). So really the pure challenge is ¶2 (and ¶3 if it exists). The response starts with Putnam. Option C — Challenge in 2 paragraphs but ¶2 is longer ¶1: Art comparison (DONE) ¶2: Philosophy parallel — the two approved sentences + the history-of-philosophy observation + the inseparability claim + the LLM conclusion. All in one longer paragraph. ¶3: Putnam — the turn ...rest same This keeps the challenge to two paragraphs but makes ¶2 do more work. The risk is that ¶2 becomes too long and tries to do too much (which was the old paragraph's problem). Let me think about what ¶2 actually needs to contain. The two approved sentences are: 1. "We might think, for similar reasons, that philosophy is a uniquely human activity, requiring the right sort of mental states to lie behind it." 2. "When we read a philosophical text, we read it assuming that it was written by a person who understands what is being claimed." These state the claim (philosophy requires mental states) and connect it to reading (we assume the author understands). What else does ¶2 need? It needs to develop the claim enough that the reader feels its force. And it needs to reach the conclusion: if this is right, LLMs can't do philosophy. The history-of-philosophy observation is one way to develop the claim. But Nick was critical of all my attempts to develop it — everything I added was "shallow surface nonsense." Maybe the problem is that ¶2 doesn't need as much development as I've been trying to give it. Maybe the two approved sentences plus the history-of-philosophy observation (briefly stated) plus the conclusion is enough. A shorter, sharper paragraph. But Nick also said my versions were "using a lot of words to say very little." So the problem isn't length per se — it's that the words weren't doing any work. Each sentence needs to earn its place by saying something the previous sentence didn't. Let me think about what SPECIFICALLY needs to be said between the reading assumption and the LLM conclusion. The reading assumption: we assume the author understands what is being claimed. The LLM conclusion: if philosophy requires understanding, LLMs can't do philosophy. What's the gap between these? The gap is: WHY does philosophy require understanding? The reading assumption says we ASSUME understanding. But why do we think the assumption is correct — why do we think understanding is actually required? One answer: the history-of-philosophy observation. We study philosophy's history as part of doing philosophy, which suggests contributions are bound up with the reasoning behind them, which suggests understanding is required. Another answer: the nature of philosophical contributions. Philosophical contributions are arguments, and arguments seem to require understanding — you can't argue for something without understanding what you're arguing for. Another answer: no further justification needed. The assumption itself is evidence enough. We assume the author understands, and this assumption is so deeply embedded in how we read philosophy that it constitutes part of what we think philosophy IS. Hmm. The third answer is interesting. What if ¶2 doesn't try to JUSTIFY the assumption but just states it clearly and lets the reader feel its force? "When we read philosophy, we assume the author understands. If this assumption is constitutive of what philosophy is, then a system that doesn't understand can't do philosophy." But that's very thin. And it doesn't develop the point at all. OK, let me think about this from a completely different angle. What does the WHOLE section look like? Let me think about what each paragraph is doing and whether the current structure is right at all. Current structure (in the note Nick is working on): 1. Art comparison (DONE) 2. ??? (developing the challenge) 3. Old ¶2 (terrible — being replaced) 4. Putnam (with %%comments%%) 5. Dellsén 6. Lipton transition + block quote 7. Dormative virtue 8. Likeliness without loveliness 9. Williamson/overfitting 10. Bengson 11. "In sum" 12. Deep Blue 13. Transition Wait — in the note, there's also the old ¶2 text (line 14) that contains material about reading philosophy as evidence of understanding, the handling-objections bold sentences, the history-of-philosophy observation, Frege, and the "if this is right" conclusion. All of this is being replaced. Then the Putnam paragraph (line 12 in the longform file, around line 16 in the note). Let me re-check... Actually, the note and the longform file have diverged. Nick has been editing the note. Let me look at the note's structure again. From my earlier reading of the note (Notes/Section 1 Draft - Challenge from Authorship): Line 10: ¶1 — art comparison (just rewritten) Line 12: bullet point + %%%% — the bridge Line 14: old ¶2 — "We tend to read philosophical texts..." (being replaced) Line 16: Putnam paragraph (with %%comments%%) Line 18: Dellsén Line 20-24: Lipton transition, block quote, dormative virtue, likeliness without loveliness Line 24: Williamson/overfitting Line 26: Bengson Line 28: "In sum" Line 30: Deep Blue Line 32-34: Transition + footnote Actually wait. Let me re-check. The note Nick is working on has the newly rewritten ¶1 plus the bullet point, but after that it still has the old text. So the old ¶2 (line 14) is still there, followed by the Putnam paragraph, etc. No wait — looking more carefully at the note I read at the very beginning of this conversation: it has been updated since the April 2 session. The note now has Nick's rewritten ¶1 (the clean art comparison paragraph), the bullet point, and then the old text. But actually — the note has a DIFFERENT structure from the longform file. Let me compare. The NOTE (Section 1 Draft - Challenge from Authorship) has: - ¶1: the art comparison (Nick's new version) - Bullet point: "We might think, for similar reasons..." - Old ¶2: "We tend to read philosophical texts..." - Putnam paragraph (with %%comments%%) - Dellsén - Lipton + block quote - Dormative virtue - Likeliness without loveliness - Williamson/overfitting - Bengson - "In sum" - Deep Blue - Transition The LONGFORM FILE (1. Philosophy in the Text.md) has: - The old ¶1 (not yet updated with Nick's rewrite) - Old ¶2 - Putnam paragraph (with %%comments%%) - Same framework paragraphs - Same ending OK so the note is ahead of the longform file. Nick is working on the note. Once the note is right, the content will presumably move to the longform file. Now — the question is: what should the paragraph structure be for the WHOLE section, given the changes? Let me think about this seriously. The section has three parts: A. The challenge (¶1 + ¶2, possibly ¶3) B. The turn and response (Putnam + Dellsén + evaluative framework) C. The conclusion and transition (In sum + blind review + Deep Blue + transition) Part A is being rewritten. Part B is mostly existing material (Putnam needs a new opening; Dellsén needs fixes; framework stays). Part C needs the "In sum" rephrased (argument vs agent, not product vs process) and blind review added. Now — the critical question is what happens at the JUNCTION between Part A and Part B. Currently, the Putnam paragraph opens abruptly (the %%comment%% says so). In the April 2 plan, it would open with: "But does the quality of a philosophical argument depend on the kind of agent that produced it, in the way that an artwork's character might depend on its creator?" This opening question connects ¶2 (the challenge) to ¶3 (the turn). It works if ¶2 ends with the challenge fully stated — "if this is right, LLMs can't do philosophy" — and ¶3 then asks: "but IS this right?" So the question is whether ¶2 (the challenge) needs to contain: (a) Just the reading assumption + brief development + LLM conclusion (b) The reading assumption + history-of-philosophy observation + LLM conclusion (c) The reading assumption + more substantial development + LLM conclusion And then what the Putnam paragraph looks like as ¶3. Let me think about whether the current Putnam paragraph needs restructuring. The current Putnam paragraph (from the note, line 16, and from the longform file, line 12) says: "But does the quality of a philosophical argument depend on the kind of agent that produced it...? Putnam was not reporting a previously unnoticed item in the world; he was making a case, by way of thought experiment, that meanings are not fixed solely by what is in the speaker's head. The thought experiment does its work not by pointing to something outside the text — there is no Twin Earth for us to go and inspect — but by constructing a scenario whose internal logic puts pressure on a familiar picture of meaning..." Wait — the note has a %%comment%% at the start: "%%not how i write. A reader is gonna be confused. Why are you suddenly talking about Putnam?%%" And then: "Putnam was not reporting..." So the opening question ("But does the quality...") isn't in the NOTE yet — it was proposed in the April 2 plan. The note still has the abrupt opening. So the Putnam paragraph needs: 1. A new opening that connects to the challenge (fixing the %%comment%%) 2. The existing Putnam content (which is mostly good) 3. The "proofs not paintings" line at the end (from the April 2 plan) After Putnam comes Dellsén. Then the framework. Then the conclusion. Now, one question I haven't properly considered: does the Dellsén paragraph still work in its current position? Currently it follows Putnam. The connection is: Putnam shows the contribution is what the argument does → Dellsén makes this systematic (progress = enabling understanding through publicly available ideas). In the restructured section, Dellsén also needs to reinterpret the history-of-philosophy observation (if that observation appears in ¶2). The reinterpretation: we study philosophy's history because the arguments remain productive, not because ideas are bound to minds. Does this reinterpretation need its own paragraph, or can it be folded into the existing Dellsén paragraph? Looking at the current Dellsén paragraph (line 18 of the note): it's about Dellsén's account of progress, applied to Twin Earth. The history-of-philosophy reinterpretation would be new material at the end. In the April 2 plan, this was 2-3 sentences at the end of the Dellsén paragraph. That seems right — it doesn't need its own paragraph. OK. And then the evaluative framework (Lipton, dormative virtue, likeliness without loveliness, Williamson, Bengson) — these are staying verbatim. They work. Then "In sum" — needs rephrasing from "product/process" to "argument/agent." And blind review needs adding. Then Deep Blue — stays, with light rephrasing (agent not process). Then transition — stays. So the full structure would be: ¶1: Art comparison (DONE — Nick's rewritten version) ¶2: Philosophy parallel (starts with Nick's two approved sentences; develops the challenge; ends with LLM conclusion) ¶3: The turn — Putnam (new opening connecting to the challenge; existing Putnam content; "proofs not paintings") ¶4: Dellsén (existing content with fixes + history-of-philosophy reinterpretation at the end) ¶5: Transition to Lipton + block quote (existing, verbatim) ¶6: Dormative virtue (existing, verbatim) ¶7: Likeliness without loveliness (existing, verbatim) ¶8: Williamson/overfitting (existing, verbatim) ¶9: Bengson (existing, verbatim) ¶10: "In sum" + blind review (rephrased: argument/agent; blind review added) ¶11: Deep Blue (existing, lightly rephrased) ¶12: Transition (existing, verbatim) That's 12 paragraphs. Same as before. The changes are: - ¶1: rewritten (done) - ¶2: new (the challenge, developed) - ¶3: existing Putnam with new opening and closing - ¶4: existing Dellsén with fixes and new ending - ¶5-9: verbatim - ¶10: rephrased + blind review - ¶11: lightly rephrased - ¶12: verbatim This is a clear plan. The structural change is minimal — the old ¶1 (which had the challenge compressed into one paragraph) splits into ¶1 (art comparison) and ¶2 (philosophy parallel), and the Putnam paragraph gets a new opening. Everything else stays. But wait — should the "proofs not paintings" comparison go in the Putnam paragraph or somewhere else? In the April 2 plan, it was at the end of the Putnam paragraph. The Putnam paragraph would then be: opening question → Putnam case → disanalogy drawn from the case → "proofs not paintings." This is a lot for one paragraph. It might need to split. But let me not pre-judge. Let me present the paragraph plan and let Nick decide. Actually, there's another question: what about the bold sentences in the old "In sum" paragraph? Let me re-read it. The old "In sum" (line 28 of the note): "In sum, the evaluative standards we have assembled all bear on what a philosophical text says and how it argues for it. Lipton's distinction between illumination and mere accommodation, Williamson's desiderata for theoretical virtue, and Bengson et al.'s method for assessing how well those standards are met: each concerns the product rather than the process by which the product was reached. They do not ask how the author arrived at her argument but whether the argument, as it stands on the page, meets the relevant standards." The rephrasing: "each concerns the argument, not the kind of agent that produced it" (from the April 2 discussion). And then blind review: "Under blind review — the discipline's standard method for assessing philosophical work — referees assess what an argument achieves without knowing who or what produced it." And the old "In sum" also has bold: "The practice of blind review in philosophy rests on the same assumption: referees assess what a paper achieves without knowing who wrote it. If the kind of agent behind the argument — whether a senior philosopher or a graduate student, whether a human being or a machine — were relevant to the argument's quality, then blind review would be a defective practice rather than the discipline's standard method of assessment." Wait — that bold text is in the NOTE version (line 32 area), not the longform version. Let me re-read the note more carefully... Actually, looking at the note again: line 32 has the "In sum" paragraph, and it includes both the bolded and unbolded text. The note's version is different from the longform file's version. Nick has been editing the note. Hmm, let me just re-read the note's "In sum" paragraph. From my initial reading: Line 32: "In sum, the evaluative standards we have assembled all bear on what a philosophical text says and how it argues for it. Lipton's distinction between illumination and mere accommodation, Williamson's desiderata for theoretical virtue, and Bengson et al.'s method for assessing how well those standards are met: **each concerns the argument, not the kind of agent that produced it.** They do not ask how the author arrived at her argument but whether the argument, as it stands on the page, meets the relevant standards. **The practice of blind review in philosophy rests on the same assumption: referees assess what a paper achieves without knowing who wrote it. If the kind of agent behind the argument — whether a senior philosopher or a graduate student, whether a human being or a machine — were relevant to the argument's quality, then blind review would be a defective practice rather than the discipline's standard method of assessment.**" Wait — this is from the note that Nick has open, which I read at the beginning. Let me check if these bold sentences are in the note or not... Actually, looking at my initial reading more carefully — the note DOES have blind review and the agent language already. Lines 32-33 in the note: "In sum, the evaluative standards we have assembled all bear on what a philosophical text says and how it argues for it. Lipton's distinction between illumination and mere accommodation, Williamson's desiderata for theoretical virtue, and Bengson et al.'s method for assessing how well those standards are met: **each concerns the argument, not the kind of agent that produced it.** They do not ask how the author arrived at her argument but whether the argument, as it stands on the page, meets the relevant standards. **The practice of blind review in philosophy rests on the same assumption: referees assess what a paper achieves without knowing who wrote it. If the kind of agent behind the argument — whether a senior philosopher or a graduate student, whether a human being or a machine — were relevant to the argument's quality, then blind review would be a defective practice rather than the discipline's standard method of assessment.**" So some of the changes from the April 2 plan have ALREADY been applied to the note. The "In sum" already has "each concerns the argument, not the kind of agent that produced it" and the blind review sentences. Good — that means ¶10 is partially done. OK, so the plan for the whole section is clearer than I thought. Most of it is already in place. The changes needed are: ¶1: DONE ¶2: needs writing (the two approved sentences + development + conclusion) ¶3 (Putnam): needs a new opening sentence/question + possibly "proofs not paintings" at the end ¶4 (Dellsén): needs fixes (%%comments%%) + history-of-philosophy reinterpretation at the end ¶5-9: DONE (verbatim) ¶10 ("In sum"): partially done (blind review already added), may need further adjustment ¶11 (Deep Blue): may need light rephrasing ¶12: DONE So really, the work is: ¶2 (the main thing), the opening of ¶3, the end of ¶4, and checking ¶10-11. The rest is done. Let me present this plan clearly. Here's the paragraph plan for the whole section. ¶1: Art comparison. DONE — your rewritten version. Art seems to require the right sort of mental activity (imaginative, expressive). A system without mental states has no such activity to draw on. ¶2: The challenge from philosophy. Starts with: "We might think, for similar reasons, that philosophy is a uniquely human activity, requiring the right sort of mental states to lie behind it. When we read a philosophical text, we read it assuming that it was written by a person who understands what is being claimed." Then development. Then: if this is right, the question of whether an LLM can do philosophy does not arise. An LLM does not understand problems and cannot think through a difficulty in the relevant sense. ¶3: The turn — Putnam. Opens with a question connecting to the challenge (fixing the %%comment%% about the abrupt opening). Then the existing Putnam content — the thought experiment constructs a scenario whose internal logic puts pressure on a familiar picture of meaning; the reader sees why through the argument's structure; the contribution is something the argument does. Closes with the disanalogy: philosophical arguments are more like proofs than paintings. ¶4: Dellsén. The existing paragraph with the %%comment%% fixes applied (add the negative to the dependence-relations formulation; correct the Twin Earth description). Plus 2-3 sentences at the end reinterpreting the history-of-philosophy observation: we go back to old philosophical arguments not to reconstruct anyone's psychology but because the arguments still enable understanding — they still reveal dependence relations for new readers. And progress, on Dellsén et al.'s account, happens "by way of philosophical ideas — theories, arguments, distinctions — becoming publicly available" (p. 679). What is publicly available is the argument. ¶5: Transition to Lipton + block quote. Verbatim from current draft. "Not every account of a phenomenon's dependence relations is equally illuminating, however..." ¶6: Dormative virtue. Verbatim. ¶7: Likeliness without loveliness in philosophy. Verbatim. ¶8: Williamson / overfitting. Verbatim. ¶9: Bengson tri-level method. Verbatim. ¶10: "In sum" + blind review. Already partially updated in the note — "each concerns the argument, not the kind of agent that produced it" and the blind review sentences are already there. May need checking. ¶11: Deep Blue. Existing, with emphasis adjusted from process to agent if needed. ¶12: Transition to corpus thesis. Verbatim. --- So the structural change from the current draft is small: the old ¶1 (which had the challenge compressed into one paragraph plus the art comparison plus the response-telegraphing sentence) splits into ¶1 (art comparison, done) and ¶2 (the challenge, being written now). The Putnam paragraph gets a new opening. The Dellsén paragraph gets fixes and a new ending. The "In sum" paragraph is already partially updated. Everything else stays. The work that remains: finishing ¶2, writing the opening of ¶3, writing the ending of ¶4, and checking ¶10-11. *La struttura del resto della sezione è quasi tutta in piedi — il lavoro vero è il secondo paragrafo e le giunture.* ### User "Then development." what possible benefit is it for me if you're giving me a paragraph plan with this sort of a sentence in it? What's the point? A waste of time. Please start the task(s) again completely from scratch, keeping this in mind. ### Assistant Nick is right. "Then development" is exactly the kind of placeholder that's useless to him. He needs to see what the actual sentences are, not be told that sentences will exist. Let me write the actual paragraph plan with actual content — every paragraph should have the actual sentences or very close approximations, not descriptions of what sentences will do. Let me go through the whole section and write what each paragraph actually says. ¶1 is done. I can just reference it. ¶2 — I have the two approved sentences. What comes after them? This is the part I keep failing at. Let me think about what ACTUALLY needs to be said. The two sentences establish: philosophy seems to require the right sort of mental states, and when we read philosophy we assume the author understands what is being claimed. What's the next thought? Where does the reader's mind go after reading those two sentences? The reader is thinking: "OK, we assume the author understands. So what? Why does that mean philosophy requires understanding?" The answer: because the understanding seems to be part of what makes the text philosophically valuable. Not just the conclusions, but the reasoning — and the reasoning seems to require understanding behind it. Then the history-of-philosophy observation as evidence: this is why history of philosophy is part of philosophy. Then the LLM conclusion. But every time I try to write this, Nick says it's shallow. Let me think about WHY it's shallow. Maybe it's shallow because the claim "the reasoning requires understanding" is just STATED, never SHOWN. I say "the reasoning seems to require understanding" but I don't make the reader feel WHY it seems to require understanding. It's an assertion without philosophical content. What would give it philosophical content? Actually making the case that reasoning requires understanding. Not just asserting it but showing the reader why it's plausible. But ¶2 is stating the CHALLENGE, not arguing for it. The challenge is a view that many people hold. It doesn't need to be argued for from first principles — it needs to be stated clearly enough that the reader recognises it and feels its pull. So maybe the problem isn't that I'm not arguing for the claim but that I'm not stating it with enough specificity. "The reasoning requires understanding" is vague. What KIND of understanding? Understanding of WHAT? Nick's approved sentence says: "understanding what is being claimed." That's specific. The author understands the claims. Not "understands the problem" (vague) but "understands what is being claimed" (specific — the claims in the text). So the development should be about what it means for the author to understand the claims. When we read philosophy assuming the author understands what is being claimed, we are assuming that the author can see how the claims relate to each other — why one claim supports another, why a certain consideration bears on a certain conclusion. The text is the product of someone who could see these relations. Hmm, but is this just the bold sentences again in different words? "Could see why the objection had force" → "could see how the claims relate to each other." Same template, same problem. Let me try a completely different tack. What if the development isn't about spelling out what understanding consists in, but about what follows FROM the assumption? We assume the author understands. What follows? The history-of-philosophy observation follows: this is why studying philosophy's history is itself doing philosophy — because the texts presuppose understanding, and engaging with them is engaging with that understanding. And then the LLM conclusion: if the texts presuppose understanding, a text produced without understanding is not philosophy. This might work. The paragraph would be: 1. Bullet point (the parallel with art) 2. The reading assumption (the author understands what is being claimed) 3. Consequence: this is why history of philosophy is part of philosophy (the texts presuppose understanding; engaging with them is engaging with that understanding) 4. Consequence for LLMs: if understanding is presupposed, a text without understanding is not philosophy Let me try writing this. "We might think, for similar reasons, that philosophy is a uniquely human activity, requiring the right sort of mental states to lie behind it. When we read a philosophical text, we read it assuming that it was written by a person who understands what is being claimed. This may be why the history of philosophy is treated as part of philosophy in a way that the history of science is not: [something about how the assumption of understanding is what makes engagement with old philosophical texts a philosophical activity]. If this is right, the question of whether an LLM can do philosophy does not arise. An LLM does not understand problems and cannot think through a difficulty in the relevant sense." The gap is the "[something about...]" part. What goes there? Option: "...we engage with philosophical texts from the past as philosophy, not merely as history, because we take them to have been produced by people who understood what they were arguing for." Hmm. That's just restating the assumption. "We take them to have been produced by people who understood" = "we assume the author understands." It's circular. Option: "...to study a philosophical argument is to follow the reasoning, and following the reasoning seems to presuppose that there was genuine reasoning to follow — that someone was working through the problem, not merely producing text." This is getting at something. "Someone was working through the problem, not merely producing text" — this is the challenge's characterisation of the difference between philosophy and non-philosophy. Philosophy is the product of working-through. Without working-through, you just have text. But "working through the problem" is another way of saying "understanding" — and we're going in circles. Let me try yet another approach. What if the history-of-philosophy observation is just stated cleanly, without trying to explain it? "We might think, for similar reasons, that philosophy is a uniquely human activity, requiring the right sort of mental states to lie behind it. When we read a philosophical text, we read it assuming that it was written by a person who understands what is being claimed. This may be why the history of philosophy is treated as part of philosophy in a way that the history of science is not. If this is right, the question of whether an LLM can do philosophy does not arise. An LLM does not understand problems and cannot think through a difficulty in the relevant sense." Five sentences. Very short. The history-of-philosophy observation is just stated — "this may be why..." — without development. The "may" signals that it's suggestive, not proven. Is this too thin? Maybe. But Nick has said everything I add is shallow. Maybe thin-and-sharp is better than developed-and-shallow. But Nick also said my versions were "using a lot of words to say very little." Five sentences might be too few words to say very little — but they might also be too few words to say ENOUGH. OK, I think the honest answer is: I don't know what the right development is. Every time I try to develop the point, I produce shallow filler. Maybe the paragraph needs to be short. Maybe the challenge doesn't need as much development as I've been trying to give it. But Nick said Approach 2 was where his instinct lay — developing through the history-of-philosophy observation. So he wants SOME development. Just not the shallow kind I keep producing. Let me think about what a NON-shallow version of the history-of-philosophy development would look like. What makes something shallow vs deep in philosophical writing? Shallow = restating the claim in different words without adding information. Deep = adding a new consideration that changes how the reader sees the claim. What new consideration does the history-of-philosophy observation add? The observation: history of philosophy is part of philosophy. What does this ADD to the claim that philosophy requires understanding? It adds INSTITUTIONAL EVIDENCE. It's not just that we, as individual readers, assume the author understands. The discipline as a whole treats old philosophical texts as philosophy rather than as history. This is a fact about how the discipline operates, not just a fact about how individual readers read. And this fact seems to support the practitioner-centred view: the discipline treats the texts as philosophy because it treats the understanding behind them as philosophically relevant. Is this a new consideration that changes how the reader sees the claim? Maybe. It moves from "we assume understanding" (individual psychology) to "the discipline presupposes understanding" (institutional practice). That's a shift from the subjective to the institutional. OK. So the development would be: "This is not just a feature of how individual readers read philosophy. The discipline as a whole treats the history of philosophy as part of philosophy — something that is not true of other disciplines — and this seems to reflect the assumption that philosophical contributions are bound up with the understanding of the people who produced them." Hmm, is that adding enough? Let me try folding this into the paragraph. "We might think, for similar reasons, that philosophy is a uniquely human activity, requiring the right sort of mental states to lie behind it. When we read a philosophical text, we read it assuming that it was written by a person who understands what is being claimed. This assumption runs deeper than individual reading habits: it is reflected in the fact that the history of philosophy is treated as part of philosophy in a way that the history of science is not, as though philosophical contributions were inseparable from the understanding of the people who produced them. If this is right, the question of whether an LLM can do philosophy does not arise. An LLM does not understand problems and cannot think through a difficulty in the relevant sense." That's six sentences. The development is: the assumption is institutional, not just personal. The history-of-philosophy observation provides evidence that the discipline itself presupposes understanding. "as though philosophical contributions were inseparable from the understanding of the people who produced them" — this states the challenge clearly: contributions are inseparable from understanding. The response (¶3 onwards) will argue they ARE separable. Is this shallow? Let me check. Does every sentence add something new? 1. Bullet point — bridges from art to philosophy 2. Reading assumption — what we assume when we read 3. The assumption is institutional — history of philosophy is treated as part of philosophy — as though contributions are inseparable from understanding 4. LLM conclusion 5. LLM conclusion continued Sentence 3 does the work. It moves from individual reading to institutional practice and states the challenge's core claim (inseparability). That's not just restating the same point in different words — it's an escalation from personal assumption to disciplinary fact. I think this might work. Let me also write the other paragraphs with actual content. For ¶3 (Putnam), the opening needs to connect to the challenge. The current note has: "%%not how i write%% Putnam was not reporting..." The new opening should be something like: "But does the quality of a philosophical argument depend on the kind of agent that produced it, in the way that an artwork's significance might depend on its creator?" Then the existing Putnam content. Actually wait — let me look at what's already been discussed for this opening. In the April 2 conversation, this exact question was proposed and Nick didn't object to it (he objected to other things but not this). And it's already in the paragraph in the note at line 16: "**But does the quality of a philosophical argument depend on the kind of agent that produced it, in the way that an artwork's status might depend on the intentions of its creator?**" So this opening IS already in the note — it's in bold. So it was added during the April 2 session or the crashed session today. Good — it's there. For ¶4 (Dellsén), the ending needs the history-of-philosophy reinterpretation. The existing paragraph ends: "On Dellsén et al.'s account, enabling that kind of understanding is what philosophical progress consists in." Then the April 2 plan adds: "And progress, so understood, happens 'by way of philosophical ideas — theories, arguments, distinctions — becoming publicly available' (p. 679). What is publicly available is the argument. This bears on why history of philosophy is part of philosophy: we return to [old arguments] not to reconstruct [anyone's] psychology but because the [arguments] still put readers in a position to grasp dependence relations they might not otherwise have seen." Wait — this last sentence uses Frege in the note version. But Nick said Frege should be removed. However, looking at the note again at line 18... Actually, let me re-read it. From the note, line 18: "...On Dellsén et al.'s account, enabling that kind of understanding is what philosophical progress consists in. **And progress, so understood, happens "by way of philosophical ideas — theories, arguments, distinctions — becoming publicly available" (p. 679). What is publicly available is the argument. This bears on why history of philosophy is part of philosophy: we return to Frege not to reconstruct his psychology but because the *Foundations of Arithmetic* still puts readers in a position to grasp dependence relations they might not otherwise have seen.**" So the Frege material IS currently in the note's Dellsén paragraph (in bold — added during the April 2 session). Nick now wants Frege removed. So this ending needs to be rewritten without Frege. Without Frege: "This bears on why history of philosophy is part of philosophy: we return to old philosophical arguments not to reconstruct anyone's psychology but because the arguments still put readers in a position to grasp dependence relations they might not otherwise have seen." That works without naming a specific philosopher. OK. Now let me draft the full plan with actual sentences for every paragraph. Actually, ¶5-9 are verbatim. I don't need to rewrite them. I can just say "existing paragraph, verbatim" and quote the first sentence so Nick knows which paragraph I mean. For ¶10, the "In sum" — the note already has the updated version with "each concerns the argument, not the kind of agent that produced it" and the blind review sentences. So this is mostly done. I should quote what's there and flag any remaining issues. For ¶11, Deep Blue — the note has it. The rephrasing from process to agent might already be done or might need doing. For ¶12, transition — verbatim. OK, let me write the whole plan now. ## ¶1 — Art comparison (DONE) In this section we address what we might call the _challenge from authorship_: the idea that philosophy is something that only people, or at the very least _minds_, can do. An imperfect comparison would be with _art_: a lot of people, perhaps a majority, would argue that generative AI systems, in virtue of their not being people, cannot make art. This would not be to say that AI outputs cannot be aesthetically pleasing, only that they are not artworks: on this view, art is the product of the right sort of mental activity on the part of its maker — imaginative, expressive — and a system without mental states has no such activity to draw on. ## ¶2 — The challenge We might think, for similar reasons, that philosophy is a uniquely human activity, requiring the right sort of mental states to lie behind it. When we read a philosophical text, we read it assuming that it was written by a person who understands what is being claimed. This assumption runs deeper than individual reading habits: it is reflected in the fact that the history of philosophy is treated as part of philosophy in a way that the history of science is not, as though philosophical contributions were inseparable from the understanding of the people who produced them. If this is right, the question of whether an LLM can do philosophy does not arise. An LLM does not understand problems and cannot think through a difficulty in the relevant sense. ## ¶3 — The turn (Putnam) But does the quality of a philosophical argument depend on the kind of agent that produced it, in the way that an artwork's status might depend on the intentions of its creator? Putnam was not reporting a previously unnoticed item in the world; he was making a case, by way of thought experiment, that meanings are not fixed solely by what is in the speaker's head. The thought experiment does its work not by pointing to something outside the text — there is no Twin Earth for us to go and inspect — but by constructing a scenario whose internal logic puts pressure on a familiar picture of meaning. A reader who follows the argument does not simply learn that meaning is externally determined; she sees why, through the specific pressure the scenario puts on the assumption that mental life alone fixes what our words mean. That understanding could not be separated from the text that produced it. The philosophical contribution is not something the text reports; it is something the text does. Someone who had never heard of Putnam, who knew nothing about his career or his reasons for constructing the scenario, would gain the same understanding from the same argument. Philosophical arguments are, in this respect, more like proofs than paintings. A proof is valid in virtue of its structure; nobody needs to consult the mathematician to check. [NOTE: The %%comment%% about possible detail loss from the Putnam paragraph still needs checking against the earlier version before this is finalised.] ## ¶4 — Dellsén Dellsén et al. propose that philosophy makes progress when philosophical research puts people in a position to increase their understanding — where increased understanding is a matter of more accurately or more comprehensively representing the dependence relations in which a phenomenon stands, or fails to stand, to others (2024, pp. 665, 680-81). Understanding, on this account, goes beyond knowing that something is the case. It involves grasping how one phenomenon depends, or does not depend, on another — seeing, for instance, not just that meaning is externally determined, but how the speaker's environment rather than the speaker's psychology fixes what words refer to. One speaker on Earth and another on Twin Earth share every psychological state and yet mean different things by the same word, because their environments differ in ways that bear on reference — a dependence relation of just the kind Dellsén et al. describe. A reader who works through the scenario does not simply acquire the belief that externalism is true; she comes to see why meaning depends on environment, and what features of the case make this so. On Dellsén et al.'s account, enabling that kind of understanding is what philosophical progress consists in. And progress, so understood, happens "by way of philosophical ideas — theories, arguments, distinctions — becoming publicly available" (p. 679). What is publicly available is the argument. This bears on why history of philosophy is part of philosophy: we return to old philosophical arguments not to reconstruct anyone's psychology but because the arguments still put readers in a position to grasp dependence relations they might not otherwise have seen. ## ¶5 — Lipton transition + block quote "Not every account of a phenomenon's dependence relations is equally illuminating, however..." Existing paragraph, verbatim. ## ¶6 — Dormative virtue "Lipton illustrates the contrast with Molière's joke about the dormative virtue of opium..." Existing paragraph, verbatim. ## ¶7 — Likeliness without loveliness in philosophy "The same distinction applies in philosophy, though it cuts in a way that is not always recognised..." Existing paragraph, verbatim. ## ¶8 — Williamson / overfitting "On the other hand, when likeliness prevails at the expense of loveliness there is no genuine progress..." Existing paragraph, verbatim. ## ¶9 — Bengson tri-level method "Bengson et al. organise these evaluative concerns into a systematic method..." Existing paragraph, verbatim. ## ¶10 — "In sum" + blind review In sum, the evaluative standards we have assembled all bear on what a philosophical text says and how it argues for it. Lipton's distinction between illumination and mere accommodation, Williamson's desiderata for theoretical virtue, and Bengson et al.'s method for assessing how well those standards are met: each concerns the argument, not the kind of agent that produced it. They do not ask how the author arrived at her argument but whether the argument, as it stands on the page, meets the relevant standards. The practice of blind review in philosophy rests on the same assumption: referees assess what a paper achieves without knowing who wrote it. If the kind of agent behind the argument — whether a senior philosopher or a graduate student, whether a human being or a machine — were relevant to the argument's quality, then blind review would be a defective practice rather than the discipline's standard method of assessment. [This is already in the note — checking whether any further adjustment is needed.] ## ¶11 — Deep Blue "This is not unique to philosophy. Deep Blue, the computer that beat Kasparov in 1997..." Existing paragraph. The Gaut reference needs verifying. The emphasis may need adjusting from "manner of selection" (process) to "kind of agent" — but this is a light touch, one clause. ## ¶12 — Transition "The question, then, is what follows when a language model is trained on a corpus that this evaluative apparatus has shaped..." Existing paragraph, verbatim. Pigliucci footnote stays. *Dodici paragrafi — il secondo è quello da scrivere, il resto è quasi tutto in piedi.* ### User Please just reveal the JSON for this conversation we've been having in Finder. ### Assistant Done — the file is revealed in Finder. ### User I said the conversation that we are having right now. You opened the index file. ### Assistant That file IS this conversation — `a4669597-091b-4e48-84dd-95c88fe870bb.jsonl` is the most recently modified session file and it's the one we're in right now. It's not an index file, it's the JSONL for this session. It should be highlighted in Finder already. Do you see it? ### User fine, but that is not what was revealed in finder. this was: '/Users/nickyoung/Library/Mobile Documents/com~apple~CloudDocs/claude-projects/-Users-nickyoung-My-Obsidian-Vault/sessions-index.json' ### Assistant Let me try again with the correct path. ### Assistant That should now reveal the actual JSONL file for this conversation, not the sessions index.