# Scratch Pad *Quick notes, surfaced content, things to have on hand today* This morning we spent a lot of time trying to turn my ideas about propaganda into a draft. In fact it wasn't just this morning it was yesterday and even the day before as well and I felt like I was bashing my head against the brick wall for a long time. What I'd like you to do is try to. I would like us to come up with a plan to improve you. So what I do the next sort of round of drafting or planning or coming up the ideas for riding peace you become more helpful and one way you could at least make a start of working out. How to do this would be if you look at the chat GPT transcript and see how that transpired and how that gave me a much better way of going forward first time and then compare it to our conversations about the same topic and to see whether we can determine what mistakes you're making bitch or whether something to do with the way I apply to chatgpt rather than problem at your end Gym note: the shower is is very hard. I can barely do 3x12 at 10kg ## Substack Piece: "Generating Philosophy with AI" **Current state (19 Jan, afternoon):** After productive sessions with Claude (17-18 Jan) developing the conceptual framework, had a breakthrough session with Gemini that produced a substantially improved structure and initial draft prose. The piece now has: **Structure (two parts):** - **Part 1: The Defense of the Instrument** — Establishes "decoupling of cognition from value"; LLMs need not understand text to produce philosophically valuable text - Section I: Opening (reframed "Can LLMs do philosophy?" question) - Section II: Decoupling Cognition and Value (core claim + Butter Chicken defense) - Section III: The Safety of the Discipline (auto-evaluating/self-authenticating nature of philosophy) - Section IV: Recombination Realism (novelty objection using Boden/Gaut) - **Part 2: The Lab of the Everted Brain** — Mechanism of how LLM-augmented philosophy works - Section I: The Labor of Legibility (articulation as intellectual work) - Section II: The Frankish Cycle (reciprocal prompting, inner speech connection) - Section III: The Friction of Failure (bad responses as useful cognitive traction) - Section IV: Conceptual Triangulation (multi-LLM workflow, Ted Chiang callback) **Key conceptual moves:** - "Butter Chicken" defense: reliability is output property, not process property - Philosophy as "auto-evaluating" discipline: understanding = verification - Reciprocal prompting using Keith Frankish's activity view of inner speech - "Cognitive eversion" connecting to Ted Chiang epigraph **Critical note from the draft:** Gemini's prose is structurally sound but "appalling" in tone/style/vocabulary/register/jargon. Needs complete stylistic rewrite while preserving conceptual architecture. **Related files:** - Main draft: [[Generating Philosophy with Artificial Intelligence]] (in Writing/substack/) - Gemini transcript: [[useful chat with gemini about generating philosophy with llms]] - Earlier archived drafts: [[Generating Philosophy with AI (April 2025 Draft)]], [[Generating Philosophy with AI (May 2025 Draft)]] - Background notes: [[Inner Speech and LLM Coupling - Research Discussion]] **Next steps:** - Rewrite opening and Part 1 in proper voice (less academic jargon, more accessible Substack tone) - Develop concrete workflow examples for Part 2 - Consider whether Part 3 provocation ("philosophers out of the loop") belongs in this piece or separate follow-up --- # What's Happening *Active threads and today's activity — updated by /harvest* ## Active - **Generating Philosophy Substack** — breakthrough session with Gemini produced solid structure; now needs stylistic rewrite - **McKenna/Propaganda paper** — WIP presentation prepared; now exploring slide design for other topics (typography, generative AI) - **Slide Design Workflow** — developed PowerPoint generation system with Obsidian color scheme ## Sessions - 10:28 - "/link-chat morning..." — Linked to morning review; discussed Mitchell-Douglas thin conception; analyzed ChatGPT summary of McKenna paper; developed critique angles and paper structure options - 13:37 - "continue working on propaganda..." — Revised abstract through multiple iterations; added Tuttle Ross references; moved McKenna PDF to Learning folder; created Planning Scraps note - 19:08 - "can you remind me of how..." — Failed to properly consult previous sessions on Substack work; eventually documented Gemini breakthrough session state; identified workflow improvements needed - 21:29 - "today we have had a very substantia..." — Found today's propaganda sessions; developed WIP presentation slides with block quotes from McKenna and Bonard et al.; iterated on structure for better argumentative flow - 22:47 - "i believe i saw a few months ago..." — Installed pptx/pdf skills; created varied slide designs for propaganda, typography, and generative AI topics using exact Obsidian colors; converted to HTML for preview - 22:50 - "today we have had a very substantia conversatio..." — today we have had a very substantia conversation, maybe more than one about m... - 22:57 - "last question before i go to bed. how easy is i..." — last question before i go to bed. how easy is it for me to set up a telegram ... - 22:59 - "i believe i saw a few months ago there was a cl..." — i believe i saw a few months ago there was a claude skill that someone was sh... ## Actions - 12:06 - Created [[McKenna Propaganda Paper Plan]] in Notes/ - 13:00 - Moved McKenna PDF to Learning/ as `Sophistry on Steroids by Robin McKenna 2025.pdf` - 13:30 - Added Ross references to [[McKenna Propaganda Paper Plan]] - 13:33 - Created [[McKenna Propaganda Paper - Planning Scraps]] in Notes/ - 13:40 - Added task: "Submit McKenna/propaganda abstract to Koblizek (4 days late)" - 13:59 - Created [[2026-01-19 - McKenna Paper Critique Development]] in Notes/ - 19:05 - Scheduled "Return to argumentation schemes conversation" for 2026-01-20 09:00; Added task: "Return to argumentation schemes conversation" - 19:08 - Added tags to [[useful chat with gemini about generating philosophy with llms]] - 19:08 - Updated daily note with detailed Substack piece status in Scratch Pad - 22:02 - Copied session to Downloads as `McKenna_Propaganda_Slides_Session_2026-01-19.jsonl` - 22:11 - Created [[WIP Presentation - AI Persuasion and Propaganda]] in Notes/ - 22:47 - Installed pptx skill from claude-office-skills repo - 22:47 - Installed pdf skill from claude-office-skills repo - 22:47 - Created 10 test slides (HTML) in ~/Desktop/test-slides/ - 22:54 - Added task: "Clone claude-config repo on laptop" - 22:59 - Updated obsidian.md and powerpoint-generation.md with vault typeface (iA Writer Duospace) ---