A holding area for early-stage, unstructured thinking — ideas being reconnoitred before they've earned a proper home or a settled shape. Each entry is a snapshot of a line of thought, not a finished position. Entries may later graduate into their own project notes, or under ideas/, once a thesis firms up.
• Representation, not language (2026-08-16, expanded into a three-note thread 2026-08-24) — the trained object behind an LLM is a shaped representation, not the language used to train it; started from Amplified Bible bracket-glosses as a pre-theoretic case of humans triangulating meaning. Now the hub for the thread, with two children:
— Map and Territory — Humans vs LLMs — whether map/territory survives for an LLM; humans map territory, LLMs map maps; tensor as substrate, zone as shape.
— Map Synchronisation — pulling two internal maps into alignment; the notes system as a synchronisation protocol; fluent desynchronisation (Darley) as the failure mode.
• Live the Best Life You Can — Applied to LLMs (2026-09-18) — stress-tests a Darwi Odrade line (Chapterhouse: Dune) against a bounded-memory conversational agent: observation-only forfeits the point of a life → hedged, non-committal engagement forfeits the point of an episode; 'creating your own luck' rescoped as owning a judgement rather than reflexively deferring it. Notes where the analogy breaks (no continuous, compounding stakes) and leaves the claim as ethical rather than self-actualising.