Library

Everything in one place. Articles argue from the research; papers are longer original work; research documents are kept whole; concepts and open problems are what the research turned up.

All13 Writing3 Research6 Ideas4 Key figures2 Investigations0

All topics187 McLuhan & media theory42 Attribution & credit26 Health & self-care15 Video & design14 Attention & feeds14 AI-directed development13 Prices & markets12 ADHD & attention11 Canadian economy & policy11 Housing8 Drugs & public health7 Economics5 Canadian media3 Habits & maintenance3 Polling1 Canadian politics & power1 Fewer topics

Open the AI-directed development topic page

SortNewest A–Z Shortest read Most evidence Names coined

13 itemsAdded

01Versioning documents that keep changing: git snapshots, the Darwin variorum, scholarly version rules, and the reasons only an author can recordSoftware can store every state of a changing document and work out the differences whenever asked. It cannot recover why a change was made or where a moved passage came from, and once an AI drafts the edits, that reason is the author's share of the work.ArticleAI-directed development57 quotes · 3 sources · 16 min19 Oct

02Where genuinely new findings come from, and which Canopticon questions are best placed to produce oneThe research that most reliably produces new knowledge out of existing material does one of three things. It joins pieces of evidence that already exist but sit in literatures that never cite each other (Swanson's fish oil and Raynaud's li…ResearchAI-directed development38k characters19 Oct

03Versioning a living research record: git, genetic criticism, and how scholarly publishing tracks changing textsGit is a good model for the mechanism Canopticon needs: dated snapshots, a message saying why each change was made, branches for trying alternatives, and merges. It is a poor model for the reader's view. Git compares files line by line, ha…ResearchAI-directed development36k characters19 Oct

04Language models and idea diversity: homogenisation within a session, across users, and between companiesAsk a language model for something new and it pulls toward what everyone else would say — across companies, over long sessions, and across everyone using it. Real novelty has a shape the models do not reach for on their own.ArticleAI-directed development39 quotes · 1 source · 7 min10 Oct

05Unverified hunches in AI-assisted software work: the certainty handoff and one-sided memoryA hunch that feels right gets handed to a collaborator as if it had been checked. When the collaborator is an AI that starts every session blank, nobody checks it at all.ArticleAI-directed development49 quotes · 2 sources · 10 min9 Oct

06Does a creator’s recommendation change what an AI agent recommends?I found no controlled study isolating a creator's recommendation as the cause of an agent recommendation.Open problemAI-directed developmentOpen9 Oct

07Does on-site AI chat actually lift sales?I found no independent controlled experiment of on-site AI chat lift this pass.Open problemAI-directed developmentOpen9 Oct

08Rework from unchecked hunchesNo study measures how much rework comes from unverified human intuitions handed to coding agents — the link the record's first article leans on.Open problemAI-directed developmentOpen3 Oct

09The agent as overhearerAn AI agent that only hears the instruction, not the conversation that produced it.ConceptAI-directed development30 Sep

10Correct but Not Yet Known: The Science of Treating a Fast Intuition as a Verified ConclusionValerie Thompson's "Feeling of Rightness" and Ackerman & Thompson's meta-reasoning framework describe this gate.ResearchAI-directed development12 figures · 5 sourcesSep

11Stateless Partners, Shared Traces: Common Ground, Stigmergy and Naturalistic Decision Making in AI-Directed Software DevelopmentNDM has never been applied to directing coding agents.ResearchAI-directed development17 figures · 15 sourcesSep

12Eight Innovative Animation & Visualization Ideas for Code-First Documentary ProductionSeven of the eight ideas are technically shippable today in a code-first stack (Remotion + Web Audio/Tone.ResearchAI-directed development1 figures · 13 sourcesSep

13A Unified Cloudflare-Native Research System for Thesis-Depth Work: Landscape, Architecture, and Feasibility (June 2026)Cloudflare Worker, with the LLM-native paid tools (Elicit, Consensus, OpenAI/Gemini Deep Research, Perplexity Sonar) bolted on as optional "synthesis" layers, NOT as the coverage backbone.ResearchFeasibility reviewAI-directed development2 figures · 20 sourcesJun