Field of Sheaves
Pete Kaminski's meta-project of using AI to synthesize frameworks that address similar topics but don't typically talk to each other.
Core Concept
"Smashing together frameworks" that ought to go together but never have before.
Each Synthesis = A "Sheaf"
- 10-20 page websites
- Combine multiple frameworks/theories
- Address the same part of the world from different angles
- Authors/creators often don't know about each other's work
The Meta-Collection = "Field of Sheaves"
- Collection of all these syntheses
- Working on framing "what is good about these things, and what is bad"
- Exploring "how do humans work with them"
- "Why is it okay that the AI has kind of organized things in a way that makes sense"
Examples Created
Pete has created syntheses on:
- Collective efficacy
- Social cyclic theories
- Intention-action gap
- Strategy and management frameworks
- Government governance principles
- Commons stewardship by communities
"I've got, you know, 6 or 8 sheaves of whatever these things knowledge together"
The Process
Inspiration Sources
Often triggered by OGM call mentions:
Example: Stacey Druss mentioned something about "people in her building working together" creating a "little society that protects itself from outsiders"
Pete's response: "Huh, that reminds me of:
- The broken windows theory
- This
- That"
Then: "Hey, AI, help me put together a wiki, a 20-page, kind of, paper slash thesis slash framework slash overview"
The AI Partnership
"Sometimes things that are a little bit different, that people have kind of been looking at the same part of the world and coming up with different ways to do it, and they don't talk to each other. So, the most fun for me is when me and the AI can kind of smash frameworks together..."
Pete's Ambivalence
The Good
- "Half, like, amazing"
- Has pointed out a couple in the OGM list
- Successfully synthesizing disparate knowledge
- Creating useful overviews
The Embarrassing
- "Half I'm embarrassed by it"
- "Because it's mostly built by AI"
- Too embarrassed to release publicly yet
- Still working on the framing
The Meta-Work
Working on:
- "What is good about these things, and what is bad"
- "How do humans work with them"
- "Why is it okay that the AI has organized things in a way that makes sense"
Call for Collaborators
"If this is something that's like, oh my gosh, Pete, could I help, you know, could I do that? Could you help me do one of those myself? I'd love to hear from you."
Relation to Agentic AI
Built using Agentic AI and Document Synthesis tools:
The same tools Pete uses for software development, now applied to knowledge synthesis.
Connection to Remixability
Relates to Scott Moehring's Remixability concept:
- Both about creating "ingredients" rather than finished products
- Both about enabling others to create
- Both about modular, recombinant approaches
- Difference: Scott creating original components, Pete synthesizing existing frameworks
Philosophical Questions
Attribution and AI
"Why is it okay that the AI has kind of organized things"?
When work is "mostly built by AI," what's the human's role? What's the value?
Human-AI Collaboration
"How do humans work with them?"
Not replacing human thinking but augmenting/organizing it in new ways.
Quality and Authority
What makes an AI synthesis valuable vs questionable?
How do you evaluate when you're "half amazed, half embarrassed"?
Comparison to Traditional Synthesis
Traditional:
- Human reads all sources
- Human synthesizes
- Takes months/years
- Clear attribution
Field of Sheaves:
- AI reads sources
- AI organizes and drafts
- Human guides and refines
- Takes days/weeks
- Murky attribution
Future Vision
Pete: "Everybody's gonna be doing this kind of stuff in 18 months, 24 months"
Currently working on:
- Better framing of the approach
- Understanding what's valuable vs problematic
- Making it ready for broader sharing
Related Concepts
- Agentic AI and Document Synthesis
- Claude Code
- Droid
- Remixability
- Knowledge Synthesis
- Human-AI Collaboration
Participants
- Pete Kaminski (creator)
- Doug Breitbart (expressed interest: "Count me in on that invite. Would love to try my hand and see yours")
- Stacey Druss (unwitting inspiration for one synthesis)
Back to README