Knowledge Graph's need to be Disposable like AI Codebases
Date: 2026-09-09
I used to describe The Daemon, the precursor to CGFS as A medium to think through
Understanding a topic is like creating a series of PR's until you have one that fits the requirements and or the world.
Every time we read something, tell a story, or even look at a meme the way our mind interprets the information is unique to those circumstances.
Sure there needs to be this beautiful ontology like wikidata that describes absolutely everything that we pull from. But wikidata is just like the dictionary, the lived life of people is expressed with words not edited, debated, and peer reviewed perfection.
I find with AI it is sometimes better to start a project from scratch with a slightly different pronot. Just like someone might reread a book from the beginning or get someone to explain something to them.
Maintaining the Provenance of live thinking thinking and creating A Medium to Cultivate Personal and Collective Thinking
Design Consideration
I am still thinking that Meme Maps are going to be Nostr Events. But how do we link and cache them? Maps are like Nostr Kind 1 threaded notes where everything referencing the origional note. Instead we can have something similar but instead of a thread we get a graph. And instead of everything being new text we can NIP-59 Gift Wrap events we link into the graph.
Also each Nostr Event is basically a SQL Table. But each row needs to be updated independently like DynamoDB but if we get DID pointers like NIP-05 DNS Identities we can have retarded linked lists. Then you also have the problem of who regulates if a DID get's updated or not. But this is how thinking is done, someone might think Plato's forms is a great idea then add on a bunch of fluff over here, or be like there is this exception over here.
The fact that there is no meme mapping cultivating medium yet is sorta weird to me.