Knowledge's purpose is to help us understand the world so that we can act in it, but "knowledge itself" is trapped within our minds. We need to formalize and transcribe knowledge into a medium so that it can be ingested by others. This process is noisy and will always be different from "knowledge itself", there will always be a loss of information. So as technology changes, so should knowledge bases, to more accurately capture the representation that is held in our minds.

1) don't use pdfs.

2) Not everything in a paper is relevant to a reader, but you can imagine a paper or work that might be 90% irrelevant, but the other is 10% diamond to a specific reader. How do you find the gold in something like a full textbook? You create atomic units -> turn the sections of a paper into individual events and chain them together. The chain provides context within the larger work if you care to dive in, the individual events help isolate the diamonds and enable not just remixing, but literal stand in's for a section. Why rewrite a full introduction justifying your work if someone else stated it way better than you can articulate? Just slot in their section, maybe fork it and add it to whatever your working with. This modularization of long form text has been formalized as NIP-62 Curated Publications https://github.com/nostr-protocol/nips/pull/1600

Want to automatically parse content as events? identify sections and then upload them.

want to automatically convert videos into articles with images?

https://habla.news/u/liminal@gitcitadel.com/1729213258345

https://github.com/IllDepence/unarXive

unarXiv has an open subset of papers that could be used.