3 ms·

A Git-powered Knowledge Network. Goal: Empowers humans to explore and iterate knowledge faster Features: - Upload your own knowledge base, wiki to join the n

A Git-powered Knowledge Network. Goal: Empowers humans to explore and iterate knowledge faster Features: - Upload your own knowledge base, wiki to join the network - Search network-based knowledge - 'Clone' a knowledge branch - Edit on it then 'Push' it back to the network - The branch owner / other users can choose to 'Pull' the update to their branch - 'Merge' different knowledge branch Steps: 1. Build a search engine for network-driven knowledge 2. The users iterate knowledge with other knowledge branch owners in the network 3. Build an efficient machine learning data pipeline on top of the structured knowledge input & output system 4. Use the unique data to train the customizable personal knowledge AI

anilgulecha 6y ago
I'm working on a branching/merging authoring framework and application. Git, specifically, is bad as a backing store - as merge conflict resolution is not something you want end users to do - it's not friendly. p2p sync (OT/CRDT) is what I'm betting and building on. If anyone is interested in this space, I'd like to discuss/geek-out on this with you. Please reach out.
- allenleein 6y ago
  Great points, will reach out soon :)
- billconan 6y ago
  any good tutorial on implementing crdt? Most of them are for plain text. How to implement crdt for structured data?
  - anilgulecha 6y ago
    Looks at yjs - it's probably the project that's furthest along here.
prerok 6y ago
I don't understand how this is different from a wiki. I think that wikipedia already contains the collective knowledge from a vast amount of people and, dare I say it, is already the knowledge base of humanity. Can you explain a bit more what the difference is?
- allenleein 6y ago
  Good question. It's very different. For example, if you want to install/learn Haskell, Lisp...on your machine, you don't want to follow Wikipedia. Instead, you go community-generated docs or personal blogs for the updated docs.
raghava 6y ago
Just curious. Isn't wikipedia (sans git - in some sort) doing this?

anilgulecha 6y ago

I'm working on a branching/merging authoring framework and application. Git, specifically, is bad as a backing store - as merge conflict resolution is not something you want end users to do - it's not friendly. p2p sync (OT/CRDT) is what I'm betting and building on. If anyone is interested in this space, I'd like to discuss/geek-out on this with you. Please reach out.

allenleein 6y ago
Great points, will reach out soon :)
billconan 6y ago
any good tutorial on implementing crdt? Most of them are for plain text. How to implement crdt for structured data?
- anilgulecha 6y ago
  Looks at yjs - it's probably the project that's furthest along here.

prerok 6y ago

I don't understand how this is different from a wiki. I think that wikipedia already contains the collective knowledge from a vast amount of people and, dare I say it, is already the knowledge base of humanity. Can you explain a bit more what the difference is?

allenleein 6y ago
Good question. It's very different. For example, if you want to install/learn Haskell, Lisp...on your machine, you don't want to follow Wikipedia. Instead, you go community-generated docs or personal blogs for the updated docs.

raghava 6y ago

Just curious. Isn't wikipedia (sans git - in some sort) doing this?