Conversation

The floor for new “note-taking apps” now includes bi-directional links, local data/self-hosted, complex block types, transclusions, advanced queries, graph viz, spaced repetition systems, multiplayer, and a slice of cake. Some think we’re in hell. Seems pretty great to me 😎

453

supachaidev

@supachaihimself

Feb 8

The problem is it should be an easy way to tell us that the new note we are writing is the same idea that we have created before so we do not duplicate the note. Or at least link it automatically. The unlinked mention of Obsidian is ok but not seemless yet.

Ari T.

@also_tan

Feb 8

I agree this is a real problem with actually USING your KM system. It pops up in GTD as well. The issue is, how can you know you already have this knowledge and should thus search the database first? (so as not to reinvent the wheel if you encounter a problem)

Abraham

@ABSamma

Feb 8

That happens a lot even in Wikipedia. This is we're curators and maintainers can be extremely helpful in proposing merges etc. Unfortunately, I don't think there's a reliable way of solving this. You still need humans to poke around and maintain.

Replying to

and

A nice meaty problem! Roam accommodates it well as you can search inline with ((brackets)) and unlinked mentions catches 95% of repeats But some system that automatically looks for blocks with duplicate phrasing and concepts (maybe some ML/AI helpers) and flags them is feasible

7:31 AM · Feb 8, 2022Twitter for iPhone

Replying to

and

It would be interesting to see a note-taking app using ML/AI helpers to solve this problem. Maybe while we're typing a keyword that is closed to the old notes, it will warn us without us typing [[ first.

Replying to

I've been back and forth on the issues on AI helping with this kind of task. On the one hand it could act as a nice prompt + prevent duplication, but it could also rob of the ability to form those connections yourself which could be detrimental to learning/ deep understanding.

pngwn

@evilpingwin

Feb 8

But we also need to consider the limits of the human brain, we can't remember everything. I'd be curious to know if there has been any research in this area: does the prompt + recognition work the same as manually making the connection yourself, cognitively speaking.

Show replies