Definitely. You have a nice touch with complexity. The only helpful aspect would be concrete examples of some of the context where this configuration starts breaking down.
1. Size. Under a few hundred notes it's fine. After that, search starts pulling notes that look related but aren't.
2. Questions with more than one step. "Why did we drop X, and what broke because of it" sits across three notes. Search finds the ones that mention X, not the ones that explain it.
3. Old facts. Reverse a decision or swap a tool, and the old note stays. The system then quotes the wrong half.
Great article. Thanks for sharing your ideas.
Thank you!
Just read some of your posts and liked a lot. Just subscribed you.
Keep on the good work.
Thank you, i also subscribed back, let’s learn from each other.
Great. Thank you.
Let’s keep in touch.
Definitely. You have a nice touch with complexity. The only helpful aspect would be concrete examples of some of the context where this configuration starts breaking down.
Thanks Arne-Per. A few spots where it breaks:
1. Size. Under a few hundred notes it's fine. After that, search starts pulling notes that look related but aren't.
2. Questions with more than one step. "Why did we drop X, and what broke because of it" sits across three notes. Search finds the ones that mention X, not the ones that explain it.
3. Old facts. Reverse a decision or swap a tool, and the old note stays. The system then quotes the wrong half.
The first two are what got me into graphs. Wrote it up here if it helps: https://theaioperator.io/p/what-is-graph-engineering-a-field