This talk is good: https://youtu.be/8pTEmbeENF4?t=748
This part (marked above) in which he talks about how "APIs" are crap is very enlightening. Unfortunately he doesn't say how we can get rid of APIs. But I think Nostr can and is solving at least part of the problem.
I will take a look at it. But first I will take a nap. GN 🫂
sounded like he was talking about something like autogpt + copilot.
but…this talk was only 10 years ago? sounds like it was 30 years ago. and…transparencies?!
Interesting train of thoughts.
I see a potential conflict. While we have to be open for new ways of doing things, we also have to look at software development as a trade, that requires learning and mastering.
So, digging into the details and training ourselfes as much as possible but still be "open".
That sounds like becoming a master pianist and still looking for new musical instruments to learn.
Very good. Loved this one of his as well.
APIs can be replaced by a large tube that sucks data from one place to another. Simples.
We are fortunate that when compared to alternatives it's relatively easy to start implementing nostr protocol. It's great to have mini-specs (NIPs) which are explaining the most important aspects of certain topic. And you can implement subset of them and still get something useful. But we are still using primitive tools to do this development, as Bret's video shows.
Two weeks ago I started to watch again Alan Kay's presentations. He also doesn't like APIs. It's too rigid for the scale he is interested in like we are too. You need some sort of auto-negotiation between communicating parties which are complex and getting more and more autonomous (not just automated as we humans are making them rn). His thinking is inspired by biological systems. Encapsulation of internal state of the cell (we're already ok here). Its interaction with environment (message passing, good too). DNK - so far I didn't see any executable code (like wasm? or even better something higher level) transported by nostr events (somebody pls correct me). Simulation within cell of some subset of external environment (could be other cells or more complex structures). And use of simulation results for reasoning and adaptation. I don't see evidence for last two.
Microapps idea is fine but is there something better? He was thinking a lot about the ways to empower users to do custom programing. I think that's important for the future. As long as we have this division between developers and users, it's asymmetry which could impact decentralisation. Things could be better in the long run. Store for microapps or repository of composable bloat-free building blocks with cryptographically proven ancestry stored on relays accessed from mega app which is combination of OS and IDE?
Protocols are like APIs. Can we grow them automatically or at least more efficiently, for now with humans in the loop guiding their development..
Ele menciona paradigmas de programação e computação completamente esquecidos. Ideias que poderiam colocar de cabeça pra baixo o mundo tecnológico atual
