Daniel CohenMar 8
Hailey
Computers can be understood https://blog.nelhage.com/post/computers-can-be-understood/
Computers can be understood

Computers and computer systems are build up from deterministic, comprehensible, building blocks. Their operations and behaviors can be understood and reasoned about. I relate my personal beliefs and mindset on this point, and explore some manifestations and ramifications of this philosophy.

Made of Bugs
Feb 10 at 8:21amWeb
Show threadWolf480plFeb 10

@hailey
> Willingness and skill for reading source reduces your reliance on documentation

Unfortunately IME it also reduces your reliance on community forums for help, which 1. makes those forums miss out on your ability to help folks there, and 2. makes you lonelier

Show threadWolf480plFeb 10
@hailey
Another drawback of this approach I've noticed is if I read the source code of a dependency before using it, I might see sth I don't like, and then I won't want to use that dependency anymore, even though in practice it would've been fine.
Show threadCliff'sEsportCornerFeb 10
@hailey so does anyone get exposed to basic logic gates and circuits anymore? Eons ago had years of electronics in HS, well 1st year was house wiring (literally wiring a house was goal) but after that was electronics breadboarding and electronics repair.
Show threadTedFeb 10

@hailey
This is one of my biggest hurdles with LLMs. It's a machine that can't be understood.

I have a predictable mental model of what I work on and the tools that I use. And if I find that it doesn't hold, I've found a bug that I can fix or need to update my mental model. I can build a model by playing with inputs and observing their effect of outputs.

Meanwhile I can't predict an LLM, as the output changes with no input change. There's no way to model what effect a input has.

Show threadsarnaFeb 10
@hailey good post. @ the AWS bit - I think these systems are intentionally complex and hard to understand. *you're not supposed to understand them*. and that's a feature. these companies want buy-in, and they want people doing stuff via trial and error, so that the cloud bills are fatter and they can sell you consultants. I feel the AI coding thing is an extension of that :/