Egregoros · Phoenix Framework

Post

Remote status

Context

wigbert

I dont actually know for sure if having an LLM written readme file is a good way to tell if someone is a bad programmer, but my gut tells me to immediately put that person in a hole and now they really have to prove to me they aren't retarded

johnny peligro

@mischievoustomato@0.5dollah.click remote

isnt writing a readme one of the most LLM-worthy tasks? it can read the entire codebase and write it...

wigbert

@RustyCrab@clubcyberia.co remote

@mischievoustomato bad programmer spotted

In all seriousness LLMs should NEVER write english for you. The entire point of written language is to take a slice of your brain as best as you can represent it and put it in paper. Every word choice is meaningful. It also shows YOU understand the code personally.

You can have a bot fact check your readme and integrate that feedback thats perfectly fine.

johnny peligro

@mischievoustomato@0.5dollah.click remote

bad programmer spotted

im not good, and not bad. I blend in and listen, and dont trust anyone.

Eh... I don't know. I guess README == writer-knows-how-the-stuff-really-is but you can't really assume that. Things can be hidden for XYZ reason, forgotten and more. Yeah sure integrate feedback and whatever but I don't know. I'd rather focus on programming myself and leave the README stuff for the LLM, it'll do a nicer job at it than I could, and if something's amiss, it can be corrected. Or maybe not because again, human imperfection.

this makes me wonder, how would you view a project where the code was vibecoded, but the readme was hand-written?

For my project, I like that it can put a lot of detail into the readme about things I don't know (http setup details for example), and minute things about implementation and whatnot.