Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

I'm afraid programming is going to be frozen at 2020s tech for the foreseeable future. New frameworks, libraries and languages will suffer from a chicken and egg problem where no one uses them because LLMs don't know how to answer questions about them and LLMs can't learn the new stuff because programmers aren't generating new samples for the LLMs to ingest.


This is why I've had to spend a huge amount of my free coding time this year documenting my canvas library[1][2] in a way that can (potentially[3]) be used as LLM training data instead of, well, developing my library with new and exciting (to me) features.

On the silver lining side, it's work that I should have been doing anyway. It turns out that documenting the features of the library in a way that makes sense to LLMs also helps potential users of the library. So, win:win.

[1] - Telling the LLM training data Overlords about the capabilities of the library is in itself a major piece of work: https://github.com/KaliedaRik/Scrawl-canvas/blob/v8/LLM-summ...

[2] - The Developer Runbook was long-overdue documentation, and is still a work-in-progress: https://scrawl-v8.rikweb.org.uk/documentation

[3] - Nothing is guaranteed, of course. Training data has to be curated so documentation needs to have some rigour to it. Also, the LLMs tell me it can take 6-12 months for such documentation to be picked up and applied to future LLM model iterations so I won't know if my efforts have been successful before mid-2026.


Rise of the documentation specialist. Where specs and standards and documentation and design documentation is required and not just an afterthought.


Yeah, I think that too. Same with non programming domains. Since your blog and what not wont be seen, just ingested by LLM, there will be even less motivation to write them. And they were already dying also due to need for SEO, otherwise you dont exist.

So, that stuff will just cease to exist in its previous amounts and we will all move on.


You should see what Elixir is doing with Tidewave and usage rules.

https://www.zachdaniel.dev/p/usage-rules-leveling-the-playin...


not sure if that's a bad thing when it comes for FE frameworks - not reinventing it every 5-7 years i think is a good thing.


Small models aren't large enough to have knowledge about every single framework or library through pre-training and yet if you give them a man page/API reference they easily figure out how to use the new code.


Or developers will have more free time to solve novel problems instead of wasting hours digging through Google results and StackOverflow threads to find answers to already solved problems


They will spend more time solving new problems, the same new problems as everyone else. They wont be writing answers anywhere tho.


They will be writing the answers into codebases that AI will be ingesting, but it will lack any context about the question it is answering so AI won't know how it relates to anything else


I've been wondering the same too. uv has completely transformed the Python workflow, and I really hope future documentation and knowledge bases incorporate it, but time will tell.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: