← Back to context

Comment by dnautics

1 day ago

> is there any advantages of using untyped programming language

without any evidence, i claim the corpus might have higher quality variable names and conventions that are "human crutches" around not having types.

LLM knowledge in your non public codebase must be strictly local, and so checking on details and identities of types incurs a cost for the LLM to go fetch that info. if the LLM can "just know" (guess with very high confidence) then thats better for the LLM.

> non-typed languages has more traning data

as per anthropic "poisoning llms with 250 examples" finding, i suspect that corpus size does not really matter that much for any language that is reasonably well used.

Rather than having the LLM and human devs all guessing from verbose variable names, can't they both use a language server that observes the code and can surface that kind of structure info cheaply?

Part of the point of types is enforcing more of the contract at various code boundaries (function, module, etc), and that enforcement is specifically so that you don't have to keep the whole codebase in your head / context window in order to be able to work on it.