Note: An LLM was not used in writing this microblog entry.
This entry is part of a microblog series called LLMs
I’ve had what I’d describe as “hole in one” experience where I generated two Haskell packages, sse-conduit and llm-conduit, from a single Claude prompt, which both compiled and worked the first time when pointing it at llama.cpp’s server. This gave me pause as in the past I would enjoyed sitting down to write these and publishing them. Given that the whole thing took about 10 minutes, I didn’t do either.
I call this a funny name like “hole in one” because it’s not a normal experience. I had a few weeks prior implemented the same thing manually with Emacs Lisp, so I wasn’t entirely robbed of the experience. I believe I’m arriving at a new standpoint which is that one may “vibe” projects, but only those things they could do themselves readily. The exercise of writing is often as valuable as the written thing.
The more things that you vibe that you do not understand, the more your knowledge will stand still and that muscle will not be exercised, and atrophy. By contrast, if you only vibe things that you already understand fully, but do the hard work on everything else, you’ll continue growing, but will also get the accelerative edge of LLM vibing.