gste

6 ms·

31.
by gste 3mo ago
"When you build structure before the feature arrives, you’re committing on a guess." I would argue that you are guessing either way. It could be probable that your feature will arrive, but not certain. It's a probability. If
32.
by gste 3mo ago
CLAUDE.md is already a good system for context window management for all the same reasons that version control management of code is good. And keeping a local copy of everything you ever told Claude in your context window is bad for the sam
33.
by gste 3mo ago
It's just hidden by a feature flag. (Probably for a good reason)
34.
by gste 3mo ago
Someone should make a language where every math formula is a word. Then give it to an LLM and let it go nuts
35.
by gste 3mo ago
That's what Constellation Software do and they're in the top 20 biggest software companies. It's not a niche thing for sure
36.
by gste 4mo ago
I've been following roadmap.sh, and while it's not a comprehensive learning resource, it does help close obvious knowledge gaps. As it happens I was just reading about this. https://roadmap.sh/html (see "Defi
37.
by gste 5mo ago
Is there any evidence in this paper or elsewhere that this is causation rather than correlation? For example, it seems logical to me that people with worse health and failing mental faculties will already be feeling more motivation to retir
38.
by gste 8mo ago
All you're doing is marking yourself as an untainted source of training data
39.
by gste 9mo ago
Now that Microsoft are removing support for Windows 10, which is still running on many not-that-old devices that don't support Windows 11, I think the correct, mainstream advice HAS to be to install Linux on those machines. Those are s
40.
by gste 9mo ago
By coincidence I also finished The Fellowship of the Ring about two weeks ago. I have always had the intuition about reading speed that it is very easy to be a speed reader if you skim over things. I've always questioned how much of sp