mh-

5 ms·

61.
by mh- 3mo ago
It might! In my prompt, I'm supposing you have some priors here about what's possible. Try saying something like "materially faster". But you need to also make sure it knows how to measure that endpoint's performanc
62.
by mh- 3mo ago
It's just a way of defining what "done" means, and passing off evaluation of whether "it's done" is true to a separate context and/or model. If false, it prompts the original context accordingly.
63.
by mh- 3mo ago
/goal open a draft PR that makes <endpoint> ~200ms faster while still passing tests, and adhering to nearby style+pattern It's obviously not perfect, but it's not like you're going to YOLO it into prod with
64.
by mh- 3mo ago
> In the US today, about 15% of adults have tried GLP-1s, with 7% currently taking them, according to research based on Bain’s proprietary GLP-1 adoption model. Is the 12% from another source?
65.
by mh- 3mo ago
I didn't realize that re-timestamped comments . That's a pretty unexpected behavior, if true. I wouldn't like if that happened to a comment I left - a lot can happen in a week.
66.
by mh- 3mo ago
The concern usually only comes when you fall on the "outside" of the prevailing opinion within the group of people voting on comments. edit: my response isn't about lobste.rs. I don't have an account there.
67.
by mh- 3mo ago
They're also shuttering their (US) community site in 30 days. Q: When will the communityus.oneplus.com close? A: The communityus.oneplus.com will close on 11:59PM ET,August 16, 2026. https://www.oneplus.com/us&
68.
by mh- 3mo ago
This isn't just measuring the speed of tokens that are written to a file (code). Your human partner is thinking a lot faster than 5 tok/s.
69.
by mh- 3mo ago
> Then on execute it could even redraw the terminal to pretend you pasted X all along. Diabolical idea. Anyone know if there's been anything in the wild that did this?
70.
by mh- 3mo ago
I would expect that it does, as well as some of the other directives I've seen in this thread ("never repeat the question"). It's one thing to tell it to do that in outputs , but I wouldn't at all be surprised to f
71.
by mh- 3mo ago
I have, but it's definitely very rare. And might have been manual intervention as described above.
72.
by mh- 3mo ago
I agree with that, the ones I see flagged are frequently espousing viewpoints I disagree with. But they should be debated then, and flagging them denies everyone that opportunity.
73.
by mh- 3mo ago
lol, fair. It wasn't a great example. Have a good night.
74.
by mh- 3mo ago
In the meantime, here's one I saw recently: https://news.ycombinator.com/item?id=48660796
75.
by mh- 3mo ago
I always vouch them but have never bothered to email. I figure the mods are seeing the same HN I am. But maybe I spend too much time here some evenings.. But yes, it's the "going against the zeitgeist" ones I have in mind. It
76.
by mh- 3mo ago
Happy to find and email you some when I'm at my desk. (I'm well aware of the genAI policy, and very supportive of it.)
77.
by mh- 3mo ago
The anti-AI crowd on Steam only really seems to care about it being used for art, from what I've seen in reviews.
78.
by mh- 3mo ago
I believe we're in a fairly limited window of time right now where that's even detectable via tools like Pangram. For small models it'll, maybe not the case. But for anything worth paying for.
79.
by mh- 3mo ago
It's just a collection of comments pushing back on LLM-generated prose. 'dang sends it in replies to emails that ask him why their comments were flagged when that's the reason. He's offering examples of the broad communi
80.
by mh- 3mo ago
(Apologies if you've commented on the following already, but I've not seen it.) The most concerning growing trend I see is comments that don't violate any guidelines, but are flagged to dead. I'm not talking about poster
81.
by mh- 3mo ago
Only the fixed portion is. So about $20 of my $400-800 monthly electric bill is income-based.
82.
by mh- 3mo ago
Indeed. That's why I think it depends on the individual's workflow where it should live.
83.
by mh- 3mo ago
Ah that makes sense, wasn't trying to be pedantic. Thanks for clarifying.
84.
by mh- 3mo ago
A smarter approach (progressive disclosure) for tools has been implemented by (I presume all) the harnesses over recent months, but you're 100% right in any case. I enable tools specific to each project only in that project , and have
85.
by mh- 3mo ago
Maybe, depends on their workflow. In my human workflow, I tend to use commits as checkpoints and then squash before pushing. I'd usually only run time-consuming tests before squash+push. But yes, anything you want to ensure really ne
86.
by mh- 3mo ago
This is interesting, because if I start a fresh session of Claude Code right now and run /context, I see the following: Opus 4.8 (1M context) claude-opus-4-8[1m] 23k/1m tokens (2%) Estimated usage by category Sy
87.
by mh- 3mo ago
> [..] my rules do say: verify all changes with tests I am a bit surprised that you're disappointed that it does exactly what you told it to - people usually have the opposite complaint. If you're using it interactively and w
88.
by mh- 3mo ago
I'm unclear who you're arguing with. I don't fly GA and I live under the approach to a GA airport, which is why I'm well-read on this subject.
89.
by mh- 3mo ago
Correct. For others reading this though: virtually all piston-engine GA aircraft in the US today are still burning 100LL (leaded), and there are nearly 200,000 of them actively flying. There is a timeline to transition to UL, but very low c
90.
by mh- 3mo ago
The vast, vast majority of indie game devs will not see $1m in revenue in multiple lifetimes.

More ›