jnovek

6 ms·

61.
by jnovek 4mo ago
> Why would a company do any of these things? What? It’s actually insane that they haven’t yet. I don’t like changing tools. What engineer does? I want to learn one tool and tune it to my exact preferences. Proprietary vendor tools are
62.
by jnovek 4mo ago
Hallucination rates go down by a few % with each new model generation, although some milestones have seen regression e.g. reasoning models are just overall worse. It is a bit hard to measure and compare across generations because the tests
63.
by jnovek 4mo ago
They don’t need to be eliminated or “fixed”, they just need to be less frequent than human error.
64.
by jnovek 4mo ago
If you set temperature to 0.0 you almost have a key-value store, but finding the right key for your value might take some effort.
65.
by jnovek 4mo ago
LLMs will absolutely be like that. The speed this technology is moving at makes me certain, especially over a period of 10-20 years; 20 years ago I was bugging friends for a GMail invite and AI was a joke left to academics. I think it will
66.
by jnovek 4mo ago
If we’re talking about credibility, unfounded accusations of lying tend to harm the credibility of the accuser more than the accused.
67.
by jnovek 4mo ago
Your original comment was about pricing ethics, does Anthropic’s connection to the DoD have anything to do with pricing ethics? They’re in no way coupled, one can be ethical while the other is not.
68.
by jnovek 5mo ago
> Every dumbass America bad trope is here. Disregarding wealth inequality is the most trope-ey American thing you could possibly do.
69.
by jnovek 5mo ago
There are entire categories of ads that operate on insecurity, they just don’t come out and scream “this is because you’re insecure” as that would make for bad copy. E.g. you think adult diapers advertise on anything other than insecurity
70.
by jnovek 5mo ago
This might be survivorship bias, the people who read and enjoy a difficult book for the sake of it probably don’t need social media validation to do so.
71.
by jnovek 5mo ago
I think so, I think you’re getting downvoted because of how you’ve framed it. People don’t like it when you tell them that you think they’ve made a stupid purchase. :-P I do think more expensive models are valuable in some cases. For exampl
72.
by jnovek 5mo ago
Having a lead in a new market even for a short time creates initial conditions that are in your favor. That’s why startups get all fired up over being first to market.
73.
by jnovek 5mo ago
Correct output is table stakes. Your test only shows that the products work as advertised, it doesn’t reveal reasons why people prefer one to the other. You’re guessing that it’s a result of advertising, and I agree that that’s probably
74.
by jnovek 5mo ago
> A colleague was convinced Claude is better That’s actually what my comment was based on; raw code output isn’t the only measure of quality. Engineers write better code if they have the tools they prefer.
75.
by jnovek 5mo ago
Opus 4.5 was definitely stronger than DeepSeek V4 for me, specifically with large context. I’m being pedantic/splitting hairs, though. I’ve obviously switched to DeepSeek full-time because it makes more sense to me pragmatically — I sp
76.
by jnovek 5mo ago
Im pretty sure this is a category error. The correct comparison would be: - A person writing code vs a bot writing code - A person designing systems vs a bot designing systems The comparison in the article is between a bot writing
77.
by jnovek 5mo ago
I’ve been using DeepSeek V4 in OpenCode exclusively for about a month. I think it’s great, but coming from Claude Code it did feel like going back in time by ~6 months in model capabilities. This isn’t a big deal to me for what I do, but th
78.
by jnovek 5mo ago
I can’t tell the difference between code written in vim or vs code but it matters substantially to the person writing the code. There’s stuff beyond just the output that goes into tool choice.
79.
by jnovek 5mo ago
Seeing yourself as immune to propaganda probably makes you more susceptible to propaganda. Edit: Oh they’re trolling, nm. :-/
80.
by jnovek 5mo ago
What web stuff are you guys working on that 16GB didn’t cut it a few years ago? I’m not questioning your statement, it’s just completely different world from my day-to-day and I’m curious.
81.
by jnovek 5mo ago
You mean like bending SNAP data to somehow fit your picture of what poor people are like?
82.
by jnovek 5mo ago
You, personally, haven’t been poor, though, so you you don’t know what it’s like to have to balance all the things a poor person does as an adult . Your confidence that you can speak on this topic because your wife was poor as a child come
83.
by jnovek 5mo ago
I prefer the results of studies to the anecdotes of some people you know.
84.
by jnovek 5mo ago
I think it’s a great thing to be legislating on. I would like to see it on the federal level, myself.
85.
by jnovek 5mo ago
Perhaps they should just write games in a way where they can actually release server code when they it shuts down?
86.
by jnovek 5mo ago
What will the negative impact of this law be, exactly? Hurting live service games which are already cancer?
87.
by jnovek 5mo ago
Don’t a bunch of shooters also follow this model?
88.
by jnovek 5mo ago
Release a spec. Release a binary distribution. I’m sure they could find a way to make it happen if it was in the studios’ interest.
89.
by jnovek 5mo ago
There is so much context here that you’re missing — have you ever been poor before?
90.
by jnovek 5mo ago
Is this supposed to be an analogy?

More ›