Viewing profile — furyofantares

yegge has been famously writing ultra long blog posts for decades now

LLMs definitely write worse code for LLM consumption than humans can. In my experience your claim can't be further from the truth. I can get much further with an LLM starting from …

Looks like gpt-image-1 to me, from the piss filter to some of the mistakes. I mean, image-2 or nano banana also still would have been dogshit. But I find it curious that image-1 wa…

You're right, for example photography was controversial.

> there is no AI publication that I am aware of that is good I think you're right that there isn't any that's any good right now, but that's beside the point that if that changes (…

Small games. One thing is I have it doing a LOT of visual verification of its results, which uses a lot of tokens. I also tend to throw some huge tasks it can't do yet and watch it…

I'm having Claude/Codex continue each others' sessions pretty frequently without issue. I have max plans on both and still hit limits, whenever I do I open the other and say "resum…

Try a little NE of Seattle.

There's definitely exceptions! People can make personal choices of course. Plus if you're early to something you can capture the gains before business and markets adapt.

Some of it is fun and new, I'm spending a lot of time just making stuff I wanna make. But regarding work produced for the marketplace, I think it is naive to think any productivity…

Haha, I misread as "word games" and was very confused about the claim.

They also would have produced a team of engineers that knows the Rust codebase.

> I have used LLMs to create stuff like word clones just for fun. It was a disaster. Sure, it had the basic functionality. But the moment you started with page structure (harder th…

If you pay anything for a service you expect a refund on the order of hundreds of hours of credit (168 in a week) for every hour of downtime?

I'm shocked to be enjoying this read as much as I am, given how filled it is with LLM-tics, repeated information, and excessive useless verbiage. I sure wish they had posted whatve…

I'm inclined toward your same viewpoint here, except OpenAI's announcement of this came with a blog post that gave off a real bragging/marketing vibe. To me it's clear that even if…

If I'm using medium or low reasoning, I use Sonnet 5. If high or above, I use Opus 4.8. (Before 5, I was never using Sonnet. This is a Sonnet 5 vs Opus 4.8 comparison.) Sonnet 5 an…

OK, I guess my confusion was that the effort to ban open weights models is what I would categorize as "groveling at the white house to try to stop their competition".

I'm very confused by this comment, I don't know who you're referring to.

I've seen plenty of fucked-up guardrails that vehicles have passed through. We know that with enough momentum they will be penetrated. It seems like exactly the correct terminology…

Is turning the alignment features off more like multiplying "don't be evil" by 0, or by -1?