Hacker Newsnew | past | comments | ask | show | jobs | submit | jdiff's commentslogin

A merge to main itself is pretty substantial, especially a week after saying, "[This] code that does not work. We haven’t committed to rewriting. There’s a very high chance all this code gets thrown out completely."

I think any instruction tuned model is going to "know" it's an LLM.

The same crowd who always vote against their own interests. To paraphrase, "give them somebody to look down on, and they'll empty their pockets for you."

It's going to get a lot harder to convince people that $outgroup took your job when $outgroup are all obviously unemployed too.

Have you used Jujutsu before? It's git-backed and it sounds like it incorporates a lot of these niceties from Mercurial. I find it an awful lot more intuitive than Git to use and the stable identifiers are absolutely lovely to have.

Of course I use jj every day. In fact it improves on mercurial by eliminating the need to run `hg evolve` because it just auto-evolves for you.

I've never had much contact with Mercurial myself, so are there any features from Mercurial that JJ doesn't already incorporate? Or any differences you find interesting?

Memory bandwidth is completely different on any browser from measured results on my M2 Pro machine. Weirdly, the estimated performance levels and even exact product name differ between Chrome and Firefox. Firefox calls it an M2 Pro and overshoots measured memory bandwidth by 40GB/S, Chrome calls it an "Apple M2 Pro" and overshoots by 80.

I don't think this has been the case for at least 39 days. The news is slowing down. The big headlines now, besides unverified marketing claims, are efficiency gains. Which are fantastic, but don't seem to be met with matching performance gains.

A single million line commit overwriting all of main is pretty flashy and high profile. Anyone who was working in this repo previously can throw whatever the heck they were doing out the window. This isn't in a branch, this commit isn't marked WIP, this is now the committed, active, current state of the bun project. That's quite a statement to make, even if there isn't a blog post somewhere (yet).

If your assertion of falsehood were true, the current top story on HN wouldn't be Turso shutting down their bug bounty due to overwhelming slop.

This is not good reasoning. You're offloading your thinking to "Turso" for some reason.

You're also assuming that they haven't made the alternative judgement that instead of triaging the haystack of slop that they get in order to potentially pay out to someone, they should instead be spending that cash and effort on tokens to find bugs in their own codebase.


You should read the mentioned article. They have hired some of the people they paid out to, and some of those people were LLM-assisted.

The claim I'm rebutting is "in the past few months nearly every LLM generated report is real." If that were true, there would be no need to close the bounty. The bounty is to address approaches that they themselves may not have considered, so would still hold value if the claim held true, as outside individuals may still hold unique LLM-assisted approaches and perspectives.


Everyone is going to see different results. Overall the general trend is AI is getting better. Although that might be partially people are shutting down their bounty programs which gives the incentive to generate slop.

> Looking at the email headers it has a bunch of Exchange-specific headers so it's probably actually from Outlook for Mac.

My guess would be some sort of internal forwarding mailing list, yeah.


the lighting effects are very brogue and like nothing I've seen in angband, which is very very barebones ASCII by comparison. brogue-likes push into ANSI art territory with their abuse of terminal formatting.

Some of the many variants did expand on the ASCII graphics a bit, but yeah, I see what you mean.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: