kajman

I've found LLMs useful for surfacing popular recommendations. I also get the overwhelming feeling that it's all very early days still when the machine mixes together whatever was crawled into the weights with a web search or two and dumps it into a markdown blurb.

I totally agree with the above that a more polished and less obvious use of LLMs integrated back into search engines may be more useful, but will definitely be more usable.

You may (or may not) find Paul McKenney's free book on parallel programming interesting - it has an unusual tone for a Computer Science textbook that made it less dry for me. There's a substantial section covering lock free structures.

https://mirrors.edge.kernel.org/pub/linux/kernel/people/paul...

The Don is too busy selling exclusive TLDs to private companies that won't use them to care what the enforcers are up to.

I always figured telegram got the screws turned on them all the time because their lack of E2E encryption meant it was viable to demand they proactively police the platform in the first place. Maybe Signal would just be outright blocked in these locales if it was anywhere near as popular, though.

I assumed they just wanted to cultivate FOMO to sell an even more expensive version to researchers later on.

Especially if you are looking for where the fun is, the idiosyncrasies of your own engine gives you a world with it's own flavour if you incorporate that flavour into your design process you could create a feedback loop that turns into something special.

I loved the old STALKER games, and the wackiness of their engines was a lot of the charm. I ended up buying the new one out of nostalgic dedication and it's probably the worst example of "Unreal slop" I've experienced, having not bought many newer games. I'm sure the butchers running Xbox have run the numbers and think they'll make even more money throwing armies of contractors with allegedly fungible skills at the next Doom games, but I'll leave others to bankroll that while I enjoy games I don't need frame generation for.

Counter question: where's the useful production-grade software that's been made entirely with "agent swarms"?

Personally, I don't think useful business problems can be solved entirely with LLM loops. If you choose not to pay others to handle part of that then it's all on you.

"I am eighteen years old, have a good set of passkeys, and believe in Sam Altman, the star-spangled banner, and the fourth of July. I have taken up a BLM lot, cleared up eighteen acres last year, and placed top of it a bitcoin mine. My vibe coded drop-shipping startup looks first-rate, and the conversion rate and total addressable market are bully.

I would be very surprised to see any "make it safe" solution that does not render these models useless for programming. I assume posts like this are mostly clout-motivated, but I really dislike the implied vision for what LLMs should offer here.

The U.S. is becoming like the place that keeps releasing the best open weight models?

I'd be really happy to come across this in a project I were interested in. So much hobby OSS is infested with slop that I don't even want to skim the code if I pick up a hint that there's no humans at the wheel.

Stars are a really bad metric for quality/popularity but they're something here. I think some sort of impact score when combined with repo age, # contributors, etc would be pretty valuable.

I worry the sort of core classes that are in massive lecture halls (and everyone remembers hating) would be more of a "blind leading the blind" situation if run like this. Maybe not with lavish funding and small classes. I had a couple classes like you describe, but they were on specific topics I was very interested in near the end of my undergrad program.

I'm sure investors thought one or two of the ISPs laying all that fiber would be collecting fat rents on them until the sun burnt out. I'm glad they got so much in the ground before there was a reckoning. I hope this industry ships more very expensive models, ASAP.

You aren't ready to get that job until you've truly accepted your role as a genderless machine that turns salary into personal validation for the hiring manager.

I never thought I'd defend Electron, but I'd rather use the bloated web UI than a vibe coded Qt/GTK version I'm positive will not have seen any human QA.

zero signal: resume & cover letter. applicants will mass-apply with ai-tuned resumes that happen to perfectly match our listing

How often do you interview candidates with bad or mediocre resumes, though? My impression is that a perfectly tuned (and almost certainly dishonest) resume is just table stakes now.

That is interesting. Half a year is not nothing and I expect it's harder to keep a project functioning when the base is vibe coded rather than having mature abstractions and architecture already.

I am still skeptical on this method's ability to deliver polished products though. I've kept an eye out on it in the OSS world and don't think I've seen big anything yet.

I'm just a petty conspiracy theorist, but my assumption is their use of Persona was part of "the deal" they've been encouraged to make, and what you're seeing is a company being brought to heel.

I'm always curious when I see these stories. How long have you been doing this, for what sort of work, and was the codebase mature before you began working like this?

I think email is a pretty poor example of decentralization in the modern internet. I pay a company to not have to worry about hosting my own and I've still had emails blackholed by the massive providers for unknown reasons I can only assume are some combination of the custom domain* and the provider. I went back to using Gmail on my resume and for applications after nearly missing out on an interview because of it.

*: Not an .xyz, has proper SPF and DKIM records

I haven't used Mastodon, but my experience with Lemmy was a lot of petty fighting about who should be defederated. The OP's reference to warring fiefdoms was very spot-on for that scene. I imagine it's a huge turn off to casual, well-adjusted people exploring the space. But I wouldn't know.

There's, of course, another sort of grossness to corporate moderation from above, but at least with ATProto you can take your identity and content to another AppView, if it wasn't shown there already. AFAIK, any fediverse migration tooling requires a cooperative host server you haven't already been banned from.

So, anyone who wants to index it gets a copy and you have no control over what they do.

How does this differ from scrapers hitting the blog directly?

Is the idea that we're just one OTA update from them turning into bombs? Considering the quality of software in the auto industry, I would be about as worried about any domestically assembled EV.

I'm cautiously optimistic that anti-conpetitive action against hardware will fail. There's a lot of money willing to fight for cheaper inference. The same can't be said for providing consumers with cheaper cars.

I can't say I'm as optimistic about there continuing to be an open market for foreign LLMs.

I'm not excited to see that the smart glasses are apparently competing on their number of conspicuous cameras now. I used to be really excited about wearable displays, but the current regime of mass surveillance and its consequences is way too vile. I hope the negative sentiment from Meta's glasses keeps rolling, because it might be one of few places where consumer distaste can actually harm mass surveillance.

I really wanted to like Gleam because the Rust-like syntax and static types looked familiar and robust, but seeing how Lustre does HTML templating versus how Phoenix does Heex was my deciding factor to try the latter. My understanding is this is because of a current lack of any macro system.

I probably didn't give it a fair shake, but it all looks very early days, which isn't for me. Kind of a shame, because the dynamic types are still kind of driving me up the wall in Elixir. Elixir structs and dialyzer type specs feel so much worse than what statically typed languages provide.

It sounds like these people would otherwise be laid off if they were not parked in this department instead. I imagine they are quite unhappy, but getting paid 200k+ to do tasks that are usually gig work is hardly the most heart wrenching story in this tech market.

I am indeed sworn to not reveal lots of data I knocked on doors for. My memory isn't that good, especially compared to the database it went into, anyway.

I hope it's not a baseline for individual records, but my assumption was that the census data would be pretty useful as a baseline for aggregate information, especially when it comes to comparing to private sets they're working with.

There's not many cases of enforcement. Non-response is taken about as seriously as the Robinson–Patman act. I think the Census Bureau is very reliant on people thinking there will be enforcement, however, which is why the materials they send all have a threatening aura. I don't know about the ACS, but for the decennial census I often felt like my job as an enumerator was just to bother people until they'd answer. The case would keep being recycled until we got at least (IIRC) a head count.