JeremyNT

Cost per token is going to tank; it's a race to the bottom. The frontier model providers have no moat between each other and eventually even the open weight models will be "good enough" for almost every purpose.

Right now the market is expecting something a lot more than "sells a lot of tokens" out of these things to justify their valuation.

> Intel reports PC sales are down. That's likely related to RAM costs, which made price/performance of new PCs drop.

How is this not related to "AI cost fears"?

The amount of cash being dumped into AI datacenters has driven up the cost of a lot of other things in adjacent industries.

The binance link still works for me? Maybe it's region blocked.

If this isn't a damning indictment of Cursor, I don't know what is. Perhaps it will be the death knell.

What value is this VS fork actually adding given today's advanced harnesses? Surely whatever benefits it offers are not worth the risks going forward.

I use Fastmail which has been a very boring and high quality product. The company is not American, it is Australian, but any country with close ties to the US probably offers little in the way of surveillance avoidance.

It's best to think of email as fully public information anyway, use Signal for anything you care about.

I reproduced it about 10 hours ago, but it seems they've "fixed" the issue so it now generates the correct results.

You could imagine a sane policy regime to unwind the "look the other way" approach to illegal immigration and the under-the-table low wage labor market that it created.

Instead we get... well, whatever the fuck this gestapo shit is.

I guess regulatory reform doesn't really play with the kind of people who will gleefully vote for mass deportation on the grounds that they just want immigrants gone.

PXE boot enabled and attempted by default on an internet-facing network interface is certainly a... choice.

This is something that shouldn't be possible in the real world because any sysadmin should know not to do this, but guaranteed there are at least some small time operations with a couple of colo'd servers that only have single interfaces serving public traffic, and the intern who set them up didn't know what a boot order was or what "PXE" even did.

> Realistically there were only ever two possible outcomes: you do more work for the same amount of money, or you’re one of the ones who were laid off and your work was transferred to the former.

Workers need an answer to this problem and we shouldn't wait until all the deleterious effects stack up to make the demand.

A 32 hour work week would be a really good start.

We need to remember that there is a mechanism to address this kind of corruption, which is impeachment and removal, and by failing to even consider it Congress has endorsed the behavior as well.

If there is ever any reconciliation they also need to be held accountable.

I don't! I think it was a great idea. But I think people commonly misunderstand the way the program operates, assuming Parton herself was funding it, and I'm glad somebody highlighted its structure.

Really glad to see this as the top comment.

Parton did a lot of very nice things and was a great contributor to society in good and decent ways that many of her wealth peers totally fail.

But there has since her passing been an element of myth and folklore that has elevated those accomplishments beyond reality.

Her library is a great idea and her support was notable, but the funding in large part comes from government grants and other donors.

To a large degree, the library program is also your tax dollars at work.

You're right but the trick LLM-backed tools have is that they can just write traditional code on the fly.

There are usability and performance problems right now which is why we have this interregnum. But as these systems evolve, the tools can offload "expensive" operations to traditional compute by writing the glue code on demand, and persist the code for future use (so regenerating it won't be required again later) when the user confirms it's doing what it's supposed to do. You can A/B test it in real time and ask the user to pick the "better" result.

Same is true for e.g. GUI elements, widgets, etc. Write it the first time, have your user verify it, persist it, retrieve it.

And maybe you have a couple of people in your org who work through the process of generating a bespoke app that processes data with a GUI optimized for whatever they're doing, but it's all mediated directly through your vendor. The users run through the process all in the web UI of whatever LLM vendor the org's data is silo'd with, the output is a mix of traditional computation and LLM contexts that anybody can maintain via asking the tool to change itself.

There's a huge cohort of people who are convinced they're somehow immune to right wing propaganda and continue posting information exclusively to sites owned by reactionary oligarchs. Sometimes that information is useful or important, which is why nitter and its ilk are sorely needed.