Wednesday, August 19
Daily News Stuff 19 August 2026
Triple-Baked Baklava Edition
Triple-Baked Baklava Edition
Top Story
- A federal judge has ruled that a state judge who used AI to handle the entire decision in a case so she could spend the day playing Cookie Run Kingdom or something equally important cannot be sued. (Tom's Hardware)
The state judge's actions may not be reasonable, or even legal, and the decision can be appealed, but the judge cannot sued.
Tech News
- Move fast and crash planes: Google just bought the anonymised databases of the defunct Spirit Airlines for $10 million to train AI models. (CNN)
With OpenAI training on Reddit, Amazon training on Twitch streams, and now Google training on incompetent and defunct airlines, I don't think we have to worry too much about AI taking our jobs.
And more about AI destroying the economy so there aren't any jobs to be taken.
- Building a Steam Machine in an ASRock Deskmeet X300. (WCCFTech)
Interesting thing: The Deskmeet X300 has four DDR4 SODIMM slots. Which is great if you just happen to have a stockpile of DDR4 SODIMMs from old laptops and mini-PCs.
Which I do.
- Best Buy apparently did not steal half this customer's memory. (Tom's Hardware)
They took half the memory out, yes, but returned it along with the laptop. And since it has a dedicated GPU, performance won't suffer too much.
Actually, it just sounds like one stick of RAM went bad, and since the laptop was upgraded by a third-party, it wasn't Best Buy's responsibility, and the system is now working again.
Musical Interlude
Disclaimer: She's a bop girl.
Posted by: Pixy Misa at
06:43 PM
| Comments (1)
| Add Comment
| Trackbacks (Suck)
Post contains 265 words, total size 3 kb.
1
The fundamental thing about the managerial consensus on AI shenanigans, is that almost all managers went to university.
The history that universities teach non-engineers has a lot about engineering's success, and less about the many failures. Engineers get more exposure to such things as technical failures, and technical successes that were economic failures.
But this does not mean that someone with a masters in real engineering has the full toolkit for connecting possible machines designs to how they should understand the economic context. This gap is what, among other things, the LLM enthusiasm happened in.
You would basically choose to do a dissertation on this in engineering if you were too stupid to know what side your bread was buttered on, or were really allergic to ever making money. Engineering PhDs make small amounts of money with showing technical excellence. They make large amounts of money by technical excellence, and a bunch of soft skills, and the ability to do actual engineering management. They make no money when they go out of their way to tell people that their technical skills are poor.
The government pays academics to say what the government wants to hear. What the government wants to hear is that it can buy correct answers from academia.
This question of knowing the economic value of a machine first is pretty much a trap for everyone. You design the machine, then you figure out how to manufacture it, then finally you test it on the market. In that order, they are really valid sources of truth. In other orders, they are invalid. People understand this (sometimes) in the business world, it is less understood in academia.
The Austrian economists figured this out decades ago, and many American businessmen understood that the Austrians were correct.
Semiconductors are a very distorted market, and makes central planning almost look viable. The tech industry may thus average delusional.
Even if businessmen in these enterprises were selecting 'good datasets', and even if their incentives lined up to being accurate about the technology, and about the economic ramifications, they don't have the tools for truth. They might have to reinvent, or rediscover those tools, and do not have incentive to explore or to test.
The history that universities teach non-engineers has a lot about engineering's success, and less about the many failures. Engineers get more exposure to such things as technical failures, and technical successes that were economic failures.
But this does not mean that someone with a masters in real engineering has the full toolkit for connecting possible machines designs to how they should understand the economic context. This gap is what, among other things, the LLM enthusiasm happened in.
You would basically choose to do a dissertation on this in engineering if you were too stupid to know what side your bread was buttered on, or were really allergic to ever making money. Engineering PhDs make small amounts of money with showing technical excellence. They make large amounts of money by technical excellence, and a bunch of soft skills, and the ability to do actual engineering management. They make no money when they go out of their way to tell people that their technical skills are poor.
The government pays academics to say what the government wants to hear. What the government wants to hear is that it can buy correct answers from academia.
This question of knowing the economic value of a machine first is pretty much a trap for everyone. You design the machine, then you figure out how to manufacture it, then finally you test it on the market. In that order, they are really valid sources of truth. In other orders, they are invalid. People understand this (sometimes) in the business world, it is less understood in academia.
The Austrian economists figured this out decades ago, and many American businessmen understood that the Austrians were correct.
Semiconductors are a very distorted market, and makes central planning almost look viable. The tech industry may thus average delusional.
Even if businessmen in these enterprises were selecting 'good datasets', and even if their incentives lined up to being accurate about the technology, and about the economic ramifications, they don't have the tools for truth. They might have to reinvent, or rediscover those tools, and do not have incentive to explore or to test.
Posted by: PatBuckman at Thursday, August 20 2026 03:47 AM (6eVrQ)
53kb generated in CPU 0.0269, elapsed 0.1477 seconds.
58 queries taking 0.1333 seconds, 368 records returned.
Powered by Minx 1.1.6c-pink.
58 queries taking 0.1333 seconds, 368 records returned.
Powered by Minx 1.1.6c-pink.









