High schools, at least in the US, are strongly incentivized to teach only what's on the standardized tests (english and math).

Makes sense. Companies have been investing for a few years now in ramping up production. And at the same time battery companies have been investing in refining their designs to reduce the need for rare minerals. For example, LFP batteries, which Tesla uses for all their "standard range" vehicles, for example, contains no Cobolt. Which means that while the number of EVs being produced is up, so is the supply of the required materials, and the quantity required per vehicle is dropping. So it's not shocking that prices would be dropping back towards normal.

These large language models (ChatGPT, etc.) aren't always 100% precisely right, but their "wrong" outputs can be quite useful. For example, if you ask ChatGPT to write a program to do X, it might not get the entire program completely right, but it's generally a very useful starting point, getting the right libraries pulled in, solving the basic problem, etc., and it just needs some tweaking to get the subtleties right. That's quite useful - it saves you a lot of time, even if it's not 100% perfect. Of course, sometimes it is 100% correct. And sometimes it's "confidently wrong". Luckily for us humans, even though AI can be a huge accellerator, it can't replace us 100% of the time yet. :-) Speaking as a human, that's not a bad thing.

That all sounds good, but why has productivity gone up in Europe, where they're far more regulated, and not in the US?

I this is it - the US massively overbuilds high end housing, which cost a lot more to build, and underbuilds affordable housing, because the market is optimized for ROI not for producing houses efficiently, and they'd rather spend more time making a house that's 4x larger and sells for 4x as much, because it gets more money out of the single lot, which means that there's far more labor per house built and thus fewer houses produced for unit of labor, which is less output as measured by this productivity metric, but making the builder far more money per lot, which is what the builder cares about. A huge factor on top of that is that private equity firms have been distorting the market to buy up all the lower cost housing and taking it off the market as rentals, leaving only much more expensive housing, which they buy and flip at inflated prices. So, basically, the US massively underbuilds housing, making far too few houses that are far too large, because that's most profitable to the builders. In other countries they have more rational housing policies, and crank out many affordable houses instead of forcing people to either rent or buy far more house than they want, which yes, is wildly unproductive by any rational standards, but which makes investors and builders much more money. And what's more important - people being able to buy starter homes, or making investors richer?

There's FinCEN that has the mission of collecting financial transactions for research purposes by law enforcement. Specifically authorized by laws https://www.fincen.gov/fincens... .

I agree that the whole "land of the free" self-image here in the US falls apart once you learn how the rest of the world actually works. Privacy laws, etc., outside the US are a lot more effective, etc.

Financial transactions above a threshold ($10k until recently, now $600) are reported to both the IRS and FinCen https://www.fincen.gov/ . The police very much do use FinCen to look for fraud, money laundering, terrorism, etc.

Tax havens are used to conceal individual and corporate revenue in order to avoid paying taxes legally owed. For example, by US law you owe taxes on your income, and tax havens are illegal concealment of income to avoid paying taxes owed. They're mainly used by the very rich and corporations to avoid paying taxes, creating a significant loss of tax revenue that means that those of us who properly pay taxes owed have to pay more taxes, or to finance more debt. Supporting tax dodges just screws everyone honest about their taxes.

Linux dominates mobile phones and tablets, utility devices (chromebooks, steamdeck, etc.), etc., and WINE and Proton keep improving to the point where you can run pretty much all Windows apps on Linux easily, to the point where they are increasingly displacing traditional desktops. I wonder if we're hit "the year of the linux desktop" the same way that we hit "the year of networking" because everything was increasingly networked and eventually, we looked back and realized that we were all networked and could stop talking about "the year of the network".

There is a real challenge that you often see companies manage by things that they can measure easily, instead of things that are actually of value. A common example is measuring developers and teams' productivity by lines of code generated, when often high value work is a small number of lines that required deep thought to get right, while cranking out tons of lines of code might be a low value activity, so treating lines of code as meaningful can create a perverse incentive to reward people doing easy work over people doing hard work. I don't know any companies paying developers literally by line of code, but I've certainly seen management teams looking at KLOC generated by different teams or vendors and equating that with delivered value, which is unfortunate.

Interestingly, for most applications the unified RAM performance plus the SSD storage means that OS and apps perform well with quite a bit less RAM than on other machines, because they do less data copying in RAM, and can swap faster when they need to, and are optimized to run in less RAM, and have many operations optimized in specialized silicon (video transcoding, neural nets, etc.). You might want to try out the base M2 Mini to see how it performs for you!

Sure, but integrating the RAM on chip gives Apple Silicon much better performance (lower latency) than off-chip RAM. That's the tradeoff. Want that amazing performance, lose some flexibility. Now keep in mind that the large majority of computer sold never upgrade storage or RAM.

Memory that's shared between CPU and GPU is a feature - it means that you don't need to send copies of data from the CPU to the GPU and back, because they can both reference the same RAM, speeding system performance. As a result, data processing that uses GPUs is much faster on Apple Silicon than on PCs with separate CPU and CPU, if it's optimized for Apple Silicon so its not making unnecessary copies but just working in a unified pipeline.

ChatGPT doesn't literally do an internet search - it only knows about a finite set of content, processed in 2021. But even so, it's a very useful first step in finding information, because it does a pretty good job of understanding your request for information and responding with well-structured info. It's got drawbacks - it can't go very deep and when it's asked about topics that require expertise it can be "confidently wrong" which can be dangerous because it all sounds good to the user who isn't an expert, But even with those limitations, it's a much more intelligent first step in learning about a topic than web search.

If Google can add a comparable layer of intelligence, and then provide links drilling down to specific sources of deeper info, that'd be a willing combination. Google search does have some intelligence, and does understand some info (e.g. flight schedules, restaurant reviews) but ChatGPT is dramatically smarter, and Google had better catch up if they want to keep owning all that search traffic / revenue.

According to MSG's statement, the ban is on specific attorneys actively engaged in litigation against MSG, not all employees of a law firm that has someone working on such a case. She's not engaged in a case against MSG.