Which desktops did you try, and what issues blew it for you?

I had a hard time finding a linux I could live with, and I first started looking over 25 years ago. It's only been about six years now since it's become sufficiently stable and complete. And implementations vary wildly. I prefer the KDE desktop as being the most functional (and least annoying), but KDE on Kubuntu is not nearly as slick as KDE on PCLinuxOS.

But at the far end, IMO current Gnome makes Win10 look stellar.... good gods, who thought a cellphone makes a good desktop??

Oh, the story has one good function. It points up how in isolation, such metrics are garbage.

Otherwise.... what you said.

XP64 here. Same philosophy. Block the garbage, don't be stupid, glory in my lack of visitors, and remember that attack vectors are mostly discovered by reverse-engineering the patches. No patches, way fewer clues.

Whatever small risks are well offset by an OS that doesn't continually make me long to reach through my monitor and throttle a UI developer.

California brings water from a lot farther than that -- the Aqueduct runs about 500 miles to service Los Angeles. (And has turned the formerly-lush Owens Valley into a desert.)

Destroying the first dam was environmentally catastrophic, and produced a massive fish kill. Ooops.

https://californiaglobe.com/ar...

Why? Search algorithms are enormously efficient and easily parallelizable, and what you can describe can be divided into stages and pipelined as well.

You also don't know all that many people, maybe a hundred, maybe a thousand tops. Even sequentially going through a list of all the external characteristics you know about a thousand people is sheer triviality for a modern computer. The dataset would be small because you're not trawling through petabytes of data but looking through already digested and organized information.

Everything works better when specifically designed for the application. A CPU can render in software, but a GPU does it better. A GPU may mine crypto well, but an ASIC designed for crypto and nothing else will do a much better job.

AI turns out to have some quite specific needs that GPUs don't ideally satisfy: https://www.youtube.com/watch?...

All that, but I suspect the real issue is that freight is a lot more profitable in that hold than is baggage, so they're trying to reduce the amount of baggage by charging it out of existence.

No one gets 45 years for smuggling drugs, or for that matter, for smuggling live bodies. If you look at the list of charges, they're a damned long stretch, making it sound like he was a massive smuggling op, not someone who had the wrong tank in the back of his truck.

If you move your household from Mexico to Arizona and bring along your fridge, have you committed a crime?

Well that's stupid.

The entire history and purpose of "230" was to encourage sites to engage in moderation, alternatively favoring or suppressing different content, and to give them the freedom to do so without being treated anything like newspapers or other media.

I doubt that there's anything interesting that happened, and I certainly don't believe your take on it, but as a general rule there's nothing at all wrong with the government offering advice or asking people to do things and for people to agree or to voluntarily do those things.

For example: If the government puts out an Amber Alert, you don't have to read it, you don't have to watch for the child who has apparently been kidnapped, and you don't have to report sightings. You can ignore the whole thing and go about your day. You can even deliberately notice the kid and the kidnappers and not lift a finger. That's not illegal. You're committing no crime by letting kidnappings happen where you lack a duty to stop them.

But it's nice to help rescue children, so why not do what the government is asking you to volunteer to do?

Apparently the reason why is that you are opposed to anti-kidnapping, pro-saving-children government conspiracies of that sort.

They're almost always the same. If there are any that aren't, I'd be shocked. He occupies the same sort of 'designated target of hate' that the Rothschilds did. In fact, that's really where it all starts -- a couple of political consultants working for Victor Orban, the Hungarian dictator, decided that a useful political tactic would be to have an enemy to demonize, so they rather arbitrarily decided it would be Soros. Read all about it.

And so we wound up with Hungary being thoroughly fucked up, Hungary impairing the functioning of the EU and NATO, increases in anti-semitism and fascism, probably daily death threats against a guy who did nothing wrong, and all to score some cheap political points.

It's disgusting.

Still works fine, actually. And it's been closer to three decades now. 47 USC 230 is from 1996.

Good thing that wasn't the argument, then. In fact, your summary of it doesn't even make sense -- middlemen don't get in trouble for taking things down, they get in trouble for not taking things down.

What actually happened was that just before the Internet got big, two cases came down concerning different online services. CompuServe got sued for user-posted content, but was found not to be liable because they had not moderated anything and were just a middleman. Prodigy got sued for user-posted content and was found to be liable because they moderated their boards (for things like bad language; they wanted to be family friendly) but had failed to moderate every post perfectly. By letting one bad thing through, they were liable for it -- and by extension, anything else they had failed to catch.

Since Congress wanted sites to moderate user content -- they were really concerned about porn -- they passed a law that encouraged sites to do moderation but did not hold them responsible for failing to moderate every single little thing perfectly in every respect. Further, sites got to choose what they were moderating for -- could be porn, but could just as easily be off-topic posts, like talking about carrots when everyone else is talking about money.

In practice, sites don't like to moderate much -- it takes effort, it may lose engaged users, it costs money, it can't please everyone -- but they certainly can, and there's nothing wrong with it. Get rid of the protection of the CDA and sites won't be able to do mandatory moderation sufficiently, so they'll fall back on none. This is apparently okay with scum who get kicked off of boards left and right, but should not be okay with people who have standards and don't want to put up with that crap.

I assume that you're replying to the part about fair use turning prima facie infringements into non-infringing uses, but it's difficult to tell.

If so, well, that's the statute at work:

No factor is determinative in fair use, even though often enough the fourth factor is. Always do the full test, every time. And recognize that much comes down to which side of the bed the judge got up on. It's not as bad as the copyright utility doctrine and conceptual separability, but it's not great. Look at time shifting; people extoll it as a classic fair use but really it fails on three out of four factors and is middling at best on the fourth, and was originally going to go the other way. Prof. Litman has a nice write-up about it. And for every example of fair use, there's always a counter example where the same sort of thing came out differently due to a minor factual difference.

Anyway, got a cite where a court held that a prima facie infringing use was a lawful fair use and which was nevertheless still an unlawfully infringing use meriting damages and/or injunctive relief? I'd love to read it. Closest thing that comes to mind was some of the section 1201 stuff from ages ago, distinguishing fair use from circumvention.

Doesn't sound like a problem to me.

Oh boy, guess what? Even have an LL.M in IP.

There's really no such thing as an obvious fair use. It's always fact-intensive, always case-by-case. And there's always the risk of times changing. Format shifting comes to us from the RIAA v. Diamond case, and it's terrifying to think of how differently that gone had it been litigated just a few years later when the iTunes Music Store was open. I'll agree though that the use of pirated books for training was a bad idea, in that it does not help how a court will perceive the AI developers. Never a good idea to piss off a judge, and it's clear these guys were not thinking.

Still, there's no market for AI analysis of books as far as I know, and the use was transformative in nature given that no one appeared to be reading them, they were just grist for the mill. This is actually more favorable to fair use than Google Books, which is intended for some human to ultimately use to get to read snippets of text in search results. Imagine how strong Google's case would've been if they'd said that searching for a string of text points you to a book with no snippet or anything else to tell you what's inside. Maybe a page or chapter number, for all the good that does.

Once you get over the training hurdle though, it looks like pretty smooth sailing. The model is demonstrably lossy. It's too abstract to mesh with the concept of an abridgement or condensation. You might be able to get snippets here and there out of it, but I would imagine that given a few examples to dive deep into analyizing, OpenAI can show that they appear because they're built out of common chains of text from multiple works, or are overrepresented due to multiple copies of the work. (Kind of like how the image-generating AIs are said to like to make fake Getty Images watermarks because it saw so many of them in training data, so it must be important)

Personally, I like their chances. Either way, it'll be interesting to see.