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.

Users are allowed more freedom with regard to non-fiction, since the facts therein are uncopyrightable, and the general organization may be as well. Fiction, being more creative, enjoys a bit more protection from potential fair users.

I'm a lawyer. I would suggest you take a look at the court opinions concerning Google Book Search. You may be surprised. Try Authors Guild, Inc. v. Google Inc., 954 F. Supp. 2d 282 (S.D.N.Y. 2013) which was at the trial level, and then Authors Guild, Inc. v. Google Inc., 804 F.3d 202 (2d Cir. 2015), when it went up on appeal.

If an infringing use is also a fair use, it is rendered non-infringing, and that's the end of the analysis. There are no additional steps after fair use.

Gotcha covered:

So there's your answer.

A quote is not a derivative work because it's not based on a preexisting work. Instead, that's a reproduction of part of the work (a separate exclusive right under 106, however). A literary analysis is not a derivative work, but if you dug too deep and merely produced an annotated work or adaptation, then it would be. It's not too hard to stay on the correct side of that line.

Ideas can always be used. Facts -- or things claimed to be a real-life fact -- can always be used. But be cautious again of digging in too deeply. Castle Rock Entertainment Inc. v. Carol Publishing Group, 150 F.3d 132 (2d Cir. 1998) was a case where someone wrote a book of Seinfeld trivia questions. The questions weren't about the show, per se, but about the contents of the show. (That is, they did not ask on what soundstage it was filmed, but things like what is the number of the apartment, or what tagline did character X say repeatedly in a particular episode, that sort of thing) This was held to be copying many little fragments of the show itself, and thus infringing. And it wasn't held to be fair use.

Memorizing a book is not infringing. Reciting to a large group of people -- by any means, whether from memory or not -- would be, if the book is copyrighted.

I remember this being a big area of interest about 15 years ago, but there doesn't seem to have been much about it lately. So long as it's done in real time and just for the reader, and not an audience, I would be inclined to say that it's not infringing the reproduction right, derivative right, or most significantly, the public performance right.

You've never had to buy a license to read a book. Perhaps if you're doing other things in association with your book reading, you might need a license for that but not for just reading.

We've yet to invent a general-purpose AI. These things aren't reading books the way we do. My understanding is that they're basically compiling statistical models, and figuring out how each word or part of word relates to others. Kind of a more complicated version of playing the autocomplete game on your phone, where you input one word in a text chat and then just choose whatever the next suggested word is. The sentences you get make sense for a little while, but become run-ons and start to loop.

Analyzing a work isn't a copyright infringement. Predicate work needed to reach the point of conducting an analysis might, though. And the output of the software might as well.

That and also not wanting machines to be able to write things that sound like something she might write. Licensing isn't all that relevant, as in many cases a license would simply not be granted.

Like I said, if a disk needs to be ejected at boot, you hold down the mouse button during boot and it should be ejected by the drive. This is even mentioned in the 128k's manual. The paperclip thing is only if nothing else works.

If the machine is already up, ejecting the disk normally should work fine as long as it isn't in use. If it is, you can close out whatever's using it or if it's already the boot disk, well, too late given that your issue was not wanting it to boot off that disk.

And yes, there were two originally hidden, but later sometimes subtle but immediately available "programmer's keys" for reset and interrupt. Given the flaky OS of the day, though, dropping into MACSbug was usually pretty pointless.

On older machines, you needed to order a small plastic clip that fit on the side and would poke the buttons hidden behind some of the vents. Eventually they were provided as small buttons on the case. Or you could just use the right sort of many-button keyboard shortcut.

The Xerox Alto was the internally used research project. The Xerox Star was marketed commercially but had way too steep a price to assemble a useful system -- since it basically required a small LAN of multiple machines and a laser printer -- for hardly anyone to buy.