It pisses me off to reflect that they can sit on works until copyright expires, keeping them out of print. There's no real need for any of these so-called rare books to be rare while they're under copyright.
And related to this, the books that are in print are mostly only in print in the shittiest way. I often see well-made books from the 17th or 18th centuries which are still in good nick. It's ridiculous that in the 21st century, publication standards have fallen to the point where for most works a disposable format is the only type available - where no amount of money could buy a truly decent hardback copy.
If we have to have copyright laws, I'd like to see two changes to them.
When a publisher has no incentive to keep an edition in print, it should be available to any other publisher to print, without compensation to the original publisher, and with renegotiated royalties for the author.
And if the publisher keeps a book in print - but only in bestseller-grade materials, bogroll paper that furrows in any humidity and perfect binding that molts its pages a couple of dry seasons later - and if it refuses to print a durable hardback copy with signatures, good paper and decent print - something that will still be readable in several generations' time - any other publisher keen to have a crack at it should be able to, again without any compensation for the original publisher, though perhaps in this case, with matching royalties for the author.
Books in the 17th and 18th century often didnt get bound by the publisher. They were done by independent binders for _custom_ orders. You’d see whole libraries with the owners binding/cover standards rather than per book.
Books in that time were _luxury_ goods. Most people could not afford them. One of the ways that was changed was to introduce cheap, mass produced bindings that were lower quality than the bespoke artisianal bindings done by specialist craftsmen.
You can still get custom bindings done. There exists whole niches on the internet of crafters that will take a production run book and strip its binding and make you extremely high quality and custom bindings and covers.
My wife worked for a company that specialized in rebinding paperbacks for schools and libraries. It makes more economic sense to do that for niche use cases rather than make all print runs more expensive.
> When a publisher has no incentive to keep an edition in print, it should be available to any other publisher to print
How do you adjudicate that? And wouldn't it just lead to loopholes such as "Ghost Printings" (cf. Ghost Flights https://en.wikipedia.org/wiki/Ghost_flight_(commercial_aviat... ) where the books are technically printed in the required volume but practically unavailable to customers through one method or another. Because, the cost of wastefully printing a few books to warehouse, is less than the potential losses of the IP rights, probably.
We reprint old books after checking out copyrights (for all books, this means pre-1930, but for some (I'd actually say most) it also means ones published up to 1964 and 1973, depending on how the rightsholders did (or didn't) do the renewals).
We use a special guillotine type cutter to cut off the binding and then store the pages in a sealed plastic bag which goes in the archives; they're stored there indefinitely in case the book needs rescanned for some reason. We also keep the original, uncompressed copies of the books on magnetic disks.
We also go out of our way to try to find rare books published in 1931, 1932, etc. so they are ready to go once the copyright expires.
And no, no AI company has ever come to us and asked to run training on all of our scanned copies.
> We reprint old books after checking out copyrights
Who is we?
> then store the pages in a sealed plastic bag which goes in the archives; they're stored there indefinitely
Is that the best thing for archival storage? Like could things like chemical breakdown increase the humidity in the sealed bag or concentrate corrosive chemical vapors? I was under the impression the best environment was an actively climate-controlled environment.
It's easier to get good quality scans from individual pages than it is from pages in a complete book. Imagine laying a book flat, then the page is distorted in the area around the spine. You can work around this (either by trying to correct for the distortion in software or with clever scanners that position the book more advantageously) but it adds complexity compared to chopping off the spine and just dealing with flat sheets of paper.
Easier to send them through a duplex scanner (or put them in a flatbed one if they are fragile). Cheaper than buying the automated ones with the page turning robot arm.
I have done it at home for my books since the mid-00s.
That's why archive.org should have never been sued for lending books they had physical copy of. This is the result. Publishers should be more careful what they wish for.
The archive.org story was more nuanced than that. If I recall correctly the full story was that they used to lend digital versions of books they physically bought and scanned with DRM to enforce a sort of one to one at a time restriction.
But during covid archive.org decided to just remove the limit and lend unlimited copies concurrently which started the debacle with the publishers.
> But during covid archive.org decided to just remove the limit and lend unlimited copies concurrently
IIRC, this was 100% it. Lending one digital version of one physical asset was likely already a violation copyright. Lending UNLIMITED digital versions of one physical copy was DEFINITELY a blatant violation of copyright.
"IA maintains that it delivers each Work “only to one already entitled to view [it]”―i.e., the one person who would be entitled to check out the physical copy of each Work. But this characterization confuses IA’s practices with traditional library lending of print books. IA does not perform the traditional functions of a library; it prepares derivatives of Publishers’ Works and delivers those derivatives to its users in full. That Section 108 allows libraries to make a small number of copies for preservation and replacement purposes does not mean that IA can prepare and distribute derivative works en masse and assert that it is simply performing the traditional functions of a library. 17 U.S.C. § 108; see also, e.g., ReDigi, 910 F.3d at 658 (“We are not free to disregard the terms of the statute merely because the entity performing an unauthorized reproduction makes efforts to nullify its consequences by the counterbalancing destruction of the preexisting phonorecords.”)."
Yeah that was it; if I got this right, US libraries got the right to lend out one digital version of a book that they had in their inventory. Archive.org combined those digital versions so that people could check out a digital book if any library in the US had it (digitally) available. But during the 'rona they removed this limit and just lent out books regardless of it being "checked out" digitally from a library.
This wasn't a very smart move of them. I get why they did it but they put themselves at a huge legal risk.
And, regrettably, The Archive lent books regardless of physical possession.
Publishers had accepted the prior arrangement before The Archive decided to push it, if not explicitly then implicitly by not suing.
I'm a believer in The Archive's mission, and I wish they had treated the goodwill they'd accumulated as something worth preserving and not a currency to be spent.
It has been stated by many before me: lending books should have been handled by a separate entity, especially when they removed the physical backing requirement.
Just to be clear, publishers hadn't accepted the "controlled digital lending" (CDL) premise, not even with the one-to-one ratio. Their position was always "first sale ends when the atoms do". There was even controlling precedent: a few years before IA tried their online lending library thing, there was an "MP3 resale" company called ReDigi that had lost on very similar grounds. The publishers suing IA even made sure to sue in the same venue that had decided the ReDigi case so it'd be controlling precedent.
Furthermore, in the discovery for the Internet Archive case, publishers had already found a case where IA had lent out books despite knowing their partner libraries wasn't actually withdrawing loaned-out copies from circulation. The CDL premise was always just a suggestion, and IA would have still lost their case if they hadn't done the National Emergency Library (NEL) stunt or if they'd been sued in another venue that hadn't had the ReDigi case as precedent.
It's important to note that whenever a company decides to sue for copyright, it is often late, because the company is banking infringements up to the 3-year statute of limitations and because building a meritorious case takes time. The lack of a timely lawsuit proves almost nothing about the intent of a publisher with a valid case against you.
The thing is, I don't even think the whole stunt damaged much of the IA's goodwill? I know of a few people who withheld donations to IA, but that was mainly under the assumption that publishers would be getting a billion-dollar damage award that would immediately bankrupt IA and result in it's archives being sold off to Lexis-Nexis or something. The funny thing is, IA wound up settling for a sum so small they had to promise never to reveal it, and the danger is gone, so the only thing people complain about now is just that the NEL stunt maybe pushed them "above the radar" or something.
It's still insane that shredding books for AI training is legal, but this isn't.
> You can reprint a bestseller. You can't replace the last three copies of an 18th-century botanical text once someone shreds them for training data. And the judge said it's legal. So it's going to accelerate.
Aren't they shredding only the books still under copyright protection? How is an 18th century botanical text still under copyright?
IDK about the shredding, it's not nice, but it's more a problem with copyright law than AI companies.
Scanning books you own should be legal from a copyright point of view, and not require shredding.
Second, one should think about abandoned property provisions for copyright works published more than 50 years ago and in danger of being forgotten: once challenged, either you as the owner have to prove that the work is preserved for future generations (e.g. in various libraries around the world), or you have to authorize further copies, or you give up copyright on the work.
Scanning books by taking them apart into singular pages and scanning those pages is faster and cheaper. AI training is a numbers game, so they want faster and cheaper.
What happens to the pages after? No one needs them anymore, so they get mulched and recycled.
That would be the dominant scanning method even if copyright wasn't a thing. But then again - if copyright wasn't a thing, there would be much less need to scan any physical media.
The reason why OpenAI can't just go on Amazon, buy a "digital edition" of a 2018 book and use that is that it would violate the license in ten ways, and then the DMCA laws that forbid breaking DRM on top of it.
Think about that last point for a moment. Our “rights to read” are diminished significantly with digital works as compared to printed works. Right of resale. Right to lend.
In the end, digital publishing just isn’t right and will lead to massive gap in our historical records. They require active curation and cannot be preserved simply by resting on a dusty shelf.
Every innovation since the microprocessor isn't worth saving in the grand scheme of things.
When today's algae evolve enough into tomorrow's sentient creatures, they're really only going to need up to the industrial revolution and should probably stop right before that.
Empirically, probably not. We had lots and lots of content before copyright, and people seem to produce lots of content even in jurisdictions with weaker copyright.
Old rare books where there are single digit copies should enjoy some sort of patrimonial protection just like museum pieces. You can own them but have the state have the option to buy it if you’re about to significantly deface it or destroy it.
We manage to do this for endangered wildlife too without anyone counting every single specimen; why shouldn’t we be able to estimate how rare a book is?
The publishers sued AI companies for training on shadow library data, hoping to negotiate content deals for big $$$ down the line. Instead, they got analog hole'd.
Turns out that buying an old book for $5 and destructively scanning it for $25 is way cheaper than paying extortion fees to the copyright-mongers.
What I don't buy is it being "rare, precious books". First, they're not after ancient texts - they're after the books that there's still copyright on. Second, when it comes to books, "old" doesn't mean "valuable" - plenty of libraries destroy old books because there's no demand for them, and storage costs you. This is how those scanning companies get books for so cheap.
They are paying a fair amount. In the ballpark of $5 per book.
You know, piracy online is nice and simple - but it's kind of hard to get physical media without paying what the previous owner considers "a fair amount" to part with it.
I think they are in the legal sense of right, and I think they only discarded the remains of these dissected books because previous rulings (e.g. archive.org's lending practices of digital copies of books they physically owned) gave rise to a situation where destruction bore less legal risk. As for the moral case, I don't have much to say on that, we all have our own lines in that sand.
I would say that especially for books out of copyright, they are unquestionably legally in the right. There is no legal standing for "I like books and it makes me feel squicky when someone disrespects them". Nothing stops you from buying old books and using them as decoupage[1] fodder, using them to start a campfire, or making paper airplanes out of their paper.
Moreover, while it is emotionally appealing to some people to want to add some sort of "responsibility to society" to people who own the old books, it's a very emotional plea that can't really be manifested in the real world. As already pointed out in other places, "an old book" itself doesn't really mean much in terms of what its value is in any particular dimension. Plus, I am always deeply suspicious of anything that expands to "Other people, who are not me, should expend vast quantities of resources so that in the next five or ten times I think about this issue for the rest of my life I feel slightly better about this issue" which is what this really amounts to. I think that as superficially appealing as that may be, it's really a very hostile and demanding position to take.
Personally, to the extent that I would want to lay a "social responsibility" on the AI companies, I'd like to see something like they are either obligated, or ideally, just do it of their own free will, to make the scans of the books that are out of copyright available for some reasonable fee (ideally, "free because we like the PR", but given the scope demanding it be free is not reasonable), and without them trying to lay any further claims on the public-domain results. Trading "one old book somewhere, inaccessible to the world" for "a scan of the book and an OCR of it" I would judge a net win for society for rather a lot of these old books, which are by no means "worthless" sitting in some old collection somewhere but would be a lot more useful for being available.
They could give a copy of the data once to some third party organisation that then seeds it in bittorrent or something like that. Basically, what I want to say is that this doesn't need to be an ongoing obligation for the scanner to be worthwhile for society.
yes. in the sense that they probably wouldnt want to do this but the law forces them to do an extremely dumb thing.
they're paying for the books, no shady things going on there. whether the publishers should deserve more than a single copy's worth is a separate question.
having the law such that its illegal to scan a book and then keep it, but legal to scan it and destroy it - gg no re there, law people retardmaxxed themselves as they tend to do with anything related to digital data.
Law aside, destructive scanning is cheaper and easier. Cutting the book into individual pages would be happening regardless, because that is the easiest, fastest, cheapest way to scan it.
Buying up old books is legal. Digitizing and distilling them arguably is too. And also:
> paying extortion fees to the copyright-mongers
Yes yes, greedy fat cat publishing oligarchs treading on the poor put-upon scrappy AI underdogs. /s
Back in reality, the fraction of people who got into publishing books to get rich collecting rents is… not large. There’s so many other fields that are likely to reward participants with more wealth that it’s absurd — even with all the passion for the work in tech it’s probably relatively less pure.
And whatever the excesses of copyright have been, the whole bargain has always been on more pro-social foundations and stronger intellectual foundations than “extortion” sneers. It recognizes that incentives matter and work that’s valuable should be rewarded and incentivized.
A culture that takes a Robin Hood approach to low marginal cost billing points but fawns over the hypercapitalized distribution King Johns isn’t creating a freer or richer society or fighting the real cartel center, it’s indulging resentment and caricature.
I see mentions of Bradbury's "Fahrenheit 451" in that thread but what this really seems to be mostly like is Vernor Vinge's "shred and scan" factory in his novel "Rainbows End".
Perhaps the last great near-term predictor. I often wish he'd written more. To remind us all, he predicted shred and scan would be a short stop over done by villains on the way to nondestructive scanning.
That said, supporting Anna's archive is one of the best things you could do for humanity long term in my opinion.
I think some villainy is implied by "billionaire-backed-woodchipping of a library," but it's just my interpretation, no actual knowledge of Vinge's perspective.
That’s digital though, so it requires the continued survival of readers for the data that is stored. The best thing you could for the long term is probably to buy a few hundred physical books to keep in a bookcase in your home.
Speaking as someone with dozens of bookshelves and tens of thousands of books... I kind of prefer the idea that continued survival means getting a bunch of 14TB drives and, you know, hosting certain files obtained from certain places. The reality is that most people's book collections are simply going into dumpsters, speaking as one of the people who go and try to buy these book collections at estate sales. (We can't do anything with the sheer volume of these books so most of them go into the dumpster. I cannot store hundreds of thousands of books, or millions, and no libraries want them.)
Also, holy cow, hard disks (as in the magnetic oxide kind) got a lot more expensive.
Yeah agreed. I have on my long-term project list a hardware 'oracle' that would have everything and a local good model as a librarian/assistant and be solar powered in a pinch.
They should be forced to publicly release the books as an Ebook.
Leave it up for anyone to download and then compensate the copyright holders later.
In fact if ingesting these books for LLMs is fair use, us commoners should be able to read them for free. Maybe restrict commercial redistribution though.
Maybe we can kill two birds with one stone: digitize rare books and reverse the damage from Authors Guild v. Google [1].
Let AI companies do this. But require them to make the digital copies public. Maybe with a multi-year delay, to give the original scanner advantage to doing it.
Let's say one of the books to be digitized and destroyed is the sole remaining copy of a book from 1850, which is now considered public domain.
On one hand, hoarding such a book, stealing its content from the public domain, locking its content behind a for-profit machine, and destroying the only remaining copy is clearly wrong. It's equivalent to stealing a public resource, just like mining minerals or oil on public lands without a permit or mineral rights. Pure extraction.
On the other hand, taking care to digitize the copy and making it available for free in perpetuity, as well as being required through regulation to provide access to that content through, let's say a public utility LLM/AI available for free through libraries and online... and perhaps after fair due diligence being required to preserve physical copies in a public archive of rare books of which there are no known remaining physical copies...
That seems much more reasonable to me at least. I can imagine there are many who would not see it that way though. Do we see it happening or gaining regulatory, moral and/or public support?
If true this is basically unforgivable. The wanton destruction of history is the stuff of the Third Reich and the Taliban. You simply cannot profess to care about knowledge, culture, or civilization while destroying the physical manifestations of the same.
And this is the beginning of the end for content creators.
Why would I spend hours creating original content if Google can extract it and present the answer directly in an AI Overview? What is the incentive to keep doing the work?
If creators stop producing high-quality original material, the information we get over the next few years will increasingly be based on recycled, low-quality garbage.
Take a look at most successful journalism today. It's behind a paywall. You get paid by the people who are interested and value your work.
Can Google steal it and present it in an AI overview? Well kinda. Today Google is doing a trick - they're saying "You can refuse to consent to being fed into the slop machine, but if you do we won't crawl you for Google so you'll get no search traffic. But you're not going to get search traffic anyway! So you might as well opt out of being fed into the slop machine. And companies are starting to do that [1]
It's really interesting, because essentially what it means is Google is turning into a walled garden, but there's nothing growing inside it so they have to continually import new plants to live in their walled garden and they're going to have to pay to do that. So soon Google will be paying news sites for the right to plumb their feed into the slop machine.
From what I understand, the rare books in question are not some historically relevant medieval manuscripts, but rather some relatively recent books (still under copyright) for which there are few print copies available for purchase.
I am not sure that physically destructing one copy of this type of book to preserve its contents digitally is so bad. Pretty much anything that is still under copyright should be valuable only for its content, not for the physical medium it's printed on.
> Pretty much anything that is still under copyright should be valuable only for its content, not for the physical medium it's printed on.
Well, there are special collectors editions with the signature of the author and gold pages or whatnot. But the AI companies are probably not using those.
From 1984: "Every record has been destroyed or falsified, every book rewritten, every picture has been repainted, every statue and street building has been renamed, every date has been altered. And the process is continuing day by day and minute by minute. History has stopped. Nothing exists except an endless present in which the Party is always right."
It seems a key contention of theirs is the possibility that rare books are being destroyed this way, yet the things they cite don't seem to suggest this (based on their paraphrasing), they just throw the following at the end to make it seem like it's occurring to irreplaceable books:
> You can't replace the last three copies of an 18th-century botanical text once someone shreds them for training data. And the judge said it's legal.
Is there evidence of this? Since otherwise they could very well be describing what is only occurring to in-print or non-rare books. (This is a genuine question since their post doesn't shed any light on it.)
Weeding is the natural process of disposing of less-demand books. Like the rest of us, libraries operate in finite space, so if they want new books, they have to remove ones their users aren't using. Most libraries will try to sell books before disposing of them in any destructive way.
People are very desperate to try to claim that something that's somewhat obviously morally wrong is actually highly nuanced, because it makes them feel uncomfortable.
This is really dishonest framing, unless you really, honestly can't tell a difference between pulping a mass market paperback romance novel that there's 3 million of in circulation, and shredding an 18th century botanical text that there's only 2 copies of in existence.
> The attachment contained 3,000 English-language titles organized by ISBN number, including books such as Distinct Element Modelling in Geomechanics by K.R. Saxena (1999); Barrett's Traditional Fairy Tales (2021), an academic study of Irish folklore; and Laser Shock Peening of Advanced Ceramics by Pratik Shukla (2018).
How is a book from 2021 considered rare in this context? There's almost certainly a digital copy of it in existence prior to Anthropic purchasing a print edition.
Niche text. It's not impossible that there was only ever under a thousand of them printed and released into circulation.
A digital copy would exist somewhere, of course. But for us, that only matters if we can buy or download it. And for AI companies, that only matters if they can get a digital copy DRM-free and licensed permissively enough.
This argument doesn't make any sense. All manner of AI companies just ingest whatever random text they can find on the internet to train their data, including copyrighted publications. Why would DRM on a digital copy of a book matter?
Yeah - it does feel a bit overly dramatic, books mentioned from a potentially related article are from like 2018 (https://nltimes.nl/2026/06/25/rare-book-dealers-fear-tech-fi...). Let's not pump the drama more than we need to, Sam Altman made another mention of the singularity over the weekend so enough of that going around.
Is this true? Rare books would very often be out of copyright for a start. What the the actual ruling that says you can scan if you destroy the original? ISBNdb is a database of book information, as you would guess from the name.
This is the opposite of a book burning. These books which only a few would ever know the names of, let alone find, let alone read, are being digitised so they can be found in electronic searches.
And knowledge from these books can be "imprinted" into a model to be used by much more people or even survive this planet once sent into space in a probe.
> being digitised so they can be found in electronic searches
You make it sound like they are running a second Project Gutenberg. They most definitely are not making these available for electronic searches. At least not searches the public can participate in.
Are they being digitised? That word implies that the book would have a digital representation of the original, which is not what an LLM is.
Can they be found in electronic searches?
I mean I kinda get what you mean, the knowledge was potentially forgotten and now it might not be, but also the authors who created that knowledge get no credit, no reward.
This is the result of our copyright law in the United States, which is extremely tilted to favor authors and publishers. The judge made exactly the right call and the companies are following the law.
The fix here is to change the law to permit training AI without destroying the original materials. But that is going to be a heavy lift.
> One professional bookseller who specializes in selling foreign language books on these marketplaces told me that, starting in April, he and other booksellers noticed a historic spike in sales. [...]
> This bookseller said his inventory is full of rare, foreign language, and low circulation books, meaning that if they are destroyed in the process of becoming training data, they’ll be even harder to obtain.
I will make a robot scanner for books. I will then scan all the books in my state libraries and make a digital copy of them (without destroying them) before these things come for them.
Yes, this is very disturbing to me. I recently discovered how great old, out of circulation/print can be. Now old abandoned libraries are like a treasure trove to me. If these books are not digitaly preserved, that sounds very sad to me.
I have zero proof for this, but just a what if: what if Anthropic's strict anti-China stance actually means the Chinese training corpus is way more valuable than people realize?
People keep bringing up these rare books but never share what they actually are. What are their names? When were they written? How many copies were in existence? Were they in libraries, or locked up in vaults? Did people have access to them prior to being scanned for AI? How many such books have actually been destroyed?
Weird to see so many of these "trust me bro" twitter stories make it to the front page and cause outrage when no one has any real information.
It’s unclear whether ISBNdb will scan books without ISBN’s, which were invented in the late 1960’s. Customers appear to be ordering books to be scanned by ISBN? Here is one book seller’s experience:
> Bulk purchases also usually reflect interest in a specific topic, whereas the recent, very large purchases were of books that had little in common, except for the fact that they all had ISBNs. This seller also sells rare books that do not have ISBNs, and none of those were part of the bulk purchases.
Article is paywalled, but I saved a few quotes here:
I don't buy this at all. Looking up some of the mentioned rare books, in other sources, on worldcat shows every single book available somewhere. Most of them in dozen or even hundreds of libraries.
So the premise here is that the books are valueless enough that they're being sold by weight, but they're also rare enough that someone will one day wonder where they went, and also that the AI companies aren't also saving the text somewhere much safer than physical media in a warehouse. k.
Not in a traditional sense, but obviously on the surface, they're using the info to regurgitate in some fashion and serve back.
The point is that even under the best intentions, hallucinations occur. Then there's the fact that most models have an ideological bias programmed into them.
The only expectation I have is for companies or anybody else to not destroy rare books. Is that such a tall order?
> they're using the info to regurgitate in some fashion and serve back.
Sure, in the same sense that they regurgitate any other text they consume. LLMs by definition do not have the full training dataset available, though. It’s far larger than the resulting model. So they can’t reliably reproduce full text without an external source (or if it’s in the training data repeatedly). ChatGPT actually refused to give me a bible quote the other day, presumably because I ran into some general “book regurgitation” safety net.
> The only expectation I have is for companies or anybody else to not destroy rare books. Is that such a tall order?
Honestly, yeah. The idea that people or corporations should hold onto books forever because of a cultural “ick” about throwing out books is a bit ridiculous. Most books end up in landfills.
They aren’t feeding Da Vinci manuscripts into this pipeline. They are feeding still-in-copyright books.
People would have at least somewhat less of a problem with this if they also put up an archive of PDFs of all these rare books if they are out of copyright.
But that would help competitors with training data, which I assume is why they don’t do this.
you wouldn't believe how much shredding your local library does in the name of space conservation and to address changing borrower preferences and demographics. I doubt any AI company comes close to the annual combined library turnover
> A federal judge ruled the practice is fair use because eliminating the original means only one copy exists at a time.
Really? That sure wasn't a thing when one startup got sued for streaming from its wall od dvd-players, and they adhered to 1 disc = maximum 1 stream at same time.
I don't see any proof of shredding here. Most book scanners I'm aware of are from Google's scanning days, and those had cameras plus page turning.
If we use 'shredding' to mean A book is laid flat, its cover is removed, and then a paper cutter cuts through the binding to create a flat stack of sheets, which are then fed to a sheet feeder, then I could maybe imagine this is better than a page-turning scanner. But, sheet feeding old paper sucks shit, bro. It's not fun.
Upshot, I think we'd like to hear from an anonymous frontier lab employee here to see what's going on -- there are a lot of books in Anna's archive available at considerably less difficulty.
The idea here is that the frontier labs are no longer willing to risk using the likes of Anna’s Archive, because they have already been held legally liable for that in the past.
Are you referring to the META lawsuit? I think the current landscape is not bad for the frontier labs -- it's settled law that it's legal to 'read' and ingest this data. Anthropic went ahead and just settled a licensing deal for content. The open issue in that META suit is whether or not any distribution happened, as I understand it. I'm certain they all have full backups of the archive somewhere in the org.
This kind of reads like a blood libel. My guess is they're buying all those books that University libraries are throwing away these days (to see other HM threads for that), kinda sad but probably aren't "rare books" in the way people are thinking
I think people are, on the whole, too precious about old things. In the case of books produced after major commercial printing began, I don't believe it is the paper that imbues the book with historical value.
Indeed, I think there's a high chance that this process increases preservation of the most relevant part of the media - the actual content!
There are tons of old rolls of film slowly rotting away in warehouses that were never digitised. Even for beloved media, the BBC occasionally tracks down an old lost episode of Dr Who.
For now these books are in corpuses of training data, but eventually I trust they will make their way to the rest of us.
Well, you probably weren't going to go and find any rare, non-digitised books to physically go and read (unless you were going to, in which case, rock on), so we can start by benchmarking relative probability there.
1. At some level of critical information-withholding mass, a leak or disclosure similar to SciHub is inevitable because of the commonly held opposition to hiding knowledge.
2. Availability via Google Books or similar.
3. Availability via AI model reference.
4. Failing any of the above, better AI models that are more capable of doing more things, at the expense of books that were likely to go unread (revealed preference, rare books are often rare for a reason). This will obviously be a nonstarter if you don't want this to happen, but I think it would be good for the world if it did.
I think category of old books that were going to be read or otherwise become important parts of human knowledge that have not yet been digitised and now will never become so because they are instead being shredded and will never make their way into the light because of AI company data hoarding is a small category.
I pre-empted this - my argument does not apply to texts where the physical object is a major part of the historical value of the thing. No, I'm not saying to destroy one of the four remaining Magna Cartas that were meticulously copied by hand. But even if I was, we're only dealing with texts here that are irrelevant enough to have never been digitised already - we tend to digitise most things of value and so the Mona Lisa and Magna Carta would never have been part of this discussion in the first place.
I am however OK with destroying one of the remaining 50 children's books of which only 300 copies were ever printed in a small town in Ohio in the 1970s as a test run for a failed book which was subsequently never commercialised.
When was the last time a formerly undigitised book that was printed but fell out of print suddenly became a success after being discovered in the last, say, 50 years? Is once in 100 years the sort of level of frequency we're talking about?
This is the kind of rationalization hoarders use. The inability to get rid of things because it could turn out to be something we want in the future for reasons that we cannot currently describe.
It's a loss avertive instinct that I think is misplaced. Treating every printed book as priceless is intractable. It's not how we treat these books at the moment. Apparently, today we don't even care enough to spend a few hours per book nondestructively scanning them in.
Let's say we, instead of destructively scanning this books, nondestructively scanned them. What would you propose doing with the copies afterwards? Sell them? To who? They're valueless individually for the overwhelming part. Warehouse them? Why? For who? Do you want to go and look through them? Why haven't you done so already? Have you ever shown interest in consuming an undigitised book a single time in your life to date?
It feels like the anti-AI crowd here have to tie themselves in knots here to make the loss minimization work.
> For instance, a painter may insist on proper attribution of their painting, and in some instances may sue the owner of the physical painting for destroying the painting even if the owner of the painting lawfully owned it.[1]
The Mona Lisa's painter isn't alive, they can't sue, and this act doesn't apply to printed books. You're expanding the hypothetical to say "also you're under a specific set of laws that forbid exactly the thing I questioned".. that's moving the goalpost.
It pisses me off to reflect that they can sit on works until copyright expires, keeping them out of print. There's no real need for any of these so-called rare books to be rare while they're under copyright.
And related to this, the books that are in print are mostly only in print in the shittiest way. I often see well-made books from the 17th or 18th centuries which are still in good nick. It's ridiculous that in the 21st century, publication standards have fallen to the point where for most works a disposable format is the only type available - where no amount of money could buy a truly decent hardback copy.
If we have to have copyright laws, I'd like to see two changes to them.
When a publisher has no incentive to keep an edition in print, it should be available to any other publisher to print, without compensation to the original publisher, and with renegotiated royalties for the author.
And if the publisher keeps a book in print - but only in bestseller-grade materials, bogroll paper that furrows in any humidity and perfect binding that molts its pages a couple of dry seasons later - and if it refuses to print a durable hardback copy with signatures, good paper and decent print - something that will still be readable in several generations' time - any other publisher keen to have a crack at it should be able to, again without any compensation for the original publisher, though perhaps in this case, with matching royalties for the author.
Books in that time were _luxury_ goods. Most people could not afford them. One of the ways that was changed was to introduce cheap, mass produced bindings that were lower quality than the bespoke artisianal bindings done by specialist craftsmen.
You can still get custom bindings done. There exists whole niches on the internet of crafters that will take a production run book and strip its binding and make you extremely high quality and custom bindings and covers.
At the risk of stating the obvious, any poorly made books from then wouldn’t have lasted this long and so you would never see them.
Those books predate the development of wood pulp paper. It isn't the publisher's fault they can't economically print on rag paper anymore.
How do you adjudicate that? And wouldn't it just lead to loopholes such as "Ghost Printings" (cf. Ghost Flights https://en.wikipedia.org/wiki/Ghost_flight_(commercial_aviat... ) where the books are technically printed in the required volume but practically unavailable to customers through one method or another. Because, the cost of wastefully printing a few books to warehouse, is less than the potential losses of the IP rights, probably.
We use a special guillotine type cutter to cut off the binding and then store the pages in a sealed plastic bag which goes in the archives; they're stored there indefinitely in case the book needs rescanned for some reason. We also keep the original, uncompressed copies of the books on magnetic disks.
We also go out of our way to try to find rare books published in 1931, 1932, etc. so they are ready to go once the copyright expires.
And no, no AI company has ever come to us and asked to run training on all of our scanned copies.
Who is we?
> then store the pages in a sealed plastic bag which goes in the archives; they're stored there indefinitely
Is that the best thing for archival storage? Like could things like chemical breakdown increase the humidity in the sealed bag or concentrate corrosive chemical vapors? I was under the impression the best environment was an actively climate-controlled environment.
Yet
As for consumer-grade solutions, look for the Fujitsu SV600.
I have done it at home for my books since the mid-00s.
But during covid archive.org decided to just remove the limit and lend unlimited copies concurrently which started the debacle with the publishers.
IIRC, this was 100% it. Lending one digital version of one physical asset was likely already a violation copyright. Lending UNLIMITED digital versions of one physical copy was DEFINITELY a blatant violation of copyright.
"IA maintains that it delivers each Work “only to one already entitled to view [it]”―i.e., the one person who would be entitled to check out the physical copy of each Work. But this characterization confuses IA’s practices with traditional library lending of print books. IA does not perform the traditional functions of a library; it prepares derivatives of Publishers’ Works and delivers those derivatives to its users in full. That Section 108 allows libraries to make a small number of copies for preservation and replacement purposes does not mean that IA can prepare and distribute derivative works en masse and assert that it is simply performing the traditional functions of a library. 17 U.S.C. § 108; see also, e.g., ReDigi, 910 F.3d at 658 (“We are not free to disregard the terms of the statute merely because the entity performing an unauthorized reproduction makes efforts to nullify its consequences by the counterbalancing destruction of the preexisting phonorecords.”)."
This wasn't a very smart move of them. I get why they did it but they put themselves at a huge legal risk.
Publishers had accepted the prior arrangement before The Archive decided to push it, if not explicitly then implicitly by not suing.
I'm a believer in The Archive's mission, and I wish they had treated the goodwill they'd accumulated as something worth preserving and not a currency to be spent.
It has been stated by many before me: lending books should have been handled by a separate entity, especially when they removed the physical backing requirement.
Furthermore, in the discovery for the Internet Archive case, publishers had already found a case where IA had lent out books despite knowing their partner libraries wasn't actually withdrawing loaned-out copies from circulation. The CDL premise was always just a suggestion, and IA would have still lost their case if they hadn't done the National Emergency Library (NEL) stunt or if they'd been sued in another venue that hadn't had the ReDigi case as precedent.
It's important to note that whenever a company decides to sue for copyright, it is often late, because the company is banking infringements up to the 3-year statute of limitations and because building a meritorious case takes time. The lack of a timely lawsuit proves almost nothing about the intent of a publisher with a valid case against you.
The thing is, I don't even think the whole stunt damaged much of the IA's goodwill? I know of a few people who withheld donations to IA, but that was mainly under the assumption that publishers would be getting a billion-dollar damage award that would immediately bankrupt IA and result in it's archives being sold off to Lexis-Nexis or something. The funny thing is, IA wound up settling for a sum so small they had to promise never to reveal it, and the danger is gone, so the only thing people complain about now is just that the NEL stunt maybe pushed them "above the radar" or something.
It's still insane that shredding books for AI training is legal, but this isn't.
How are these things remotely related? If anything, Archive.org’s callous, thoughtless approach nuked the hands of legitimate archival efforts.
Aren't they shredding only the books still under copyright protection? How is an 18th century botanical text still under copyright?
IDK about the shredding, it's not nice, but it's more a problem with copyright law than AI companies.
Scanning books you own should be legal from a copyright point of view, and not require shredding.
Second, one should think about abandoned property provisions for copyright works published more than 50 years ago and in danger of being forgotten: once challenged, either you as the owner have to prove that the work is preserved for future generations (e.g. in various libraries around the world), or you have to authorize further copies, or you give up copyright on the work.
What happens to the pages after? No one needs them anymore, so they get mulched and recycled.
That would be the dominant scanning method even if copyright wasn't a thing. But then again - if copyright wasn't a thing, there would be much less need to scan any physical media.
The reason why OpenAI can't just go on Amazon, buy a "digital edition" of a 2018 book and use that is that it would violate the license in ten ways, and then the DMCA laws that forbid breaking DRM on top of it.
In the end, digital publishing just isn’t right and will lead to massive gap in our historical records. They require active curation and cannot be preserved simply by resting on a dusty shelf.
When today's algae evolve enough into tomorrow's sentient creatures, they're really only going to need up to the industrial revolution and should probably stop right before that.
Because there’d be much less content created in any media to capture in the first place.
An 18th century book would be out of copyright so why would it be illegal to keep the original and scan it?
It's cheaper to scan the books if you do it destructively. Cost. That's why they're shredding irreplaceable texts. Nothing to do with copyright.
https://www.404media.co/ai-companies-are-buying-tons-of-old-...
Most rare books are rare because no one cared enough about them. Ie most rare books are rubbish.
Turns out that buying an old book for $5 and destructively scanning it for $25 is way cheaper than paying extortion fees to the copyright-mongers.
What I don't buy is it being "rare, precious books". First, they're not after ancient texts - they're after the books that there's still copyright on. Second, when it comes to books, "old" doesn't mean "valuable" - plenty of libraries destroy old books because there's no demand for them, and storage costs you. This is how those scanning companies get books for so cheap.
You know, piracy online is nice and simple - but it's kind of hard to get physical media without paying what the previous owner considers "a fair amount" to part with it.
they should pay the marginal value that the next buyer would buy.
Do you also think that a person dying of thirst ought to pay the maximum price they could possibly pay for water?
Moreover, while it is emotionally appealing to some people to want to add some sort of "responsibility to society" to people who own the old books, it's a very emotional plea that can't really be manifested in the real world. As already pointed out in other places, "an old book" itself doesn't really mean much in terms of what its value is in any particular dimension. Plus, I am always deeply suspicious of anything that expands to "Other people, who are not me, should expend vast quantities of resources so that in the next five or ten times I think about this issue for the rest of my life I feel slightly better about this issue" which is what this really amounts to. I think that as superficially appealing as that may be, it's really a very hostile and demanding position to take.
Personally, to the extent that I would want to lay a "social responsibility" on the AI companies, I'd like to see something like they are either obligated, or ideally, just do it of their own free will, to make the scans of the books that are out of copyright available for some reasonable fee (ideally, "free because we like the PR", but given the scope demanding it be free is not reasonable), and without them trying to lay any further claims on the public-domain results. Trading "one old book somewhere, inaccessible to the world" for "a scan of the book and an OCR of it" I would judge a net win for society for rather a lot of these old books, which are by no means "worthless" sitting in some old collection somewhere but would be a lot more useful for being available.
[1]: https://en.wikipedia.org/wiki/Decoupage , since I imagine a number of people won't know what that is.
They could give a copy of the data once to some third party organisation that then seeds it in bittorrent or something like that. Basically, what I want to say is that this doesn't need to be an ongoing obligation for the scanner to be worthwhile for society.
they're paying for the books, no shady things going on there. whether the publishers should deserve more than a single copy's worth is a separate question.
having the law such that its illegal to scan a book and then keep it, but legal to scan it and destroy it - gg no re there, law people retardmaxxed themselves as they tend to do with anything related to digital data.
> paying extortion fees to the copyright-mongers
Yes yes, greedy fat cat publishing oligarchs treading on the poor put-upon scrappy AI underdogs. /s
Back in reality, the fraction of people who got into publishing books to get rich collecting rents is… not large. There’s so many other fields that are likely to reward participants with more wealth that it’s absurd — even with all the passion for the work in tech it’s probably relatively less pure.
And whatever the excesses of copyright have been, the whole bargain has always been on more pro-social foundations and stronger intellectual foundations than “extortion” sneers. It recognizes that incentives matter and work that’s valuable should be rewarded and incentivized.
A culture that takes a Robin Hood approach to low marginal cost billing points but fawns over the hypercapitalized distribution King Johns isn’t creating a freer or richer society or fighting the real cartel center, it’s indulging resentment and caricature.
That said, supporting Anna's archive is one of the best things you could do for humanity long term in my opinion.
Also, holy cow, hard disks (as in the magnetic oxide kind) got a lot more expensive.
Leave it up for anyone to download and then compensate the copyright holders later.
In fact if ingesting these books for LLMs is fair use, us commoners should be able to read them for free. Maybe restrict commercial redistribution though.
Let AI companies do this. But require them to make the digital copies public. Maybe with a multi-year delay, to give the original scanner advantage to doing it.
[1] https://en.wikipedia.org/wiki/Authors_Guild,_Inc._v._Google,....
Let's say one of the books to be digitized and destroyed is the sole remaining copy of a book from 1850, which is now considered public domain.
On one hand, hoarding such a book, stealing its content from the public domain, locking its content behind a for-profit machine, and destroying the only remaining copy is clearly wrong. It's equivalent to stealing a public resource, just like mining minerals or oil on public lands without a permit or mineral rights. Pure extraction.
On the other hand, taking care to digitize the copy and making it available for free in perpetuity, as well as being required through regulation to provide access to that content through, let's say a public utility LLM/AI available for free through libraries and online... and perhaps after fair due diligence being required to preserve physical copies in a public archive of rare books of which there are no known remaining physical copies...
That seems much more reasonable to me at least. I can imagine there are many who would not see it that way though. Do we see it happening or gaining regulatory, moral and/or public support?
Why would I spend hours creating original content if Google can extract it and present the answer directly in an AI Overview? What is the incentive to keep doing the work?
If creators stop producing high-quality original material, the information we get over the next few years will increasingly be based on recycled, low-quality garbage.
Can Google steal it and present it in an AI overview? Well kinda. Today Google is doing a trick - they're saying "You can refuse to consent to being fed into the slop machine, but if you do we won't crawl you for Google so you'll get no search traffic. But you're not going to get search traffic anyway! So you might as well opt out of being fed into the slop machine. And companies are starting to do that [1]
It's really interesting, because essentially what it means is Google is turning into a walled garden, but there's nothing growing inside it so they have to continually import new plants to live in their walled garden and they're going to have to pay to do that. So soon Google will be paying news sites for the right to plumb their feed into the slop machine.
[1]: https://www.wsj.com/business/media/google-search-publishers-...
I am not sure that physically destructing one copy of this type of book to preserve its contents digitally is so bad. Pretty much anything that is still under copyright should be valuable only for its content, not for the physical medium it's printed on.
Well, there are special collectors editions with the signature of the author and gold pages or whatnot. But the AI companies are probably not using those.
For whom? is the relevant question
> You can't replace the last three copies of an 18th-century botanical text once someone shreds them for training data. And the judge said it's legal.
Is there evidence of this? Since otherwise they could very well be describing what is only occurring to in-print or non-rare books. (This is a genuine question since their post doesn't shed any light on it.)
Now that I think about it, The Judge is an apt metaphor for AI : "Whatever in creation exists without my knowledge exists without my consent."
https://www.ala.org/tools/challengesupport/selectionpolicyto...
Weeding is the natural process of disposing of less-demand books. Like the rest of us, libraries operate in finite space, so if they want new books, they have to remove ones their users aren't using. Most libraries will try to sell books before disposing of them in any destructive way.
What similarities do you see here?
Once that AI companies are really shredding 200 year old rare books, and once that libraries are only weeding mass market pulp fiction.
> The attachment contained 3,000 English-language titles organized by ISBN number, including books such as Distinct Element Modelling in Geomechanics by K.R. Saxena (1999); Barrett's Traditional Fairy Tales (2021), an academic study of Irish folklore; and Laser Shock Peening of Advanced Ceramics by Pratik Shukla (2018).
How is a book from 2021 considered rare in this context? There's almost certainly a digital copy of it in existence prior to Anthropic purchasing a print edition.
A digital copy would exist somewhere, of course. But for us, that only matters if we can buy or download it. And for AI companies, that only matters if they can get a digital copy DRM-free and licensed permissively enough.
(which is terrible, but I would be delighted if they breached it and got thoroughly spanked)
You make it sound like they are running a second Project Gutenberg. They most definitely are not making these available for electronic searches. At least not searches the public can participate in.
What guarantee do we have that the book contents will be served unfiltered and unaltered?
The fix here is to change the law to permit training AI without destroying the original materials. But that is going to be a heavy lift.
The segment that talks about rare books:
> One professional bookseller who specializes in selling foreign language books on these marketplaces told me that, starting in April, he and other booksellers noticed a historic spike in sales. [...]
> This bookseller said his inventory is full of rare, foreign language, and low circulation books, meaning that if they are destroyed in the process of becoming training data, they’ll be even harder to obtain.
I wish...
This is highly disturbing news; is this standard practice? What did Google Books do before?
Weird to see so many of these "trust me bro" twitter stories make it to the front page and cause outrage when no one has any real information.
> Bulk purchases also usually reflect interest in a specific topic, whereas the recent, very large purchases were of books that had little in common, except for the fact that they all had ISBNs. This seller also sells rare books that do not have ISBNs, and none of those were part of the bulk purchases.
Article is paywalled, but I saved a few quotes here:
https://skybrian-links.exe.xyz/post/1026
> It is equivalent to book burning in the past. A form of thought control
> > It is equivalent to book burning in the past. A form of thought control
That's only bullshit if you trust AI companies to serve the book contents without alteration.
The point is that even under the best intentions, hallucinations occur. Then there's the fact that most models have an ideological bias programmed into them.
The only expectation I have is for companies or anybody else to not destroy rare books. Is that such a tall order?
Sure, in the same sense that they regurgitate any other text they consume. LLMs by definition do not have the full training dataset available, though. It’s far larger than the resulting model. So they can’t reliably reproduce full text without an external source (or if it’s in the training data repeatedly). ChatGPT actually refused to give me a bible quote the other day, presumably because I ran into some general “book regurgitation” safety net.
> The only expectation I have is for companies or anybody else to not destroy rare books. Is that such a tall order?
Honestly, yeah. The idea that people or corporations should hold onto books forever because of a cultural “ick” about throwing out books is a bit ridiculous. Most books end up in landfills.
They aren’t feeding Da Vinci manuscripts into this pipeline. They are feeding still-in-copyright books.
But that would help competitors with training data, which I assume is why they don’t do this.
Really? That sure wasn't a thing when one startup got sued for streaming from its wall od dvd-players, and they adhered to 1 disc = maximum 1 stream at same time.
If we use 'shredding' to mean A book is laid flat, its cover is removed, and then a paper cutter cuts through the binding to create a flat stack of sheets, which are then fed to a sheet feeder, then I could maybe imagine this is better than a page-turning scanner. But, sheet feeding old paper sucks shit, bro. It's not fun.
Upshot, I think we'd like to hear from an anonymous frontier lab employee here to see what's going on -- there are a lot of books in Anna's archive available at considerably less difficulty.
Indeed, I think there's a high chance that this process increases preservation of the most relevant part of the media - the actual content!
There are tons of old rolls of film slowly rotting away in warehouses that were never digitised. Even for beloved media, the BBC occasionally tracks down an old lost episode of Dr Who.
For now these books are in corpuses of training data, but eventually I trust they will make their way to the rest of us.
What makes you think they will? What would be the incentives for these companies to do so?
1. At some level of critical information-withholding mass, a leak or disclosure similar to SciHub is inevitable because of the commonly held opposition to hiding knowledge.
2. Availability via Google Books or similar.
3. Availability via AI model reference.
4. Failing any of the above, better AI models that are more capable of doing more things, at the expense of books that were likely to go unread (revealed preference, rare books are often rare for a reason). This will obviously be a nonstarter if you don't want this to happen, but I think it would be good for the world if it did.
I think category of old books that were going to be read or otherwise become important parts of human knowledge that have not yet been digitised and now will never become so because they are instead being shredded and will never make their way into the light because of AI company data hoarding is a small category.
I am however OK with destroying one of the remaining 50 children's books of which only 300 copies were ever printed in a small town in Ohio in the 1970s as a test run for a failed book which was subsequently never commercialised.
What if the perception of those books change over time and are considered masterpieces later on?
Moby Dick was out of print when Melville died 1891 and not a huge success until it got a revival in the 1920s
This is the kind of rationalization hoarders use. The inability to get rid of things because it could turn out to be something we want in the future for reasons that we cannot currently describe.
It's a loss avertive instinct that I think is misplaced. Treating every printed book as priceless is intractable. It's not how we treat these books at the moment. Apparently, today we don't even care enough to spend a few hours per book nondestructively scanning them in.
Let's say we, instead of destructively scanning this books, nondestructively scanned them. What would you propose doing with the copies afterwards? Sell them? To who? They're valueless individually for the overwhelming part. Warehouse them? Why? For who? Do you want to go and look through them? Why haven't you done so already? Have you ever shown interest in consuming an undigitised book a single time in your life to date?
It feels like the anti-AI crowd here have to tie themselves in knots here to make the loss minimization work.
That’s the whole point because the book shredding is already declared legal.
https://en.wikipedia.org/wiki/Visual_Artists_Rights_Act