Hacker News Reader

Comment by tux3 | Hacker News Reader

That's not the first thing people might worry about, but this will also be a pain for archives.

When slop floods the library faster than it can expand, who will want to maintain that. I don't think we have good enough sorting and rating (or good enough AI output detection) to prevent bookspam.

This is a point in history where "record everything" stops being viable, and we have to start hand-picking the text we want to survive. Indiscriminate things like Internet Archives stop being viable.

Why keep books, anyways? You can just ask the AI to re-generate whatever it is you want to read about on the fly.

View on HN

In France there is a law called "dépôt légal" that forces all publishers to send a copy of each new book to the national library for archiving. It's been in place since 1537.

But ebooks are exempt from it. (Ebooks obviously didn't exist in 1537 but they've now been around for decades.)

So unless AI is used to make books published via traditional publishers (which is of course possible, but somewhat unlikely), the situation doesn't change much as regards to archiving books (in France).

I'm not sure of the situation in France, but this is starting to happen readily on Amazon. Lots of physical books you can buy that are completely AI-generated nonsense.

Are those books from real publishers or from Amazon's Print on Demand service?

I should have been clearer; it's not just ebooks that are exempt, but anything without an ISBN. You can make a paper book with Amazon print-on-demand that doesn't have an ISBN, and then you don't need to send it to the dépôt-légal.

Many people argue this is bad and should be changed. But it's debatable.

According to https://www.bnf.fr/fr/depot-legal-pour-quels-documents, pure ebooks are not exempt; they are just archived directly from the distribution platforms by sampling.

What scares me is that some mixed works (PDF/on-demand prints) are not archived because the authors think the system is too clunky to deal with, which is really a shame for books like https://laurent.claessens-donadello.eu/pdf/lefrido.pdf, for example.

Libraries are already flooded. Libraries regularly send older books to be pulped, due to the fact they often contain outdated perspectives. They also look at last-checkout. There's a huge recency bias in anything except national archival and university research libraries. And donations are not put on the shelves, they go into the booksale pile and then the pulping pile. Collections are almost always only added to via librarian purchases, procuring new books at their discretion and patron request.

A few put them up on third-party bookseller sites for a time, and I've been able to get a lot of rare and notable works for comparatively little money. Actually, a lot of them come from university libraries, so maybe they're not as much of an exception as I think.

Of course we have a mechanism to filter books: publishers. They

Now ebooks and on-demand printing allow you to "publish" books without going through the quality filter of a publisher. But you won't get into a book store that way, and anybody talking about the trouble of archiving those better already have an archive of the low hanging fruit of the same category: fanfiction.org and friends (and I don't mean this derogatory, there are great stories in there)

> Why keep books, anyways? You can just ask the AI to re-generate whatever it is you want to read about on the fly

Surely, there are multiple sci-fi stories exploring this exact premise. I would love some recommendations in my replies.

Reply on news.ycombinator.com