Hacker News Reader

Something not mentioned that links to both LLM training AND drop in book sales... Anna's Archive

I would be extremely surprised if piracy accounts for any serious dent in the sales of books.

The implication is that the self-help books were scraped from Anna's Archive to train the LLMs, thus the content in the books has been integrated into the LLM and may be regurgitated on command. With the content then being available straight from an LLM, there's no need to buy the books.

If this was true, they wouldn't be getting sued and aggressively pursued for takedowns

" they wouldn't be getting sued and aggressively pursued for takedowns"

I don't think so. That behavior only tells you the modest cost of sending takedown notices/threatening letters is less than the (supposed) lost revenue. Kids I know (I'm a teacher) don't seem at all aware of it when complaining about textbook costs and I kind of vaguely hint they look online for it--nothing like the popularity of napster.

To get the DOJ and FBI to move on on this takes a lot more than submitting paperwork, costs a lot more too.

Suing is a lucrative business in America, so much so that there are entire businesses dedicated to it. It doesn't have to actually protect anything.

I still buy paper books and audit books even with Anna's Archive.

Reply on news.ycombinator.com