Post

Anthropic hired the former head of Google Books to oversee its secret "Project Panama," new court docs show — quietly buying millions of used books in bulk, breaking their spines and scanning them to feed into its Claude chatbot. wapo.st/4rjXAMQ

I mean the point isn't *the physical media,* unless you're talking about a rare book. If there are many copies, each one doesn't matter. This sounds like fetishism, like you care about "books" because they are paper, but not about the information they convey

that's not true though? archive.org will very often destructively scan things, especially when they are building the OCR friendly edition for screen readers and they absolutely need the higher quality scans. they stream it, it's fun background noise

Would love your source on this. I’ve been to the IA and seen the tools they use. Their core device is the Scribe Workstation, a cradle with suspended cameras that shoot each page (digitization.archive.org/digitization/). It may be slower than other methods but it is indeed the method they use.

digitization.archive.orgDigitization » Internet Archive Digitization ServicesLearn how Internet Archive digitizes collections with non-destructive scanning, metadata integration, and global access through Archive.org's open platform.

It's possible, but slower, less accurate, more expensive, and only really useful for rare books where the physical item should be preserved. It's not necessary for every mass market paperback copy languishing in warehouses for years on end.

This a much more measured and accurate post than your previous one. Digitization is many things, and IA has shown how to do it at scale without distraction. In fact, Better World Books, who supplied many of the books to the Panama Project is supplied by books scanned at IA.

physical books are not scarce. there is no reason to 'innovate' non destructive scanning for digital text uploading, that also costs more. eg imagine anthropic instead spent more money and then had a warehouse of surplus books that it wasn't economical to resell. why would that be better