There's no legitimate purpose for destroying these books.

There are non-invasive ways scan books. But the destruction is an added bonus. These companies make a copy of these public domain books and then pulp the remains. They're rare, so that means they're hard to find (that's what rare means, after all).

Then, they can turn around and sell their dataset to AI companies.

The destruction is the point.

Capitalism always creates perverse incentives. The most profitable thing you can do is seize something readily available and make it impossible to get without paying you.

But which companies are doing this? It sounds like these are companies that supply training data to AI companies. I think we need names.