The number of "books" being published is growing exponentially. So, a significant fraction of all books ever would have been published in the recent past (which also means that they would probably be "electronic natives" that don't need scanning and digitization). I imagine that the uptick in self-publication opportunity is already (or will become) and important factor in that growth. For these reasons, I don't find it shocking that some online repository has a few percent of all books ever written.

PS: I couldn't find numbers regarding my statements. It would be great if someone can provide sources buttressing or refuting my claims.

The problem with self-publication is that your book might as well not exist[0] unless you already have a significant following (in which case you could easily secure a publisher). This isn't an issue unique to the digital age, there's a term specifically for publishers who will just publish anything: vanity presses (though the catch is that you have to pay them usually, hence the "vanity" part). Publishers provide their name, marketing, and get your books into stores which is pretty important. (I write Wikipedia articles on manga and you can easily tell the North American publishers apart by how hard they get the manga scene to write about and review—via advance copies—their books, especially with regards to press releases and staff interviews. If a manga didn't have this, it became really hard to justify a Wikipedia article for it on the basis of Notability) Also I bet for statistics purposes, only books registered with the Library of Congress are counted (how else would you find all those self-published works?).

There are services like Lulu for free self-publication,[1] but they don't carry the same "legitimacy" or reach of publishers. I think the best analogue to online self-publishing would be the zine: they were easy to reproduce (via photocopying) and distribute (at events or via post). However, none of them really had a long-lasting legacy[2] and anything successful eventually legitimizes itself as a periodical or magazine and becomes established. Thus, I think self-publishing doesn't really change much for the individual, but makes it much easier for groups[3] to gain traction. Overall, being self-published on the internet just increases your accessibility, but we should be careful confusing it with traditional publishing or counting it in statistics because a lot of it is just noise. (bringing up Wikipedia again, there are tons of "books" on Google Books that are actually just random compilations of Wikipedia articles)

Anyway just some of my random thoughts, hope I didn't digress too much.

[0]: Overused thought experiment: "If a tree falls in a forest and no one is around to hear it, does it make a sound?"

[1]: https://www.lulu.com/

[2]: An exception like the Phrack ezine might be of interest to the HN crowd. (https://en.wikipedia.org/wiki/Phrack)

[3]: Here's where the print-web distinction breaks down. A ton of blogs and amateur news websites have evolved and became taken seriously. Just because they're not published in a book format, doesn't mean they're distinct in my opinion.

Some people are all "Self-publishing is great! I get half of the price of every book I sell."

After doing it for a while I would gladly take a much lower percentage of each sale in return for much huger sales, and having people to deal with printing, distribution, publicity, advertising, and all the other parts of the process that aren't "me drawing the next page of comics". And I'm lucky to be working in comics, rather than words, where there's a significant tradition of "underground" publishing that's become legitimatized into "small-press" and "independent" publishing, rather than an epithet like "vanity press".‘

Yeah, you'd think the number would be inflated by e.g. all those much-mocked procedurally-generated "books" on Amazon (Toilet Seat Sales Trends in China and East Asia, 2004-2005, or whatever). Of course, if no human ever reads or writes a book, nor is it ever printed on paper, is it really a book?

How about "machine monograph"?

I thought that number sounded low as well, but it sort of checks out after doing back of the hand math. 130M books / 107B people that ever lived = 0.0012 books/person, 1:1000 book to person ratio.

Checks out. I'm a struggling author who has written approximately 0.0012 books

With Libgen being around, i feel like the problem mentioned in the article partially sorted itself out(assuming the higher quality/use books go to libgen). So we've got the data.

But the layer above, the application layer, could use much more work. It won't happen in the legal realm[1], but i wonder what kind of amazing things we could achieve if we found a way to create a wide and deep developer ecosystem around Libgen, dedicated to help make this data useful?

[1]Altough i think the focus of the legal project is a bit all-or-nothing, and if Google had decided to focus only of subset of books(which it can get legal rights), and create the best applications possible, they could create a ton of value.

There are a lot of duplicates on LibGen...