Draft2Digital and Barnes & Noble Changes
A few people asked what I thought about recent changes Barnes & Noble and Draft2Digital made for self-publishers.
If you’re outside of indie author world, the gist is that Barnes & Noble was putting new restrictions on indie author accounts. Draft2Digital was now charging a $20 fee for anyone opening an account, and an annual $12 free on any accounts that make less than $100 a year on their platform. There was a good deal of online consternation about this, especially among indies with only one or two books who might not make $100 a year from them.
So, my opinion is threefold.
1.) It’s unfortunate they had to do this, and I suspect they were forced into it due to circumstances we’ll discuss below.
2.) It was inevitable something like this was coming along, because:
3.) The primary reason was the flood of AI generated slop from scammers.
The ultimate source of the problem, as is the ultimate source of many recent problems, is generative AI. A small number of scammers are generating enormous quantities of AI-generated slop books and uploading them to the platforms. We’re talking like tens of thousands of books a month or even a week. And the books are absolutely low-effort as well – AI generated gibberish text, AI generated cover, and then thrown onto the store.
This has always been a problem in self-publishing (especially with Kindle Unlimited), but AI takes it to an industrial scale. With some basic LLM knowledge, you can automate the entire thing.
The figures I’ve heard are that something like 70% of new submissions to Draft2Digital in the last year have been AI generated slop books of that nature.
Obviously, this is not viable in the long-term and an existential threat to the platforms, so something had to be done. Amazon already took some steps in that direction by limiting accounts to only three uploads a day, so it was inevitable that the other platforms would have to follow suit.
The best way to reduce scams is to increase friction. That’s why it’s sensible to lock your front door and your car. If someone really wants to break into your house or car that’s not going to stop them, but it will deter lots of casual thieves (or junkies) who are strolling around looking for low-risk things to steal. In the same way, these changes won’t stop the problem of AI generated slop books, but it will help make the problem more manageable by increasing friction.
So it’s unfortunate that these changes had to happen, but I suspect it was inevitable. I also suspect that eventually we’re going to end up with a per-book publishing fee. $10 per title would severely reduce the financial viability of these scams, but perhaps we’ll be fortunate enough that the era of free and low-cost generative AI is going to collapse before that happens. The economic signs are increasingly pointing in that direction, and scamming with AI becomes a lot less attractive when you’re spending thousands of dollars a week on tokens.
Ultimately, in my opinion the primary culprits for this situation are the AI companies, who have very recklessly and irresponsibly pushed this highly destructive and often useless technology out of a combination of messianic hubris and old-fashioned greed disguised as self-righteous altruism, much like the crypto and NFT advocates before them. I’ve said before that I think the primary problem with LLM-based AI is that it comes with a whole lot of negative results and virtually no positive ones.
The changes at Draft2Digital and Barnes & Noble are yet another example of AI creating negative outcomes and no positive results.
But thankfully it is not all gloom and doom. It is heartening to see how increasingly unpopular AI is becoming with the general public, with shutting down data center construction projects becoming a hot issue in local US politics. I think the best outcome for the entire mess of LLM technology would for using it to become as socially unacceptable as, for example, smoking in front of small children or placing bets on a dogfighting ring.
-JM