Why Amazon isn’t going to plug the shit volcano, and the one revolution that will fix the problem forever

by Suw on February 6, 2014

Mike Cane wrote a blog post in response to Chuck Wendig’s and mine, saying that he thinks the self-publishing shit volcano will come to an end, because Amazon will end it. I left a comment on Cane’s blog, but it was starting to get longer than his initial blog post and I had more to say, so I’m expanding upon it here.

Cane’s thesis is that Amazon will act to remove bad ebooks that don’t sell because all that crap clogs up their site and is bad for business. He thinks that there will come a time where Amazon feels the pain so removes poor quality books and ban further submissions from terrible authors.

I wish he were right, but I don’t think Amazon will do anything within the foreseeable future. There is one circumstance which might fix this whole problem.

So, first, why won’t Amazon act?

Amazon is not a rational actor

At least, not in any way that you or I might consider rational. It’s pretty much the only company I can think of that can consistently not make a profit and not be punished by Wall Street. In the past, we’ve seen that it only takes action when it is cornered, and then it takes the smallest action it can get away with.

Take the bestiality/rape/incest/pseudoincest furore of last year. Amazon only acted when it felt cornered, and even then it did as little as it could get away with. There’s still plenty of dodgy porn on Amazon and will continue to be, because Amazon has no interest in really properly clearing it up.

Same with the sockpuppet review affair. And when Amazon did take action, it was to put in place stupid and ill-considered rules about whether Kindle authors could review or not. It has done nothing substantial about improving the quality of reviews, even though that would be something that you’d think would affect their bottom line quite significantly. After all, if you can’t trust the reviews on Amazon, how do you know whether to buy or not?

So at the moment, there is no force pushing Amazon to act, nothing making it whip out the banhammer. Yes, the shite clogs up Amazon’s arteries, but they have shown no interest in dealing with shite in other areas of their business, because clearly having heart disease isn’t producing any painful symptoms for them. Yet.

Amazon does make money out of bad books 

50 Shades of Grey. Not a masterpiece of literature, but it tapped into a market desperate for soft porn, did well, then broke out of that niche to became a cultural touchstone, bought not because it is good but because everyone wanted to know what the fuss was all about. Other areas of shitty writing, niche erotica in particular, do well again because people want stuff that the traditional publishers won’t touch with a bargepole.

So there is no 1:1 correlation between shitty self-published books and sales. The idea that self-publishing is a meritocracy where the good writing naturally floats to the top is at best a happy fairytale and at worst a delusion. If Amazon can make money out of monster porn without getting slapped about by the law, it will.

Storage is cheap and getting cheaper

Amazon has turned cloud storage into a business, and book files are small, so there’s no real reason for them to worry about how much space the long tale of self-published dross is taking up.

If your average ebook file takes up 500kb, then you can fit 2,147,483 in a single terabyte. Amazon charges $0.010 per gb per month for its “Glacier” storage. So if you’re hiring Amazon’s cloud directly, you can store 2097 averagely-sized files for a month for a cent. You could store 5 million books for just $2384 per month, which is certainly more than it actually costs Amazon, because they obviously mark up their commercial cloud storage offerings.

It is undoubtedly cheaper for Amazon to just store all ebooks uploaded than it is for them to pay someone to figure out how best to get rid of the ones that don’t sell AND are badly written, and then deal with the resultant backlash from offended authors.

That offended backlash

If there’s one thing Amazon isn’t interested in, it’s alienating hundreds of thousands of self-published authors. A few hundred noisy gasbags it can, and does, ignore. (Including the ones in the press.) But if you consider that most books don’t sell, and there is probably more than half a million self-published ebooks getting uploaded each year and growing, that’s a lot of shit and a lot of angry authors they’d have to deal with.

Whether there would be enough angry authors to hurt Amazon’s overall sales in any meaningful manner is something I couldn’t say. But it’s certainly enough to hurt Amazon’s brand (even more than they do themselves – they don’t seem to give a crap about brand), and hurt ebook and possibly paper book sales. Not to mention the deluge of angry email that would cripple their customer support department.

So whilst I would love Amazon to take a long, hard look at their self-publishing platform, I have absolutely no confidence that they will, because I cannot see any motivator big enough to push them to action.

What might change the calculation?

There is one thing that might change all this, and when it comes online it will revolutionise the book industry in ways we cannot even imagine.

Artificial Intelligence.

When we have meaningful AI, not necessarily all the way to full consciousness, but computers sophisticated enough to be able to learn to read and be programmed to develop a reliable taste, then the whole game changes. Everything. Amazon’s pathetic recommendation engine, which is the most overrated algorithm on the planet, will become utterly irrelevant. So will reader reviews. Because when we have a computer capable of reading a book and accurately scoring it for grammar, punctuation, plot, character development, style and genre, then we have a chance to be able to sift out the good from the bad.

Of course, then the question becomes, what do we mean by ‘accurately’? Or ‘good’? Whose standards will be used to draw the lines?

If past experience with technology is anything to go by, as soon as we have AI capable of doing this, we’ll have multiple interpretations of what ‘good’ is, and suddenly all books will become discoverable. Love monster porn? But really, really love velociraptor porn? AI will be able to scan the whole corpus and give you the very best in small dinosaur erotica. Want to read books that are just like Agatha Christie’s? Easy. Want to set your standards to embrace only the most obscure literary fiction? Piffle. Here I am, brain the size of a planet and you ask me to find you some literary fiction.

When we have AI, Amazon stops being the canonical catalogue of all books on the planet. Reader reviews become irrelevant. Sockpuppetry becomes impossible. Only quality – defined however the reader wants – matters.

Is this what Google is attempting with its mass book digitisation program? In 2005, Google played down that exact rumour. Last month, nearly ten years later, Google acquired Mind Deep, an artificial intelligence company based in London. I think we can all draw our own conclusions from that.

TheBrutalKremlin February 6, 2014 at 1:21 pm

Funny, how the so-called “publishing industry” is more concerned with “bad ebooks” and what Cramazon may or may not be doing – rather than selling books. Most have no clue about retailing, and sit back relying on the same type of databases being talked about here.

If Amazon might ‘clean up’ the ‘bad ebooks’ who’s going to do it for the rest of the ‘industry’.

Too bad nobody reads anymore, unless you’re a two-bit ‘celebrity’ with a tell all.

The wagons in the circle are rotting very quickly.

Suw February 6, 2014 at 1:40 pm

Hello, BK! Hm, interesting opener to your comment. Do I sense a little hostility?

The publishing industry is very concerned with selling books, as it’s somewhat essential to their survival. Though I am not an industry insider, I hear an awful lot from publishers, writers, agents and the like about selling books. Retail is indeed challenging, and Amazon is a part of why that is currently so, but you can rest assured that it’s an issue many, many people are working very hard on.

As for the crap that traditional publishers put out, yes, they do undoubtedly publish some shit. However, they don’t reach anywhere near the depths of shititude that self-publishing can manage. Seriously, the odd typo here and there and a sub-par plot and characterisation has nothing on the utter drivel to be found in Amazon.

I love the ‘nobody reads anymore’ trope. People do read, and they read a lot, even though the competition for their attention is fiercer than ever. In 2011/12:

“Nearly two thirds of adults read for pleasure in 2011/12 (64.8%). Of these people,
over three quarters read for pleasure at least once a week (78.5%), and a further
12.3% read for pleasure less often than once a week but at least once a month. Three
per cent only read once or twice in the last 12 months.

“Nearly a half of all adults had bought a novel, book of stories, poetry or plays for
themselves in the last 12 months (47.3%). Of these people, one in ten had bought a
novel at least once a week, with a further third of people buying a novel at least once
a month. Nearly one in five adults only bought a novel once or twice in the last 12

Do the maths and it turns out that half of all respondents read for pleasure at least once a week.

Source: Taking Part 2011/12 Adult and Child Report, from the DCMS.

Oh, damn those pesky facts.

Mike Cane February 6, 2014 at 2:49 pm

AI? I think that’s asking for far more than human intervention.

One point: I never made any distinction between “bad” and “good” books. I went straight to the bottom line with *non-selling* books. For all I know, there could be gems in the Kindle Store that haven’t sold more than 20 copies. But quality doesn’t matter in my scenario, just the bottom line. I generally don’t criticize what people read because as long as they’re reading, it’s all fine to me.

Second point: Yes, it’s cheap just to “store” files. But Amazon also has to process Kindle Books in terms of accounting, for payments to writers as well as internally for tax purposes. I don’t know what that overhead entails, but for Kindle Books that sell zero, they clearly cost Amazon more money than just storing a file on a server.

Suw February 6, 2014 at 3:10 pm

AI is already here, and AI capable of reading a book and drawing conclusions on quality is definitely on its way. Google has already got a neural network that, when fed YouTube videos, learnt to recognise cats without any steering from a human source. That’s not an insignificant feat. Google is doing buckets of research in AI and machine learning, and both Google and Facebook (and probably a lot of other companies) are using basic AI now. Google also has a massive corpus on which to train AI, and is almost certainly doing something with it. Much more powerful AI will become reality, and I’d bet it’s within the next decade.

Fair point about your not making a distinction between good and bad books. However, that doesn’t really matter, because it doesn’t affect any of the arguments about whether Amazon is going to have to tackle the issue or not. Whether you look at ‘bad’ books or ‘non-selling’ books, there’s still no force pushing Amazon to remove them. Much easier to let them just sit there.

As for the accounting, well, if there are no sales, there’s nothing to account for. And when there is something to account for it’s automated. Again, easier to throw a bit more computing capacity at it than it is to change it. I would guess that the cost of accounting for non-selling books is negligible, and less than it would cost to do something about locating and removing them. The Kindle store has been around a while, and if non-selling books were a serious problem, you can bet they would have already added a clause into the service agreement to the effect that any book with no sales after 1 years gets automatically removed.

So I’m afraid I still don’t see what would push Amazon into taking action. There’s no reason to believe non-selling books cause them financial issues, so the likelihood is that they’ll do nothing.

Mike Cane February 6, 2014 at 8:27 pm

Interesting that you cite Google Then you would expect AI in Google Play Books to someday surpass Amazon at least in recommendations? But would anybody care?

Suw February 6, 2014 at 8:32 pm

I can’t see any reason why Google wouldn’t eventually be able to create an AI that was the best book filter evah!, as they’ve got the corpus and the (currently nascent) technology. Amazon could probably do it too if they really wanted to, but I don’t see the same level of technical innovation coming out of Amazon, so on current evidence I’m skeptical that they could manage it. Equally, it could come from some start-up that doesn’t exist yet. But it will come. It’s a bit too obvious of a target to be ignored.

Suw February 6, 2014 at 9:35 pm

NOTE TO COMMENTERS: Comments will not now be moderated until Friday morning at the earliest, possibly later, depending.

Actually, I’m turning comments off. This weekend is going to be a really tough one for me for reasons which will become clear next week. I don’t have the emotional bandwidth to deal with the discussion, but I’d rather not just ignore the comments. If you want to continue the discussion, there are plenty of other blog posts where you can expound to your heart’s delight.

Jim May 11, 2014 at 4:24 pm

According to the following only 1.5 percent of customers, or 15 out of 1,000 actually bother to write reviews, and they tend to on be niche items, so are not representative:


I think most online customer reviews are fake. I hope Suw is right about AI book reviews.

Comments on this entry are closed.

Previous post:

Next post: