[bambax]

How much would it cost Amazon to assess the value of each book uploaded to Kindle? Some AI process could be trained on "real" vs "scam" books and attach a score, then humans could review the worst scores and then maybe investigate the authors who post a majority of bad books.

Even after the fact, once a book is generating profits in the Kindle store, one can think of many automated controls:

1. It shouldn't matter that the old Kindles can't tell the pages you've read (just the max page you reached), if the new ones can, because then, based on the data from the newer Kindles, one can detect books where people only read the first and last page (and then maybe investigate those books manually)

2. Even if no Kindle device was able to tell the pages you've read, if it's possible to measure the speed at which a reader reached the last page (date book first opened - date last page read), then by comparing to similar books of similar size it should not be difficult to identify outliers

3. Etc.

My point is, for some reason Amazon doesn't want to police itself, neither in the Kindle store nor in the general store (third-party sellers), which is bad for its brand.

This is puzzling because it's irrational, and one would think Jeff Bezos is more of the hyper-rational type...

So a piece of the puzzle is missing, but I can't see it.

[londons_explore]

Amazon doesn't really care. This is simply one kindle unlimited author stealing from the others.

Until this style of scam steals a big perportion of revenue, enough that real authors stop uploading new books, Amazon has no incentive to do anything about it.

[Bakary]

One explanation could be that he has simply not gotten around to thinking about the problem yet.

[ihsw2]

Or he's fallen into the trap that popularity (and popularity=revenue) is an adequate signal for quality, if third-party sellers are bringing in adequate sales loads then there's nothing to fix.

[Asooka]

This would catch books that are bad on purpose, which is itself a kind of art. Though maybe it can be easily solved by triggerring a manual review once you have 100 of them uplaoded?

[s2g]

> How much would it cost Amazon to assess the value of each book uploaded to Kindle? Some AI process could be trained on "real" vs "scam" books and attach a score, then humans could review the worst scores and then maybe investigate the authors who post a majority of bad books.

or someone could just read them.

[mariocesar]

I was thinking the same, there are known automated strategies to catch copy cats or just bad writing.

However I’m guessing having a big inventory is more valuable than quality. That could be amazon’s cultural motivation to dismiss quality by default

[bambax]

The scam books are not just "bad writing" and usually don't rip off other works; many of them are filled with pure garbage: random letters that form non-words.

But maybe you're right, the size of the inventory is what matters to Amazon who may be stuck in a "growth at all costs" mentality. That would explain it in part.

[scarface74]

That should make it even easier. If X number of words don’t appear in a dictionary, then it’s a scam. Amazon also has a natural language processing platform. If it can’t parse X percentage of sentences, it’s a scam.