Slashdot Mirror
Amazon Has Everything it Needs To Make Massively Popular Algorithm-Driven Fiction (qz.com)
Thu-Huong Ha, writing for Quartz: Amazon's power in books extends way beyond its ability to sell them super cheap and super fast. This year, a little over 40% of the print books sold in the US moved through the site, according to estimates from Bookstat, which tracks US online book retail. (NPD, which tracks 85% of US trade print sales, declined to provide data broken out by retailer.) In the US, Amazon dominates ebook sales and hosts hundreds of thousands of self-published ebooks on its platforms, many exclusively. It looms over the audiobook scene, in retail as well as production, and is one of the biggest marketplaces for used books in the US. Amazon also makes its own books -- more than 1,500 last year.
All that power comes with great data, which Amazon's publishing arm is well positioned to exploit in the interest of making books tailored exactly to what people want -- down to which page characters should meet on or how many lines of dialogue they should exchange. Though Amazon declined to comment specifically on whether it uses data to shape or determine the content of its own books, the company acknowledged that authors are recruited for their past sales (as is common in traditional publishing). "Amazon Publishing titles are thoughtfully acquired by our team -- made up of publishing-industry veterans and long-time Amazonians -- with many factors taken into consideration," says Amazon Publishing publisher Mikyla Bruder, "including the acquiring editor's enthusiasm, the strength of the story, quality of the writing, editorial fit for our list, and author backlist/comparable titles' sales track."
Amazon's Kindle e-reader, first released in 2007, is a data-collection device that doubles as reading material. Kindle knows the minutiae of how people read: what they highlight, the fonts they prefer, where in a book they lose interest, what kind of books they finish quickly, and which books gets skimmed rather than read all the way through. A year after the Kindle came out, Amazon acquired Audible. Audiobooks have been a rare bright spot in the publishing industry, with double-digit growth in total sales for the past few years. Audible now touts itself as the "world's largest seller and producer of downloadable audiobooks and other spoken-word entertainment," and its site has around 450,000 audio programs.
All that power comes with great data, which Amazon's publishing arm is well positioned to exploit in the interest of making books tailored exactly to what people want -- down to which page characters should meet on or how many lines of dialogue they should exchange. Though Amazon declined to comment specifically on whether it uses data to shape or determine the content of its own books, the company acknowledged that authors are recruited for their past sales (as is common in traditional publishing). "Amazon Publishing titles are thoughtfully acquired by our team -- made up of publishing-industry veterans and long-time Amazonians -- with many factors taken into consideration," says Amazon Publishing publisher Mikyla Bruder, "including the acquiring editor's enthusiasm, the strength of the story, quality of the writing, editorial fit for our list, and author backlist/comparable titles' sales track."
Amazon's Kindle e-reader, first released in 2007, is a data-collection device that doubles as reading material. Kindle knows the minutiae of how people read: what they highlight, the fonts they prefer, where in a book they lose interest, what kind of books they finish quickly, and which books gets skimmed rather than read all the way through. A year after the Kindle came out, Amazon acquired Audible. Audiobooks have been a rare bright spot in the publishing industry, with double-digit growth in total sales for the past few years. Audible now touts itself as the "world's largest seller and producer of downloadable audiobooks and other spoken-word entertainment," and its site has around 450,000 audio programs.
Yay, now books can become as predictable as Hallmark movies.
They can tell things like when people stop reading in their Kindle, but they don't have really great data to determine why the person stopped reading. They may be able to tell how long popular dialogues are, but an interesting dialogue is determined by much more than length. In fact you might say that length is one of the least important factors.
So great, they have tons of metrics, but they don't have the ability to extract the metrics they need to make a book interesting. And that is entirely the problem.
"First they came for the slanderers and i said nothing."
I believe it when I see it. Sure, companies do collect a whole truckload of data all the time and AI has made some huge progress in the last 10 years, but in terms of end user features I have seen pretty much zero of that ever being used to create something useful.
If they wanted to they could run face detection and AI on every movie out there and tell you exactly how long each character is on screen, with which characters they share a scene, what they are doing in the scene and so on. But they don't. We are still stuck with recommendation algorithms that are complete junk and provide no explorability.
In their store they can't even manage to group related items together. Want to buy a book or movie from a series? Better know what you are looking for already, since they down show you which are part of the series or in what order they go. A trip to Wikipedia is often needed to figure things out.
For most of the items they sell they can't even bother to scan the backside.
There is a lot of cool stuff one could do with all that collected data, but I have yet to see somebody actually do something with it.