This entry in the Platform Studies series (which I’ve been enjoying quite a bit, mostly for its retro tech features) focuses less on the Kindle technology, which in itself is unremarkable, and more on the marketing of the product and how it’s affected readers, reading, publishing, etc. In this regard, I find the Kindle a bit like the iPod: the technology was already there, but someone had to get the experience right.

I was a Kindle early adopter, and wrote about buying it and my early experiences, and later about the underused potential of the Kindle DX (such as for making textbooks usable) and my disappointment with the Kindle Fire which I didn’t realize at the time was primarily going to end up as a low-cost game tablet for the gounger set.

The book correctly observes that especially the early models seemed an awkward compromise between a “book reading experience” (easy to hold in one hand, turn pages, flip, etc.) and a “computing experience” (enter notes using a keyboard, create bookmarks, notion of “Location in book” rather than page or chapter numbers, etc.) and discusses a bit of the evolution of the UX.

From the beginning, the Kindle benefited from Amazon’s unusual collection of diverse expertise: cloud computing/networking, retail, huge scale, and data mining. Amazon’s size and publisher clout meant it could encourage publishers to digitize books, and often dictate terms; Kindle was the only e-reader to launch with a sizable library of desirable titles (c. 88,000). Similar to what happens with bookstores, Amazon’s huge catalog (>65M book titles) props up “loss leader” discounting practices, and customers in that long tail can be targeted by data mining.

The author notes the success of AWS as an example of how, rather than keeping a particular innovation for internal use only, Amazon has succeeded by virtualizing its innovations and infrastructure and reselling them: AWS resells its well engineered cloud services, “fulfillment by Amazon” lets third parties sell directly to consumers (rather than locking consumers to Amazon’s retail channel) while enabling those third parties to take advantage of Amazon’s remarkable fulfillment infrastructure. Similarly, although Amazon was by far the largest print book retailer by 2007, when the Kindle launched without any serious competitors, it had the chance to become the infrastructure for digital book distribution—offsetting any potential losses from selling fewer print books.

The combination of features, and lack of oversight since everything is automated, has led to some strange problems. For example, third-party retailers may sell an item that Amazon also sells directly, charging more for the item [suggesting the retailer doesn’t stock it but will obtain it at time of purchase] or getting metadata wrong. And CreateSpace (Amazon’s print-on-demand service) had lots of fake books by real authors whose SS# or other identifying info had been obtained by fraudsters, at least enough to link to that author “officially” on Amazon.

Kindle changed the way books are marketed and sold. Rather than hierarchies of categories, Amazon relies on data-mining the relationships among titles and among purchasers’ buying patterns (“Frequently bought together”, “You might also like”, etc.), though it “bootstrapped” its catalog from the Bowker Books In Print catalog in

1. In contrast, libraries and other channels have relied on such traditionally authoritative sources regarding metadata and discoverability. And because the data mining done in the Kindle Store (including reviews, downloads, etc.) is generally not made available to publishers, it establishes a positive feedback loop in which editions designed for Kindle Store distribution are more successful than those produced primarily for print distribution.

To keep control over the ecosystem, Amazon deliberately obfuscates and rapidly evolves the file format (which started as a Mobi/Epub derivative but has lots of Amazon-specific things now), and often disables “advanced” features supported by the format in order to avoid bugs or other undesirable behavior that would need oversight, such as SVG graphics that allow overlays. In short, the formats are designed to meet the needs of publishers and distributors, not readers.

Technical problems such as Unicode characters have been solved by having each ebook carry its own dictionary containing the information and glyphs to render only those UTF-8 characters actually used in that book. Still, the challenges of getting publishers to embrace and adopt ebooks were far harder than the technical problems. The speed of technology development and the publishing tools pipeline for ebooks exists in a different universe than the process of traditional publishers (as my co-author Dave Patterson and I found out when we chose to self-publish our textbook).