[davidfor]

Originally Posted by Amazondoc

Hmmm. That's an interesting point -- but I see some problems with leaping from "timing taps" to "timing book pages".

Remember, ebooks are not necessarily read on Kindle devices -- and even when they are read on Kindles, they are not necessarily read on devices that are dedicated to reading. For example, I read ebooks either on my laptop or on one of my Fires. On the laptop, the Kindle app never "sleeps" as such, and I never have to "tap" at all to advance through a book -- I can scroll continuously. And when I read on a Fire (which I can also scroll instead of tap), my "tap" may be either a tap to advance a page, or it may be a tap to switch to another app, or any of a multitude of other things.

Firstly, my mention of the e-ink ereaders was just an example of the interaction and how it can be done. But, the same does go for every application. Each application knows everything about your interaction with it. It knows when you scroll whether by tapping, using arrow or page keys or the scroll bar. It also knows things like if it has focus or is covered by other applications. So it is just a matter of deciding what interaction is significant, and what a "page view" is. Compared to the latter, the counting is easy.

But "page" count doesn't mean much with ebooks, because that will vary greatly with customized features such as font size, margin size, and so on. You'd have to calculate on something like word count or percentage instead.

An additional problem: Would you report total time to advance to the end of the book, or average time per page, or what? For instance, if I opened a book and left it open on the first page, then went off to do something else for a few hours, then came back and scrolled quickly through the rest of the book, how would that be tallied?

And another problem: If you are trying to count time spent on each and every page (or each and every given percentage point), wouldn't that require a lot of complicated record-keeping? Those files that accompany, for instance, azw-format books (mbp, phl, apnx) are small files. I'm not educated enough to know what's in them, but is there really room there for such granular timekeeping records?

And that's why I wondered if it was a simple average. Just record the number of "pages" and the time spent. Very simple and not much data needed.

But, I don't know much about the Kindles, I do know a lot about Kobos. They do keep a record of these things. They have a table in their database that includes a page turn count for each book that has been opened. And it keeps records of the reading sessions. It counts each time a book is opened or closed and when. And separate counts for starting and finishing books. Luckily for my free time, I don't have any real details on how this data is stored, so I'm not to tempted to try and read it.

It's an interesting puzzle -- and yes, a bit pointless to ponder, given that Amazon is unlikely to tell us what method it actually uses. OTOH, IMHO their very reluctance tells us that their methods are not all that impressive; there would be no need to hide them if they were.

Well, @jhowell has found some details, but whether they publish the full algorithm is another matter. A more detailed version might be available to authors under the contracts they sign. And it is likely they have an non-disclosure clause in the contract.