I originally had a huge footnote acknowledging that much of this is obvious and that the Kindle team is full of smart people who have data showing that maybe these aren't the best optimizations. I'm just describing my expectations from having used Kindles for ten (??!) years coupled with having worked on designing digital reading interfaces for a similar amount of time. I'm certainly an edge case user.
Having seen my mother (edit: and, more pertinently: myself) struggle with her Kindle, I do think the interaction model is one-degree unnecessarily complex — all of those hidden spaces.
You could distill my whole piece to: This is a single purpose device that's been designed like a multi-purpose device.
Page turn buttons + menu button would all but eliminate any interaction ambiguity. And could be done quite beautifully.
Although I don't have access to any data, I think there might be a subset of casual users who would prefer the screen to do nothing when touched.