You're not wrong. But there's a tradeoff between creating a cinematic experience (which really does resonate with people in a strong way) and adhering to standard methods. We debated both approaches and went with this one, in-part because that's part of our ethos: interface design is in a local maximum, and we've had success in our software prototypes creating new interface primitives that i.e. improve the reading experience dramatically.

In order for this current design to work smoothly across platforms, devices, and input methods, however, it'll need to be re-implemented, and it was probably a mistake to try something so ambitious in the time we had available.