One of our big concerns is marketability.  Having a purpose-built device
and cradle is no more efficient in terms of page scanning than one
assembled from stock parts (in fact, the cube is surely slower than the
Decapod, per page, since one has to lift the cube).  But in terms of
what people will actually buy, there's a big difference.  If it looks
elegant and well-suited to the task, it becomes more marketable.  So
we're planning to manufacture both the cubes and the cradles specially.

(It's the same reason car manufacturers know that the most aerodynamic
design for a car doesn't sell -- the nose is low-slung and the whole car
looks like it's tilted forward, pointing down into the road a little
bit.  Terrific for mileage, but most people don't like the way it looks,
and that's why almost no company sells a car like that.)