Speaking of the Wired iPad app, Layton Duncan has a good analysis of how it’s built:
At its core the app is a simple XML driven layout engine. Content
is split into “stacks”, each horizontal swipe in the app loads
a new stack. Each stack has associated meta data such as title,
description, tags etc. Each stack has an ordered list of
“assets” which are essentially 1024 × 768 pixel static images
of the pages which are displayed while browsing.
That explains why text isn’t selectable. (I wonder if they could switch to PDF for the stacks?)