I prefer to call this "material modeling," because before you put a pixel on the screen, you have to ask yourself how all the components fit together in the sta

I prefer to call this "material modeling," because before you put a pixel on the screen, you have to ask yourself how all the components fit together in the state machine that you are building. The analogy to the physical world breaks down, IMHO. Virtual space is completely fungible, and we aren't ready to give that up (even assuming it were a good idea): almost everything stretches, shrinks, mutates, appears and disappears. I know part of their objective (which I applaud) is move away from this kind of paradigm (or lack of one). But in practice, you have to stop at some point and say "close enough." They say elsewhere, for example, that it's better for transitions to be fast than smooth. It's in that little interval that you can sweep the impedance mismatch (between the designer's model and the user model) under the rug. Anyway, I'm glad that we're talking about the rational basis for one design or another, rather than treating it as an impenetrable art.