True, but this is just a different sort of approximation, too. Layouts where all the elements are absolutely positioned don't work in the real world, so the markup and css will have to be 60%+ rewritten and will likely not look exactly as it's laid out here.
The intent at this point is to not generate layout html/css. Humans are much better at this right now than the editor could be. And the assumption is that the programmer will want to write the layout code himself anyway, being careful to place it in nice modules that fit with their dev environment.
We could build an entire product just trying to generate semantic, reflowable markup from a design.
I agree. Let's be honest - it's kind of weird that we rely on a program that is really designed for editing photographs as our main tool for designing web sites. I'm all for exploring this further at least as a tool for doing layouts and design that CSS can handle. For everything else (e.g. designing icons), there are more appropriate tools.
It would be nice to have something that allowed you to use both. There can be so much rapid experimentation in photoshop because we know the tool so well - having that import easily into something like this for the final touchup would be ideal (in my most perfect world) as so much time goes into the touchup and the last few pixels in css.
