Wow, thats crazy... but is it useful?

I'm interested to hear if people think so too, want to give it a shot and let us know? (I think it's useful, and that there's also a lot of potential for it to be even more useful)

This sort of thing can be. At scale the trick with this sort of thing is being able to filter out mundane/routine/well-behaved parts in order to focus on the "main" part of the evaluation.