Andreas,

> What I'm curious about here is: Has anyone ever seen a (hopefully even
> usable ;) approach to model "environmental assumptions"? Are there any
> common rules/guidelines on which "environment tests" could be based?
> Anything else in this area?

  I didn't "go deep" at all, but Prof. Tokoro's (Sony's CTO's) the
Computational Field Model supposedly addresses similar problem.

http://citeseer.nj.nec.com/169461.html

  And yes, all I did was listened to his talk long before and
browsed the paper now, so I don't have too much to say...

-- Yoshiki