In a purely qualitative way, that should be easy, if you succeed in staying naive-cold with the UDA up to step 7. Imagine that I decide to copy a piece of matter.
Unlike information, where things are crisp at some point, it is already not clear what is the relevant level, so an exact copy should be defined by something like a non distinguishability with respect to some set of instruments.
Anyway, at some point, in your zooming toward finer and finer description of the piece of matter, you arrive at your own substitution level. At that level, the matter is no more made of subpart, but is undetermined, as you comp state is no more dependent of such details, and *you* diffuse on all the possible "subcomputations", where, by the FPI, all universal machines are somehow in competition (by the invariance of the 1p for the "length of the proof of the sigma_1 proposition, or computations). How could you clone that? We cannot clone an object, because an object is not a real thing, but an information pattern, which becomes necessarily fuzzy when we look at it below the substitution level. What we can see there is only an average of the many possible computations below our (first person plural) substitution level.
OK?