I'd put it to you that there are a number of
in silico A-life experiments which have shown that imperfectly replicating code in a selective environment can evolve, and these are clearly distinct from the nucleic acid based imperfect replicators which from the basis of our own evolution.
Is there actually any argument you wish to provide as to why a population of imperfect replicators in a selective environment wouldn't evolve?
The fact that parasomnium species 'imperfect' replicators implies that there are also 'perfect' replicators a population of which would not undergoe evolution. As soon as 'imperfect' replication occurs there is evolution in a strict sense, although not the sort of adaptive evolution we usually talk about.
The only reasonable objection you raise is that parasomnium didn't include extinction as an outcome and that point is trivial at best and certainly doesn't counter Parasomnium's main points.
Additions I think might have been more useful would be a caveat to the effect that the increasing complexity need not be a one way process and that it is not neccessarily a continuing process. No organisms really needs to become more complex but starting of at a wall of very minimal complexity means that a randomised drunkards walk will always initially tend to take you away from the wall and towards greater complexity (Steven Jay Gould made this point excellently in his article
'THE EVOLUTION OF LIFE ON EARTH').
TTFN,
WK