True, I guess my perspective is biased towards CS. In that field, what I've seen in terms of editing on the part of a publisher amounted to requests for minor formatting changes (to be done by the authors) and insertion of a copyright notice. I have never heard of any publisher-side "first pass screening" of submissions to conferences I was involved in or journals I know the editors of. There, desk rejects are done by the committee/editors (and personally I would be very nervous having a non-expert do this job). To the best of my knowledge, in CS, there is nothing being done that comes close to justifying, say, the open access publication fees at their current level.
