It's a big plus and minus, isn't it, swh!
But provided all the aircraft in an Area have altimeters that are within the accuracy tolerances and set to forecast QNH, the potential variations of actual QNH from forecast QNH will affect all 'nearby' aircraft equally. (I'll track down the BOM's parameters for maximum differences in QHNs as between adjacent areas.) It makes sense, and has done for a long time.
The next questions:
What is the margin of error of a 'blind' Mode C encoder output?
What is the maximum allowable 'delta' between an aircraft's Mode C encoder output and the reading on the aircraft's altimeters (when set to 1013 hPa)?