Ok, so here's the thing. I agree on the general concept that reviews should be read in their full extent, and focusing on the score alone is just worth nothing, but, if a website/magazine chooses to support a scoring system in their review policy (and puts graphical emphasis on it), then that number must be very carefully thought out.
Let's take the last Call of Duty entry, that scored an 8 on Polygon, being reviewed on "debug machines set up by Treyarch". I suppose that, after the game release, someone at Polygon kept playing the game, maybe to check everything was ok in the retail copies, maybe just because he enjoys COD (ah well, I realize now that they promised to update the review in the League Play paragraph, which they didn't, but whatever).
Let's say he tries the PS3 version, and whoa!, nothing works. Lag, disconnections, League Play broken, and all the stuff you surely know about. So what? Go back, give it TWO scores, one for Xbox and PC, and the other one (lowered to 6,5) for PS3?
This thing about updating scores just doesn't work, in my opinion.
Simcity scored 9,5, now sits on an 8, and if the servers get fixed will be probably back on 9,5.
Isn't this confusing? I mean, you, reviewer, consider the server issues a compromising factor, and I agree with you. So score an 8, and in the final thoughts explain why.
OR, give it a 9,5, and in the final thoughts clearly explain that there are servers issues, and update the review (not the score) based on what happens next.
OR, come out with the review and a blank score, explaining you are experiencing server issues, and only ofter that gets fixed you'll come out with a precise score.
A score should never be touched. You touch it, the whole scoring thing becomes a joke. Better to get rid of scores entirely at this point (but yeah, there's that Metacritic thing, right?).