In my case, were I reviewing, I'd be trying to keep the marks on more-or-less a normal distribution, with the vast majority coming between about 4 and 7 and major dropoffs on both sides. Doing that would suggest that ten out of ten should go to something a couple of standard deviations better than average - not necessarily 'perfect', but significantly better than just top ten percent...
Edge magazine seems to do something similar - because they've given out 10/10 scores to a total of twelve games in the last seventeen years. Which seems pretty reasonable to me.
no subject
no subject
no subject
no subject
no subject