1-3 - Majorly flawed
4-6 - Only try it if I like the genre (typically FPS)
7-9 - Good game, will keep an eye out in sales
10 - Outstanding and should probably try even if it takes me out of my comfort zone (FPS/Racing/RPG)
Sadly all the **** magazines and games journalists seem to have brain washed all the kids in to accepting a 4 point system of 7 as average/bad, 8 as good/average, 9 as excellent and 10 the best.
Indeed. Basically anything under 70% is deemed to be mediocre at best, and anything under 50% a plain bad game. It's kind of a self-perpetuating situation however, because games are now judged effectively against other games and then that is used to derive the score. So if game X is average and got 70-80% and game Y is also average better it will get something in a similar range.
Someone mentioned Eurogamer earlier and to be fair I see them use a wider range. They'll (sometimes) give a genuinely bad game 2-3/10 and a genuinely good game 7/10 but reading the review you can tell it's a good game rather than going "oh 7/10 must be pretty average".
I play a little metagame on the rare occasions I read reviews, whereby based on what I've read I try to predict the score it will get at the end. Oftentimes it works thusly:
AAA game, positive review = low 90s
Sleeper hit / relatively unknown studio, positive review = high 80s
Any game with understated praise = low-80s
AAA game, mediocore review = mid-70s
Relatively unknown game/studio, mediocre review = 50-65
etc