I’m not a big fan of numbered reviews.
For one, the numerical scale is arbitrary. A 94/100 doesn’t differentiate much from a 9/10 or a 4/5. Yes, those are technically different (I do understand math, after all), but the increased granularity of a percentile or base ten model doesn’t change much from the basic five point model. It’s still just variations on “really like, sorta like, neutral, sorta dislike, really dislike” spectrum, but those are statements of fuzzy opinion. We can’t have that when Metacritic is waiting in the wings. (And insanely, some game studios tie bonuses and even salaries to Metacritic scores. This industry is so messed up…)
Two, the numbers tend to be inflated. A normal distribution of scores would suggest a bulge in the middle of the data set, but game review scores tend to exhibit this data bulge around the three quarters mark. A game that scores in the middle of a number scale, a 5 or 50 (or bizarrely, a 2.5 in a five star system… why bother with halves?), will be seen as a failure, when it is precisely in the middle of the possible scores. Maybe that means the game gets a resounding “meh”, but that’s neither a “retch” nor a “rapture”. Some may well like the game, despite its warts, while others can’t get past the small bra size.
Again, it often comes back to the simple “like-dislike” scale, but since the numbers are arbitrary and weirdly shifted, they don’t correlate well to a simple scale like that.
More than that, though, there is also the concern of personal preference. I happen to really like Cogs, a brilliant little sliding tile puzzler I picked up from Steam for $2 during the Christmas blitz. Its Metacritic score of 73 puts it squarely in “meh” territory in the typical scoring of games. Braid, a game that seemed poised to bring about the second golden age of gaming simply by its mere existence, has a nice 90 score on Metacritic, but it gets a resounding “meh” from me. Grand Theft Auto IV and Halo proudly wear their “Universal Acclaim” badge bestowed by Metacritic (complete with breathless fanboy reviews flooding the internet, tallying to 98 and 97 scores, respectively), but I can’t work up more than a resounding “retch” for either of them. Maybe they mean “universal” the same way that Miss Universe is always an Earth human, or the World Series is pretty much just a United States thing.
Far more useful to me are well-written reviews that clearly state opinions and facts about a game. I may not agree with them, but if they are well-written and clearly reasoned, I can get a fair bead on what the game offers, especially if I read a spread of favorable to unfavorable reviews and parse the commonalities. I wrote about this a little bit when I suggested a Punnet Square of difficulty, and how it colored player reactions to Valkyrie Profile: Covenant of the Plume (a game I really liked).
Even some “real” journalists don’t seem to like numbers, but wind up using them anyway. What good is a system that doesn’t accurately convey information, and that the users don’t trust?
So, henceforth, if I’m going to be pinned down to giving a particular game a “rating”, I will be using the following system:
Games I really like will be given the rating of “Fudge”
Games I sorta like will be given the rating of “Peach”
Games I am ambivalent toward will be given the rating of “Celery”
Games I sorta dislike will be given the rating of “Onion”
Games I really dislike will be given the rating of “Chitlins”
If you don’t like my particular choices, well, perhaps it’s just a matter of taste after all.