I totally agree with the first post, but i'm going further. The real problem lies in metacritic.
It uses a simplistic model.
It doesn't take into account Genre, year, platform, reviewer bias, user reviews. From a statistical point of view it's easy to implement a new system with all those features. It only needs reviews to be properly formatted for it. Because sites will never stick to a new standard, the new metacritic should rely exclusively on user reviews.
I have even modelled a system where only introducing some 10 ratings per game, each contributor will be ranked by Contributions (the more reviews the better), Bias (the less their scores deviate from average the better very specially when switching platforms), Consistency (the less "exotic" scores the best) and Fame (other users give you points because they like your reviews) giving a final Contributor Level. This level will be used as a ponderation of it's scores.
So, if you entered 20 review scores, you showed no particular bias, are reviewing in different platforms and different genres, your score will ponderate MUCH MORE than say the review of an occasional fanboy that comes to put 10 to every game in his console and 0 to the other. In fact, those fanboys review scores would be easily not accounted for.
After all this, it could be very easy to browse games by all sorts of scores.
Let's say, i want to know what's the best in gameplay terms on psp this year.
Or when you look scores for a specific game, you get the Genre Score, Platform Score, Gameplay Score, Story Score, Graphics/Art Score, Sound/Music Score, Replay Value Score, Fun Factor Score, etc... Noy only the Global Score.
And while i agree it's not necessary to stick a score to a piece of art, it has it's use in the community, because we all love games, talk about them and we like scores to be fair, not the joke they are nowadays.
Alas my capabilities are limited to stats and not web programming, so i can't create this by myself.
