I'm not sure if ioi has the storage or processing power requirements to make this work (it shouldn't be too much as long as the userbase doesn't become too large, and if the data is always being processed on another thread ) ...
Everyone is biased and it is best to consider their game score in relation to their voting history as it relates both to genre and platform so when someone adds a review score they have to be grouped together as a genre-platform review set ... This review set should then have certain statistics calculated (mean, standard deviation) and the review score for every game should then be changed to how many standard deviations from the mean is this game for this person.
Every so often (whenever the thread that processes the game database gets around to it) you update the game's review score by averaging out the number of standard-deviations from the mean and converting this data to a standard voting scale ... An example voting scale is below
10: +4 SD
9.5: +3 SD
9.0: +2 SD
8.5: +1 SD
7.5: Mean
6.5: - 1 SD
5.0: - 2 SD
3.0: - 3 SD
0.0: - 4 SD
As you can see, the review scale is heavily tilted but that is because I'm making the assumption that everyone is rational and is far more likely to be playing games that are high quality than games that are lower quality
