UnSubject has an excellent post on Metacritic. I strongly recommend reading it.
The most interesting thing about Metacritic these days--and what makes it more controversial than all the other review aggregators for other media--is that some game publishers are tying bonuses to the game's Metacritic score.
This seems a little odd to me. UnSubject's explanation is that royalties are tied to Metacritic because “it’s a quantitative measure of game quality, popularity and helps forecast sales”. I don't think this is the full explanation.
For one thing, Metacritic scores are a weak proxy for what publishers really want to learn: how much money did this game make? But publishers don't need a proxy to figure out how much money a game made, they have direct access to the sales figures. It would be more sensible to base bonuses directly off gross revenue, rather than an indirect measurement such as Metacritic.
The only theory I have is that gross revenue or sales can be thought of as a function of both game quality and marketing (and the sheer fickleness of the audience). However, marketing is traditionally the responsibility of the publisher. It seems unfair to pay or withhold bonuses if the publisher did a good or bad job on the marketing.
In theory, reviewers should not be affected by marketing. Metacritic scores should filter out marketing's contribution to gross revenue, and only represent the developer's contribution.
Of course, the fact that Metacritic tends to correlate well with popularity and sales generally means that this extra indirection is unnecessary. It would be interesting to see what games are outliers. Games where the Metacritic score did not predict sales or revenue; either a poorly-rated game selling many units, or a highly-rated game selling fewer units.