I use Metacritic as a convenient place to get a lot of differing views on a game before making my own decision to try it or buy it. It saves time and energy researching a game to determine whether it's worth investing my time and money. Getting a good idea of what a game is like and whether it will appeal to me, without having to just buy it and play it, requires researching several different opinions and assessments of that game. Metacritic provides a convenient one-stop shop for doing that research.
My only real complaint is one that applies to modern reviews in general; a numerical score to the level of precision as we assign to game reviews is total and complete bullshit.
Honestly, there is no way that 1 part in 1000(scores of 1-1000 or 1-100, with a decimal part) or even 1 part in 100(scores of 1-100 or 1-10, with a decimal part) precision score can really validly be assign to complex, creative work like a video game, especially given that it is difficult to impossible to be completely objective in the review. The personal preferences and tastes of the reviewer inevitably will skew the score by an amount much greater than the precision of the score(i.e. big error bars), and it's just hard to really gauge any element of the game with such high precision(again, big error bars). Scientifically speaking, one simply cannot have large precision errors leading into the calculation of a value and then still try to claim high precision with the resultant value itself.
In my opinion, about the best one could do would be a 1 part in 10 precision, which is either scoring 1-10 or 1-5, with half-integral scores. Even this is more than is necessary for the vast majority. It is sufficient to have a 5 point system that gives an overall qualitative feel: 5 = excellent, 4 = good, 3 = mediocre/okay, 2 = bad, 1 = crappy. This is about the level of qualification and precision that a human being can reasonably be expected to deal with for a general overall feel of a game's quality, and then he would use the actual text of the review/critique to obtain more detailed information before rendering a final decision to try, buy, or skip; in fact, I'd say the text is really the most important component of the review for making the decision because it tells you about all the little nuances and idiosyncrasies(or, at least, it should, if the reviewer did he job correctly). If one wanted to give a little more detail to the specific score, you can break it into components, as is often done, where elements such as graphics, controls, sound, gameplay, and replay value may be measured separately. However, the scores on each of these components would still have the same 5-point, qualitative scoring system because it's just hard for a human being to assess creative works with any greater precision.
Also, the score does not reveal whether the content of the game is of a nature that would appeal to a given individual gamer. The game could be of excellent quality, but the gamer still may not buy it because the kind of game and the themes of the content in the game simply do not interest him. There's nothing that can be done about that, and trying to make games that appeal to everyone is precisely the wrong approach(you'll just end up with a game that no one likes). The better approach is to come to know the particular audience of the game and tune and evolve the game towards their particular tastes and interests.
EDIT: Minor edits for clarity.
Aside from the one truth kitchen thingy that made my laugh five times as much as shown above for 100 times longer than the time it took me to write, this was a relatively obvious episode. I still liked it though.
