Not all the time.  Just sometimes.  I buy a game and I think "Damn, that part was pretty cool.  I wonder if anybody else was impressed by that!" and then I go back and read the review.  Or I buy a game and I'm can't get to it for a while.  I may read a review to hype me up as to what's on the horizon.

I'd say that's a very poor example because most games have easily over 50 reviews, so even if a mediocre game gets 2 100s or 2 0s it wouldn't effect the overall average score by more than 1 or 2 points which is nothing.

Also I'm VERY sure that for every magazine that gives out too many 10s there is another one who does the exact opposite.

How does that matter? For every site that gives too many games good reviews, there's another site that rates games too harshly. The idea of a composite reveiw site is that these outliers shouldn't effect the overall score too much. Or at the very least, should influence almost every game equally.

The metascore is weighted average. From the metacritic web page:

"Metascore is a weighted average in that we assign more importance, or weight, to some critics and publications than others, based on their quality and overall stature."

So if review scores from a particular publication (like Gaming Age) are generally unreliable, chances are the publications influence on the overall metascore is low.

Then why do gamerankings always have like 3 times more review scores than metacritic. It seems like they pick and choose their reviews.

GameRankings:-
- Uncharted 2: Among Thieves - 78 reviews
http://www.gamerankings.com/ps3/955125-uncharted-2-among-thieves/index.html

- Halo: Reach - 57 reviews
http://www.gamerankings.com/xbox360/960512-halo-reach/index.html

MetaCritic:-
- Uncharted 2: Among Thieves - 105 reviews
http://www.metacritic.com/game/playstation-3/uncharted-2-among-thieves

- Halo: Reach - 87 reviews
http://www.metacritic.com/game/xbox-360/halo-reach

Metacritic is extremely useful and I usually check it before I buy a game. I don't take the score itself too seriously, but it's a great place to find some good reviews and it least see if anyone shares similar opinions on what's  good or bad about the game.

This is true iff there is a sufficient number of reviews for a game to reach the limit for what the games score should be. Since some games have as low as 5 that would instantly throw it out. Hell at 20 you could still easily have it swayed too far to one side or another. Metarankings would have to include far, far more sites than they currently do in order to achieve the critical number of reviews for every game.

The fact that they reject so many review sites cuts this argument out at the knees though. They attempt to weed out overly high and low scores completely which will always have mixed results.

ehh, disagree... I disagree with a fair amount of review scores being too high and some review scores being too low.

But them and gamesrankings compile links to a bunch of reviews... Also it gives a general perception by reviewers. So going to a singular review... doesn't help more than that... and my favorite review site is gamesradar... which is in-house metacritic reviewer anyways. Now it's not a perfect gauge by anymeans... but it's useful.

The most negative thing, imo it does, though obviously it's an idea
is it causes sites to be too neutral... and scared to give a game a too high or low score... or places like destructoid, which imo likes to put too high/low scores so it sticks out like a sore thumb on metacritic to get hits... it works both ways.

Now if I'm interested in a game, I'll Buy it no matter what the reviews are. (See Resonance of Fate Dreamfall) But if I'm on the edge... I'll check with reviews to see if my concerns were mentioned as problems in the game.

And if a game gets high reviews I won't necessarily buy it if I don't find those games interesting. See halo, and WoW.

As a hub to find multiple reviews its great - but until I know how the weight and judge different reviews I'll take their composite reviews with a pinch of salt.  I actually find looking at the scores in the list and seeing whether it's a case of all good reviews plus a couple going totally the other way vs a broad mix of scores is more useful than the simple composite average itself.