New super mario bros Wii on metacritic.

[theRepublic]

dunno001 said:

hsrob said:

dunno001 said:
Personally, I don't care what the meta is for it. Metacritic is flawed by design. If you ask me, the best way to "fix" meta is to cut the outliers on both ends. Looking at the above (at the time of this post, there are 5 reviews), 4 of the 5 games give it at least the "average" 85, but the one low point is what drags it down. Instead, cut the outliers out, and you get a better feel for where a game should be. The 70 will be removed for being the low score, and to keep it balanced, the 96 is removed for being the high score. The remaining 3 scores create an average of 89. For games with more reviews, I'd say to cut the top 10% and the bottom 10%. It's those extreme scores that pull scores away from a more accurate thought of the reviewing public. However, it still doesn't count those that are biased toward or against something. That's something that just broke in the general review system.

I do like this idea, how many outliers should be discounted? 

My leaning is top and bottom 10%, which I mentioned. On a game with 50 reviews, it cuts out the top 5 and the bottom 5. I'm not sure if that's too many, but I really wouldn't raise the percentage any higher. I've toyed around in my head with it being 5%, but that doesn't feel like it's enough. I am rather adamant on it not being a fixed number irrelevant to the review count, though.

Oh, what I did forget to mention in my original post, is that if there is weighing on the "influence" of an individual reviewer, that needs to go away. If anything, it's those "highly influential" sites that are more likely to also be influenced by the moneyhat. Wasn't there an article that mentioned how many of the early MW2 scores were from places invited to a hotel by Activision, where the game was provided in their room? I think that speaks volumes, and wish there was an easy way to discount those reviews, but by doing that, it opens a whole new can of worms...

If you want to do that, I would suggest a statistical method.  Something like throwing out values outside +/- 2 standard deviations, or +/- 3 standard deviations.

Of course, WereKitten already explained that this would most likely just push scores higher.  The mean for most games is so high that 2 or 3 standard deviations would probably be over 100, meaning none of the scores at the top would get thrown out.