This

Yes, combined with all other opinions of course, but yes. We're looking seriously at it

I can't take credit for this idea, I think it was suggested in either the article by rocketpig saying the review system is broken, or a thread about reviews.
Anyway, someone mentioned assigning a $ value to a game based on it's reviews which reflected how much they thought the game was worth paying for.

Obviously it would be based off the RRP for a game, but something like MGS4 or GeOW2 could get a $75-80, where as Too Human or Haze could get a $15-20.

I think it shows a really good example of a value of a game. It makes it pretty clear if its a day 1 purchase or wait for the bargain bin. It's also unique and could be converted to an approximate ranking if you wanted to put together a guide for gamerankings or metacritic.

Give me a moment. I made a pretty good couple of posts that I'm going to cite concerning my general reasoning for a less precise scale (the ideal being no score but the concession the necessity for a score being a 5-point system with no fractions) and I'll post them as soon as I dig them up...

http://www.vgchartz.com/forum/post.php?id=1457111

Okay I only really found one but I think it makes a pretty capable argument on its own.

Basically it's abandoning the infantile need to bow down to the aggregate score and stop trying to assign absolute values to games because absolute values imply an objective standard of quality, which we do not have.

No grades at all. The current grading system is nothing more than a means of subtly biasing reviews; people should be made to think for themselves.

Percentage is winning by a huge margin in the front page poll.

We should do it by sales of previous consoles

A = Playstation

B = Gameboy

C = Nintendo Entertainment System

D = Sega systems

E = Xbox (?)

F = Jaguar

And for +/- it is easy.

A + = Playstation 2 (sold more than 1)

B - = Gameboy Advance (sold less than original + Colour)

C - = SNES

E + = X360

F is just F no +/-

It should be quick impressions, like Kotaku. That way, people won't judge the review by the score, and people would actually bother reading it because it isn't too long, and is... well, actually fun to read instead of dragging on paragraph after paragraph.

I'd advocate a system very similar to IGN's but with a couple of caveats:

1) the scores for presentation, gameplay, etc. are averaged giving a total base score.
2) that base score can be adjusted up or down by a maximum of 1 point (in .5 increments) based upon the "sum of parts" or "overall fun" factor.

For example a game that scores an 8.5 for all it's "technical" abilities could be adjusted anywhere from a 7.5 to 9.5 based upon the overall experience, which may (or may not) exceed it's technical performance.

Yeah, that would work.