Kinda like what we do at Snackbar, where we rate a game as 'purchase', 'rent', or 'don't bother'.

I can't recall the site that did it, but I do know that I enjoyed the reviews there because they took that "pull out your wallet" idea a little bit further. Instead of giving a score in the "arbitrary value out of arbitrary value" format, each review gave a "how much would I pay for this" monetary value. In effect, the review scores told the reader how deeply they'd have to find the game in a bargain bin before a purchase was justified.

So that's an interesting method, albeit an inelegant and also rather arbitrary one. It doesn't really solve any of the issues inherent in the other scoring methods, though it does make the writer's stance on the game's value much more clear.

The current industrywide scoring standard strikes me as rather dumb and pointless. I always appreciated how Computer Gaming World used to do it (so of course by now they've caved in... sigh): 1-to-5-star rating, with half-stars in between included. 1 was the worst of the worst, 2 was highly mediocre, 3 was perfectly average, 4 was pretty darn good, and 5 was the best of the best. Simple, sensible, balanced, easy for both readers and reviewers to understand. None of this meaningless 8.8-versus-8.9 crap.

kinda like my idea of a 1-3 review score system i came up with long ago.
1 - avoid
3 - play it
2 - make up your own damn mind. based on the accompanying review text, of course. it also serves as a catch all for games that may not appeal to everyone but has value to fans of the genre/style/whatever.
the excellence of games is subjective so there's not real point in having such a fine scale.

I always liked the binary review model (with a possible third "rent" or "for fans" option). If I were to ever do any reviewing, this is the method that I've thought I would use.

I agree that rating to the tenth is crap. I suppose that this is used to distinguish which is better between similar or cross-platform games or some nonsense, but if a reader is seriously that interested in which is better, they'd just read the review to find out which and why instead of just looking at the number.

Reading reviews.. it's a lost art.

Take a page from French encyclopedist Denis Diderot and the way he used to review art in the 1800s: use words. Make the review entertaining and educational. Discuss how the work links to broader issues, trends and criticism. Answer three questions: "how did the game make me feel?", "is this game fun to play?", and "is this game a piece of crap?"

The "objective" review systems are basically just consumer reports responding to the early days when most games were, in fact, pieces of crap, and when the primary function of a review was to warn you away from that crap. The internet has replaced the "crap" warning by providing screen shots and gamepaly videos galore. Accordingly criticism itself should change to fit the times.

I wish the whole number scale for game rating was eradicated and in it's place, we would get well written, informative reviews. A summary of pros and cons could be added at the end of the review if one is too lazy to read through it. That way, it would be up to the reader to determine an appropriate numerical rating if they really needed to see things in percentages and fractions.

...this will never happen. :/

I wish Gamerankings would calculate scores to a source baseline. I.E., find out what the average for all the scores on file for EGM, find the difference between that score and 5, and then correct all the scores so they're distributed across the entire spectrum. The math would be a little more complicated than that, but I'm rusty on my stats lingo. Is 'normalizing' the word I'm thinking of? Anyway, it'd be interesting to see the results.