One aspect of the games industry that has always bugged me is the way that games are ranked, disregarding whether game journalism is credible the fact that ranking is not relative to other games just doesn’t make sense. Seeing as the majority of games are ranked in the 6+ area it’s hard to get a gauge of 1/10 game would be.

While sat in a lecture today about page ranking algorithms used by websites I started thinking back to the Elo Chess Ranking algorithm that is famously used by Mark Zuckerberg for his facemash page created while at Harvard for ranking the hotness of women. This got me thinking about an equivalent system used to rank games dynamically by creating a site that allowed you to rate a game by comparing it to a list of other games (randomly selected) and then choosing which one was better.

The calculation to rate each game would be:

R = \frac{T + 400(W - L)}{n}

R = Rank
T = sum of all comparators rankings
W = number of wins
L = number of losses
n = number of comparisons

Each game would have a base rating of 1000

This system would allow both new games and older games have their score dynamically changed to reflect all other games in the system and give a true representation of the games score. Using in the algorithm would also allow calculations of expected outcomes of comparisons to be calculated. This could be used to create a good ‘Game Recommendation section’

Leave a Reply