Distribution of ratings data on IFDB

Just a little factoid that I thought might be of interest to some people: I was looking over the data from the early March backup of IFDB, and I wanted to see how ratings data was distributed over games.

Although there is a mean average of about 4 ratings per game, the distribution is a typical power law curve in which a few games have tons and a large number of games have few. In fact, as can be seen in the table below, nearly 70% of games have fewer ratings than the mean average, and nearly a third (close to 5000 games!) have no ratings at all.

# ratings	percentile
---------	----------
	1		32.00%
	2		50.00%
	4		69.00%
	5		75.00%
	7		81.00%
	12		90.00%
	20		95.00%
	63		99.00%
	583		100.00%

The tippy top of the distribution is Adam Cadre’s 9:05. IFDB currently shows it with 551 ratings, an artifact (I think) of ratings from deleted users that remain in the ratings data but are not shown by the site.

5 Likes

Get to work, @mathbrush

Kidding!

Maybe we need stickers around the place that say something like ‘Play and rate an unrated game today.’

-Wade

5 Likes