My guess is that it is simply regression to the mean. There are more Windows contributors than Linux contributors, so their average contribution is likely to b

My guess is that it is simply regression to the mean. There are more Windows contributors than Linux contributors, so their average contribution is likely to be smaller.