For most questions I think average would be more useful. For the question of whether people upvote pg thoughtlessly or not, median is useful. If more than half of pg's comments end up with a score of 4 or less, then we can conclude that his comments that end up with 50 or more points are a select subset.
Submission scores might be a statistically different distribution then comment scores. Also while median is better than mean, it may or may not be better than mode - difficult to tell without looking at the shape of the distribution.
Comments
For most questions I think average would be more useful. For the question of whether people upvote pg thoughtlessly or not, median is useful. If more than half of pg's comments end up with a score of 4 or less, then we can conclude that his comments that end up with 50 or more points are a select subset.
Submission scores might be a statistically different distribution then comment scores. Also while median is better than mean, it may or may not be better than mode - difficult to tell without looking at the shape of the distribution.
I think he would get more if our usernames stood out more. but I can accept that I am being grossly inaccurate and exaggerating.