Hacker News new | ask | show | jobs
by xenoterracide 6635 days ago
no. that's why it IS funny, 'cause it's true.

It's sad too, though.

If pg says anything he gets at least 50 points for being pg. The rest is because pg is actually smart and doesn't usually say dumb unthoughtful things.

1 comments

pg's median comment score is 4:

http://news.ycombinator.com/item?id=166110

More people read pg's comments, so he does get more upvotes than the same words would if coming from someone else. However, to say that you can mention pg in a comment to get more votes is neither true nor particularly funny.

he mentioned posting an article related to PG, not just comments
wouldn't mode or average be more useful?
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.
Median score is not what's being discussed. It's quality vs score. Those individual comments that are of completely average quality and yet have phenomenal scores are the most telling.

Also very telling: score velocity over time. For example, if `joe` makes a good point with factual grounding and is getting lots of points. Then `bob` replies in disagreement and bob provides almost nothing to back up his position. Then suddenly the points reverse (joe goes negative and bob gets lots of points) just because bob said it it must be true and everyone blindly agreed. There are ways to measure this impulse response that would provide more concrete numerical measurements.

Median score... there are so many effects that would cause this to not relate to comment quality that that is pretty useless. For example, just commenting on more active topics can give you a much better median score.