I do think that the forecasters seem to have been starting from a bad base rate (rather than using “sympathetic lies”). Relatively few posts about original topics (rather than e.g. org updates) hit 50 karma, and comments do so even less often. But the posts and comments we are most likely to read are unusually likely to be the high-karma ones; I wouldn’t be surprised if people tend to overestimate their frequency.
This was interesting!
I do think that the forecasters seem to have been starting from a bad base rate (rather than using “sympathetic lies”). Relatively few posts about original topics (rather than e.g. org updates) hit 50 karma, and comments do so even less often. But the posts and comments we are most likely to read are unusually likely to be the high-karma ones; I wouldn’t be surprised if people tend to overestimate their frequency.
As one extra data point: When I look at the EA Forum profile of users whose Forum posts are relatively well-known, I’m often quite surprised:
That the typical karma on their posts seems notably lower than what I’d have guessed
That they wrote a bunch of posts I didn’t know about (which tend to have notably less karma than the posts of theirs which I do know about)
This seems pretty consistent with the explanation Aaron provides.
(Also, I too found this an interesting post!)
Then, there is Buck.