[corruption]

I propose an experiment rather than making an assumption that it's not important.

Randomly change the username on new posts from users that have a low average points per post (ppp)to a top 20 user in terms of ppp.

If there is no effect of username on ppp, then the difference between the users true average ppp and the randomly assigned one will be large, if there is an effect it will be approximately zero. We could test this statistically easily enough.

What say you pg?

[jacquesm]

That's a really clever idea.