At best a side effect of something else [much like Google's PR0 on YouTube a while back... Whoops! ...it wasn't suppose to do that!]

Dave's experiment [ok under advisement... it's true for ther sake of argument - what's actually true though?]

Dave's experiement didn't even test "paid links" [a series of patterns that resemble 'paid link adverts] -- it assumed that a paid link is a sidewide link, and not just that the topic of the linking domain differs from that of the receiving domain.

Consequently, if I take Dave's conclusion at face value... it isn't the "paid" part that negatively affect his experiement, and it may not even be the sitewide part... it could have been the "unrelated part"... which is it? ...

That's why when you're advising memberships of your first-hand experiment you allow scrutiny of the data... not your word.