Without a correct evaluation of it we can’t say. It’s totally possible people made too little experiment and have no idea how to calculate the error, so are drawing wrong conclusions about things (the same exact way idiots are claiming “CCP proved that killing people increases retention” while CCP claimed the opposite). However if such a bias is proven correct, then it’s possible that the gain of buying back several tickets is superior to the gain of buying back no ticket and letting the offer be complete by other players only.
as I wrote
if p (the chance to win the auction) is not equal to n/m (number of tickets you buy, divided by total number of tickets), then you have a different equation.
What’s more probable is, that CCP coded their CCP_rand poorly, typically a wrong pseudo random function that does not use entropy correctly, and as such the first tickets are a bit more likely to be winning.