A Bare bones probability can be calculated, though. Knowing there are 18 people each compiling lists of 32 that can be arranged in any way... the hard part accounting for tendency (which is to say, nobody would choose the Raiders #1, even though they HAVE the choice).

So unless I've forgotten how to do math, one person could submit any one of 2.6313083693369353016721801206e+35 (in windows calculator notation) possibilities. Ok, so that is way too much... I'll try to bring it down. I'll assume variance in groups of fours only (so the top four teams would be the same in each regardless of order, and so on down the line). Each group of four will have 24 possibilities (4x3x2x1), creating 8 groups of 24 possibilties that can be mixed and matched.

Man it's been a long time since I did anything resembling statistics or probability in a math class. Five years removed from math to begin with...

So that would work out as just 24 to the 8th power, no? Meaning one person could come up with 110,075,314,176 different combinations?

Well... it's less than before, anyway.

BUT the probability of averages levels it out! In this system there are 8 groups of 4 teams. Each team in the group of 4 is given to have an equal probability of being any spot 1-4. So the probability of a team averaging 1 (unanymous 1st place votes) is low, but the probability of a team averaging 2.5 (evenly distributed votes) is very high within an individual group. And rather than being dependent on huge numbers of combinations, it's dependednt on the tendency of those combinations to even out...

Anyway, I'm not nearly advanced enough in statistics or probability, nor am I close enough to the math experience I do have to work this out in numbers. What I will say is that I believe the rankings do GENERALLY fall into localized pockets of teams like my unfinished method uses (not as clean as groups of four, but general). Within these pockets there are further tendencies toward the top and the bottom of the order, but the probability for each team would still be to drift toward the mean of the group. I would imagine that because of this the probability of ties in the middle of each group is pretty high up to around 10 ballots (somewhat arbitrary estimation), and then of course continues to go down from there. Assuming there are like 8-10 "natural" groups in the NFL rankings (also kind of arbitrary), the fact that there are ties in only two of them (20-25%) would seem to show that by 18 ballots these chances are in fact thinning out pretty significantly...

But then again, a lot of this is semi-educated-ass-talking. If I had the capacity for these numbers I would have gladly cranked it out, but I can only go so far before I have to defer to "logic," which is significantly removed from hard fact.

Anyway, sorry for basically spending all this time coming up with very little that's worthwhile. It was fun for me, at least...