I used the winning percentages over a 5 year period and put the odds (1st, 3rd, 5th, etc.) in one division and the evens (2nd, 4th, etc.) in the other. I figured using 5 years would be enough to reduce the amount of change-over to 0-2 teams each season.
I was wrong.
The divisions based on 2005-2009
Odds- OSU, Wis, IA, NU, PU, IN
Evens - PSU, Neb, Mich, MSU, MN, IL
Going backwards, the divisions based on a 5 year period one year previous, 2004-2008
Odds - OSU, PSU, IA, NU, MSU, IN
Evens - Wis, Mich, Neb, PU, MN, IL
There are two trades (PSU/MSU for Wis/PU)
Going back another year, 2003-2007
Odds - OSU, Mich, Neb, PU, MSU, IN
Evens-Wis, IA, PSU, MN, NU, IL
This time there are three trades from the following year (Mich/Neb/PU for IA/PSU/NU). That is half of the divisions.
Even with this very small sample, it is clear that assessing each year using a 5 year period would result in more change than they would likely be willing to work through. I think it is still a fun (and probably more fair) idea, and I hope they do something to weigh the division strengths from the recent past, not just the historical past. I'm sure that any way they align the divisions, there will be something to complain about. After a couple years, those complaints will be mostly gone. It takes a long time to build tradition, but only a short time to destroy it. And after it is destroyed, it takes a short time to forget it.