Have statistics on games, then later come up with a slightly different starting algorithm that is more fair.

I can think of two problems that statistics would prove:
1.If you start out with a really large stack of 6 or more, you are much more likely to fare better than average.
2. If you start sixth or seventh seat, you are much more likely to fare worse than average.

As I always say: If you have 1 or 2 big stacks (5 or more) you WANT to go 6th or 7th. Many dice have been killed and you can spread/connect without risk - no risk to loose a fight and lesser risk to loose the terrs again to someone else with a big stack. you only need to go first if you have 3 and 4 - dice stacks, cause they are likely to get killed by the big stacks (they are a thread...). SO: All random, all evens out in the long run, no need for changes. Stats are always nice though...

agreed alea.