On Wed, 13 Mar 2002, Narins, Josh wrote: > method, but the cap. The 2000 apportionment would have been fairer (by > measurement of the standard deviation of all districts) with only 432 Reps > instead of 435. Oh well. Sorry, I might have been able to do something about > it.
I once worked on this aspect (that fairer results could be obtained by letting the total float) but when I found out that voting power is a much bigger consideration that swamps the proportionality issue, I abandonded that line of investigation. Both Banzhaf power and Shapely-Shubik power are only in rough proportion to the number of representatives. It's voting power that should be equalized as much as possible. Basically, a random voter in my state should have the same chance of electing someone who makes a pivotal vote in the House of Representatives on a random issue as the vote of a random voter in your state. Perfectly proportional representation does not usually optimize this objective. Forest
