Here's a copy of my response to Michael Welford: Forest ---------- Forwarded message ---------- Date: Wed, 17 Jan 2001 14:48:58 -0800 (PST) From: Forest Simmons <[EMAIL PROTECTED]> To: Michael Welford <[EMAIL PROTECTED]> Subject: Re:PAV It sounds like we've been thinking along the same lines. Since sums like 1 + 1/2 + ... + 1/k and 1 + 1/3 + ... + 1/(2*k-1) are asymptotic to log(k) for logarithms of various bases, it suggests that asymptotically one could almost use the geometric mean of the various k's as a guage of satisfaction of the voters. Of course, one voter with k=0 would throw everything off. Perhaps this could be taken care of by including an ideal dummy candidate in every panel or slate being guaged, and incrementing the number of winners by one so that when the ideal candidate is withdrawn from the winning coalition there are still enough left to fill the vacant seats. By the way, your idea for filling the seat of a candidate who had to withdraw after winning was clever. I like PAV as an acronymn for Proportional Approval Voting. I consider it to be one member of the class DRLR. Forest Do you mind if I forward your message to the election methods list along with this response?
