Ranked-Choice Voting in San Francisco

I originally wrote this entry on September 8, 2004, and published it on blogs.sun.com.

I'd written earlier about the mathematics of elections, with particular attention to Condercet or ranked-choice voting. Now, Professor Lawrence Lessig has pointed his readers to a demo for San Francisco's upcoming ranked-choice voting experience. I wonder whether a reduction in the number of available rankings (say from the number of candidates for an office to only three choices, as has been done in San Francisco) diminishes the probability of cyclic ambiguities. San Francisco's ranked-choice voting system was passed as proposition A in March 2002, and the first ranked-choice vote for local offices will be held during the November 2nd, 2004 election. It will be a great time to look and see how Condercet voting does in practice in Northern California.

In the text version of the official San Francisco ranked-choice voting demo, we read:

The Department of Elections cannot predict the date on which it will begin the process of elimination and transfer. The Department will do so as soon as possible, after all provisional and absentee ballots are processed. The Department intends to report final election results no later than 28 days after election day.

The "28 days" of waiting for election results seems awefully long. It could be that old counting machines are used for a physical implementation of the various elimination algorithms. As I wrote earlier, some computing power and already-implemented algorithms could help with the counting and the elimination process in ranked-choice voting in cases of result ambiguities.