Once again Markus Schulze is trying to discredit the Condorcet-Kemeny method. (See below.)
This is ironic because "his" method -- which he calls the Schulze method, and which would more meaningfully be called the Condorcet-Schulze method -- produces the same results as the Condorcet-Kemeny method in most cases. What percentage of cases produce the same overall ranking sequence? I would guess it's at least 85% to 90%, and perhaps as high as 95%, if all possible ranking combinations are considered. If only the single-winner result is considered, the results would match in an even higher percentage of cases. > Markus Schulze wrote: > does the San Francisco Bay Area Curling Club still > use the Kemeny-Young method? If not, why did it > abolish the Kemeny-Young method? This club continues to use VoteFair ranking, a subset of which is the Condorcet-Kemeny method. Specifically they used VoteFair ranking in their 2010 March election. As you imply, this is a good endorsement for VoteFair ranking. Thanks for the reminder! Clarification: I recommended that the club use VoteFair representation ranking because it provides better proportionality than VoteFair popularity ranking. So far, based on the results of the two elections I looked at, these two rankings have been the same up to the number of positions that are being filled. There have been differences at lower levels, but that has had no effect on who is elected. Therefore, the distinction between VoteFair popularity ranking and VoteFair representation ranking has not mattered. In either case, the club continues to use VoteFair ranking to elect their officers. > Its website says: "The two nominees with the most > votes will hold office for two years, the nominee > with the third highest level of votes will hold > office for one year." This wording is ambiguous to election-method experts, but meaningful to real-world voters. When pairwise counting -- which we both advocate as the fairest approach -- is used, we can say that person A had 40 votes compared to person B who had 60 votes. This meaning is slightly different from the word "votes" in regards to plurality voting. I presume the former meaning is intended here. Also note that this quote refers to the time in office, not the vote-counting algorithm. > http://www.bayareacurling.com/elections2010 > To me, this sounds like SNTV. That's a big assumption to be made from an ambiguous wording. > Markus Schulze For those who may not know, and especially for Peter Zbornik to whom Markus intended his message, there is a third similar method. It's called Ranked Pairs on Wikipedia, it's sometimes called the Tideman method, and a more meaningful name would be the Condorcet-Tideman method. The Condorcet-Schulze method and the Condorcet-Tideman method use a similar elimination approach, where one looks for the biggest pairwise numbers and the other looks for the biggest margins of victory. (Other Condorcet methods do not have characteristics that make them worth considering in real elections. One of them is used in games where its element of randomness adds entertainment value.) All three of the above-named Condorcet methods produce the same results in "most" cases, which means the difference is not noticeable in real elections. This is a good time to mention that there is an organization (I don't recall the name) that calculates the Condorcet-Schulze winner and also calculates the Condorcet-Tideman winner. If these are not the same candidate, the Kemeny scores are calculated for the two applicable sequences, and the sequence with the higher Kemeny score determines which of the two candidates is declared the winner. This is a clever way to get around the challenge of writing software to calculate the Condorcet-Kemeny method. It also demonstrates the similarities between the results calculated by the Condorcet-Schulze and Condorcet-Kemeny methods. Furthermore, it demonstrates that the person behind it is wise enough not to completely rely on the Condorcet-Schulze method. For perspective: Back around 1990, without knowing the names of voting methods and voting criteria, I was figuring out the fairest voting method. After I considered and dismissed the IRV approach, I considered and then dismissed the Condorcet-Schulze approach, which is to look at the biggest or smallest pairwise numbers. I reasoned that that approach has the same weakness as IRV and plurality. Whereas plurality and IRV look at first-choice preferences, the Condorcet-Schulze approach is better because it looks at pairwise numbers. But I knew that looking for the biggest or smallest numbers is like looking at the surface of water to figure out what's happening beneath the surface. I reasoned that a fair method needed to look deeper. So I used an approach that is similar to fitting a straight line through a number of non-aligned data points, namely to add together numbers that measure the degree of fit. The resulting sum -- which in this case is of applicable pairwise counts -- provides a way to find the best fit by finding the sequence with the maximum sum. It was years later that I found out that an academically known method (the Kemeny method) similarly used the sum of opposition (!) pairwise counts and looked for the smallest such score. This parallel development should not seem too surprising because the general approach is used in yet other situations in physics and mathematics. Interestingly, the Wikipedia definition of the Condorcet-Kemeny method now uses the kind of score I came up with, not the one specified by Kemeny (and others). The methods are mathematically equivalent. The Kemeny-specified approach is not as easy to understand, which perhaps accounts for some of the method's earlier neglect. Perhaps the main reason the Condorcet-Kemeny method was neglected is that it's so difficult to implement in software. By contrast, the Condorcet-Schulze method is easy to implement in software. Getting back to the main subject, the members of the San Francisco Bay Area Curling Club do not need to know how to write the software, they just care about the results. They like the results (see the Testimonial page at VoteFair.org), so they continue to use it. Here it seems appropriate to address a comment written by Jameson Quinn in another post within this thread: "ps. Peter, you should not be shocked by the enthusiastic response to your query. You've thrown a hunk of meat into the lion cage: the chance to see our theories applied in a real, important case. While we can give you useful feedback, we will never come to a consensus; in the end, you'll have to pick a proposal yourself. Also, while we lions may present an amusing spectacle, we still mostly respect one another underneath, and we understand that there is no guarantee that any of our proposals will be implemented in the end." This nicely characterizes what's going on. Fortunately the conflicts between voting-method experts are well-behaved compared to the conflicts that erupt into warfare, genocide, terrorism, etc. Those physically aggressive conflicts are what we are all working to eliminate through fairer voting methods. We all want to replace fighting with casting ballots. Richard Fobes
---- Election-Methods mailing list - see http://electorama.com/em for list info
