[EM] Kemeny Condorcet method. Apparently not a good choice for those of us who want to know who won in our lifetimes.

[Richard Fobes]

Warren Smith ~

If the Condorcet-Kemeny method had been used to calculate the results 
for the 135-candidate California special election that was won by Arnold 
Schwarzenegger, and even if the voters had used ranked ("1-2-3") ballots 
(or range-like ballots that allow assigning three-digit ranking levels), 
most of the candidates could be quickly identified as not being a 
possible winner.  That would reduce the calculations to perhaps as many 
as 20 possible-winner candidates.  The full ranking -- from most popular 
to least popular -- for those 20 candidates could be calculated in 
minutes.  Yet only the winning candidate needed to be identified, and 
there are programming techniques that can quickly identify the winner 
without calculating the full ranking results.  In other words, the 
Condorcet-Kemeny method could identify the winner within a few minutes 
(or possibly a few seconds), even if the voters all ranked every one of 
the 135 candidates.  A few minutes is hardly a lifetime (although there 
may be times when it might seem like it is).

You refer to randomly generated ballot preferences, which takes your 
argument out of the realm of normal elections.  Yet there is a real-life 
situation that is similar to randomly generated ballot preferences. 
Imagine a survey or poll that is ranking a "top 100" list of musical 
songs, putting them in sequence from most popular down to least popular. 
  For this purpose there is a variation of the Condorcet-Kemeny method I 
would recommend, and it would easily handle such extreme cases (and then 
the top 20 songs could be more precisely ranked using the full 
Condorcet-Kemeny method).  I have not yet publicly described this 
alternative, but I intend to later, when I have more time.

Presumably your reason for attempting to find weaknesses in the 
Condorcet-Kemeny method is that it's fairness of always identifying a 
Condorcet winner in combination with the fact that it is relatively easy 
to explain (certainly much easier to explain and understand than the 
Condorcet-Schulze method -- which is a consideration I pointed out to 
you yesterday during our declaration-editing chat) makes it seem 
uncomfortably competitive with your preferred election method, which I 
believe is Range voting.  If I have interpreted correctly, thank you for 
this compliment of the Condorcet-Kemeny method.

By the way, if a real election is likely to involve 20, or  50, or more 
choices, then I recommend using VoteFair Ranking as described in my 
book, "Ending the Hidden Unfairness in U.S. Elections".  VoteFair 
Ranking uses "VoteFair party ranking" to identify which political 
parties deserve to have two candidates in the main election, and which 
parties are not popular enough to justify having any candidates.  This 
limitation is not for the purpose of reducing calculation time, but 
rather for the purpose of giving voters a reasonable number of 
candidates to keep track of, without distractions from 
cannot-possibly-win candidates.

Richard Fobes


On 9/11/2011 9:22 PM, Warren Smith wrote:
> I have on this thread at the CES
>     http://groups.google.com/group/electionscience/t/b135bdc214c39ffa
> reviewed some known theoretical and empirical facts about the Kemeny Condorcet
> voting method.
>
> In particular, it appears based on my literature review that humanity,
> using 2006-2011 era hardware and software, is currently unable to
> reliably determine the Kemeny winner from the votes in 5-voter,
> 50-candidate test elections generated by certain reasonable kinds of
> random vote-generating processes.
>
> The Wikipedia article
>    http://en.wikipedia.org/wiki/Kemeny%E2%80%93Young_method
> is somewhat misleadingly worded on this point.   It makes it sound
> like no problem,
> but actually the very paper they cite says quite the opposite.
>
> Further comments will be welcome.