[EM] Update from a real Condorcet deployment

[Richard Lung]

Hello Prof. Myers,

Just reading your message, I would guess these Condorcet polls are 
proportionately weighted.
It can make a big difference, even over-turn a whole theoretical edifice.
I once (in a book called Beyond Numeracy)came across a contrived 
election example, using five different voting systems, all disagreeing 
with each other. (I have seen its like, even on the most prestigious of 
web-sites.) I don't blame it for being contrived. But it was obvious 
that the different out-comes were the inevitable result of using more or 
less rudimentary counting procedures, including First Past The Post, 
Supplementary Vote and Instant Run-off Voting/Alternative Vote.
Then I checked whether a fourth system of Condorcet pairings of 
candidates were weighted.And they weren't. (I had learned weighting from 
statistics. I didn't know it was already a recognised election method, 
by Kemeny, I believe.)

I did the arithmetic. Lo and behold! Weighted Condorcet now agreed with 
the fifth method, Borda method, on the most popuar candidate. These were 
the two methods that made best use of the preferential information. My 
old statistics lecturer said the best statistical test is the one that 
makes the most use of the information. Here was a demonstration of that 
truth for election methods.
Yet the whole purpose of that text example had been to prove the 
opposite, contending that there is no best election method, as 
demonstrated by the criteria of social choice theory.

I have discussed these matters in my second of two e-books on election 
reform and research, whose links I just posted on this e-mail forum.

 From Richard Lung.




On 07/11/2015 22:06, Andrew Myers wrote:
>
> We recently crossed some thresholds over at civs.cs.cornell.edu:
>
>   * more than 10,000 Condorcet polls run
>   * more than 200,000 total votes cast
>
> CIVS is regularly used by organizations whose names you might 
> recognize to make decisions with real-world impact. Here are a few of 
> them:
>
> Ubuntu
> OpenStack
> Linux Foundation
> SUNY Fredonia
> Iowa State University
> University of Mary Washington
>
> The source code for CIVS can now be found on GitHub: 
> https://github.com/andrewcmyers/civs. Contributions are generally welcome.
>
> Regards,
>
> Andrew Myers
> Professor, Department of Computer Science
> Cornell University
> http://www.cs.cornell.edu/andru
>
>
>
> ----
> Election-Methods mailing list - seehttp://electorama.com/em  for list info


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20151109/c5ca8cab/attachment.htm>