[EM] IRV failure modes, vote splitting

[robert bristow-johnson]

---------------------------- Original Message ----------------------------

Subject: Re: [EM] IRV failure modes, vote splitting

From: fdpk69p6uq at snkmail.com

Date: Fri, December 21, 2018 11:42 pm

To: election-methods at electorama.com

--------------------------------------------------------------------------



> On Fri, Dec 21, 2018 at 11:33 PM Greg Dennis wrote:

>

>> I think we may need a better theory to explain why IRV elects the

>> Condorcet candidate with near-perfect frequency in practice. We're now

>> somewhere around 215 IRV elections in the US since San Francisco started in

>> 2004, and Burlington 2009 is still the only case.
i'm just saying that i know of no other *governmental* election using RCV (and all of these RCV elections were using the IRV method of tallying votes) where the IRV winner was not the Condorcet winner.  perhaps there is another
case where the CW loses IRV, but i am not aware of it.
we know that Condorcet winner will likely have some basis of popular support simply in order to become the CW.  we also know that if the CW gets into the IRV final round, the CW wins the IRV election.
now, imagine the race is
boiled down to 3 candidates.  then say that one candidate is a spoiler (i.e. that candidate's presence in the race *changes* the outcome of the race, but that candidate does not win), another candidate is CW. 
we know that a spoiler who is *not* a plausible winner, like having
something like 20% first-choice support or less (among the top three candidates), that spoiler will not have enough support to get into the final round, then the CW and the other candidate get into the final round and the CW wins the IRV final round.


> How do you know it elected the Condorcet candidate? All of the ballots are

> public for all of those elections?
the data of the ballots were public with the Burlington 2009 election.  i still have the files.  and in the last SF mayoral election, i believe there were also files that had the individual ballot data.
then, the bitchy thing is writing a
program in whatever language you like (i did mine in C) to open, read, and parse these files and count the votes under whatever rules.  my program only calculated the pairwise defeat matrix.  Warren Smith's program calculated subtotals for 9 different ballot possibilities once it was
boiled down to 3 candidates.

--



r b-j                         rbj at audioimagination.com



"Imagination is more important than knowledge."

 
 
 
 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20181222/114520d3/attachment.html>