[EM] possible improved IRV method

[David Cary]

It seems to me that the framework and the results are underdeveloped,
or at least poorly documented, for evaluating the overall
possibilities for strategic voting for any specific election method. 
Criteria have been developed for characterizing election methods
about the results they produce when people vote sincerely and what
kinds of strategies can be used, but there seems to be much less
about what happens when those strategies are applied.  What
information is needed, what level of collusion is required, under
what circumstances does the outcome change, under what circumstances
are the opportunities exhausted for any group to change their vote
and improve the outcome.

That kind of analysis is useful whether you just want to assess how
an election method might perform in practice and what effect it might
have on the larger political process, or whether you want to
construct a super-method that reduces or eliminates opportunities for
insincere voting by making the application of strategy a part of the
method.

People have certainly thought about these issues and have developed
some results, but it seems to be fragmented.  What are good resources
to read to better understand what is currently known?

As Eric outlines below and the subequent discussion highlighted, the
issues are not easy untangle.

For example, the use of strategies could be quite different in
different contexts.  The information that is available and the
incentives and disincentives to vote strategically could vary
significantly depending on whether the balloting is secret, whether
the election was a national popular vote for president, a town vote
for mayor, a corporate election of directors, or a vote within a
legislature or a committee.  Blanket statements maybe overlooking
some important issues.

-- Dave Cary

P.S.  Thanks to Rob for the DSV link.

--- eric at ericgorr.net wrote:

> 
> I have yet to see any convincing arguments that it does. Most of
> the  
> attempts tend to center around a single group of voters attempting
> to  
> manipulate the election to their advantage. The fundamental flaws
> with  
> these particular arguments is that they assume that (a) no other  
> groups will attempt counter-strategies, (b) that a single group of 
> 
> voters will even be capable of obtaining an accurate enough picture
> to  
> figure out how to change their votes - accuracy they would need  
> appears to be far greater then one is able to obtain with a
> scientific  
> survey of voter opinion and (c) the fact that if multiple groups of
>  
> voters are involved in insincere voting when a Condorcet Method is 
> 
> used that the ultimate resolution becomes quite chaotic and the
> worst  
> possible winner may be selected. It is the potential for selected
> the  
> least favorite choice that causes me to believe that a good
> Condorcet  
> Method actually encourages sincere voting.
> 


__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com