[EM] get Condorcet results from IRV results? (was IRV in action)

[Jan Kok]

From: "Adam Tarr" <atarr at purdue.edu>
> James Gilmour wrote:
>
> > > Similarly, when IRV is being used then the voters will also accept
> > > situations where an extremist is elected because all moderate
> > > candidates have been eliminated at earlier stages.
> >
> >Are there examples of this from real public IRV elections?
>
> If you can find examples of IRV elections where there were three strong
> factions, and where actual ballot preferences are listed, then you may
very
> well find such an example.  There are two problems in finding such
results:
>
> 1)  Very few IRV elections list actual ballot totals.  At best you tend to
> get runoff numbers, which can make it difficult to trace back to the
> original ballots.
>
> 2)  Many here have argues that IRV has a tendency to keep two parties up,
> and keep the others down.  The effect is surely weaker than it is in FPTP,
> but it is present to some degree.  This reduces the probability of the
> "nightmare" scenario I described before, but at the price of keeping
> political competition down.

I think it might be possible to determine the winner that a Condorcet method
would select, given typical Australian IRV election results _plus_ the
information on the how-to-vote cards given by party representatives to the
voters.

Here is an example of some Australian election results data:
http://www.aec.gov.au/_content/what/voting/count_hor.htm
Unfortunately it shows only one election result, but at least it shows what
data is available.

I don't know where to find the how-to-vote information for any given
election.

The key that could make this possible is that Australian voters follow the
how-to-vote directions pretty faithfully.

Anyone want to do the research, and find some instances from real life where
IRV and Condorcet would likely have chosen different winners?

Cheers,
- Jan