[EM] Real IRV Ranked Ballots

[Markus Schulze]

Hallo,

Eric wrote (2 March 2004):
> I am currently in communication with a dedicated IRV supporter
> who may be claiming that they would move away from IRV if a
> real example could be given where IRV selected an obviously
> wrong winner.

Actually, I believe that it is quite impossible to use examples
to demonstrate that a given election method is lousy.

For example, Tideman's ranked pairs method and my beatpath method
violate mono-add-top, mono-remove-bottom, and IPDA. When one of
these criteria is violated then it is difficult to argue why the
used method doesn't find "an obviously wrong winner". The only
way to justify such an example is to argue that in the long run
the used method performs better than other methods and that the
used method has to select this candidate in this example because
otherwise monotonicity, independence of clones or some other
important criterion is violated.

Markus Schulze