It's definitely nice to see more confirmation that Condorcet cycles are
rare. OTOH, I think the claim about IRV rarely failing to elect Condorcet
winners is incorrect. If he's looking at Australia and Scotland, those have
a 2-party and 1-party dominant system, respectively. No wonder IRV doesn't
have Condorcet failures, then, it's playing on easy mode!

Ironically, I'd expect to see more Condorcet failures (assuming cycles
count as failures) under an actual Condorcet system, because you'd have far
more competing parties.

On Tue, Mar 26, 2024 at 1:33 PM robert bristow-johnson <
rbj@audioimagination.com> wrote:

On 2024-03-26 22:03, Closed Limelike Curves wrote:

I don't think Scotland belongs there, since it doesn't use IRV. It uses
MMP :-)

Some local elections use STV, but that's multiwinner.

-km

On 2024-03-26 21:32, robert bristow-johnson wrote:

Congrats :-)

As for Condorcet winners: if politics is 1D and there's not too much
noise, there will be a sincere CW. If opinion space grows, say by more
parties being established, then you could get sincere cycles, but not
necessarily. So the lack of cycles shouldn't be too surprising... I
guess if anything, it's most surprising in Australia, because the Senate
has multiple parties. But from its House of Reps composition, it might
just be that IRV's compromise incentive is too strong.

-km

Even without the cycles or compromise—the really big mistake in the paper
is treating the elections as though the candidates on the ballot are the
actual choices society faces. In reality, there's plenty of others, but not
one of them runs.

If you have only 1 or 2 choices, every method gives the same result. But if
IRV actually was good at electing Condorcet winners, there would be more
than 1 or 2 options.

On Tue, Mar 26, 2024 at 4:39 PM Kristofer Munsterhjelm km_elmet@t-online.de
wrote: