[aidenn0]

> shows some effects that I see as pretty disqualifying in a voting system

I think in most plausible configurations, the intersection of Condorcet and IRV is larger than the intersection of Condorcet and plurality if you ignore strategic voting.

Do we have good data about how people do or do not vote strategically in an IRV election?

[edit]

To clarify: I think both you and I would prefer approval to IRV; this is just discussing whether IRV or plurality is less-bad.

[simonthefox]

Agreed about preference to approval. My problem with IRV isn't about strategic voting though; I think predicting the outcome is too complex to be concerned about that. The worry that I have is that a voting method that encourages candidates to move public opinion away from them will lead to weird outcomes and undermines public trust. I think the 2000 election had an example of this in plurality voting, where republicans campaigned for Nader hoping to split the vote. In IRV I worry that we'll see similar effects, where candidates campaign for opponents because they think it will give them a tactical advantage. Aesthetically that doesn't sound "fair" to me, and my worry is that weird behavior like that undermines trust in the system.