A “Condorcet winner” is the candidate that can strictly beat every other candidate head-to-head.
If the IRV winner ≠ Condorcet winner, this will resolve YES.
If the IRV winner = Condorcet winner, this will resolve NO.
If there is no ballot data to determine a Condorcet winner by the end of 2022, this will resolve N/A.
If there is no Condorcet winner, this will resolve N/A.
If there is an IRV tie, the IRV winner will be whoever is chosen, random or otherwise.
Related questions
🏅 Top traders
| # | Name | Total profit |
|---|---|---|
| 1 | Ṁ150 | |
| 2 | Ṁ3 | |
| 3 | Ṁ1 | |
| 4 | Ṁ1 |
On a philosophical note - people love to bash IRV because this is a possible outcome, but it's important to remember it is fundamentally impossible to aggregate the preferences of multiple actors in a completely rational way (Arrow's theorem). Every voting system will have some flaws, it's a matter of picking one whose benefits outweigh the flaws.
Seems to me the overwhelmingly likely case in which this happens is one where there isn't actually a Condorcet winner at all - and indeed, that's the problem with using Condorcet as your voting system. It can fail to deliver a winner, because the preferences of groups aren't rational and you can get cyclical rock paper scissors stuff. Hopefully it's self-evident why a voting system used for electing democratic representatives cannot be one that could be unable to deliver a winner.
@MattP Condorcet is a class of voting systems, not a particular system. All the systems deliver a victor in cases where there is a cycle.
@MattP Overall I agree, but a few quibbles:
Note that Arrow's theorem doesn't actually apply to our normal voting scenario: we're usually trying to select only a single winner rather than to produce a societal ranking of all the candidates (https://en.wikipedia.org/wiki/Arrow%27s_impossibility_theorem#Social_choice_instead_of_social_preference). However there are similar impossibility results (like Gibbard–Satterthwaite) for single-winner systems, so overall your point that there's no perfect system is still true.
It's true that Condorcet is not a complete system in itself. But I'm not happy with leaving it as "that's the problem with using Condorcet as your voting system", any more than I'd be happy with "sometimes the vote count can be a tie; that's the problem with using plurality as your voting system". Yes ties are rarer than the case where there is no Condorcet winner, but the idea and its solution are the same: base your system around the principle you care about, but make it robust to the edge cases that might come up. So there are plenty of systems which satisfy Condorcet's criterion (such as the Schulze method) and are also well-defined systems in the cases where no Condorcet winner exists. And you can also easily make ad hoc new ones by using any other system as a fallback, e.g. "Condorcet otherwise IRV", in which you first pick a Condorcet winner if one exists, and otherwise run IRV as normal.
@MattP The main philosophical problem with irv is that increasing support for a candidate can cause them to lose.
@MattP Smith//Score (A condorcet method that uses smith set to score for tiebreaking) has shown to be one of the most accurate methods. (Source: http://votesim.usa4r.org/tactical/tactical.html)
@AndrewG Yeah, that page is awesome :) I'm unclear on how commonly those failure modes crop up in reality though