RCV is not a condorcet system. If your system doesn't elect the most-liked candidate (the "Utilitarian Winner"), it should at least elect the most-preferred candidate (the "Condorcet Winner").

RCV is not a condorcet system. If your system doesn't elect the most-liked candidate (the "Utilitarian Winner"), it should at least elect the most-preferred candidate (the "Condorcet Winner").

A:

We do agree that the Condorcet criterion is important, which is why we're happy that virtually every single RCV election ever held in the US has elected the Condorcet winner. It is not the only criterion we consider important, however. We know from on-the-ground experience educating voters about RCV around the country that they are reluctant to rank beyond their first choice due to the fear that a second choice could work against and perhaps defeat their first choice. The fear is not unfounded, because that is true of nearly every other alternative voting method in existence. This criterion is known as later-no-harm, and all Condorcet methods necessarily violate it.

Another thing we know from practice and that voters will often rank their first choice candidate's chief opponent last, even if that's not their sincere preference, under the assumption that it will help their first choice. In Maine, there were some voters who wrote-in candidates like "Mickey Mouse" as their 4th choice just to push Poliquin or Golden last on their ballot. This strategy is often called "burying" or "turkey-raising" (the Mickey Mouse candidate being the "turkey" that is insincerely raised). Fortunately, RCV is immune to to burying because Mickey Mouse is eliminated first regardless for having the fewest first choices. Consider a race where there are two strong candidates, where both sides raise the turkey candidate M above the other:
49: A > M > B

48: B > M > A
3: M > A > B

Under RCV, A wins, but under Condorcet, the turkey M (who is sincerely almost everyone's last choice) wins. In fact, it has been shown mathematically that Condorcet methods will elect turkey candidates at equilibrium.

Sure, it's theoretically possible that M is everyone's sincere second choice, but beyond that being extremely unlikely, we have to consider the political consequence of M winning. Yes, we occasionally hear from math-oriented people who are upset that M loses, but the bigger concern we hear from voters is the opposite: that RCV will elect "milquetoast" candidates who shy away from big ideas in hope of being everyone's second choice. We have to spend much more time assuring them that "everyone's second choice" cannot be elected than justifying the exclusion of the Condorcet candidate to those that understand what that is. Note, too, that the political fallout of electing M could include the supporters of A and B joining forces to repeal the system.

Lastly, RCV is tried, tested, and works well in the real world. Whatever the theoretical properties of other alternatives, they are untested in the context of real political elections. Many of the influential people we talk to, legislators in particular, are less interested in how RCV works on paper; they want to know how it works in practice. They understandably don't want Massachusetts to be a guinea pig for something untested.

 

Showing 1 reaction

Please check your e-mail for a link to activate your account.