Ranked Choice has had some notable unexpected outcomes, see https://bolson.org/~bolson/2009/20090303_burlington_vt_mayor...
See also https://www.starvoting.us/ and https://electionscience.org/library/approval-voting/
Ranked Choice has had some notable unexpected outcomes, see https://bolson.org/~bolson/2009/20090303_burlington_vt_mayor...
See also https://www.starvoting.us/ and https://electionscience.org/library/approval-voting/
Score voting collapses to approval voting under strategy.
The right answer is "Ranked choice voting", but not with runoff but rather Condorcet. Condorcet specifies that the winner is the candidate who beats every other candidate in a pairwise matching. The single drawback of Condorcet - the possibility of having no Condorcet winner - is overblown. It's the equivalent of a tie in a plurality election. If the current standard tiebreaker mechanism is not good enough (eg another election), then the tie could be broken by taking the same ranked choices and translating them into a heuristic based on approval voting, score voting, etc.
The value of STAR voting is that it reduces even further the incentive to bullet vote. By introducing a Condorcet-esque runoff of the top two finalists, every voter is strongly incentivized to use meaningful middle scores in order to have a say in the outcome of the runoff.
IRV/RCV, on the other hand, can punish voters for voting honestly. The more viable candidates there are in the field, the more likely your top ranking for your honest favorite will result in benefitting your least preferred. Given that one of the qualities most every election reform advocate wants is to see a larger field packed with viable candidates, this drawback of IRV is especially detrimental.
It's a drawback because the input model prevents voters from expressing their actual preference. If individual voters' preferences cannot be meaningfully expressed, then the outcome isn't valid regardless of how reasonable the tallying looks. (see also: Plurality)
For example, let's say that I want L/G to win, but will begrudgingly accept a D win to prevent an R win at all costs. With Approval or Score voting, I am unable to simply express this preference! With Approval voting, my two options are to leave out D and create a higher chance of R winning, or include D and have a higher chance of L/G losing. And Score voting just allows me to create a weighted blend of the two - how much do I want to hurt L/G to help D beat R?
I don't see how STAR doesn't similarly fall to the two party tyranny. The top two scores will be the two popular parties, which votes then choose between. Putting "Condorcet" after another system seems pointless.
> IRV ... The more viable candidates there are in the field, the more likely your top ranking for your honest favorite will result in benefitting your least preferred
I wholeheartedly agree about IRV (which describes a tallying method). I'm talking about reframing "RCV" (which describes the input method, rather than the tallying method) to be tallied with Condorcet rather than greedy-algorithm runoffs. We can then debate which heuristics or procedures are appropriate for breaking any Condorcet ties.
The only voter preference RCV+Condorcet prevents is the ability to express an individual circular preference (A>B>C>A). But I've seen no arguments about why that would even be a reasonable preference to want to express.