Molly Whitevery cool illustration of ranked choice voting and how it worked in Maine’s gubernatorial Democratic primary
via Reddit (https://www.reddit.com/r/Maine/comments/1ua4e7n/sankey_diagram_of_the_democratic_gubernatorial/)
acm@molly0xfff very cool! I'd love for Philadelphia to do this for mayoral primaries, as we have a tendency to have a winner with like 32% of the vote, and the outcomes might involve more buy-in this other way!
Bill Zaumen@acm_redfox @molly0xfff Don't use ranked-choice voting. To see why, read up on Arrow's impossibility theorem. We almost had a failure in California where there was a chance that all the Democrats would be eliminated in the election for Governor because there were two Republicans and a lot more Democrats in the race. Democrats far outnumber Republicans in California.
https://en.wikipedia.org/wiki/Arrow%27s_impossibility_theorem
Todd Knarr@bzdev @acm_redfox @molly0xfff Californiaca primaries don't use RCV. It's "top two go to a runoff". With RCV the outcome you fear can't happen because to win one of the Republicans would have to end up with an outright majority.
RCV does have a flaw but it's one that can affect the outcome only in exceptionally weird and rare circumstances
JTW@tknarr @bzdev @acm_redfox @molly0xfff the problem is that the flaw of instant runoff voting (which is the system being marketed as RCV) arises exactly when the "spoiler" candidate becomes viable. It happened in Alaska in 2022: Begich (center-right) was preferred over Peltola (center-left) by a majority of voters and over Palin (right wing) by a (different) majority. But because he had fewer first choice votes he was eliminated and Peltola was elected over Palin (who was the spoiler).
@jtwcornell91 @bzdev @acm_redfox @molly0xfff It seems like that is working as intended: Begich was the first choice of the fewest voters, so he was dropped and his supporters' votes went to their second-choice candidate. Had either Peltola or Palin been dropped, a larger number of voters would've been relegated to their second-choice candidate. The outcome is the candidate who's the highest-ranked choice for the largest number of voters wins.
@tknarr @jtwcornell91 @acm_redfox @molly0xfff Frankly, you guys don't understand what was proven decades ago. It was that there are cases where any procedure for this type of election fails. Those cases vary depending on the procedure being used. Sometimes it works, but there are *always* cases where it doesn't.
Citing a case where it works merely shows that either you didn't bother to read up on the theorem or that you didn't understand it.
@bzdev @jtwcornell91 @acm_redfox @molly0xfff Yes there are cases where it fails. The same applies to the current system. The failure cases for RCV are rarer and for practical purposes require collusion between voters to force the outcome. I notice you haven't suggested a system with _better_ outcomes than RCV, which is probably because per analysis there aren't any except possibly rank-weighted points.