ITHACA, N.Y. — A group of local residents are trying to change how elections are decided in the city of Ithaca. Tom Clavel and Patrick Sewell, two leaders of the […]
… it’s a concise illustration of the core problem.
RCV only cares about top votes - it can easily eliminate compromise candidates, just because they’re less popular as a first choice. Consider the following much-more-plausible election:
40% want A > B > C.
35% want C > B > A.
25% want B > C > A.
FPTP says A wins with a plurality of 40%, because FPTP sucks.
RCV says B is eliminated and C beats A. Even though everyone who wanted A > C would prefer B. And if A beat C, everyone who wanted C > A would also have preferred B.
Ranked Pairs says A vs B is 40-60 for B, A vs C is 75-25 for C, and B vs C is 65-35 for B. The Condorcet winner is B. Why should it be anyone else?
That’s such an absurd manufactured edge case as to not be worth considering.
… it’s a concise illustration of the core problem.
RCV only cares about top votes - it can easily eliminate compromise candidates, just because they’re less popular as a first choice. Consider the following much-more-plausible election:
40% want A > B > C.
35% want C > B > A.
25% want B > C > A.
FPTP says A wins with a plurality of 40%, because FPTP sucks.
RCV says B is eliminated and C beats A. Even though everyone who wanted A > C would prefer B. And if A beat C, everyone who wanted C > A would also have preferred B.
Ranked Pairs says A vs B is 40-60 for B, A vs C is 75-25 for C, and B vs C is 65-35 for B. The Condorcet winner is B. Why should it be anyone else?