suppose the democratic primary used guthrie voting.
pulling information from your article:
support (assets) for each candidates is: Cuomo 41, Mamdani 30, Lander 8.5, Adams 7.
the candidates rank each other this way: C:ALM, M:LAC, L:MAC, A:CLM.
no majority candidate.
go to contingent negotiation rounds.
assume everyone votes honestly.
in the first elimination round, last place votes: M=41(C)+7(A)=48, C=30(M)+8.5(L)=38.5.
M is eliminated.
new rankings: C:AL, M:LAC, L:AC, A:CL.
first place votes: C=41(C), L=30(M)+8.5(L)=38.5, A=7(A).
no majority.
last place votes: L=41(C)+7(A)=48, C=30(M)+8.5(L)=38.5.
L is eliminated.
new rankings: C:A, M:AC, L:AC, A:C
first place votes: C=41(C), A=30(M)+8.5(L)+7(A)=45.5.
Adams wins.
and not a single voter had to do anything other than pick their favorite candidate.
i saw a lot of chatter on bluesky. people extolling others to use some strategy. usually, don't rank cuomo at all. which suggests to me the voter's optimal strategy isn't obvious. for some, that's a big minus for an electoral system.
If what we are after is simplicity of use, then approval voting might win. No limit on your choices, and implementation just requires turning off the voting machine's alarm for casting overvotes.
agree completely that approval is pretty simple to implement. and it gives good results. and it's easy to explain.
but...
it has some weirdities. like the burr dilemma. you can harm your favorite candidate's chances of winning by approving your second choice. but if you don't support your second, your last place choice could win. voters should approve clones. but it's not obvious how to decide to approve a middle candidate.
i think approval is stealthily the *most* demanding system on the voters. first you rank them. then you score them. then what? research how the election might go if you do and don't support your second. optimal strategy seems to be to roll literal dice.
and to keep beating the horse...
i'm all for the candidates using approval voting in negotiation rounds when none have a majority from the voters.
though when i did this the naive way in simulations it didn't work as well as coombs. will revisit*.
*=gotta build a pair of cabinets for spouse's new bathroom first. ;->
since ny voters have gone to the trouble of rank-ordering the candidates...
and they appear to be behaving as if there's a single issue axis...
why use instant run-off?
why not use condorcet?
it should be well-known that irv suffers from the center-squeeze effect. ie the condorcet winner is eliminated?
and if you really want to fix voting in the us... consider guthrie voting.
it's asset voting and a coombs-like method for negotiating a winner when none have a majority.
easy to run.
gives good results - always picks the condorcet winner when there's a single issue axis.
no spoiler effect.
no center squeeze effect.
https://docs.google.com/document/d/1GL__lJMoX5Cku35h4BLXhJHQ_NxuzGaA5tN-OORVdmw/edit?tab=t.0
The main issue with Condorcet voting is that it is unavailable as a practical reform that would ever be enacted anywhere.
So, one should accept the possibility of a rule that approximates its effects. There is no payoff in being a purity troll.
we are in complete agreement.
on a voter satisfaction scale, random=0, plurality=3, irv=7, condorcet=borda=approval=range=guthrie=9.
it would be really nice if that rule of approximation was a 9 instead of a 7.
i'm not advocating for condorcet.
i'm advocating for a system that's dead stupid simple for the voters and for the people who run the election.
and at the same time gives results comparable to the best the pure mathematicians have been able to devise.
and if voters behave as if there's a condorcet ordering then all of the 9s listed above pick the condorcet winner.
on the other hand, you're probably correct.
guthrie is a form of asset voting. which has to overcome the not-how-we-do-things-around-here barrier.
suppose the democratic primary used guthrie voting.
pulling information from your article:
support (assets) for each candidates is: Cuomo 41, Mamdani 30, Lander 8.5, Adams 7.
the candidates rank each other this way: C:ALM, M:LAC, L:MAC, A:CLM.
no majority candidate.
go to contingent negotiation rounds.
assume everyone votes honestly.
in the first elimination round, last place votes: M=41(C)+7(A)=48, C=30(M)+8.5(L)=38.5.
M is eliminated.
new rankings: C:AL, M:LAC, L:AC, A:CL.
first place votes: C=41(C), L=30(M)+8.5(L)=38.5, A=7(A).
no majority.
last place votes: L=41(C)+7(A)=48, C=30(M)+8.5(L)=38.5.
L is eliminated.
new rankings: C:A, M:AC, L:AC, A:C
first place votes: C=41(C), A=30(M)+8.5(L)+7(A)=45.5.
Adams wins.
and not a single voter had to do anything other than pick their favorite candidate.
i saw a lot of chatter on bluesky. people extolling others to use some strategy. usually, don't rank cuomo at all. which suggests to me the voter's optimal strategy isn't obvious. for some, that's a big minus for an electoral system.
If what we are after is simplicity of use, then approval voting might win. No limit on your choices, and implementation just requires turning off the voting machine's alarm for casting overvotes.
Francois Durand has done work on approval voting suggesting that it generally elects the Condorcet winner and is hard to game. See generally https://scholar.google.com/citations?view_op=list_works&hl=en&hl=en&user=pg5Gqt4AAAAJ&sortby=pubdate and recently, https://scholar.google.com/scholar?oi=bibs&hl=en&cluster=501272500949351472
agree completely that approval is pretty simple to implement. and it gives good results. and it's easy to explain.
but...
it has some weirdities. like the burr dilemma. you can harm your favorite candidate's chances of winning by approving your second choice. but if you don't support your second, your last place choice could win. voters should approve clones. but it's not obvious how to decide to approve a middle candidate.
i think approval is stealthily the *most* demanding system on the voters. first you rank them. then you score them. then what? research how the election might go if you do and don't support your second. optimal strategy seems to be to roll literal dice.
and to keep beating the horse...
i'm all for the candidates using approval voting in negotiation rounds when none have a majority from the voters.
though when i did this the naive way in simulations it didn't work as well as coombs. will revisit*.
*=gotta build a pair of cabinets for spouse's new bathroom first. ;->