Ranked choice voting fundamentally shifted the dynamics of New York's Mayoral primary race – imagine what it could do on a national scale in the US or UK.
Completely support electoral reform - but I'm aware of lots of criticisms with RCV, especially how despite being better than FPTP it can still be very distorting.
e.g.
• If your second choice candidate gets eliminated early on, when your first choice is eliminated your votes are wasted.
• Ballots have many more opportunities to be spoiled (same ranks, skipped ranks, multiple ranks)
• Ballots have to be tallied centrally and are harder to verify in batches.
They make a persuasive case for STAR voting. I think if we're going to reform our electoral system we need STAR as a more bulletproof alternative to minimise controversies and potential backsliding.
STAR voting looks like a variant of the Borda count (like the Eurovision Song Contest voting system) except that a voter can score multiple candidates the same. The problem with this system is it is easily gamable. In a traditional Borda count (where the voter has to give each candidate a different score) there's incentive to insincerely rank highest the candidate the voter thinks is best able to beat the candidate they hate but who could win. In STAR voting this could be taken to the extreme, because if there are any candidates you really don't want to win, then the best tactical vote is to simply give the candidates you hate scores of 0, and everyone else the highest available score. The system is also prone to tied winners.
RCV is extremely difficult to game. There are theoretical scenarios providing incentives to vote tactically or strategically, but they almost invariably involve such precise knowledge of the voting patterns of other voters that they would not work in reality and could easily backfire. The most effective vote under RCV is a sincere ranking of the voter's own preferences among the candidates. If you no longer care, just stop. Incidentally, if your 2nd choice is eliminated before your 1st choice, then your vote passes onto your 3rd choice, or 4th or 5th and so on. You never know, one of those could be the winner.
I'll also mention in passing "Ranked Robin", aka Condorcet. The biggest problem with this one, apart from the exponentially rising number of comparisons required with increasing number of candidates, is that it doesn't always produce a definitive "winner".
STAR is quite different from a Borda count and other forms of score-based voting because of the automatic runoff element.
Without the runoff, STAR would just be score voting where the scores given for each candidate are added up, and whoever scores highest overall is the winner. With that system you’d be incentivised to score any candidate you approve of 5/5, and any you disapprove 0/5 and you effectively have approval voting instead, but you’re penalising anyone who doesn’t optimise for this strategy.
I think we’d agree that doesn’t really work.
But, introducing an automatic runoff, where your full vote goes to whichever of the leading two candidates you prefer (or is counted as a vote of no preference), adds an extra incentive to only score candidates equally if you’d actually be happy for your vote to be one of no preference. In a lot of cases, you might genuinely just want any, for example, left wing party to beat any right-wing party, in which case giving them all 5/5 is slightly strategic but reflects your genuine strong preference to avoid any right-wing party.
However, if you only really want REF to win over CON but CON over anyone else, your best option is to give REF 5/5 and CON 4/5 so your vote is counted in the runoff between the two if they both do well. Extend that further, and maybe you really want CON to beat LAB, but would want LAB to beat any more left-wing party. You might then give LAB 1/5. All of a sudden, your ballot is starting to reflect your genuine political preferences. Only giving CON 4/5 does penalise them slightly but they’ll still get your vote if they make it to a runoff, and if they don’t I think that would just reflect the electorate’s genuinely stronger support for the other side.
You don’t need to give everyone except your hated candidate 5/5, you just need to give them more than 0 but still get to express all sorts of preferences between them. The runoff forces you to make a tough call as to whether you’d rather give stronger support in the first phase or express a preference in the runoff. Voters would make that decision based on how cautious they feel and how acceptable they find their lesser evils. I would argue that means the strategy is just to think carefully about your vote and be honest, capturing that is good design.
Under RCV I think a lot of the strength of voters’ preferences aren’t properly accounted for. I might be disappointed to have my 2nd, 3rd, 4th, etc choices get eliminated before my vote eventually transfers straight to my 6th choice. I’d probably feel plenty of voter regret about not putting my 2nd or 3rd choice first instead and helping them stay in the running longer to pick up more support. Now I’d be thinking about strategically changing my 1st choice vote. Under STAR at least I can always give my favourite 5/5 with no repercussions, even if it might be sensible to also give higher scores to my 2nd or 3rd choices, perhaps not always ideal, but I don’t see that strategy as being quite as distortive in elections with lots of similarly performing candidates.
don't necessarily disagree (for Britain, we advocate for a commission on electoral reform to deliberate on which system makes most sense) but the difference between RCV and FPTP is palpable
A great read. Thanks! Gives a lot of food for thought.
Anathema to Trump, Farage and, er, Starmer and Badenoch.
Completely support electoral reform - but I'm aware of lots of criticisms with RCV, especially how despite being better than FPTP it can still be very distorting.
e.g.
• If your second choice candidate gets eliminated early on, when your first choice is eliminated your votes are wasted.
• Ballots have many more opportunities to be spoiled (same ranks, skipped ranks, multiple ranks)
• Ballots have to be tallied centrally and are harder to verify in batches.
This page by Equal Vote explains it much better detail: https://www.equal.vote/rcv_v_star
***
They make a persuasive case for STAR voting. I think if we're going to reform our electoral system we need STAR as a more bulletproof alternative to minimise controversies and potential backsliding.
STAR voting looks like a variant of the Borda count (like the Eurovision Song Contest voting system) except that a voter can score multiple candidates the same. The problem with this system is it is easily gamable. In a traditional Borda count (where the voter has to give each candidate a different score) there's incentive to insincerely rank highest the candidate the voter thinks is best able to beat the candidate they hate but who could win. In STAR voting this could be taken to the extreme, because if there are any candidates you really don't want to win, then the best tactical vote is to simply give the candidates you hate scores of 0, and everyone else the highest available score. The system is also prone to tied winners.
RCV is extremely difficult to game. There are theoretical scenarios providing incentives to vote tactically or strategically, but they almost invariably involve such precise knowledge of the voting patterns of other voters that they would not work in reality and could easily backfire. The most effective vote under RCV is a sincere ranking of the voter's own preferences among the candidates. If you no longer care, just stop. Incidentally, if your 2nd choice is eliminated before your 1st choice, then your vote passes onto your 3rd choice, or 4th or 5th and so on. You never know, one of those could be the winner.
I'll also mention in passing "Ranked Robin", aka Condorcet. The biggest problem with this one, apart from the exponentially rising number of comparisons required with increasing number of candidates, is that it doesn't always produce a definitive "winner".
Sorry, I have to disagree a little.
STAR is quite different from a Borda count and other forms of score-based voting because of the automatic runoff element.
Without the runoff, STAR would just be score voting where the scores given for each candidate are added up, and whoever scores highest overall is the winner. With that system you’d be incentivised to score any candidate you approve of 5/5, and any you disapprove 0/5 and you effectively have approval voting instead, but you’re penalising anyone who doesn’t optimise for this strategy.
I think we’d agree that doesn’t really work.
But, introducing an automatic runoff, where your full vote goes to whichever of the leading two candidates you prefer (or is counted as a vote of no preference), adds an extra incentive to only score candidates equally if you’d actually be happy for your vote to be one of no preference. In a lot of cases, you might genuinely just want any, for example, left wing party to beat any right-wing party, in which case giving them all 5/5 is slightly strategic but reflects your genuine strong preference to avoid any right-wing party.
However, if you only really want REF to win over CON but CON over anyone else, your best option is to give REF 5/5 and CON 4/5 so your vote is counted in the runoff between the two if they both do well. Extend that further, and maybe you really want CON to beat LAB, but would want LAB to beat any more left-wing party. You might then give LAB 1/5. All of a sudden, your ballot is starting to reflect your genuine political preferences. Only giving CON 4/5 does penalise them slightly but they’ll still get your vote if they make it to a runoff, and if they don’t I think that would just reflect the electorate’s genuinely stronger support for the other side.
You don’t need to give everyone except your hated candidate 5/5, you just need to give them more than 0 but still get to express all sorts of preferences between them. The runoff forces you to make a tough call as to whether you’d rather give stronger support in the first phase or express a preference in the runoff. Voters would make that decision based on how cautious they feel and how acceptable they find their lesser evils. I would argue that means the strategy is just to think carefully about your vote and be honest, capturing that is good design.
Under RCV I think a lot of the strength of voters’ preferences aren’t properly accounted for. I might be disappointed to have my 2nd, 3rd, 4th, etc choices get eliminated before my vote eventually transfers straight to my 6th choice. I’d probably feel plenty of voter regret about not putting my 2nd or 3rd choice first instead and helping them stay in the running longer to pick up more support. Now I’d be thinking about strategically changing my 1st choice vote. Under STAR at least I can always give my favourite 5/5 with no repercussions, even if it might be sensible to also give higher scores to my 2nd or 3rd choices, perhaps not always ideal, but I don’t see that strategy as being quite as distortive in elections with lots of similarly performing candidates.
don't necessarily disagree (for Britain, we advocate for a commission on electoral reform to deliberate on which system makes most sense) but the difference between RCV and FPTP is palpable