As one data point: I’m very positive about CES, and think they’re one of the best marginal uses of funding right now. (Note that Aaron didn’t ask me to write this.)
(Ties: I’ve recommended a grant to CES from Open Phil before, and a further grant is under consideration at OP right now; even given this possible grant, CES would have need for further funding for the coming years.)
I think it would be very helpful to do an explicit cost effectiveness model for approval voting. Of course there would be a lot of uncertainty, but not necessarily more than AI or alternate foods. It could be for the present generation (like this or this for alternate foods) or for the long term future, like this for alternate foods and AI. Then we would have at least some quantitative way of comparing.
Congrats on your win in Fargo! Here are some thoughts.
Like you, I favor approval voting over IRV (based on a cursory assessment). However, I actually like the idea of some states using approval voting and some states using IRV. My feeling is that every voting method, including approval voting/score voting, likely has its own set of flaws. Some flaws may only become apparent when the method is used in a sequence of large elections over an extended period of time, as strategic voting, polling behavior, party politics, and election coverage reach an equilibrium.
In machine learning, a common strategy to improve performance is to make use of a collection of different machine learning algorithms. The idea is that any individual algorithm makes mistakes, but by taking the average output across all the different algorithms, individual mistakes often cancel each other out. In the same way, if different states make use of different voting methods, hopefully the flaws in each individual voting method will tend to cancel each other out. The best strategy becomes being the best candidate possible instead of trying to game an array of different voting systems simultaneously.
(This doesn’t diminish the urgency of making sure that a decent proportion of the electorate makes use of approval voting. Just saying it might not be a tragedy if IRV gets a decent fraction of the vote share, and IRV/approval voting advocates should work together to expand use of better voting systems, especially in swing states! Actually, speaking of swing states, is there any technique that could be used by a state to allocate its electoral votes strategically? Suppose Gary Johnson gets the most approvals in Florida. Seems like Florida residents could be pretty upset by their electoral votes being “thrown away” if Johnson isn’t given electoral votes by any other state. Is their any legal provision for Florida waiting until other states have had their electoral votes allocated before they get allocated in Florida? Though, even if there is, this gets complicated if multiple states were using such a procedure...)
---
Another random thought. Has anyone studied the properties of the following “instant elimination voting” procedure:
Figure out which candidate was ranked last by the largest number of voters.
Eliminate that candidate. For every voter who ranked that candidate last, their second-to-last choice is now treated as their last choice.
Repeat until all but one candidate has been eliminated.
I presented it as IRV in reverse, but one can imagine variants. The objective here is to minimize our probability of electing a heavily disliked candidate as opposed to maximize our probability of electing a heavily liked candidate. IMO this is a reasonable objective—I suspect in general, it’s easier for a bad candidate to do a lot of harm than a good candidate to do a lot of good.
I know “disapproval voting” is mathematically equivalent to approval voting, but it seems like from a behavioral economics standpoint, they might not be equivalent. If disapproval voting causes voters to “default approve” candidates they’re unfamiliar with, and approval voting causes voters to “default disapprove” candidates they’re unfamiliar with (see: research into opt-in vs opt-out organ donation), that suggests obscure 3rd party or independent compromise candidates will do better in disapproval voting. Not sure if this is a good or bad thing. And I suppose a reddit style upvote/downvote scheme would average out the biases of disapproval and approval voting.
For every voter who ranked that candidate last, their second-to-last choice is now treated as their last choice.
That’s not Coomb’s method, but might be equivalent? I’m a little confused by this. Remember that after that candidate is eliminated, they are removed from all ballots, not just the ones that listed them last.
On a mix of voting methods being used across states
I think this mixed approach is good. I wouldn’t worry about IRV not getting a foothold here though. My worry is IRV taking over the map. Also, there are other places that use IRV like Australia. To my knowledge, there’s been limited research on international and US IRV data. That could change once we have our director of research position.
Approval voting in presidential elections.
I’d point to the FAQ on presidential elections here. If you’re using approval voting, it should be part of the national popular vote. Otherwise, you get strange tactics coming up.
IRV variants & variants in general
One way to evaluate the performance of a voting method in electing a good winner is to use simulations. You have a computer take different conditions like candidate scenarios or different kinds of voters (you might call these dials) and then run these elections millions of times. Then you can see how much of the maximum utility was captured by the voting method.
We’re dealing with a model here, so there are assumptions that will vary from simulation to simulation. But in the models I’ve seen, approval voting fares well even with tactical voters, particularly against IRV and way better than the choose-one method. Also, there is a “magical best” in these simulations. That is, the “magical best” voting method magically picks the winner in each election that maximizes voter utility. “Magical best” is the unattainable ceiling of voting method performance in choosing a good winner.
That “magical best” mark is not terribly far away from approval voting’s mark. And you could get some small but likely meaningful increases by going from approval to score/range voting. Once you’re at score voting though, there’s not a lot to be gained.
And it’s because there’s little utility to be gained beyond score voting (and not much even beyond approval) that I don’t get particularly excited about the fanciest idea of a new voting method. And believe me, I hear all kinds of those ideas. I’ve actually gotten phone calls from random people on the weekend concerning this before.
The point is, we have some voting methods like approval that do really well in electing high-utility winners that are so easy. And not only is it easy but it has perks like giving an accurate reflection of support for losing candidates. And it is precinct summable and easily auditable. These factors are important. It’s not just about maximizing utility from the winner. A voting method has other jobs, too. (See the FAQ on how to evaluate a voting method.)
In terms of CES mission strategy, score voting is really the only other single-winner voting method that makes sense for us to try because it has so much simplicity going for it as well. It just has some small implementation hurdles and slightly more complexity that approval voting doesn’t have. But it’ll likely be a little bit before we consider anything with score voting, and it’ll have to be a strategic target. One step at a time, as they say.
This sounds very interesting! I’m trying to understand it better and have a few questions:
1a. You mentioned that other groups are trying to implement IRV and may not even understand that approval voting is superior. Can you explain why you think other people think this and even advocate for apparently inferior methods? Your article seemed convincing at first glance and I don’t think this is a particularly partisan issue.
1b. “We also haven’t faced organized opposition.” What kind of opposition do you anticipate facing? IRV supporters? Elected officials?
2. Since legislative reform is a nonstarter according to you and ballot initiatives for changing voting methods are present in less than half of US states, what is the medium-long term plan? Get as many cities on approval voting as possible and hope that this builds pressure for approval voting nationally?
3. What factors led to the convincing margin of victory in Fargo despite it being seen as a long shot by the media?
4. What would make you change your mind about approval voting being the best option to advocate for?
1a. You mentioned that other groups are trying to implement IRV and may not even understand that approval voting is superior. Can you explain why you think other people think this and even advocate for apparently inferior methods? Your article seemed convincing at first glance and I don’t think this is a particularly partisan issue.
I think the relative support issue is a matter of those IRV advocates deweighting the likelihood of failures in IRV and overweighting the value of IRV’s existing use compared to approval voting. There’s also a lack of knowledge on many of the nuances, which is just a product of voting theory being so complicated.
1b. “We also haven’t faced organized opposition.” What kind of opposition do you anticipate facing? IRV supporters? Elected officials?
With ballot initiatives, the most likely opposition will come from those who currently benefit from a choose-one voting method. If an official or party would have likely won under a choose-one method versus approval voting, then they’ll likely oppose approval voting.
We don’t anticipate IRV supporters to oppose specific approval voting ballot initiatives, though we have seen articles from those groups publicly attacking approval voting. In public venues, we’re often bumped to keep room for IRV speakers even if there’s redundancy. This exclusion can affect our perceived legitimacy with donors, media, and other people in the reform network. We’ll likely have to continue with large ballot initiative wins before excluding us becomes unacceptable.
As an opposition example, there was a piece of state legislation that had enabling language permitting cities to use approval voting and IRV. A left-leaning organization opposed the inclusion of approval voting because they thought (rightfully) that approval voting would elect a more moderate government. In my conversation with the left-leaning organization, they told me that with IRV, they’d at least have some wins—even if there was some back and forth losses in the complete opposite direction. I told the person that if they wanted more partisan ideologies to be represented then they should support including proportional methods in the enacting language, but they seemed uninterested in listening to me at that point.
2. Since legislative reform is a nonstarter according to you and ballot initiatives for changing voting methods are present in less than half of US states, what is the medium-long term plan? Get as many cities on approval voting as possible and hope that this builds pressure for approval voting nationally?
I think yes. It’s hard to understate how frantic our pace is compared to how long it took IRV to move with reforms. That said, if we don’t have sufficient funding to run large initiatives, then approval voting could get shut out early. There’s a lot riding on us having that immediate momentum.
Also, while legislative reform is most likely not on the table now, it may be in the future. But that’s likely only if approval voting has more of a track record and the choose-one method is more regularly publicly scorned.
3. What factors led to the convincing margin of victory in Fargo despite it being seen as a long shot by the media?
This is a technical subject, and I think it just takes too much energy for media to gather the information necessary to make a more accurate prediction. For instance, I would classify “long shot” as an event less than 10% likely to occur. Yet, using base rates from similar initiatives and eventually polling, there was never an indication that the odds were ever anywhere near that low. In my calculation, I don’t think my assessment ever dipped to 50%.
Most initiatives focusing on single-winner voting methods pass. Not all of them, as we’ve learned since, but the vast majority do. We also had a convenient narrative in Fargo. Their commission created a task force that recommended approval voting—which the commission then ignored. We also had strong support on the ground. I also had the benefit of talking with lots of other people who successfully ran initiatives at a conference in earlier 2018.
So long as there’s sufficient funding, we’ll only get better at this.
4. What would make you change your mind about approval voting being the best option to advocate for?
Part of the strategic rationale for going with approval voting versus a higher-utility method that’s more complicated is that a more complicated method is less likely to get enacted. Another higher-utility method would be range/score voting where voters score each candidate on a scale. It also has a number of desirable qualities in terms of practicality (though it may take some effort to have it work on the worst of US voting machines).
There appears to be a small but measurable gain in utility going from approval to score where there is little added complexity. If we were to advance score voting in the future, for instance, we’d have to repeat the same process we did with approval voting (i.e.: proof of concept, replication, then scale).
Beyond score voting, there is little to be gained in utility between where score voting lands and where an unattainable magic best voting method would be. There are lots of other variations that scatter around this space, but those methods often add extra complexity and present practical implementation burdens that may also reduce their impact. Keep in mind, a voting method’s simplicity also helps its ability to do other jobs like convey support for other candidates who didn’t win.
I’m at a loss to imagine what might be present for the current choose-one method to be preferred over approval voting. If it somehow showed evidence of consistently electing worse candidates, that would be evidence. The same would be true compared to other voting methods. These are the types of empirical questions we can ask by being able to fund a director of research position. It’s clear that without our work that those polls comparing different voting methods just wouldn’t be done.
As one data point: I’m very positive about CES, and think they’re one of the best marginal uses of funding right now. (Note that Aaron didn’t ask me to write this.)
(Ties: I’ve recommended a grant to CES from Open Phil before, and a further grant is under consideration at OP right now; even given this possible grant, CES would have need for further funding for the coming years.)
I think it would be very helpful to do an explicit cost effectiveness model for approval voting. Of course there would be a lot of uncertainty, but not necessarily more than AI or alternate foods. It could be for the present generation (like this or this for alternate foods) or for the long term future, like this for alternate foods and AI. Then we would have at least some quantitative way of comparing.
Congrats on your win in Fargo! Here are some thoughts.
Like you, I favor approval voting over IRV (based on a cursory assessment). However, I actually like the idea of some states using approval voting and some states using IRV. My feeling is that every voting method, including approval voting/score voting, likely has its own set of flaws. Some flaws may only become apparent when the method is used in a sequence of large elections over an extended period of time, as strategic voting, polling behavior, party politics, and election coverage reach an equilibrium.
In machine learning, a common strategy to improve performance is to make use of a collection of different machine learning algorithms. The idea is that any individual algorithm makes mistakes, but by taking the average output across all the different algorithms, individual mistakes often cancel each other out. In the same way, if different states make use of different voting methods, hopefully the flaws in each individual voting method will tend to cancel each other out. The best strategy becomes being the best candidate possible instead of trying to game an array of different voting systems simultaneously.
(This doesn’t diminish the urgency of making sure that a decent proportion of the electorate makes use of approval voting. Just saying it might not be a tragedy if IRV gets a decent fraction of the vote share, and IRV/approval voting advocates should work together to expand use of better voting systems, especially in swing states! Actually, speaking of swing states, is there any technique that could be used by a state to allocate its electoral votes strategically? Suppose Gary Johnson gets the most approvals in Florida. Seems like Florida residents could be pretty upset by their electoral votes being “thrown away” if Johnson isn’t given electoral votes by any other state. Is their any legal provision for Florida waiting until other states have had their electoral votes allocated before they get allocated in Florida? Though, even if there is, this gets complicated if multiple states were using such a procedure...)
---
Another random thought. Has anyone studied the properties of the following “instant elimination voting” procedure:
Figure out which candidate was ranked last by the largest number of voters.
Eliminate that candidate. For every voter who ranked that candidate last, their second-to-last choice is now treated as their last choice.
Repeat until all but one candidate has been eliminated.
I presented it as IRV in reverse, but one can imagine variants. The objective here is to minimize our probability of electing a heavily disliked candidate as opposed to maximize our probability of electing a heavily liked candidate. IMO this is a reasonable objective—I suspect in general, it’s easier for a bad candidate to do a lot of harm than a good candidate to do a lot of good.
I know “disapproval voting” is mathematically equivalent to approval voting, but it seems like from a behavioral economics standpoint, they might not be equivalent. If disapproval voting causes voters to “default approve” candidates they’re unfamiliar with, and approval voting causes voters to “default disapprove” candidates they’re unfamiliar with (see: research into opt-in vs opt-out organ donation), that suggests obscure 3rd party or independent compromise candidates will do better in disapproval voting. Not sure if this is a good or bad thing. And I suppose a reddit style upvote/downvote scheme would average out the biases of disapproval and approval voting.
That’s Coomb’s method, and it generally works better than IRV, but has strategy-susceptibility problems.
That’s not Coomb’s method, but might be equivalent? I’m a little confused by this. Remember that after that candidate is eliminated, they are removed from all ballots, not just the ones that listed them last.
On a mix of voting methods being used across states
I think this mixed approach is good. I wouldn’t worry about IRV not getting a foothold here though. My worry is IRV taking over the map. Also, there are other places that use IRV like Australia. To my knowledge, there’s been limited research on international and US IRV data. That could change once we have our director of research position.
Approval voting in presidential elections.
I’d point to the FAQ on presidential elections here. If you’re using approval voting, it should be part of the national popular vote. Otherwise, you get strange tactics coming up.
IRV variants & variants in general
One way to evaluate the performance of a voting method in electing a good winner is to use simulations. You have a computer take different conditions like candidate scenarios or different kinds of voters (you might call these dials) and then run these elections millions of times. Then you can see how much of the maximum utility was captured by the voting method.
We’re dealing with a model here, so there are assumptions that will vary from simulation to simulation. But in the models I’ve seen, approval voting fares well even with tactical voters, particularly against IRV and way better than the choose-one method. Also, there is a “magical best” in these simulations. That is, the “magical best” voting method magically picks the winner in each election that maximizes voter utility. “Magical best” is the unattainable ceiling of voting method performance in choosing a good winner.
That “magical best” mark is not terribly far away from approval voting’s mark. And you could get some small but likely meaningful increases by going from approval to score/range voting. Once you’re at score voting though, there’s not a lot to be gained.
And it’s because there’s little utility to be gained beyond score voting (and not much even beyond approval) that I don’t get particularly excited about the fanciest idea of a new voting method. And believe me, I hear all kinds of those ideas. I’ve actually gotten phone calls from random people on the weekend concerning this before.
The point is, we have some voting methods like approval that do really well in electing high-utility winners that are so easy. And not only is it easy but it has perks like giving an accurate reflection of support for losing candidates. And it is precinct summable and easily auditable. These factors are important. It’s not just about maximizing utility from the winner. A voting method has other jobs, too. (See the FAQ on how to evaluate a voting method.)
In terms of CES mission strategy, score voting is really the only other single-winner voting method that makes sense for us to try because it has so much simplicity going for it as well. It just has some small implementation hurdles and slightly more complexity that approval voting doesn’t have. But it’ll likely be a little bit before we consider anything with score voting, and it’ll have to be a strategic target. One step at a time, as they say.
(Bayesian Regret example for reference)
This sounds very interesting! I’m trying to understand it better and have a few questions:
1a. You mentioned that other groups are trying to implement IRV and may not even understand that approval voting is superior. Can you explain why you think other people think this and even advocate for apparently inferior methods? Your article seemed convincing at first glance and I don’t think this is a particularly partisan issue.
1b. “We also haven’t faced organized opposition.” What kind of opposition do you anticipate facing? IRV supporters? Elected officials?
2. Since legislative reform is a nonstarter according to you and ballot initiatives for changing voting methods are present in less than half of US states, what is the medium-long term plan? Get as many cities on approval voting as possible and hope that this builds pressure for approval voting nationally?
3. What factors led to the convincing margin of victory in Fargo despite it being seen as a long shot by the media?
4. What would make you change your mind about approval voting being the best option to advocate for?
1a. You mentioned that other groups are trying to implement IRV and may not even understand that approval voting is superior. Can you explain why you think other people think this and even advocate for apparently inferior methods? Your article seemed convincing at first glance and I don’t think this is a particularly partisan issue.
I think the relative support issue is a matter of those IRV advocates deweighting the likelihood of failures in IRV and overweighting the value of IRV’s existing use compared to approval voting. There’s also a lack of knowledge on many of the nuances, which is just a product of voting theory being so complicated.
1b. “We also haven’t faced organized opposition.” What kind of opposition do you anticipate facing? IRV supporters? Elected officials?
With ballot initiatives, the most likely opposition will come from those who currently benefit from a choose-one voting method. If an official or party would have likely won under a choose-one method versus approval voting, then they’ll likely oppose approval voting.
We don’t anticipate IRV supporters to oppose specific approval voting ballot initiatives, though we have seen articles from those groups publicly attacking approval voting. In public venues, we’re often bumped to keep room for IRV speakers even if there’s redundancy. This exclusion can affect our perceived legitimacy with donors, media, and other people in the reform network. We’ll likely have to continue with large ballot initiative wins before excluding us becomes unacceptable.
As an opposition example, there was a piece of state legislation that had enabling language permitting cities to use approval voting and IRV. A left-leaning organization opposed the inclusion of approval voting because they thought (rightfully) that approval voting would elect a more moderate government. In my conversation with the left-leaning organization, they told me that with IRV, they’d at least have some wins—even if there was some back and forth losses in the complete opposite direction. I told the person that if they wanted more partisan ideologies to be represented then they should support including proportional methods in the enacting language, but they seemed uninterested in listening to me at that point.
2. Since legislative reform is a nonstarter according to you and ballot initiatives for changing voting methods are present in less than half of US states, what is the medium-long term plan? Get as many cities on approval voting as possible and hope that this builds pressure for approval voting nationally?
I think yes. It’s hard to understate how frantic our pace is compared to how long it took IRV to move with reforms. That said, if we don’t have sufficient funding to run large initiatives, then approval voting could get shut out early. There’s a lot riding on us having that immediate momentum.
Also, while legislative reform is most likely not on the table now, it may be in the future. But that’s likely only if approval voting has more of a track record and the choose-one method is more regularly publicly scorned.
3. What factors led to the convincing margin of victory in Fargo despite it being seen as a long shot by the media?
This is a technical subject, and I think it just takes too much energy for media to gather the information necessary to make a more accurate prediction. For instance, I would classify “long shot” as an event less than 10% likely to occur. Yet, using base rates from similar initiatives and eventually polling, there was never an indication that the odds were ever anywhere near that low. In my calculation, I don’t think my assessment ever dipped to 50%.
Most initiatives focusing on single-winner voting methods pass. Not all of them, as we’ve learned since, but the vast majority do. We also had a convenient narrative in Fargo. Their commission created a task force that recommended approval voting—which the commission then ignored. We also had strong support on the ground. I also had the benefit of talking with lots of other people who successfully ran initiatives at a conference in earlier 2018.
So long as there’s sufficient funding, we’ll only get better at this.
4. What would make you change your mind about approval voting being the best option to advocate for?
Part of the strategic rationale for going with approval voting versus a higher-utility method that’s more complicated is that a more complicated method is less likely to get enacted. Another higher-utility method would be range/score voting where voters score each candidate on a scale. It also has a number of desirable qualities in terms of practicality (though it may take some effort to have it work on the worst of US voting machines).
There appears to be a small but measurable gain in utility going from approval to score where there is little added complexity. If we were to advance score voting in the future, for instance, we’d have to repeat the same process we did with approval voting (i.e.: proof of concept, replication, then scale).
Beyond score voting, there is little to be gained in utility between where score voting lands and where an unattainable magic best voting method would be. There are lots of other variations that scatter around this space, but those methods often add extra complexity and present practical implementation burdens that may also reduce their impact. Keep in mind, a voting method’s simplicity also helps its ability to do other jobs like convey support for other candidates who didn’t win.
I’m at a loss to imagine what might be present for the current choose-one method to be preferred over approval voting. If it somehow showed evidence of consistently electing worse candidates, that would be evidence. The same would be true compared to other voting methods. These are the types of empirical questions we can ask by being able to fund a director of research position. It’s clear that without our work that those polls comparing different voting methods just wouldn’t be done.
Thank you! I appreciate these response and the thought behind your initiative. Best of luck! :)