I’m going to stick my neck out and say that approval voting is the best option here. Why?
It avoids almost all of the problems with plurality voting. In non-pathological arrangments of voter preferences and candidates, it will produce the ‘intuitively’ correct option—see here for some fun visualisations.
It has EA cred, see Aaron Hamlin’s interview on 80k here
And most importantly, it’s understandable and legible—you don’t need people to trust an underlying apportionment algorithm or send the flyers explaining theD’Hondt method to voters or whatever. Just vote for the options you approve of on the ballot. One person, one ballot. Most approvals wins. Simple.
I fear that EAs who are really into this sort-of thing are going to nerd-snipe the whole thing into a discussion/natural experiment about optimal voting systems instead of what would be most practical for this Donation Election. A lot of potential voters and donors may not be interested in using a super fancy optimal but technically involved voting method, and be the kind of small inconvenience that might turn people off the whole enterprise.
Now, before all you Seeing Like a State fans come at me saying how legibility is the devil’s work I think I’m just going to disagree with you pre-emptively.[1] Sometimes there is a tradeoff between fidelity and legibility, and too much weighting on illegible technocracy can engender a lack of trust and have severe negative consequences.
Actually it’s interesting that Glen references Scott as on his side, I think there’s actually some tension between their positions. But that’s probably a topic for another post/discussion
Won’t people be motivated to disapprove vote orgs in all cause areas but their preferred one? That would seemingly reduce approval voting to FPTP as between cause areas in effect.
Well, the top 3 charities will get chosen, so there’s no benefit to you only selecting 1 option alone unless you really do believe only that 1 charity ought to get funded. I think AV may be more robust to these concerns than some think,[1] all I think all voting systems will have these edge cases.
I also may be willing simply bite the bullet here and trade-off a bit of strategic voting for legibility. But again, I don’t think approval is worse than this than many other voting methods.
But my fundamental objection is that this is primarily a normative problem, where we want to be a community who’ll vote honestly and not strategically. If GWWC endorse approval voting, then when you submit your votes there could be a pop-up with “I pledge not to vote strategically” or something like that.
I don’t think any voting system is immune to that—Democracy works well because of the norms it spreads and trust it instills, as opposed to being the optimal transmission mechanism of individual preferences to a social welfare function imho.
Thanks. I assume there will be at least 3 orgs for each cause area.
If we can assume the forum is a “community who’ll vote honestly and not strategically,” approval voting would work—but we shouldn’t limit the winners to three in that case. Proportional representation among all orgs with net positive approval would be the fullest extent of the community’s views, although some floor on support or cap on winners would be necessary for logistical reasons.
I’m going to stick my neck out and say that approval voting is the best option here. Why?
It avoids almost all of the problems with plurality voting. In non-pathological arrangments of voter preferences and candidates, it will produce the ‘intuitively’ correct option—see here for some fun visualisations.
It has EA cred, see Aaron Hamlin’s interview on 80k here
And most importantly, it’s understandable and legible—you don’t need people to trust an underlying apportionment algorithm or send the flyers explaining the D’Hondt method to voters or whatever. Just vote for the options you approve of on the ballot. One person, one ballot. Most approvals wins. Simple.
I fear that EAs who are really into this sort-of thing are going to nerd-snipe the whole thing into a discussion/natural experiment about optimal voting systems instead of what would be most practical for this Donation Election. A lot of potential voters and donors may not be interested in using a super fancy optimal but technically involved voting method, and be the kind of small inconvenience that might turn people off the whole enterprise.
Now, before all you Seeing Like a State fans come at me saying how legibility is the devil’s work I think I’m just going to disagree with you pre-emptively.[1] Sometimes there is a tradeoff between fidelity and legibility, and too much weighting on illegible technocracy can engender a lack of trust and have severe negative consequences.
Actually it’s interesting that Glen references Scott as on his side, I think there’s actually some tension between their positions. But that’s probably a topic for another post/discussion
Won’t people be motivated to disapprove vote orgs in all cause areas but their preferred one? That would seemingly reduce approval voting to FPTP as between cause areas in effect.
Well, the top 3 charities will get chosen, so there’s no benefit to you only selecting 1 option alone unless you really do believe only that 1 charity ought to get funded. I think AV may be more robust to these concerns than some think,[1] all I think all voting systems will have these edge cases.
I also may be willing simply bite the bullet here and trade-off a bit of strategic voting for legibility. But again, I don’t think approval is worse than this than many other voting methods.
But my fundamental objection is that this is primarily a normative problem, where we want to be a community who’ll vote honestly and not strategically. If GWWC endorse approval voting, then when you submit your votes there could be a pop-up with “I pledge not to vote strategically” or something like that.
I don’t think any voting system is immune to that—Democracy works well because of the norms it spreads and trust it instills, as opposed to being the optimal transmission mechanism of individual preferences to a social welfare function imho.
or here: https://link.springer.com/chapter/10.1007/978-1-4419-7539-3_2
Thanks. I assume there will be at least 3 orgs for each cause area.
If we can assume the forum is a “community who’ll vote honestly and not strategically,” approval voting would work—but we shouldn’t limit the winners to three in that case. Proportional representation among all orgs with net positive approval would be the fullest extent of the community’s views, although some floor on support or cap on winners would be necessary for logistical reasons.