I recall there was some discussion about different voting systems for the Donation Election. I think it would be interesting to see whether/âhow different the results would have been with a different system. We canât really retrofit the data to quadratic voting, but I think it would be possible for approval voting and rank-choice voting (for approval voting we could just say that any candidate receiving points from a person is approved, unless they voted for many candidates in which case just the top 12). I am not sure what codebase you have at the backend for this, but I could imagine switching the voting system could be fairly easy; maybe it isnât worth it though.
Thanks for bringing this up! I actually explored this a bit, but hid it in a footnote:
Interestingly, I think approval voting would have yielded similar winners (with a big assumption about how to extrapolate what people would have âapprovedâ of) â if you rank projects by the number of voters who gave it at least 10%, for instance, the picture doesnât change much (if you increase from 10% to 20%, LTFF starts showing up in the top 3 again).
What would have happened with ranked-choice voting is [more] unclear to me; the voting system was in fact very different, and my quick attempts at trying to see what would happen if I tried to interpret the scores as ranks were extremely contingent on minor tweaks in how I interpret things and on decisions like whether I used ârankingsâ of all the candidates or only the top 10, etc. If I only rank points given when they amount to over 5% of a voterâs total points and add some randomness to produce rankings out of equal scores, the results look fairly similar, but I didnât do things carefully.
I was initially lazier with ranked-choice voting, but now finished a rough ~run of the elimination process and got very similar results as what we got with our voting system. Note that I chose the 5% cutoff as the thing that seemed most reasonable given how people voted before I saw the results with this translation into RCV, but things might in fact still change quite a bit if you choose a different translation into RCV (e.g. only the top 5 projects people gave points to, instead of the projects people gave at least 5% of their points to). The ranking I got was:
RP
CEâs Fund
AWF (very close call; I had added some randomness to get orderings and on some reloads it was beating CE, but that seemed rarer than CE beating AWF)
LTFF
GW
GD
AMF
THL
GFI
ALLFED
Etc.
(Please donât trust these results too much, though; I was not careful here.)
Whoops my mistake. OK thanks, interesting! Maybe next year we can have an informal meta-vote beforehand on which voting system we want to use ;) I think currently I am in favour of RCV but maybe I am biased by being Australian and the fact that we use that here, so it seems especially intuitive and nice to me.
Someone else I talked to is also in favor of RCV, and I agree that it has benefits (e.g. easier to use than this system), but I also think it has some downsides â e.g. I think itâs a worse exercise for voters than this system is. Btw, you might also be interested in the discussion that happened on my quick take before we decided on a voting system.
(My current top, low-resilience guess about changes we should make to the voting system, if we ran this again, is that we should remove the 3-winner restriction and that we should think about trying to get people to vote on cause/âproblem areas â separate from voting on charities.)
Nice, wow there was lots of engagement on this beforehand! I think I am now leaning towards abrahamroweâs suggestion to just take the average of everyoneâs distributions, possibly with some minimum threshold to avoid the hassle of disbursing small amounts of money. But so many considerationsâa more complicated decision than initially meets the eye I think.
I object to your translation of actual-votes into approval-votes and RCV-votes, at least in the case of my vote. I gave almost all of my points to my top pick, almost all of the rest to my second pick, almost all of the rest to my third pick, and so forth until I was sure I had chosen something that would make top 3. But e.g. I would have approved of multiple. (Sidenote: I claim my strategy is optimal under very reasonable assumptions/âapproximations. You shouldnât distribute points like youâre trying to build a diverse portfolio.)
Thanks! I agree that the approach you describe is optimal under very reasonable assumptions, but I think in practice few people used it (the median ratio between someoneâs top choice and their second choice was 2, the mean if you throw out one outlier was ~20; only 7 people voted for at least 2 candidates and had ratios between their top two that were at least 20). Moreover, we had some[1] voters who didnât vote the way you describe, but who did assign a fairly big number of projects similar small point values â I think kind of throwing in some points for charities they donât favor that much, and I didnât want to overweight their votes in the way I tallied the RCV-translated (or approval-translated) scores.
Still, I agree that my translations are bad â I should at least represent scores from people who basically approximated RCV in the current voting method the way they would be counted in RCV. I might try this (and think about what translation actually makes sense â just the top 10 charities people voted for?) later, but might not prioritize doing it.
For approval voting, you could also just look at the number of voters who gave a charity any (positive) number of points; these counts are included in this post and wouldnât have changed the top 3.
Quickly estimating: there were 90 voters who voted for at least 4 candidates whose last two votes differed by a ratio of less than 1.5. There were 27 if instead of requiring at least 4 candidates, you require that the smallest point assignment is <5.
Or looking at it another way: across all ratios (across all voters) between what a given voter gave the candidate they ranked N and the candidate they ranked N+1, if we remove only the top 1 percentile of ratios (removing because a few people did use an approximation of RCVâequivalent in this case to removing ratios higher than 100:1), the mean is 2. Across all ~12K ratios, about 500 are exactly 1.
There are many different voting methods that use ranked ballots, and itâs frustrating that one of the worst gets all the attention, and is considered synonymous with the term âranked choice votingâ.
Do the ranked ballots produce a Condorcet winner? A strict Condorcet ranking of the rest?
I recall there was some discussion about different voting systems for the Donation Election. I think it would be interesting to see whether/âhow different the results would have been with a different system. We canât really retrofit the data to quadratic voting, but I think it would be possible for approval voting and rank-choice voting (for approval voting we could just say that any candidate receiving points from a person is approved, unless they voted for many candidates in which case just the top 12). I am not sure what codebase you have at the backend for this, but I could imagine switching the voting system could be fairly easy; maybe it isnât worth it though.
Thanks for bringing this up! I actually explored this a bit, but hid it in a footnote:
I was initially lazier with ranked-choice voting, but now finished a rough ~run of the elimination process and got very similar results as what we got with our voting system. Note that I chose the 5% cutoff as the thing that seemed most reasonable given how people voted before I saw the results with this translation into RCV, but things might in fact still change quite a bit if you choose a different translation into RCV (e.g. only the top 5 projects people gave points to, instead of the projects people gave at least 5% of their points to). The ranking I got was:
RP
CEâs Fund
AWF (very close call; I had added some randomness to get orderings and on some reloads it was beating CE, but that seemed rarer than CE beating AWF)
LTFF
GW
GD
AMF
THL
GFI
ALLFED
Etc.
(Please donât trust these results too much, though; I was not careful here.)
Re quadratic voting, see also here.
Whoops my mistake. OK thanks, interesting! Maybe next year we can have an informal meta-vote beforehand on which voting system we want to use ;) I think currently I am in favour of RCV but maybe I am biased by being Australian and the fact that we use that here, so it seems especially intuitive and nice to me.
Someone else I talked to is also in favor of RCV, and I agree that it has benefits (e.g. easier to use than this system), but I also think it has some downsides â e.g. I think itâs a worse exercise for voters than this system is. Btw, you might also be interested in the discussion that happened on my quick take before we decided on a voting system.
(My current top, low-resilience guess about changes we should make to the voting system, if we ran this again, is that we should remove the 3-winner restriction and that we should think about trying to get people to vote on cause/âproblem areas â separate from voting on charities.)
Nice, wow there was lots of engagement on this beforehand! I think I am now leaning towards abrahamroweâs suggestion to just take the average of everyoneâs distributions, possibly with some minimum threshold to avoid the hassle of disbursing small amounts of money. But so many considerationsâa more complicated decision than initially meets the eye I think.
I object to your translation of actual-votes into approval-votes and RCV-votes, at least in the case of my vote. I gave almost all of my points to my top pick, almost all of the rest to my second pick, almost all of the rest to my third pick, and so forth until I was sure I had chosen something that would make top 3. But e.g. I would have approved of multiple. (Sidenote: I claim my strategy is optimal under very reasonable assumptions/âapproximations. You shouldnât distribute points like youâre trying to build a diverse portfolio.)
Thanks! I agree that the approach you describe is optimal under very reasonable assumptions, but I think in practice few people used it (the median ratio between someoneâs top choice and their second choice was 2, the mean if you throw out one outlier was ~20; only 7 people voted for at least 2 candidates and had ratios between their top two that were at least 20). Moreover, we had some[1] voters who didnât vote the way you describe, but who did assign a fairly big number of projects similar small point values â I think kind of throwing in some points for charities they donât favor that much, and I didnât want to overweight their votes in the way I tallied the RCV-translated (or approval-translated) scores.
Still, I agree that my translations are bad â I should at least represent scores from people who basically approximated RCV in the current voting method the way they would be counted in RCV. I might try this (and think about what translation actually makes sense â just the top 10 charities people voted for?) later, but might not prioritize doing it.
For approval voting, you could also just look at the number of voters who gave a charity any (positive) number of points; these counts are included in this post and wouldnât have changed the top 3.
Quickly estimating: there were 90 voters who voted for at least 4 candidates whose last two votes differed by a ratio of less than 1.5. There were 27 if instead of requiring at least 4 candidates, you require that the smallest point assignment is <5.
Or looking at it another way: across all ratios (across all voters) between what a given voter gave the candidate they ranked N and the candidate they ranked N+1, if we remove only the top 1 percentile of ratios (removing because a few people did use an approximation of RCVâequivalent in this case to removing ratios higher than 100:1), the mean is 2. Across all ~12K ratios, about 500 are exactly 1.
There are many different voting methods that use ranked ballots, and itâs frustrating that one of the worst gets all the attention, and is considered synonymous with the term âranked choice votingâ.
Do the ranked ballots produce a Condorcet winner? A strict Condorcet ranking of the rest?