Quadratic Voting

cfrank

Reading about this system, it appears very nice. It was apparently proposed by Vickrey among other scientists, if you are familiar with a Vickrey auction.

https://en.wikipedia.org/wiki/Quadratic_voting

There was also a recent experiment in Quadratic Voting performed by the caucus in the Colorado House of Representatives in 2019. I am just learning about this now, pretty exciting I think.

Keith Edmonds

@cfrank What are the upsides of this? It seems like a worse version of cumulative voting and that is already a pretty terrible system.

rob

@cfrank Sounds kind of cool in theory, assuming you aren't proposing it for actual election of humans for office.

I see it as automating/streamlining the sort of horse trading that can happen in a legislative body, such as where one representative might say "I'll vote for the bill to give funds to reinforce dikes against flooding, if you agree to vote for the bill to give funds for earthquake proofing bridges" (simple example assuming some locales care about some issues more than other issues)

I haven't thought in detail about it, but my initial impression of such ideas is simply that it is an extremely hard problem, especially in partisan bodies where the representatives themselves are elected by plurality voting.

cfrank

@Keith-Edmonds I mean, I'm still learning about it myself. It seems promising that a state-level house of representatives was willing to use it, and that it was proposed by Nobel Laureate professor of economics William Vickrey. Vickrey auctions were a discussion of interest in the previous forum. There are plenty of resources to learn about it if you are actually interested in investigating upsides.

cfrank

@rob It isn't totally clear to me how QV works in the context of any specific voting system, it seems to be that preferences are indicated with the votes that get quadratically more expensive as they are allocated on a single indicator. Or maybe it's just that for example exactly 2 votes on one indicator costs 4 credits total, and exactly 3 votes on one indicator costs 9 credits total, etc. Rank-order systems always seem to be too much work for the voters in my opinion. I just wanted to throw the concept out there because it appears to have some backing and research behind it, and I didn't see it being discussed here.

I'm still watching this lecture: https://www.youtube.com/watch?v=mgU55B_wAas

And this one: https://www.youtube.com/watch?v=PVS7dpgKohY

So far it seems as though QV encourages thoughtful engagement of voters on issues, and leads to sensible indications of social preferences. It also seems to be robust against collusion. I think it seems to mitigate bullet voting as well. Burial as @brozai points out is a sort of annoying issue, I'm not sure how that can be mitigated in a score system.

Keith Edmonds

@cfrank said in Quadratic Voting:

It seems promising that a state-level house of representatives was willing to use it, and that it was proposed by Nobel Laureate professor of economics William Vickrey.

Governments make terrible decisions all the time and I would not take medical advice from an engineer any more than voting theory advice from an economist.

I have talked about this a lot in the past and it does not seem like a reasonable system at all. A quadratic penalty in a machine learning or economics framework makes sense but this is the wrong tool for this case. Its a bad attempt to fix a bad system. I would spend your time elsewhere

cfrank

@keith-edmonds that doesn't make sense to me. Social choice theory is a subfield of economics, Kenneth Arrow was an economist. I have a textbook called "Economics and Computation," almost half the book is dedicated to voting theory. Maybe the analogy would be better if the engineer was a general practitioner and you were looking for a dermatologist, but in any case somebody who is trained as an economist is very well-suited to analyze voting systems.

I would like to hear more about what you have said in the past about why this system is not at all reasonable. I'm not sure what is so terrible about it, especially in a multi-winner context.

Keith Edmonds

@cfrank said in Quadratic Voting:

Social choice theory is a subfield of economics, Kenneth Arrow was an economist.

It is a specialized field. Thomas Hare was a political scientist. I am a physicist. Warren Smith is a com Sci person. It is its own field.

I would like to hear more about what you have said in the past about why this system is not at all reasonable.

There was a discussion I started about Quadratic voting on the now decommissioned CES forum a few years ago. It is now archived on the here
https://www.votingtheory.org/archive/posts?where={"topic_id"%3A541}

I'm not sure what is so terrible about it, especially in a multi-winner context.

It has vote splitting and gives low PR. It is not "terrible" but it is not worth considering given that you have simpler systems which do not have vote splitting and give high PR.

Marylander

@cfrank The strategy is going to end up being the same as FPTP though. It doesn't seem to mix with contexts where the outcomes for candidates are binary (i.e. you're either a winner or a loser) very well.

It seems to make more sense in a context where there is a scarce resource to be distributed, like time given to different speakers. (Perhaps this makes it attractive to economists.)

cfrank

@keith-edmonds that makes total sense, I didn't consider that and wasn't aware that this had been discussed before, thank you for sharing the discussion. I still think that voting theory is intimately connected with economics, computer science, and political science, more than physics, say. That is not to say anything about your expertise as a voting theorist, because the field is definitely interdisciplinary and requires thinkers from diverse backgrounds.

I know that Vickrey auctions were a previous topic of discussion that marginally related to the concept of an "ideal" voting system, and that Vickrey specifically was a very intelligent guy with a relevant background.

I also wonder about the extent of vote splitting if QV is incorporated into a range-voting context, and also about whether proportional representation in terms of the fraction of individuals with an interest is important if representation of interests is what is desired, because many people may be marginally invested in some issue, whereas there may be a smaller group with intense interests that should perhaps be represented. It's a touchy subject because special interest groups can be problematic.

@cfrank There is a large body of research on proportional methods with approval ballots. Proportionality with score is likely approximated by approval when the population is large.

cfrank

@marylander yes that is sensible, I am not necessarily promoting this system, I just wanted to start a discussion about it. I do think that if voters incur costs for supporting a candidate they are more likely to indicate their support thoughtfully and in some sense perhaps more honestly, though that's not a rigorous argument.

cfrank

@brozai Yes for sure, I started reading the text you supplied. It's very interesting. I have been focused on single winner systems but I am getting a lot more keen on approval voting, Condorcet methods and multi-winner systems. Thanks for your continued engagement, I feel like I am learning a lot.

@cfrank regarding Condorcet methods, some professors I've been in contact with did a series of three works on the methods "Split Cycle" and "Stable Voting" you may like to read.

https://arxiv.org/abs/2004.02350
https://arxiv.org/abs/2008.08451
https://arxiv.org/abs/2108.00542

Regarding multiwinner rules with ranked ballots there is a nice analysis here

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7089675/

One of the main takeaways of that one is that choose-one is actually not too bad when there are multiple winners (although as we know, it is quite bad for a single-winner election).