Quadratic Voting in the Field

The Democratic caucus in the Colorado state legislature wanted to get their member’s feedback on the bills most important to them. That’s hard to do because each member has an incentive to claim that their pet bill is by far the most important bill to them. Thus, Chris Hansen, the chair of the House Appropriations Committee, who also happens to have a PhD in economics, decided to use a modified form of quadratic voting. Each voter was given 100 tokens to vote and the price of x votes for a policy was x^2 so you could buy 10 votes on your favorite policy for 100 but you could also buy 5 votes on each of your four favorite policies (5^2+5^2+5^2+5^2=100).

Wired: So in mid-April, the representatives voted. Sure, each one could have put ten tokens on their pet project. But consider the or: Nine votes on one (cost: 81 tokens) but then three votes on another (cost: nine tokens). Or five votes each (25 tokens) on four different bills!

In Colorado at least, it worked, kind of. “There was a pretty clear signal on which items, which bills, were the most important for the caucus to fund,” Hansen says. The winner was Senate Bill 85, the Equal Pay for Equal Work Act, with 60 votes. “And then there’s kind of a long tail,” Hansen says. “The difference was much more clear with quadratic voting.”


In a steady state these systems work well, but can be subverted as the various parties learn how to work the system.
A good parallel is the NBA, where rule changes a few years ago made the three-point shot incredibly valuable. One team, Golden State, figured this out and became dominant and other teams are only now catching up.

Absurd comment. "A few years ago" was more than 30 years ago. Golden State's strategy was not just a simple "figuring out" but actually developing a really rare skill (shooting 3's at will). Nor would I classify Golden State's strategy as a clever trick to "work the system".

...there are many methods of "voting". Which is the optimum ?

the Colorado legislature is somehow assumed to represent the views of the majority portion of the Colorado citizen Electorate -- how often does that actually happen in formal legisiaive decisions ?

you mean "the price of x votes for a policy was x^2"

Fixed. Thanks.

Interesing. But vulnerable to collusion.
Suppose, I have pet project "A" and don't care much about anything else.
I can either give 10 votes to "A".
Or I find 3 partners with pet projects B, C an D. And we agree to split our votes among these 4 projects. In this case A,B,C,D each get 20 votes total.

So collaborating is bad...?

If 3 partners split their tokens between A,B,C,D then each project gets 25 tokens per person 5 votes or 15 votes in total for each.

Two people really dedicated to project A, though, could put all of their tokens on it thereby yield A 20 votes.

It's not really collusion. It makes the spending of political capital explicit. If you really want one thing, then you can *probably* get it but at the cost of having almost no say on everything else. Your influence can be wide but shallow or narrow but deep. Not really seeing a problem.

It IS collusion. It is called "logrolling." But that doesnt always imply logrolling is effective because of competing coalitions. It may be the case though that logrolling always outperforms non-logrolling, all else equal. Clearly it is preference dependent. If everyone has the same most-favorite policy, it will be the winner regardless of coalitions. Logrolling gets interesting when preferences are not so coordinated.

So a bill that polls extremely well with the voters won. This sounds revolutionary.

That it polls well with anyone is an indicator of Idiocracy.

That ... was the winner? The sort of thing my local city council passes, unanimously, every week, in the early kumbaya moments of the meeting?

Thanks, this is indeed very interesting, and I'll remember it, though perhaps not precisely in the way quadratic voting boosters are hoping.

Garbage in equals garbage out.

Subject (crack (allusion), Description (rhetorical point), modifier (metaphor to photography)

What's stopping people from "selling" one vote at a time to 100 other people to get them each to vote for your favorite policy once, in exchange for voting for their favorite policy once?

carrés déchiquetés de fudge pâle

Nothing stops them. But as Boonton's example explains above, not every strategy is coalition-proof. You're getting more than one offer for your vote.

IOW, their top priority is a bill to lard onto the business sector in Colorado more compliance costs. Because, you know, when you want to figure out the wage or salary offer that best meets the needs of your business, you just must consult a labor lawyer.

Democrats are just sh!ts.

Yeah, that's one depressing finding.

Isn't this the classically libertarian solution? No more government mandarins enforcing arbitrary rules. If you think you are injured, sue in court.

That one of your employees is dissatisfied with his wages or salary is not a tort. It properly is only when you're in arrears on his promised compensation.

The bill also defines what the legitimate bases are for wage differentials between male and female employees, and "that's what the parties agreed upon after a negotiation" is conspicuously absent. You could also argue that it shifts the burden of proof from the employee to the employer when there is a dispute involving alleged discrimination.

I would certainly prefer a bid and ask system. Maybe that with wage transparency is a better option.

This is actually fine. It doesn't have to be perfect, nor the best, nor the last.

Optimization requires change.

To reject such experiments you must either think you are already in heaven or hell. Not a very centrist nor pragmatic approach to take.

Centrism isn't a goal, bro.

Keeping our individual rights is the goal.

As Arrow explained, there is no such thing as perfect in voting. But there are clearly outcomes that are objectively better than others.

Years ago The Atlantic, I think it was, had a contest on its backpage - they'd set out a definition or scenario and ask readers to coin a word for that phenomenon. Maybe we need a word for when intense pragmatism is deployed in search of solutions to problems that are other than the actual problems we have, or at least among the least salient. I think it would get some use in the workplace too.

[I only entered once - you actually had to mail it in! - I think the prompt was: a term for when young people are absorbed in their mypods listening to music, indifferent to their surroundings. We entered "sonic womb" and, for good measure, "My Own Private iPodaho," neither of which merited even Honorable Mention. Anyway, the prize was like a copy of "The Funny Little Atlantic Book of Made-Up Contest Words." Super-lame, didn't care about winning anyway, you can keep your stupid contest!]

Sonic womb is pretty good.

The bill prohibits an employer from:

Relying on a prior wage rate to determine a wage rate;

So no more X% raises, since to calculate new wage, you need to rely on the old wage.

Interesting, I think Oregon passed some labor laws a year or two ago that included that provision. As well as a bunch of others describing what are legitimate reasons for paying differing wages for seemingly the same job, and what are illegal reasons.

Prior wage is on the no-no list, although seniority or years of experience is okay.


Which journal articles discuss Quadratic voting?

At first glance, some things jump out at me:

- the scheme seems sensitive to the number of coins you allocate. 385 coins would allow you to perfectly rank your top 10 choices. 10 to first, 9 to second, 8 to third. Having only 100 coins adds an arbitrary constraint with no apparently mathematical basis. A single coin could make a large difference in the utility of your allocation.

2. Many allocations are strictly dominated by others. Only those allocations with zero remainder are admissible.

3. The shape of your preference function still matters greatly in how you allocate coins.

4. Different people's preferences will drive any possible coalition and determine if it is coalition-proof.

5. I dont immediately see a marked improvement of quadratic voting over other voting schemes such as Approval, Haare, Coombs or even plurality with runoff. How often would quadratic voting result in a Condorcet winner if one exists?

6. What about higher order exponents?

Comments for this post are closed