a/democracy· 14 October 2026 · 7 min read

What happened when Colorado let lawmakers vote by how much they cared?

Cover of Quadratic Voting (Omxus Press)

In April 2019, Colorado lawmakers had more than 100 bills asking for $120 million and only $40 million to give. So they tried a strange voting rule, one where some of them had to stop and remind themselves what a square was. Could a vote tell the difference between "I don't mind" and "this decides my life"?

A pile of bills

By the start of April 2019 the appropriations committees of the Colorado legislature held more than 100 bills asking for more than $120 million between them. Budget writers had set aside $40 million (Eason, 2019). Somebody had to choose.

The House Democrats had tried to choose fairly the year before. Each member got 15 votes to give to 15 bills. Chris Hansen, who chaired the House Appropriations Committee, said it produced "this big blob" of bills with roughly the same number of votes and no clear preferences (Eason, 2019). Everyone had a say, and the say told them nothing.

A price on caring

In 2019 the caucus tried something new. Each of the caucus's 41 members got 100 tokens to spread across 107 bills (Coy, 2019; Rogers, 2019). One vote on a bill cost one token. Two votes on the same bill cost four. Five cost twenty-five. Votes cost the square of their number, so each extra vote on one bill cost more than the last. A member could spread tokens thinly across many bills, or pile them on the one that mattered most, and lose the chance to back much else.

Members voted anonymously, on a computer that did the maths. Even so, some "had to remind themselves what a square was" (Eason, 2019).

The blob became a list. A bill to close the gender pay gap, the Equal Pay for Equal Work Act, came first with 60 votes (Rogers, 2019). Health measures took five of the top ten places. Bills nobody fought for dropped away. "We got a better signal with less noise in that signal," Hansen said, "and we were able to capture people's intensity, and that was what was really new. We never had a way to do that before" (Eason, 2019).

Why the square

A normal vote can't tell "I don't mind" from "this decides my life". If 51 people mildly prefer one option and 49 would be badly hurt by it, the 51 win. Under quadratic voting, with 100 credits each, the 51 might cast one vote apiece and the 49 five votes apiece. The 49 win, 245 to 51, and pay for it: each has 75 credits left for every other question, against 99 for each of the 51.

Votes on one questionCredits it costsCost of that last vote
111
243
5259
1010019

Why the square and not some other price? Economists Steven Lalley and Glen Weyl showed that only a square price makes people buy votes in proportion to how much the outcome matters to them, so the total adds up everyone's real stakes (Lalley & Weyl, 2018). A flatter price lets anyone who cares at all pour everything into one question. A steeper one drowns out strong feeling.

Can money buy it?

The fair worry is that the well-off or the organised will find a way to buy a louder voice. The largest quadratic system on public record lets you test that. Gitcoin Grants funds open-source and public-goods projects with a cousin of quadratic voting, quadratic funding: a matching pool rewards projects by how many people give, not just how much (Buterin et al., 2019). Gitcoin had distributed over US$50 million this way by 2023, and more than US$60 million since (Gitcoin, 2023, 2026).

Its first twelve rounds, 2019 to 2021, are public: 4,714 grants, $9.36 million donated, $6.73 million in matching (Burmeister, 2021). So what did the matching pay for? Holding the money fixed, doubling a project's donors multiplied its matching by 2.23. Holding the donors fixed, doubling their money multiplied it by 1.14. The donor effect beat the money effect in all twelve rounds.

In plain words: the quadratic rule paid for how many people cared, not the size of the cheques. There's a catch worth knowing. Because it rewards breadth so strongly, matching flowed to the most widely backed projects: the top tenth of grants took a median 78% of the matching, against 69% of the donations. Funds that want to protect small, niche work can cap matches or keep a separate pool.

Where it has been tried

Taiwan used it in 2019. The Presidential Hackathon handed 30% of its scoring to the public, each person allocating 99 points quadratically. Digital Minister Audrey Tang explained the odd number: "It's 99 rather than 100 points — as in Colorado — because then people won't even have the possibility of just going there and voting everything on the same project" (Huang, 2019). The aim was to stop a small panel of judges following each other.

Researchers have used it in surveys, too. Compared with ordinary agree/disagree scales, quadratic surveys gave a clearer reading of the people who cared most (Quarfoot et al., 2017).

Then the court

Five years later the method met the law. In January 2024 a Denver District Court judge ruled that the caucuses' anonymous ballot broke the state's open meetings law, which bans formal action by secret ballot. The judge found the secrecy deprived the public of the ability to hold representatives to account for how they voted. The ruling did not ban the method: the caucuses could keep using quadratic voting, the judge said, if they made members' votes public (Paul & Eason, 2024).

The Denver court did not reject quadratic voting. It rejected secret quadratic voting by elected representatives. Citizens need a secret ballot so they cannot be bought or bullied; representatives need a public record so they can be held to account. A system built for citizens keeps the ballot secret; a system built for legislators publishes every allocation (Paul & Eason, 2024).

What it would take here

The Australian lower house has used preferential voting for a century (Farrell & McAllister, 2006). A ranking tells the count which option you prefer. It still can't tell it how much. No one has run quadratic voting for a public election: the deployments so far are a caucus, a hackathon, surveys and grant rounds. It needs a strong answer to fake identities (one actor posing as many), to organised blocs, and to secrecy. It also has a real weakness: credits only measure what people will spend, and someone who's never been asked to think about a question may not spend well. The likeliest path runs from organisations and participatory budgets to councils and then further. In a participatory budget, a resident who badly needs a bus route can concentrate credits there and outweigh several casual supporters of a park upgrade.

Closing the loop

Saying what you want is half of it. The other half is finding out whether you got it. Software engineers make public, numbered promises about how reliable a service will be, with an agreed margin for failure, and when the margin runs out, new changes stop until it's fixed. Imagine a government doing that for housing or waiting times: say what people want and how much, publish the target, and when it's missed, stop and fix it. Nobody has built that yet. It's easier to imagine than most of what we've tried.

What you can do

You don't need a parliament. Try this on your club, your strata, your staff meeting or your kids. Everyone gets 100 credits. One vote on an idea costs 1, two cost 4, three cost 9, ten cost 100. Add it up on the back of an envelope, in the open. You'll be surprised who cares about what, and how quiet people finally get heard.

If your council runs a participatory budget, ask them to try it. And when someone tells you a thing passed because most people wanted it, ask how much. Ask who it hurt, and how badly. Fifty-one people who mildly prefer something aren't the same as forty-nine who'd lose their homes. Now we've got a way of showing it.

Sources

  1. Burmeister, T. (2021). Gitcoin Grants rounds 1–12 [Data set and analysis code]. GitHub. https://github.com/Tburm/gitcoin-grant-analysis
  2. Buterin, V., Hitzig, Z., & Weyl, E. G. (2019). A flexible design for funding public goods. Management Science, 65(11), 5171–5187. https://doi.org/10.1287/mnsc.2019.3337
  3. Coy, P. (2019, May 1). A new way of voting that makes zealotry expensive. Bloomberg Businessweek. https://www.bloomberg.com/news/articles/2019-05-01/a-new-way-of-voting-that-makes-zealotry-expensive
  4. Eason, B. (2019, May 28). Colorado's experiment in quadratic voting tried to remove politics from budget decisions. But did it work? The Colorado Sun. https://coloradosun.com/2019/05/28/quadratic-voting-colorado-house-budget/
  5. Farrell, D. M., & McAllister, I. (2006). The Australian electoral system: Origins, variations and consequences. UNSW Press.
  6. Gitcoin. (2023). Gitcoin [Home page, archived 2 June 2023]. https://web.archive.org/web/20230602035335/https://www.gitcoin.co/
  7. Gitcoin. (2026). Quadratic funding. https://gitcoin.co/mechanisms/quadratic-funding
  8. Huang, Y. (2019, October 2). Highlights from first RadicalxChange Taipei meetup. Medium. https://yahsinhuangtw.medium.com/highlights-from-first-radicalxchange-taipei-meetup-f2a9c3b797ab
  9. Lalley, S. P., & Weyl, E. G. (2018). Quadratic voting: How mechanism design can radicalize democracy. AEA Papers and Proceedings, 108, 33–37. https://doi.org/10.1257/pandp.20181002
  10. Paul, J., & Eason, B. (2024, January 5). Judge orders Democrats at the Colorado Capitol to stop using secret voting system to allocate state budget. The Colorado Sun. https://coloradosun.com/2024/01/05/colorado-legisalture-quadratic-voting-lawsuit/
  11. Quarfoot, D., von Kohorn, D., Slavin, K., Sutherland, R., Goldstein, D., & Konar, E. (2017). Quadratic voting in the wild: Real people, real votes. Public Choice, 172(1–2), 283–303. https://doi.org/10.1007/s11127-017-0416-1
  12. Rogers, A. (2019, April 16). Colorado tried a new way to vote: Make people pay—quadratically. Wired. https://www.wired.com/story/colorado-quadratic-voting-experiment/

Keep reading