This year we've launched Ballot Lab, an independent research nonprofit for voting methods. This post covers what we do, where we are, and what more money would let us do. Ask me anything in the comments.
Our biggest uncertainty: whether neutral evidence changes which reforms are picked. We explain why we think it does, and what we’re unsure about.
Cities and states keep changing how they vote. In the recent elections, we tracked 97 ballot initiative efforts in 27 states to change election rules. Those choices decide who governs and how trillions of public dollars get spent.
These reforms are made on thin evidence. Most accessible research comes from advocacy groups who write it to win campaigns. Academic work is more careful, but it's slow, often paywalled, and rarely built around the questions jurisdictions face. Reform campaigns spend millions. Unfortunately, almost nobody funds the neutral work that tells them which reforms actually deliver.
For people focused on the long-term future: much of the risk we face runs through how well governments make decisions. Better voting methods are one plausible lever on that. Ballot Lab exists to measure that, before groups spend hundreds of millions more on their best guesses. If a method turns out to be meaningless, that's worth knowing early too.
Every experimental study is:
Here's how a study moves through those steps. We write a research brief and a draft design, and outside reviewers check the design before anything runs. We then preregister the design and analysis plan on the Open Science Framework (OSF), our default registry, before we collect data. After the study is written up, a second outside review is recorded publicly and we revise. The final paper, data, and code are archived on Zenodo with a DOI, posted on our site, and then submitted to an open-access journal.
Our research currently falls into three programs:
What we've done so far: incorporated in March, got our IRS determination in July, set up the site and studies page, and published How 2026 Ballot Measures Show Who Gets to Change the Rules, with interactive charts and a searchable dataset. Two studies are outlined at ballotlab.org/studies.
Ballot Lab is new, so it doesn’t have a track record of its own yet. I do though. I led The Center for Election Science, an advocacy organization for approval voting, from 2013 to 2023. We built it from scratch to over $1M a year and brought approval voting to Fargo (2018) and St. Louis (2020), its first two US cities. We also ran one of the first comparative studies on voting methods that used a control measure. I also founded Male Contraceptive Initiative and grew it past $1M; its funding mechanism later funded a human clinical trial of a long-acting reversible contraceptive. My academic training is in social sciences and law.
Our academic advisors are Steven Brams (NYU), Jean-François Laslier (CNRS and Paris School of Economics), Herrade Igersheim (CNRS, University of Strasbourg), Lindsey Cormack (Stevens Institute of Technology), and Marc Kilgour (Wilfrid Laurier University). We bring in mathematicians, game theorists, and political scientists project by project. We'll also be working with The Unjournal as a partner on open, outside review of our studies. See our team page for more.
To avoid bias in experimental studies, we use preregistered analysis, outside reviewers for the design and write-up, and we publish no matter what. If a study finds approval voting underperforming, for example, we publish it like anything else.
We're operating well below capacity. Until new funding comes in, we expect to spend about $6,300 a month. Here's our planned budget:
| 2027 budget | Budget, 2027 through mid-2028 | |
|---|---|---|
| Programs: studies and research (collaborators, data, survey panels) | $104K | $225K |
| Programs: content development (reports, explainers, publishing) | $83K | $91K |
| Management and general (building the organization, accounting, insurance, compliance) | $82K | $144K |
| Fundraising | $31K | $40K |
| Total budget | $300K | $500K |
Costs are grouped by function, the way nonprofits report expenses on IRS Form 990. We're treating 2027 as a building year, so compared with 2028, it puts more into content (reports and explainers) and into building the organization, with Study 1 designed and ready to run when a suitable election comes up. Staff time in 2027 is split about 40% studies, 25% content, 20% management and general, and 15% fundraising. In the first half of 2028, more of it shifts to studies.
At today's level, we’re limited in how much writing, analysis, and study design is possible. We can't field studies, cover publication fees, or get to the conferences where the funding and the audience are.
Gap for 2027 alone: about $265K. ($300K minus about $36K: about $24K in projected cash on January 1 plus about $12K in recurring gifts.)
Gap for our full plan through mid-2028: about $465K. ($500K minus the same $36K.) This is the number we'd most like to close. This gives us the runway to plan and execute our programs better.
The $300K and $500K figures come from budgets we've already built and sent to funders.
Each layer assumes the one before it is funded.
$10K: Internal analysis. We write up analysis and reports much like the ballot initiative article for the 2026 election.
$100K: an evidence-grading report, reference work, and explainers, plus Study 1 designed and preregistered.
Alternative: Because experimental studies rely on actual elections fitting an interesting scenario, we may instead run more exploratory research. This can be done in tandem with explainers and analysis on the site.
$300K: the full 2027 plan.
$500K: the full program through mid-2028.
$1M: two full years, and a real research institute.
At this level, Ballot Lab is a standing institution that jurisdictions, journalists, and funders can rely on for neutral answers. Note as well that individual projects are projections based on information available now and may change if better opportunities arise.
Single-jurisdiction reform campaigns often spend seven figures. State campaigns can go well into eight figures. Since 2016, campaigns for and against statewide voting-method measures have raised more than $150 million, by our tally of Ballotpedia’s figures. Alaska alone accounts for about $36 million across three measures. Repeals of poorly understood reforms cost millions more and set things back years. A few hundred thousand dollars of neutral evidence can steer where that larger money goes and which reforms hold up once voters live with them.
It was a combination of media through 80,000 Hours, conversations with local advocates, and being able to cite research that persuaded St. Louis advocates to push for approval voting. There is a segment of reformers and philanthropists who want to make the right decision. But voting theory is hard and many gaps still need to be filled. Ballot Lab exists to help key people make the right decisions on these electoral reforms.
Isn't this approval voting advocacy with a new name? Ballot Lab runs no campaigns. Experimental research will run approval alongside plurality and instant runoff on the same voters, and the analysis is locked before we see data. We will make evidence-based recommendations, and they'll follow the results.
Why not leave this to academics? Academics do good work here, and our advisors and collaborators are academics. But journals reward novelty, move slowly, and often sit behind paywalls. Jurisdictions need answers to practical questions on a policy timeline. We work with academic collaborators and publish openly and quickly.
Why give now instead of waiting for more work? Because the initial work is what this money produces. At $100K, funding turns our outline into the evidence-grading report, reference work and explainers, and Study 1 designed and ready to run.
What happens if you raise less than $500K? We do the tiers in order. Every layer above produces something published and useful on its own.
How will your upcoming book affect this timeline? I’m expecting to publish my book on voting methods early in 2027. Additionally, any promotion for the book will include mention of Ballot Lab. That process has also helped me identify more collaborators and have a better understanding of the research.
I used to give to The Center for Election Science. Where should that support go now? CES closed this year and no longer accepts donations. I led it from 2013 to 2023, and it ran under new leadership for about three years after I left. If you gave to CES to learn which voting methods actually work, that’s the question Ballot Lab exists to answer and share.
Experiment timing. Experimental studies depend on a real election that fits an interesting scenario, and we don't know when those elections will occur, so their timing has a lot more variability. If no suitable election comes up, there are lots of other exploratory studies we can do instead, alongside explainers and analysis on the site. To build attention and additional value, we’ll have to build those components up anyway.
Our near-term goal is $100K, enough for reference work, explainers, and an evidence-grading report, and to get Study 1 designed and preregistered. If this work makes sense to you, please give at a level that fits your capacity.
To the people who have already given this year: thank you. You got Ballot Lab off the ground.