This is a model — a starting point, not a final text. In-state partners will shape and own the version that goes on the ballot in their state. We're developing it in the open because we want to get the design right before anyone starts collecting signatures.
A model for the roughly 24 states with the citizen initiative process — see which states qualify →
Every year, a randomly selected group of ordinary citizens takes up one important policy question, deliberates it thoroughly, and sends a real bill to the legislature. The legislature must vote — up or down. If they say no, it goes to the people on the next ballot. Citizens always have the final word.
Each year, a panel of randomly selected registered citizens — drawn to reflect the demographic diversity of the state — convenes to work on one significant policy question. Like a jury: chosen by lot, not by ambition or appointment — and trusted for exactly that reason. They choose the experts they hear from. They control the process. They vote on the result. If they reach a decision, legislative lawyers draft it into a complete, enforceable bill — and it goes to the legislature, which must vote yes or no. If they say no, it goes directly to the people on the next general election ballot.
No report. No advisory opinion. No government panel that can be quietly shelved. A real bill, a required vote, and a guaranteed path to the people.
If you're thinking "this sounds great, but politicians and special interests will find a way in" — that's exactly the right question. We've thought about it too. Every structural choice in this initiative was made with that threat in mind.
Here's what we've built in specifically to prevent capture:
A separate, randomly selected Rules Assembly designs the process each year — how participants are chosen, how the assembly runs, how disputes are resolved. They are explicitly prohibited from considering likely policy outcomes. No politician sets the rules.
A separate Topic Selection Assembly chooses the policy question from public nominations — asking only: what issue is most important to people's lives and least addressed? They cannot consider likely outcomes. No organized interest chooses what gets deliberated.
Assembly members are prohibited from accepting anything of value from anyone with a stake in the outcome. Attempts to influence members outside the official process are a violation. Outreach channels are controlled by the assembly, not by outside interests.
All proceedings, testimony, deliberations, and votes are public and recorded. The selection process is open-source and auditable. There are no back rooms. Everything is on the record.
The initiative itself is reviewed at 2-year and 4-year intervals by a citizen panel — not the legislature. If something isn't working, citizens can fix it. If politicians try to weaken the process through legislation, the review assembly can flag it and the people can correct it.
No safeguard is perfect. But a process that distributes authority across multiple independently-selected citizen panels — with everything public and auditable — is far harder to capture than a single government commission or a legislature that answers to donors.
A smaller citizens panel designs the process for each annual cycle — how participants are selected, how the assemblies run, how disputes are resolved. Their only job: make it fair, equitable, and effective.
A separate citizens panel selects the policy topic for the year from public nominations — asking: which issue, within scope for the state legislature, is least addressed and would make the biggest difference in people’s lives?
Every registered voter in the state (identity verified) is invited to participate — through an open online platform or in person at libraries and community centers.
Over roughly 120 days, the assembly deliberates under facilitated, citizen-directed process, working toward a proposal with real legal teeth.
It is unlawful to privately contact or attempt to influence assembly members outside of official proceedings. Any attempt — calls, texts, gifts, pressure — must be reported within 48 hours and becomes public record. Members who report are fully protected. Official channels, including expert testimony and public comment, are open and unrestricted.
After two years of the pilot, voters cast a straight yes/no ballot on whether to continue the citizen assembly process. If they approve, a Review Assembly convenes to evaluate the program — continue as-is, expand to handle more topics per year, or sunset future cycles. Their recommendation goes to the next general election ballot.
All software — the sortition algorithm, the public input platform, the selection system — must be published under an OSI-approved open source license and publicly auditable before each cycle begins. No black-box tools in any function that affects selection, deliberation, or results.
Citizens assemblies aren't a new idea — they've been run in dozens of countries over the past 20 years. Here are a few clear, positive places to dig in.
Citizens assemblies have been studied seriously for decades. The findings are consistent enough to be useful.
Misinformation spreads because it travels to people who can’t ask it a question. A deliberation room is its worst environment: participants get direct access to experts, push back in real time, and have to persuade the people sitting next to them. Organizations fighting misinformation at the millions-of-users scale are in a nearly impossible battle. At the scale of 150 randomly selected citizens, the same problem becomes tractable.
James Fishkin at Stanford has run Deliberative Polls across 28+ countries since the 1990s — in every case, when ordinary citizens spend time with balanced information and expert testimony, their views shift measurably toward positions closer to real-world evidence and data. The gap between what people initially believe and what the evidence actually shows shrinks substantially — not because anyone told them what to think, but because they had time to ask questions and push back.
One well-studied example: France’s Citizens’ Convention on Climate (2020). 150 randomly-selected French citizens spent nine months hearing from climate scientists, economists, industry representatives, and union leaders. Their 149 proposals were notable for spanning the political spectrum — market-based mechanisms alongside regulation, worker transition support alongside emissions targets, community approaches alongside national mandates. The participants arrived with varied politics and the proposals reflected it: something no lobbied legislature had managed to produce.
Sources: Stanford Center for Deliberative Democracy · Fishkin, When the People Speak (Oxford UP, 2009) · France Citizens' Convention for Climate final report (2020)
Ireland's Citizens' Assembly on abortion rights (2016–2018) is the most studied case of citizen satisfaction with a deliberative process. At the end, nearly all members said they would recommend the experience to others — including members who arrived with strong opinions that were challenged by the process. The final recommendations, developed over 18 months, were described by participants as genuinely their own, not something handed to them.
This pattern holds across many countries. Germany's Citizens' Council on Democracy (2021), Australia's Citizens' Parliament (2009), and Iceland's National Forum (Þjóðfundur, 2010) — which brought together 1,000 randomly selected citizens to set national priorities after the financial crisis — all report high satisfaction and strong perceived legitimacy among participants.
Sources: citizensassembly.ie · OECD, Innovative Citizen Participation and New Democratic Institutions (2020) · Landemore, Open Democracy (Princeton UP, 2020)
Research consistently finds that when people deliberate together — across political, economic, and demographic differences — polarization within the group decreases. People arrive primed for conflict and leave describing the experience as one of genuine learning. This doesn't mean they agree; it means the disagreements become more substantive and less tribal.
On legitimacy: political scientist Hélène Landemore and others have found that randomly selected assemblies are often viewed as more legitimate than elected representatives on contentious issues, particularly where voters distrust elected officials' motives. The mechanism is obvious: a participant was chosen by lot, not by donors and party gatekeepers, and has no electoral incentive to reach a particular conclusion.
Sources: Landemore, Open Democracy (Princeton UP, 2020) · Gastil & Levine (eds.), The Deliberative Democracy Handbook (2005) · OECD 2020
More than 20 states have the citizen initiative process needed to put this on the ballot. See viability ratings, signature requirements, and cost estimates for all 50.
Share your feedback on the initiative — what works, what doesn't, what's missing. This is a living design.
We're building this initiative and a founding community. Get updates as it develops.
Get UpdatesSupport the state groups working to put this initiative on the ballot.
Sign the Pledge →