Hyperlocal democratic decision-making tools are out there, ready to be applied.
“Do I contradict myself? Very well then I contradict myself, (I am large, I contain multitudes.),” goes a line from Walt Whitman’s “Song of Myself.” Each of us immediately recognizes its truth. What would our democracy look like if it, too, contained multitudes — embracing not only the multiplicity among us but the multiplicity within us?
Sixty-four percent of Americans believe we are too polarized to solve our problems, a stark increase from just five years ago. Yet freed from rigid categories — progressive/conservative, urban/rural, migrant/native, Democrat/Republican — we are unpredictable and eccentric. Like the man who showed up to a mayoral rally in New York City, my home, wearing a “MAGA for Mamdani” hat. He was one of the estimated 20% of New York City Trump voters who supported Zohran Mamdani for mayor.
Our political machinery flattens the many complexities and internal contradictions of the electorate. Traditional polling methods, referendums and even our ranked-choice voting system tend to force complex human preferences into reductive choices. The 2016 Brexit referendum is a striking example: a binary vote — in or out — that forced people into opposing camps when their actual feelings about European Union membership were far more nuanced.
In New York City today, we have a chance to do things differently. Mamdani won the mayoral election by mobilizing New Yorkers from all walks of life. His newly created Office of Mass Engagement aspires to sustain that energy beyond Election Day by embedding public feedback directly into city policy. But how do you genuinely engage a city of more than 8 million residents who speak an estimated 750 languages, hold diverse religious and cultural beliefs and live in radically different economic circumstances? And how do you do this in ways that surface real preferences — rather than amplifying voices that happen to align with the administration’s predetermined goals?
To his credit, Mamdani has already made a genuine effort to hear what his constituents want. Last December, he spent 12 hours listening to 142 New Yorkers at his 'The Mayor is Listening' event. As he goes about his daily business, he is — at least so far — considerably more approachable than many past leaders of the city. But events like this won’t scale to millions. And one-way listening, where many speak to one, can’t capture the collective wisdom that emerges when people both speak and listen to each other.
For close to a decade, I’ve overseen the deployment of a tool that offers a path forward: a nonprofit, open-source technology called Polis, designed to understand public opinion in its complexity. I’ve helped governments worldwide use Polis to gather input on everything from climate policy and urban planning to food pantries and mental health services. Across all these deployments, one pattern emerges consistently: Common ground exists even in the most fraught terrain. New York City is no exception — beneath the noise of competing interests and divided communities lies the same potential for unexpected consensus.
Polis inverts the logic of typical social media. It’s a tool for broad listening, rather than broad casting of opinions. It works like this: On each topic raised on Polis, participants — from thousands to tens of thousands, and scalable to hundreds of thousands — share their perspectives in tweet-length statements and vote “agree,” “disagree,” or “pass” on each other’s statements. The Polis algorithm then breaks people out of their bubbles by showing them not just the statements they are likely to agree with, but also those they are likely to disagree with. The conversation is displayed as an interactive map and summarized by a large language model in real time, showing areas of agreement and disagreement. Consensus is reached when there is broad agreement across otherwise divided groups.
Polis's most famous application is in Taiwan. After the 2014 Sunflower Movement opened space for citizen participation, the government invited g0v (pronounced gov-zero) — a hacker community that rose to prominence during the movement — to prototype a process for mass engagement in policymaking. The result was vTaiwan, which combines in-person deliberation with virtual participation via Polis. Audrey Tang, a g0v member who helped design vTaiwan, later became Taiwan's digital minister.
vTaiwan was not a patronizing experiment to make people feel like they were being heard. It solved real policy deadlocks. For example, the government had been trying for years to legalize online liquor sales, but alcohol merchants, e-commerce platforms and people concerned about underage drinking couldn’t find common ground. When the issue went through vTaiwan, 450 stakeholders reached consensus within weeks and formulated recommendations the government went on to implement. The alcohol case wasn’t unique. Under Digital Minister Audrey Tang’s leadership from 2016 to 2024, vTaiwan tackled more than 28 issues ranging from Uber regulation to revenge porn, with 80% resulting in government action.
vTaiwan is often framed as an example of how AI can invigorate democracy. The truth is actually the opposite: Democracy invigorated AI. The 2011 Occupy Wall Street movement is what initially inspired a group of Seattle-based technologists — led by Colin Megill — to build Polis. It was proprietary software at first, but Tang insisted that any tool used for public deliberation be fully transparent. So in 2016, Polis became open source, and it has remained so ever since.
Yet in all of America, only Bowling Green, Kentucky, has used Polis at scale. Last year, I helped thousands of residents come together for citywide planning — proving they’re not bowling alone in Bowling Green.
Once the provisional capital of Confederate Kentucky, Bowling Green — a Republican-led city of about 75,000 — is now home to Western Kentucky University, two major Japanese aluminum plants and one of the highest per capita refugee populations in the nation. It’s about an hour’s drive from Nashville, one of the fastest-growing hubs in the South. City leaders were convening to plan for a doubling of Bowling Green’s population in the coming decades. My Polis colleagues and I — with funding and support from Google — were invited to help. It was our first public-private partnership, but Polis’s open-source license meant even Google’s code contributions had to remain public.
In March 2024, Bowling Green held a three-week digital town hall using Polis, and approximately 10% of the city’s population participated online to share their perspectives on what the city could become by 2050. To make sure the participants were as representative as possible, volunteers canvassed at local businesses and international grocery stores. They took iPads to soup kitchens, rural diners and flea markets. They talked to bored people in the DMV line, farmers at the agricultural gala, Afghan Uber drivers, fans at a college basketball game and congregants at local churches and the Islamic Center. They reached longtime locals, the recently arrived Burmese refugees and the Bosnian refugees who had arrived in the early 1990s. The Bowling Green Daily News pitched in by creating a mock issue from the year 2050, which was hand-delivered at 5 a.m. by a crew of volunteers.
The city and county administration went all in. Doug Gorman, the judge executive of Warren County, a Republican and the highest elected official in the region, appeared on T.V., on the radio and at events promising residents that their elected officials would listen to what they had to say.
“My perspective is growth can either happen to us, or it can happen for us,” I heard Gorman say repeatedly.
The multiplicity of views in Bowling Green — both individual and collective — were reflected on Polis. This happened because residents weren’t choosing between preset options. Instead, they were sharing their own visions for the city’s future and voting on each other’s contributions — from what they valued in community life to what they hoped local leaders might address.
Take two issues that emerged as salient to residents: more locally-sourced food options — meaning more farm-to-table restaurants, farmers markets and farm-based tourism — and allowing drag shows in public spaces. Locally-sourced food brought people together. Drag shows divided them. The most common stance was supporting both. But nearly as many favored locally-sourced food while opposing drag shows. Smaller groups opposed both, supported drag shows while opposing farm-to-table food, or didn't feel strongly about one or both issues. Every possible combination existed. Knowing someone's stance on local food gave you no information about their views on drag shows. People were not following a progressive or conservative script.
And amid the multiplicity, shared priorities emerged: farm-to-table restaurants and year-round farmers' markets among them, but also protected farmland and support for small businesses. More green spaces and a well-maintained historic district. More social services, particularly for the elderly. More child care, venues for family activities, better schools and vocational training — especially in trades like electrical, plumbing and carpentry. More affordable housing, with stricter oversight of developers and incentives to reuse vacant buildings. More bike lanes, an enhanced bus system and a stop on the Amtrak line. More jobs that pay a living wage.
Alongside all this, there were sharp disagreements about LGBTQ rights, about immigration, about who the town was for, and who it was not — statements that, when mapped, clustered with discussions about the role of churches in the community (issues Bowling Green will have to grapple with outside of Polis).
City and county leaders are now using these and other insights — drawn from nearly 8,000 perspectives and over a million votes — to develop a comprehensive plan expected to be ready in 2027. Ben Peterson, who leads the City-County Planning Commission, told me that hearing from the larger community gave them “the confidence for having the right conversations” with officials and developers alike. Unprompted, he noted that the amount of consensus surprised him.
Zohran Mamdani won New York City not just by having strong opinions, but by finding common ground across divides. Governing will require expanding that common ground across neighborhoods, across boroughs, across countless competing priorities.
The city could launch a pilot within the first 100 days on any mayoral priority — housing, child care, public safety or city-owned supermarkets. For example, CUNY's Brooklyn College used Polis during the pandemic to design food pantries — identifying what the community most needed, who needed it most and how to connect resources like neighborhood fridges and local gardens.
Imagine the Office of Mass Engagement using this approach to augment the Rental Ripoff Hearings to be held in every borough: engaging tenants, landlords, housing advocates and developers to find where even divided stakeholders agree.
City Council members could deploy Polis in their own districts to understand constituent priorities, inform legislative decisions or find common ground on local issues. This would complement participatory budgeting programs, which give residents decision-making power over capital projects like playground upgrades and library renovations.
Participatory democracy can be tricky — any serious attempt to open new channels for direct public engagement upsets existing power structures. But as has been widely reported, Mamdani’s listening event felt different. The participants gave it their all. They didn’t come simply to be heard or to blow off steam; each came to help their mayor make policy. The Office of Mass Engagement can build on this momentum, helping build a real listening infrastructure that transforms how the city governs.
If individuals contain multitudes, cities do too — and so must the democracies that govern them.