The wrong metric
When cities say they want to "modernize their town halls," they usually mean capacity. Can more people attend? Can we livestream? Can we run parallel sessions? Can we accept written comments through the portal?
That's the wrong metric.
The problem with the town hall as an institution isn't that its bandwidth is too low. It's that its access pattern is filtered — and the filter systematically excludes exactly the residents whose input matters most for equitable policy. Livestreams don't fix that. Written comments don't fix that. Adding more meetings doesn't fix that.
Fixing it requires a different question: what kind of participation are we actually asking for?
What a town hall filters out
The typical Tuesday-evening 6:30pm town hall filters out:
- ◇Working parents. Bedtime routines often run until around 8pm for young families. A meeting that runs 6:30–9pm is a hard conflict.
- ◇Shift workers. Nurses, retail, hospitality, food service. Many "essential worker" jobs are shift jobs. Weekday evenings are working hours for them.
- ◇Caregivers. People caring for aging parents or disabled family members can't plan around a specific meeting time.
- ◇Non-English speakers. Simultaneous translation exists at some meetings; most don't offer it, and when they do, it's often for one language.
- ◇Residents who feel unwelcome. People who've had bad experiences with municipal staff, or who don't see themselves reflected in the room, self-select out.
- ◇Introverts and processors. People who need time to think before speaking rarely benefit from a two-minute-comment format.
Add these up and you're looking at a large share of the residents of most cities. Which means: the residents who show up to town halls are a biased sample, not a random sample, of your community.
Any policy decision made from town-hall input reflects the values of the people who could show up. That's not the same as reflecting the community's values.
Three failure modes of the town-hall alternative
Most "town hall alternatives" don't fix the filter. They just move it.
1. The digital town hall
Replace the room with a Zoom. This helps the parents-with-bedtimes problem slightly (they can join from home). It doesn't help shift workers, caregivers, non-English speakers, or residents who feel unwelcome. And it introduces a new filter: digital access, comfort with video calls.
2. The public comment portal
Replace the meeting with a form. This changes the shape of the filter — now it selects for people willing to write. That's not obviously a broader group than the room. And written comment participation tends to skew wealthier, older, and more politically engaged than the general population.
3. The survey
Replace the discussion with a Likert scale. This scales, technically. It also flattens human perspective into a binary or a 1–5. A "No" could mean anything from "I hate this" to "I like this but not this way." The signal is unusable for anything but a headline number.
What "scaling access" actually looks like
If the point is to hear from residents who don't show up to town halls, the alternative has to:
- ◇Meet them on their schedule. Not a scheduled event. Not a comment period with a deadline. A conversation they can start at 11pm from their phone.
- ◇Meet them in their language and comfort. Voice input, translated interfaces, a low writing burden.
- ◇Meet them where an opinion is welcome, even if half-formed. A structured conversation is much lower-friction than "write a formal public comment."
- ◇Show them what happened with what they said. A versioned public document that reflects their input closes the feedback loop that town halls almost never close.
Civic Filament is one implementation of this pattern — an AI mediator conducts one-on-one interviews with residents, value-profile clustering surfaces where values align across the community, and a Living Requirement Document is the artifact the city and residents both reference. The interview happens on the resident's schedule, from a phone or computer browser, with voice input available. (It's English-only for now; multi-language support isn't built yet.)
We are not the only implementation of this pattern. Decidim, CitizenLab, and Consider.it each solve pieces of the same problem in different ways. What they share is a rejection of the town-hall bandwidth problem as the real problem.
If you're considering scrapping the town hall entirely
Don't.
Town halls do things a chat interface can't: they let residents see each other, they create a public record of who cared enough to show up, and they produce the friction that sometimes leads to political action. What they don't do is produce a representative signal about what the community wants. (Neither does any self-selected tool, ours included — but a wider door should mean a less-skewed sample.)
The move is to complement, not replace: keep the town hall for its ceremonial and political function, and use a scaled-access alternative to produce the input that policy actually depends on. That's what our pilot program is set up for — one issue, free, alongside your existing engagement process, so you can see whether the extra signal is worth the recurring cost.
The buyer questions we get most
If you're a city clerk or engagement lead evaluating this class of tool, the same three questions come up:
- ◇What happens to interview data? We answer this on the Participant Rights page, including the records-law posture we take for governmental deployments.
- ◇How do we know the AI isn't steering residents? The mediator's explicit instructions — what it must do and never do — are published on the Methodology page. Deviations are treated as bugs.
- ◇What does this cost? Community tier is free under 500 participants. Municipal is $2,500/mo. Detailed comparison against consultant-led engagement studies is in the FAQ.
---
If you're running a town hall this month and this hit a nerve, request a walkthrough — we'll show you what a parallel Civic Filament run would produce alongside it.