Slow Feeds, Small Rooms: Why Good Community Tech Should Make You Wait

There's a moment in every community platform's life when someone suggests adding a streak. It's usually framed as a growth idea. It's usually the beginning of the end.

Not because streaks are evil, but because they change what the space is for. The moment you optimise for returns per day, the people who came to find a babysitter or a book group start to feel like an audience being farmed. And the loudest voices — the ones who post eleven times before lunch — become the product.

The metric that ate the mission

Most community tools inherit their measurement habits from advertising. Daily actives. Session length. Posts per user. Notification open rates. These numbers are easy to move and almost never tell you whether a street is friendlier than it was last spring.

Ask a residents' association what success looks like and you'll hear different things. The arguments stopped. Three new people came to the meeting. Someone finally fixed the bin store. None of that shows up in a dashboard, which is why it gets ignored in favour of whatever is easy to chart.

The 2023 Online Safety Act and the codes now being phased in by Ofcom have pushed the conversation in an unexpected direction. Platforms are being asked, formally, to think about how their design affects users — not just what users post. That's a big shift. It used to be enough to say the content was the problem. Increasingly, the question is whether the shape of the thing is the problem.

What friction actually looks like

Friction gets a bad name because it's often just badly built software — captchas, broken logins, forms that time out. That's not what we're talking about. Deliberate friction is a design choice that slows a person down at the exact moment they're about to do something they might regret, or that makes belonging mean something.

Three practical forms stand out.

Verified local membership. Not a national ID check — just enough to know that someone claiming to live on your road probably does. A posted card, a neighbour vouching, a council tax letter, an invite from an existing member. Each has trade-offs around privacy and inclusion, and none is perfect, but all of them change the tone. People behave differently when they're recognisably themselves in a known place.

Slower posting. A compose window that asks you to pick a category, an audience and an expiry date. Thirty seconds of deliberate decision-making removes a surprising amount of rubbish. Some UK groups have gone further and batched posts into a daily digest — one notification, one look, done. The engagement numbers drop. The complaints about noise drop faster.

Human moderation with a face. Volunteers or paid staff whose names are visible, working to a short published rulebook, with a stated response time. Slow, unscalable and worth every hour. Automated filters catch slurs; they don't catch the person who makes every thread slightly worse.

Three places it's happening

A tenants' forum in Leeds moved from an open Facebook group to a closed board with postal verification and a once-a-week digest. Membership halved. Attendance at the actual monthly meeting doubled.

A repair café network in Bristol switched off comments entirely, replacing them with a single 'I can help with this' button that routes to a coordinator. Arguments about whether a kettle was worth fixing disappeared; the queue of people wanting to learn soldering got longer.

A timebank in Cardiff introduced a two-step process for new members: an online form, then a phone call with a volunteer. It sounds like a barrier. In practice it's the thing members cite most often as why they stayed — someone spoke to them before they were allowed in.

None of these are scalable in the Silicon Valley sense. All of them are replicable.

The 2026 argument about feeds

The wider debate has caught up. Age assurance, addictive feed design, the DSS/friction question, and the slow grind of getting the Online Safety Act's codes into force have put design choice on the political agenda in a way it wasn't five years ago. The coroner's findings in the Molly Russell inquest back in 2022 — that algorithmic curation contributed to her death — still sit at the centre of that argument, and rightly so.

But there's a risk in framing all of this as a child-safety problem only. Most adults don't love infinite feeds either. They've just been told that's what the internet is. Anyone building for a village, an estate, a hobby group or a union branch has an opening here: design the small, slow version and let people notice how much better it feels.

Principles worth stealing

• Measure belonging, not minutes. Survey three questions a quarter: do you know more neighbours than last time, would you ask them for help, would you recommend this to someone new.
• Make joining take a day, not a second. Not a fortnight — a day. Enough time for a human to say hello.
• Give every post a shelf life. Expiring content quietly solves the archive problem and discourages performance.
• Cap notifications at one a day by default. Let people raise it if they want. They mostly won't.
• Publish the rules, the moderators and the response time. Legitimacy comes from visibility, not secrecy.
• Design the offline handoff. Every feature should have an answer to 'what does this do for the room?'

The boring conclusion

The uncomfortable truth for anyone building social tech is that the healthiest communities produce worse numbers. Fewer sessions. Lower frequency. Longer silences. If your board deck rewards the opposite, friction will keep losing — no matter how many times you write 'community' in the strategy doc.

The alternative is to be honest about what you're measuring and then defend it. A platform that helps forty people on one street trust each other a little more is a success even if it never trends. That's not a growth story. It's a living one.