A social platform for your neighbourhood
Start with something useful, end up with something that matters
2026-03-25 — 2026-03-25
Wherein a fee-funded, resident-verified listings board is proposed, and, when a February heatwave cuts power, generators and insulin fridges are coordinated without algorithmic filtering.
Interested? Get in touch or sign up for updates.
This is the third in a series of posts about practical community infrastructure building. The first proposed neo-friendly societies for financial mutual aid. The second explored community-owned AI compute. This one is about something that sounds simpler but might be harder: building a local social platform that people actually use.
This post is even less definite than its companions, were that possible. A friendly society has a clear value proposition (money when you need it) and a well-understood legal form. A compute collective has a concrete asset (a machine) and a measurable output (tokens per second). A local social platform has … vibes? Community? The warm feeling of knowing your neighbours?
These are real goods. But they’re harder to price, harder to pitch, and much harder to bootstrap, because the value of a social platform is almost entirely a function of whether other people are on it—the cold start problem that has killed a thousand well-intentioned alternatives to Facebook.
So why am I writing about it? Because I think the returns, if it works, are enormous—and because the other two projects in this series might naturally seed this kind of platform as a side effect, so we might as well be intentional about it. A friendly society is a social institution. A compute collective is a community of users. Both need trust, communication, and a shared sense of identity that doesn’t currently have a natural digital home. The social platform is the connective tissue.
1 Why now?
Two things have changed that make this more tractable than it was five years ago.
1.1 The incumbents are falling apart
This is the enshittification thesis, and it’s moved from theory to lived experience. Cory Doctorow’s account of the platform lifecycle—subsidise users to build lock-in, extract value once they’re trapped, then pivot to serving advertisers and shareholders at users’ expense—is something that we’ve all seenplay out so many times that it is getting embarrassing that we do not plan around it. It’s a description of what Facebook, Instagram, and Twitter have already done. Paul Krugman has argued that this logic extends to any business with network effects, not just social media specifically.
The practical consequence is that the platforms people relied on for local coordination—Facebook Groups, Facebook Marketplace, Facebook Events—are getting worse, not by accident but by design. Meta is optimising for ad revenue, not for helping us find a second-hand couch or organize a street party. The gap between what these platforms provide and what communities actually need is widening.
That gap is a market niche. It’s a small one, and it won’t make anyone rich, but it’s real.
1.2 LLMs lower the development cost
Building a social platform used to require a well-funded startup. Now a competent developer with access to AI coding tools—or a small collective with a shared sovereign compute instance—can stand up a functional web application in weeks rather than months. The technical barrier has dropped dramatically.
Robin Sloan has written beautifully about this: an app can be a home-cooked meal. Not everything has to be a startup. Not everything has to scale. Sometimes we make a thing for the people we know, and it’s good because it’s for them specifically, in a way that a platform serving two billion users can never be.
Chris McCormick calls this sub-Dunbar software—software that works precisely because it doesn’t try to scale beyond the number of people who can actually know each other.
The question is whether we can build something that stays in that sweet spot: big enough to be useful (a marketplace needs liquidity; an events board needs events), small enough that trust is personal. I think the answer is neighbourhood-scale—somewhere between 500 and 2,000 people, roughly the population of a few city blocks or a small suburb.
2 Start with an itch
The graveyard of alternative social platforms is vast, and nearly all of them died of the same cause: they asked people to join a new social network for its own sake, on the promise that it would be better. It turns out that “better” is not a compelling reason to abandon the place where our friends already are.
So don’t start with “a social network”. Start with something concretely useful—something that scratches an itch that the incumbents have stopped scratching.
2.1 A local marketplace that doesn’t suck
Facebook Marketplace is notoriously terrible, and it’s getting worse. It’s flooded with scam listings, drop-shipped junk, and interstate sellers who aren’t actually local. The search is bad. The messaging is bad. There’s no meaningful reputation system, so we’re constantly fielding queries from people who ghost us. And Meta’s content moderation is simultaneously too aggressive (flagging legitimate listings for inscrutable policy violations) and too permissive (doing nothing about obvious fraud).
A neighbourhood marketplace that verifies residents (even loosely—a postcode check, a vouching system from existing members) and maintains a local reputation score would be immediately, obviously better than Facebook Marketplace for the specific use case of buying and selling things to people near us.
Is this a good business model? No, almost certainly not, if “business model” means “thing that generates venture-scale returns.” Classified advertising was once a goldmine, but the margins have been competed away, and a neighbourhood platform will never have the volume to make advertising work.
But “business model” can also mean “thing that sustains itself.” An incorporated association with modest membership fees—say $5-10/month per household, or $50-100/year—could cover hosting and basic development costs for a platform serving a few hundred to a couple of thousand users. That’s the friendly society funding model applied to digital infrastructure: we pay for it because it’s ours and it’s useful, not because someone is extracting profit from our attention.
2.2 Other wedges
The marketplace is just one entry point. Depending on what the community actually wants, the seed might be:
A local events calendar. I’ve ranted about this before—Facebook is a strong incumbent, and Facebook does it badly. A simple, non-algorithmic calendar of what’s happening in our area, maintained by the people who live there, would fill a real gap.
A skills and services directory. Who in the neighbourhood can fix a tap? Who does tutoring? Who has a ute and will help you move for a slab of beer? This information exists socially—people know it about their friends—but it’s not discoverable. A lightweight directory, backed by reputation, makes it discoverable.
A tool and resource library. Do I really need to own a pressure washer I use twice a year? Tool libraries exist in physical form (Melbourne has several), but a digital layer makes them more accessible.
A mutual aid coordination board. When the heatwave hits or the storm knocks out power, people need to know who needs help and who can provide it. This is the crisis function from the scenario above—but it works best if the social infrastructure already exists before the crisis.
I think the right wedge is very important — we need to scratch an actual itch. But we would need to choose something concrete, see if people use it, and let the platform grow in the directions the community pulls it. This is the “home-cooked meal” philosophy: we’re making this for specific people— ourselves, so ask each other what we’re hungry for.
What I’m hungry for is community sovereignty, but that’s more of an ingredient than a meal.
4 What about the attention economy?
The platforms we’re trying to escape are attention-extraction machines. Their business model is to capture as much of our time and cognitive bandwidth as possible, because that’s what they sell to advertisers. This produces the behaviours we’ve all learned to recognize: infinite scroll, notification bombardment, algorithmic rage-bait, the manufactured sense of urgency about things that don’t matter.
A neighbourhood platform funded by membership fees has no reason to do any of this. There’s no advertiser to sell attention to. The incentive is reversed: the platform should help us get what we need and then go do something else. Check the marketplace, find the bookshelf, arrange pickup, done. Check the calendar, see there’s a park clean-up on Saturday, RSVP, done.
This is a radical proposition in the current landscape, and it’s worth being explicit about: we are building a platform that wants us to spend as little time as possible on it. The value isn’t in engagement; it’s in the social infrastructure that exists because the platform exists.
5 Technical sketch
I don’t want to go deep on technology here—that’s for a companion post—but a rough sketch helps ground the costs.
The simplest version of this is a web app (mobile-friendly, not a native app—app store review is a bottleneck we don’t need) running on a modest VPS. Off-the-shelf options include Bonfire (a modular, federatable social platform built for community governance), Discourse (which despite being “forum software” is increasingly used as community infrastructure), or even a custom build on top of something like Elixir/Phoenix or Django with real-time features.
Ballpark costs for a neighbourhood of ~1,000 active users:
- Hosting: $20–50/month for a VPS (Hetzner, DigitalOcean, or a local Australian provider like VentraIP)
- Domain and email: $50–100/year
- Development: This is the wild card. An initial build using AI coding tools and volunteer labour could be near-zero in cash terms, or a few thousand dollars for a contractor to set up and customise a Bonfire or Discourse instance.
- Ongoing maintenance: 5–10 hours/month of volunteer or modestly-compensated time
At $5/month per household and 200 participating households, that’s $1,000/month—more than enough to cover hosting and a small maintenance stipend, with surplus for improvements.
The federation question is interesting: platforms like Bonfire speak ActivityPub, which means a neighbourhood instance could connect to the wider Fediverse (Mastodon, etc.) for broader discovery while keeping local governance local. Whether this is desirable depends on whether the community wants to be a walled garden or a node in a network. I’d bet on starting closed and opening up deliberately, rather than starting open and trying to close down later.
6 How this connects to the other projects
This platform is the social layer that the other mutual-aid projects need.
The neo-friendly society needs a communication channel that isn’t Facebook, isn’t email, and isn’t yet another WhatsApp group. It needs a place where members can discuss proposals, vote on decisions, and coordinate activities. A local social platform provides this.
The compute collective needs a community of users. The local platform’s members are the natural user base, and the AI infrastructure can in turn power features of the platform—content moderation assistance, search, translation for multilingual neighbourhoods, smart matching for the marketplace.
Together, these three projects form something like a stack: financial resilience (the friendly society), computational sovereignty (the compute collective), and social infrastructure (the local platform). Each is useful on its own; together, they’re the skeleton of a genuinely self-reliant local community.
7 Prior art
7.1 Platform cooperatives
Nathan Schneider’s work at the Platform Cooperativism Consortium is the most developed theory of community-owned digital platforms. His argument is that community ownership won’t emerge from open protocols alone—you need to deliberately invest in ownership ecosystems with real governance structures. The Platform Co-op School has highlighted Hylo as a working example of community-owned social infrastructure.
7.3 RadicalxChange governance
The RadicalxChange movement—associated with Glen Weyl, Audrey Tang, and the Plurality project—offers interesting governance mechanisms that could apply to a community platform: quadratic funding for deciding which features to build, data dignity frameworks for how member data is handled, Harberger-style mechanisms for allocating scarce resources (like prominent placement on the marketplace). These are more speculative, but they point toward forms of community governance that go beyond simple majority voting.
7.4 Existing local platforms
There are precedents, though none have broken out. Nextdoor started with a local-trust model and then seems to have enshittified in exactly the way Doctorow’s theory predicts. Various attempts at “local social” have launched and died—but the ones that survived longest were the ones attached to a concrete use case (marketplace, events) rather than generic “local social networking.” The lesson is: wedge in with utility, grow into community.
8 What’s still missing from the literature
Geek time! I would also like to research how and why people join these things, and learn how to make local platform development easier. How do users actually migrate from an enshittified platform to a community alternative when network effects penalise first movers? This is a coordination game with heterogeneous switching costs, and there’s almost no formal modelling of it. The RadicalxChange Plurality book gestures at this but doesn’t model it. If someone wants a research problem that matters, hit me up.
9 Open questions
- What’s the right first wedge for a specific neighbourhood? Marketplace? Events? Something else? This probably varies by community and the only way to find out is to ask.
- How do you verify “local” without being creepy about it? Postcode checks? Vouching? Physical meetups?
- What’s the right governance structure for a platform that’s too big for consensus but too small for representative democracy?
- How do you handle the transition when someone abuses the platform—doxxing, harassment, commercial spam? The norms for a 500-person neighbourhood platform are different from the norms for Twitter.
- Is federation (ActivityPub, etc.) an asset or a liability at this scale? Does connecting to the global Fediverse dilute the local trust that makes the platform valuable?
- Can this platform model be replicated—published as a template, the way the friendly society template is meant to be?
As with the other posts in this series: if you know about any of this, if you’ve tried something like this, or if you’re in Melbourne and want to help build it, I’d love to hear from you.
The technology is the easy part. The institutional design is the interesting part. The actual community is the hard part. Let’s start.