When will the Human Web be captured?

Before platforms arrived, contribution was local, fragmented, and mostly invisible. You helped a neighbour. You volunteered at a community centre. You supported a cause. None of it accumulated into a record. None of it carried across contexts. None of it made you more findable, more trusted, more recognised anywhere else.

Then platforms arrived, and they did something genuinely useful. They aggregated. They matched. They tracked participation. They surfaced relevant connections. They made contribution visible in ways individuals and small organisations never could.

And then, over time, they captured it.

This is not a moral story about bad companies. It is a structural story about how certain kinds of infrastructure inevitably accrue power. The platforms that won did not win by being exploitative. They won by being better. And they became better through a specific mechanism: they extracted signals from a distributed ecosystem and fed them back as improved services. Search engines extracted the link graph and gave us better discovery. Social platforms extracted the social graph and gave us better connection. Marketplaces extracted transaction data and gave us better matching. In each case, the platform began as the helpful participant and ended as the ecosystem.

A new layer of the internet — what the technology community has begun calling the agentic web — now faces the same pattern. The human-centred articulation of that layer is what we call The Human Web. The first is the broader space being built; the second is the position on what it should be for.

The thing being built right now

A new layer of the internet is forming. AI agents, acting on behalf of individuals and organisations, will increasingly mediate access to opportunity, services, and coordination. This is not speculative. The infrastructure is being laid: agent-to-agent protocols, context-passing standards, verifiable credentials, the early shapes of cross-service identity.

The question is not whether this layer will exist. It is who builds it, under what rules, and who ends up owning its central coordination functions.

The optimistic story — the one being told in manifestos, including mine — goes like this. The new layer will be open. Individuals will hold their own identity. Agents will represent people, not platforms. Contribution will be portable. Value will flow back to those who produce it. A civic, human-first alternative to the closed-platform model of the last twenty years is possible.

What the optimistic story is finally specific enough to say — and what makes it different from the open-architecture promises of the last two decades — is what the alternative is for. The Human Web exists to serve human flourishing: what people do, what they become, the lives that follow from the doing. Beneath that named purpose sit two paired architectural commitments — to human agency, and to the conditions for collective wellbeing — which together unpack what taking flourishing seriously actually means at the level of the architecture. Not as decoration on a technology project, but as the substrate the architecture rests on. This is the proposition that the wellbeing-economy movement has been making in policy and academic terms for twenty years and that operations have, until recently, lacked.

I believe that story. I am also worried that "open" and "agentic" by themselves are not enough to make it true.

The mechanism of capture

Decentralised systems struggle with three specific problems.

Trust and safety at scale. Any given node in an open network can verify that it is itself trustworthy, but it cannot, on its own, tell which other nodes are safe to interact with. Scams, low-quality providers, bad-actor organisations, compromised agents — these are not individual-node problems. They are ecosystem problems.

Discovery. Open networks make it possible for an agent to exist and publish an offering. They do not, on their own, make it possible for the right agent to find the right opportunity at the right moment. Discovery at scale requires signal aggregation.

Signal interpretation. Millions of interactions produce patterns no individual node can see. What kinds of volunteering lead to sustained community connection? Which coordination patterns predict successful carer networks? Which credentials actually correlate with what they claim to? Answering these questions requires data at ecosystem scale.

These are not solvable at the individual-node level. They require a layer of cross-ecosystem intelligence.

And that layer is exactly where platforms have always captured networks in the past. If we do nothing, it is where they will capture this one. The sequence is already visible. A large player will arrive offering "trusted agents," "verified coordination," "smart matching across services." The service will be genuinely better than anything available in a pure node-to-node architecture. Users will adopt it rationally. And once it is embedded in enough interactions, it will define the environment everyone else has to operate in.

The capture will look, as it always does, like a helpful service becoming indispensable.

The missing layer

If the problem is clear, the required response is clear too. The Human Web needs its own signal layer — a shared, open, independently governed layer that can do the things decentralised networks cannot do on their own: process trust, enable discovery, interpret patterns, coordinate at scale.

Not a platform. Not a closed service. Not a single organisation's product. A shared layer that operates alongside and on behalf of the ecosystem of nodes, processing signals without owning them, returning insight without capturing the underlying data.

This is structurally different from a platform in one critical respect: it does not hold the user relationship. The user's identity stays in their wallet. Their data stays in their personal environment. Their credentials stay theirs. The signal layer sees enough, in anonymised and aggregated form, to make the ecosystem work — and no more.

If we build this layer, the Human Web can coordinate at the scale that matters without becoming a platform in disguise. If we don't, a platform will build it for us, and everything else we're building becomes window-dressing on someone else's infrastructure.

Why open standards matter more than they sound like they do

Every ambitious digital-infrastructure document reaches a moment where it says "open standards are critical," and a reader's eyes glaze over. So let me be specific about why open standards are not merely a nice-to-have here.

If signals are going to be aggregated across a distributed ecosystem, every node needs to speak a common language about what a signal is. What counts as a verified participation event? What does a credential look like when it's issued by a charity versus by a council versus by a neighbour? How does an agent from one service trust a credential minted by another? Without shared schemas, the signal layer cannot function — and the only organisation with enough power to impose a schema on everyone else would be, predictably, the one that captures the ecosystem.

Shared standards, developed in public, adopted by independent alliances, are the mechanism by which the signal layer stays open rather than becoming a single organisation's asset. This is why the work of bodies like the Open Data Institute is foundational rather than supplementary to the Human Web vision.

The window

This is not a theoretical problem. It is a timing problem.

There is a window — perhaps eighteen to thirty months — in which the signal layer can be built under rules that prevent capture rather than under platform logic. The window is open because:

• The ecosystem of nodes is still forming. Patterns have not yet locked in.
• Users are not yet dependent on any one signal provider.
• Standards are still being negotiated.
• Institutional actors (governments, civic bodies, standards organisations) are paying attention.

The window will close when:

• A single large provider becomes the de facto signal layer for a key domain.
• Dependency accrues.
• The cost of switching infrastructure exceeds the benefit of doing so.
• Institutions quietly give up on building alternatives.

We cannot know precisely when that happens. We can reasonably assume it is sooner than we'd like.

What needs to happen

Concretely, four things.

Shared signal infrastructure. Layers for trust, discovery, and coordination, operating under independent governance and open standards. The point is not that any single organisation owns them — the point is that whoever operates them is bound by rules that prevent capture, with accountability that survives any change of operator.

Open standards adopted before the capture window closes. Participation credentials. Contribution events. Agent interoperability. Trust attestations. Each of these needs a schema, a governance body, and live adoption in real pilots.

Data portability enforced architecturally, not just regulated. A user must be able to leave any Human Web service and take their identity, credentials, agents, and history with them. Not as a legal right they would have to sue to enforce — as a property of the infrastructure itself.

Real deployment in real domains. This is where most infrastructure work has failed. Abstract standards that never meet operational reality produce nothing. The first domains have to be chosen carefully: places where the need is already acute, where trusted coordination across many actors already matters, where the value of the infrastructure is visible quickly.

Volunteering, wellbeing, civic participation, mutual aid — the domain of human contribution — is where the signal layer can be built first. The fragmentation is already there. The actors are already motivated. The gap between what people do and what gets recognised is already wide enough that closing it is self-evidently valuable.

A concrete example

The Open Participation Alliance, forming with civic-tech partners across the UK volunteering ecosystem, is the first domain-level partner initiative built on the Human Web architecture. Its founding pilot — Lifted, Do IT, Team Kinetic, Lifepass — is built to be functionally complete by design: each partner does distinct work the others don't duplicate, covering the primary functions an agent ecosystem in the participation domain needs. The alliance commits to shared standards, independent governance, and a Charter that protects against the failure modes the Human Web's other manifestos are vulnerable to. It also articulates an architectural pattern for sponsorship — the way organisations (employers, councils, foundations) can fund or invite populations they have standing with into the Web, without becoming platforms themselves.

I mention this not to close the essay with a product pitch but to make the point that the window is not theoretical. Initiatives of this kind have to form, and have to start operating, now. If they do, the signal layer for at least one domain gets built in the open. If they don't, we are relying on Big Tech and competing regulators to accidentally produce something better than what they have produced so far — which has been, roughly, nothing.

The line we need to hold

The Human Web will not stay open just because it is built on open protocols. Open protocols do not, on their own, solve the trust, discovery, and signal-interpretation problems that any functioning ecosystem needs solved. If we don't build the signal layer under rules that prevent capture — independent governance, portability, transparency, accountable standards — we will discover, probably quietly, over three or four years, that we have built another platform layer. Just with better branding, and with the addition of having wasted the one window we had to do this differently.

The question in the title of this essay is not rhetorical. The Human Web will be captured — unless enough people building it realise, in time, that openness without shared infrastructure is not enough. The urgent question is whether the layer gets built, in time, to rules that survive it.

And the rules we build it under are for something specific: an internet whose central concern is what people do and the lives that follow from the doing. Anything less is the old pattern with new vocabulary.

Build the signal layer. Now. Before it is built for us.