RSA Connect: the Royal Society of Arts as a mind of minds

I began this enquiry partly out of frustration. I have been an RSA Fellow for nearly 20 years. It is a unique and wonderful institution, and I am proud to be part of it. Yet sometimes, at John Adam Street or at an event outside London, I get a sense of missing out: there are people I would enjoy talking to, that I would learn from, people I might collaborate with, and no easy way to find them or make a connection. The talks and discussions are often interesting, but the real power of the organisation is its people. Fellow-to-Fellow connection is episodic, discovery is often accidental, and consequently the scale of collaboration is limited. Our most recent attempt to address this is Circle, a moderated discussion platform with around 6,700 members, roughly 22% of the Fellowship. A typical post receives between 10 and 15 replies from a potential audience of thousands. Circle functions more like a structured digital commons than a network: it requires moderation, cannot self-organise, and carries no deep understanding of who Fellows are or what they might offer each other. It is a limited technology.

The RSA is a child of revolution, and we should draw inspiration from that. Born in 1754 at the beginning of a transformation in the way we live, work, create value, travel, and socialise that we are still living through, it was formed precisely to channel the energies of that revolution, to encourage the arts, manufactures and commerce in an age of transformative change. It was not a passive bystander but a catalyst for prosperity. While the organisation has deepened the power of fellowship over the centuries, it could be argued that it has drifted from that original catalytic ambition, becoming more a forum for ideas than a driver of change. Klaus Schwab, founder of the World Economic Forum, popularised the idea of four distinct industrial revolutions, first, second, third, and fourth, each separated by a break in technology or energy source. There is something useful in that periodisation, but also something misleading. Steam, electricity, digital computing, and now artificial intelligence are better understood as successive accelerations of a single continuous transformation. We are not entering a new revolution. We are living through the same one, in its latest and perhaps most consequential phase. The RSA was born at the start of that transformation. The question is whether it can return to its original role, not as a witness to change but as a shaper of it.

There is something fitting here, because the technologies now emerging from this latest phase may provide exactly the tools to make that possible. So my question is: could we do things differently? Could we harness them to better connect the 30,000 Fellows, create a network functioning at a higher level than it currently does, and form, as a kind of super organism, a mind of minds? Could AI be the means by which the RSA returns to the heart of the debate?

The Fellowship numbers over 30,000, a reservoir of experience, talent, and creativity unmatched by most institutions. Yet this resource remains largely latent. The paradox is that the more Fellows we have, the more value exists in principle, but the harder it becomes to find and use that value in practice. Density produces richness, but also inaccessibility. Without a new architecture of connection, the Fellowship remains an underused asset while the world's challenges accelerate around it. RSA Connect is my proposed answer to that problem.

What a network is — and what it is not

The RSA Fellowship is extraordinary: experienced, knowledgeable, creative, spanning every field of human endeavour. It is one of the richest concentrations of talent and purpose in the world. And it is not a network.

A network is not a set of things that could be connected to each other. It is a set of things that are connected, with measurable flow between them. The test is not membership or proximity or shared values; it is flow. Can a signal travel from any node to any other node through the system? If yes, it is a network. If not, it is not.

Think of the telephone system. It is not a network simply because telephones exist and people are registered as subscribers. It becomes a network when every subscriber is connected to every other, when any number can call any other number. But even then, connection alone is not enough. The difference between being connected and making a call is information flow. The line exists. The call is the flow. RSA Connect first establishes the lines. Then enables the calls.

Within the Fellowship, there are many small-world networks, clusters of Fellows who know each other, collaborate in overlapping fields, share social circles. Flow happens within those clusters and it is real and valuable. But flow between clusters is accidental, governed by proximity and chance rather than relevance and intent. Fellow A in one cluster cannot reliably reach Fellow B in another, even when B is precisely the person A needs. The Fellowship as a whole, as a single network with measurable flow across all its nodes, does not yet exist. RSA Connect proposes to create it for the first time.

[INSERT GRAPHIC 4 — small-world vs full mesh. The toggle shows the difference between the current RSA's isolated clusters and RSA Connect's fully meshed network, with the network test result visible in the stat cards.]

The mathematics of why this is hard

What appears on the surface to be an organisational problem, how to connect people who share values and purpose, turns out on examination to be a mathematical one. The base network condition between any two Fellows is a directed connection: A→B is one connection, B→A is a separate connection. They are counted separately because reaching out and being reached are different acts. In a fully meshed network of N Fellows, the total number of directed connections is N(N−1). For 30,000 Fellows, that is approximately 900 million directed connections.

This number immediately reveals the problem. Metcalfe's Law shows that the value of a network grows quadratically with the number of nodes: each new Fellow creates connections with every existing Fellow, multiplying potential value rapidly. Reed's Law goes further: the number of possible meaningful subgroups within a network grows as 2ᵎ. At 30,000 Fellows, that number exceeds the atoms in the observable universe.

The mathematics reveals something important: the richness of the network is simultaneously its value and the primary impediment to realising that value. This is the Density-Value-Accessibility Paradox, the DVAP. As the network grows, latent potential grows without limit. But realisable value, the value that can actually be accessed and used, rises, peaks, and then falls as complexity overwhelms the navigational capacity of human cognition and current tools. Events, a CRM database, a discussion forum: these support flow within small-world clusters. They cannot generate flow across a network of 900 million directed connections. The complexity problem is not a failure of will or organisation. It is a mathematical condition. And it is precisely why the challenge has resisted solution for so long: the tools required to address it did not exist.

Human intelligence hits limit conditions at three nested levels: the individual, where Dunbar's research establishes that human cognition can maintain approximately 150 active social relationships before the cost exceeds capacity; the collective social, where even groups of individuals using digital tools cannot navigate a Fellowship of 30,000; and the organisational, where the history of complex institutional failures, from the infected blood scandal to Grenfell, from King's Cross to the Post Office Horizon case, shows repeatedly that information needed to act well existed somewhere in the system but could not be assembled into a coherent, shared, actionable picture. The problem is not communication. It is the absence of an architecture capable of turning dispersed signals into connected intelligence.

AI, specifically graph neural networks, large language models, and semantic heuristic tools, naturally operates at precisely these limit conditions. It was built for high-dimensional combinatorial problems that exceed human navigational capacity. The same class of solution that found Move 37 in the 2016 Go match, a move with a one-in-ten-thousand probability of being played by any human expert, is what RSA Connect applies to the Fellowship's latent intelligence.

The graph below models this. The x-axis is the number of Fellows N. The purple curve is latent potential, N(N−1) directed connections, always growing, indifferent to whether any connection is ever used. The amber dashed curve is realisable value without tools: it rises as the early network forms, peaks, then falls as complexity overwhelms navigation. The two vertical markers show the proof of concept at 1,000 Fellows and the full Fellowship at 30,000. The green curve shows what RSA Connect does, lifting realisable value away from the amber collapse through four stages toward the latent potential ceiling. The two sliders let you adjust heuristic tool strength and system maturity independently.

[INSERT DVAP GRAPH — latent potential, realisable value without tools, RSA Connect curve. Sliders: heuristic tool strength and system maturity. Markers: PoC at 1,000 Fellows, full Fellowship at 30,000.]

What RSA Connect does — and does not do

RSA Connect does not create value. It creates the conditions for value to emerge. That distinction matters. The value in the Fellowship is already there, in the knowledge, experience, and creative capacity of 30,000 Fellows. What RSA Connect does is make that value realisable, by converting latent potential into navigable connections.

It does this through four stages. The stepper below walks through each one.

[INSERT GRAPHIC 3 — four network conditions stepper. Click through: full mesh, semantic profiles, heuristic enquiry, system learns.]

At the heart of the system is the Fellow dashboard, the personal portal into the network. Each Fellow adjusts heuristic sliders to weight different search strategies, directing their enquiries by relevance and intent. The result is a ranked list of Fellows most likely to be useful for a given enquiry, drawn from across the whole Fellowship rather than from within existing social circles.

[INSERT GRAPHIC 1 — Fellow dashboard. Live heuristic sliders rerank suggested Fellows in real time as discovery priorities are adjusted.]

Each Fellow also has a semantic profile, a representation of their intellectual character across multiple dimensions, including the cognitive modes through which they prefer to work. This profile shapes how RSA Connect routes enquiries and surfaces connections. Two Fellows with complementary profiles, one strong in systems thinking and mathematical reasoning, another in linguistic and interpersonal intelligence, may be exactly what a given enquiry needs, even if they have never met.

[INSERT GRAPHIC 2 — semantic profile radar. Adjustable sliders for each of Gardner's seven intelligences, live radar chart, and a generated fingerprint.]

From epistemology to ontology — and a question worth holding

The fourth network condition is where something genuinely new begins to emerge. As the system accumulates enquiry patterns, it builds what might be called an epistemology, a model of what the Fellowship knows and how it knows it. From that epistemology, over time, an ontology begins to form: the system develops categories, relationships, and affinities that are its own, grown from the actual intellectual life of the Fellowship rather than imposed from outside. This is the mind of minds, not a metaphor but a description of what a learning network with memory, connection, heuristic tools, and an epistemological structure begins to become. It is not general intelligence. It is situated intelligence, specific to the RSA's community, values, and purpose.

Whether a system of this kind could, at some point, be described as conscious is a question I raise rather than answer. It is not a frivolous question. A system with persistent memory, self-referential learning, and an emergent ontology begins to share some of the structural properties that philosophers associate with mind. I leave that question open, with the observation that it is exactly the kind of question the RSA is placed to explore.

[INSERT 3D NETWORK VISUALISER — rotating fully meshed network, three semantic clusters, flow strength sliders per cluster, rotation speed control.]

A prototype for something larger

The proof of concept is 1,000 Fellows: 999,000 directed connections, a Reed space of possible clusters running into the billions, and a real community of people with genuine intellectual purposes. If RSA Connect works at that scale, the architecture scales further.

And the architecture is general. The same design that addresses the Fellowship's complexity problem is, in principle, applicable wherever complexity and uncertainty exceed the navigational capacity of existing systems, in public health, in infrastructure delivery, in governance. In each of these domains, the same pattern recurs: information existed, but there was no architecture capable of assembling it into a coherent, shared, actionable model. RSA Connect is envisaged first as a working prototype within the Fellowship. What it demonstrates, if it works, is that the problem is solvable.

The RSA was formed in 1754 to be a catalyst for prosperity in an age of transformative change. That is still the job. The question is whether it will meet this moment with the architecture it requires.

Stephen Feber FRSA