Why You Cannot Control a Startup Community and What Guiding One Actually Looks Like

There is a familiar failure mode in ecosystem building that has played out in enough cities to constitute a pattern. A government body, or a university, or a well-funded foundation decides to create a startup hub. Committees are formed. A strategy is commissioned. Targets are set, usually involving some combination of startups founded, jobs created, and capital raised. Physical infrastructure is built. Two or three years later, the facilities are underused, the startups that benefited have largely graduated or left, and the people who were supposed to be galvanised by the initiative are politely sceptical of the next one.

Brad Feld and Ian Hathaway wrote The Startup Community Way partly as an attempt to explain why this happens and what to do instead. The full summary of The Startup Community Way covers the complete framework. This piece focuses on one idea that runs through the whole book: that the instinct to control an entrepreneurial ecosystem is not merely ineffective but is actively counterproductive, and that the alternative is guidance rather than management.

The complex systems argument

Feld and Hathaway’s theoretical move is to classify startup communities as complex adaptive systems, a category drawn from the natural sciences and from the work of institutions like the Santa Fe Institute. A complex adaptive system is one in which behaviour emerges from the interactions of many agents, none of whom can see or control the whole. Think of ant colonies, immune systems, or financial markets. The whole behaves in ways that cannot be read off from the behaviour of any individual part.

The key implication for ecosystem builders is that the system cannot be optimised from the outside. There is no vantage point from which a planner can see the whole network and adjust its behaviour the way an engineer adjusts a machine. Interventions produce effects that ripple through feedback loops in ways that are genuinely difficult to predict. This is not a counsel of despair; it is a description of how the system actually works, and ignoring it is what produces the pattern described above.

It is worth noting that this argument has respectable intellectual ancestry. W. Brian Arthur’s work on increasing returns and path dependence, and Jane Jacobs’s earlier observations about how urban economies self-organise, are doing a great deal of the underlying work here, even where they are not explicitly cited. Feld and Hathaway’s contribution is to apply this tradition specifically to entrepreneurial ecosystems and to draw practical conclusions from it.

The illusion of control

The book is crisp on what goes wrong when ecosystem leaders treat a complex system as though it were a complicated one. Complicated systems, like an aeroplane or a nuclear plant, can be understood through their components, modelled in detail, and managed by experts who have the right information. Complex systems cannot. Their behaviour is irreducibly emergent. The same input can produce very different outputs depending on initial conditions and the feedback dynamics of the network.

Policymakers who build “startup master plans” are typically treating a complex system as a complicated one. They assume that if you build the right infrastructure, fund the right programmes, and measure the right outputs, the ecosystem will respond predictably. It will not, because the ecosystem’s behaviour is a product of the relationships and cultural norms inside it, which are neither visible in a spreadsheet nor responsive to institutional authority.

This is not simply a criticism of government policy. The same error appears inside large organisations when innovation labs are created with carefully designed processes and quarterly milestones, and produce very little genuine innovation. The error is structural, not political.

What guidance actually means

Feld uses the gardening metaphor throughout his writing on this subject, and it is useful enough to be worth unpacking carefully. A gardener does not tell plants how to grow. The gardener tends the soil, manages light and water, removes obstacles, and observes. The growth is the plant’s. The gardener’s job is to create the conditions under which that growth is possible and to stay attentive enough to notice when something is going wrong.

Applied to ecosystem building, guidance means several things. It means connecting people who would benefit from knowing each other, without requiring any particular outcome from the connection. It means creating forums where founders can speak honestly about failure, which spreads institutional knowledge through the community faster than any training programme. It means being consistent and available over long periods, because trust in networks accumulates through repeated interaction rather than through formal authority. And it means resisting the urge to extract credit or control from the relationships that are formed.

The practical guide to applying the complexity mindset covers the operational side of this in some depth. What is worth emphasising here is that guidance is not passive. It requires sustained attention, careful observation, and the willingness to act at the right moment rather than the scheduled moment.

The role of trust

Control-based approaches to ecosystem building tend to produce compliance. Guidance-based approaches, done well, produce trust. The distinction matters because trust is what makes the system work. In a high-trust network, founders share competitive intelligence with potential rivals, because they believe in the long-term benefits of a healthy community. Investors take meetings with founders who do not fit their thesis, because they know that referrals compound. Experienced entrepreneurs mentor the next cohort without being paid to do so, because they received similar investment earlier and the culture expects reciprocity.

None of this happens under institutional command. Trust of this kind is built through repeated voluntary interactions over extended periods. It is fragile in the short run and extraordinarily durable in the long run. The ecosystems that have it (Boulder, Tel Aviv, Cambridge) look qualitatively different from those that do not, in ways that raw startup counts and capital figures do not capture.

This connects directly to the book’s companion principle, covered in the piece on why quality and connectivity matter more than quantity. Guidance creates the conditions for deep connections. Control creates the conditions for shallow ones, because in a controlled environment, relationships are formed around the institution rather than around genuine mutual interest.

Where this is hardest to apply

It is tempting to read the guidance-over-control argument and conclude that it is mainly a critique of government policy or institutional overreach. That reading lets too many people off the hook. The same dynamic plays out inside companies when product teams are over-managed, when innovation processes are audited by people who were not present for the conversations that produced the insight, or when senior leadership tries to manufacture a “startup culture” through formal initiatives and branded values statements.

The hardest version of the guidance argument is the internal one: accepting that if you want genuinely creative output from a team or a community, you have to relinquish the kind of oversight that would make you feel comfortable. This is not a popular message in most institutions, which is presumably why so much ecosystem building continues to be done through control mechanisms despite considerable evidence that they do not work particularly well.

Common misconceptions about the argument

That guidance means doing nothing. The gardening metaphor is occasionally read as a licence for passivity. It is not. Tending an ecosystem requires consistent, skilled effort. The difference is that the effort is directed at conditions rather than outcomes.

That control fails only in public sector contexts. Private sector actors, including corporate innovation labs and venture-backed accelerators, make the same structural error with the same results.

That trust-based networks are naively idealistic. The most commercially successful startup ecosystems on earth are trust-based networks. The argument is empirical as much as it is normative.

That the framework only applies to large ecosystems. The same principles apply at much smaller scales: a team of ten people trying to build something genuinely new inside a large organisation is a miniature complex adaptive system, and it responds to guidance and control in the same ways.

A measured verdict on the book’s contribution

Feld and Hathaway are right about the central problem and broadly right about the alternative. The complex systems framing is genuinely useful because it gives practitioners a vocabulary for things they observe but have struggled to articulate: why their programmes are less effective than their inputs would suggest, why trust seems to compound in some communities and decay in others, why copying successful ecosystems tends to produce imitations rather than equivalents.

Where the book is less satisfying is in its guidance on what specifically to do. The diagnostic half is much stronger than the prescriptive half. Knowing that you should guide rather than control does not, by itself, tell you which connections to make, which forums to create, or how to intervene when a promising community is losing momentum. That work, the genuinely difficult applied work, is left largely to the reader’s judgement.

That is, in fairness, partly the point. A book that claimed to offer a detailed method for guiding a complex adaptive system would be claiming to solve a problem that its own framework says is unsolvable. The honest version is that the principles are clear and the application is context-dependent. The Startup Community Way provides the former. The latter is the practitioner’s job.