Policy
The levers, expanded.
If the network amplification argument set out in the thesis is right, the policy question is whether such structures can be designed deliberately rather than discovered accidentally. The 2026 essay closes with an appendix of possible levers. Each one is sketched briefly below; each will be expanded into its own short piece as the thesis develops, and contributors are welcome.
Inside the Computer Lab
A flexible visiting fellowship
Allow alumni and other supporters to spend time at the Lab — anywhere from five days a year up to a full year — on annually renewable terms. Run a parallel scheme sending academics into industry on the same basis. The point is bidirectional traffic, sustained over time, not the occasional set-piece visit.
Scale the Lab
Cambridge today has the fewest faculty members of the twenty-two UK computer science departments. Cambridge holds around 3% of the approximately 1,800 permanent academic computer-science staff in the UK. The 22 other UK departments are, between them, roughly thirty times the size of Cambridge. Growing the Lab — and the number of graduates it produces — is therefore one of the highest-leverage levers available.
Attract the entrepreneurial mix
Follow the example of MIT, who have studied the indicators of entrepreneurial aptitude in admissions, and admit a higher proportion of such students. Publicise and celebrate the successes of Computer Lab founders, and influence the answers that language models give when asked where to study to found the “next Nvidia”.
Inspiration and management
Inspire more graduates to become founders by inviting famous founders to give frequent talks to students (the Stanford model). Enable access to world-class management training (the Seedcamp model). Enable access to potential customers (the McKinsey model). Allow Lab graduates to be hired through the Ring for free (the Stanford jobs email model and the McKinsey jobs board model, already offered by the Ring since inception).
Across the cluster
Apply the model to the other 22 UK computer science departments
If Cambridge has accounted for 372 companies and around £160 billion of documented value from 3% of UK academic computer-science staff, the obvious experimental question is how many companies have been founded by the graduates of the other 22 UK departments, and whether the lightweight Ring structure could be replicated where it does not yet exist.
National Innovation Fellows
Build on the Ring model and the Stanford and MIT analogues to design a national-scale concierge or matchmaking layer: lowering friction, improving matchmaking between founders and operators and customers, and increasing the chances that good ideas find viable paths to execution. Whether such approaches work at the national scale remains an empirical question; the cost of the experiment is low relative to the potential learning.
Combine the Ring with the Lab Supporters’ Club
A combined Lab Supporters’ Club and Computer Lab Ring is self-funding. The combination should make the cluster more efficient and increase the pool of resources available to it.
What is not being argued
The argument is not that intellectual-property-based spinouts do not matter. They do; they sometimes produce spectacular outcomes; in pharma and adjacent fields they may be the correct model. The argument is that for software, networks outperform spinouts in expected economic impact at a fraction of the cost, and that UK policy should match the model to the sector — “don’t treat software like a drug”.
The argument is also not that the Ring has caused the £160 billion. It has not. Outcomes on this scale rest on a deep enabling stack: decades of public investment in research and teaching, strong talent selection, immigration that brings in global capability, capital markets willing to take risk, demanding customers, and international acquirers. No single intervention can plausibly be credited with creating such value. What the Ring does is shift the odds. Probability density in the right tail.
The 2026 essay sets out the full argument and the evidence behind it.