A Brief — 2026-07-02
Accelerator interviews ask for Sprunginnovationspotential — breakthrough innovation potential. The implicit model: somewhere in the applicant pool is the next steam engine, the next Gutenberg press, and the right interview question will surface it.
It won’t. Breakthroughs on that scale are not a supply problem you can select for in a cohort interview. They are rare in a way that has nothing to do with how well the search is run.
The “advanced tech” fields these calls point toward — bio-engineering, quantum cryptography — are not neutral ground where the best idea wins. They are heavily gate-kept: capital-intensive, credential-intensive, lab-intensive. Selecting for breakthrough potential inside gate-kept fields doesn’t select for the best ideas. It re-selects for who already had access before the interview started.
Funding bodies build lighthouses: flagship funded projects, flagship PIs, flagship labs that the rest of the field is meant to navigate by. But breakthroughs are more likely to come from solving commons-inspired coordination problems — shared data, shared standards, shared infrastructure — than from funding one more lighthouse. The lighthouse model photographs well. It is not where the leverage is.
Here is the uncomfortable part: today’s science system has already adapted to calls like this. It can absorb “help us find breakthrough potential” without changing anything structural, because the interview format itself reproduces the existing gates. Asking gate-kept fields to nominate their breakthroughs is not a disruption of the status quo. It is the status quo, dressed as a search for one.
Science policy has a polite term for the actual mechanism: Reputationsgemeinschaften — reputation communities. The less polite and more accurate term is cliques. Grants, invitations, citations, and who gets asked to interview for “breakthrough potential” all run through informal reputation networks before they run through any formal review criteria. A reputation community isn’t inherently corrupt. It’s closed by construction — and closed-by-construction is exactly the property that makes a field gate-kept, whatever the application form says about openness.
None of this is an argument for patience — it’s the opposite. Political leadership is trending toward the authoritarian, and in places openly toward the totalitarian, and that trend moves faster than a science funding cycle does. Commons-governed infrastructure isn’t just a fairness argument at that point; it’s a structural safeguard. Distributed, collectively governed compute is harder to seize than a handful of centrally funded lighthouses. If the infrastructure that can resist concentrated control isn’t built while building it is still possible, it won’t get built once it isn’t. That’s the actual urgency: not “innovate faster,” but “decentralize before centralizing power makes decentralizing the kind of thing you get stopped from doing.”
Compute is the clearest case, same as in the trespassing argument: publicly funded compute, governed as a commons rather than gate-kept by institution or platform, is what actually removes the barrier — not another accelerator selecting for who already had the access to build a pitch deck. Herumlungern im Gelände — strategic, interest-driven wandering through tools and APIs before the fences go up — is a counter-culture, not a universal option. Not everyone can afford to trespass. That’s the argument for funding the commons instead of funding the trespassers.
Not everyone can afford to trespass. That’s the argument for funding the commons instead of funding the trespassers.
One exchange is worth naming on its own: asked how philanthropic capital (Renaissance Philanthropy) and public capital (SPRIND — tax money) fuse in a program like this, the answer was that the philanthropic side takes the training component. That’s a real division of labor, and worth being explicit about: public money for infrastructure and research, philanthropic money for developing the people who’ll use it. It’s a reasonable split in principle. Whether it actually reaches the people without access — rather than the people who already knew where to apply — is a separate question, and the same gate-keeping problem this whole brief is about.
Companion piece to Trespassing — A Brief. Both trace back to the same 2026-06-15 conversation; this one covers the half of it that didn’t make it into the first brief. See also Seeking Out the Hans Zimmer of AI and The Champion Who Never Wrote a Line.