To fund breakthrough ideas, don't seek consensus. Instead, identify proposals that are highly polarizing among experts—where half think it's brilliant and the other half thinks it's terrible. This indicates a departure from the norm and holds the potential for true innovation.

Industrial strategy is more effective when focused on solving big problems, like creating healthy school lunches or landing on the moon. This "mission-oriented" approach stimulates innovation across many sectors, unlike traditional policy that just hands subsidies to favored industries.

The default assumption for any 'moonshot' idea is that it is likely wrong. The team's immediate goal is to find the fatal flaw as fast as possible. This counterintuitive approach avoids emotional attachment and speeds up the overall innovation cycle by prioritizing learning over being right.

The fund backs underfunded, high-risk ideas that others pass on. The goal isn't just to find a unicorn; it's to contribute to science by definitively disproving a hypothesis. A failure is viewed as "crossing out a wrong answer" for the entire field.

The most effective government role in innovation is to act as a catalyst for high-risk, foundational R&D (like DARPA creating the internet). Once a technology is viable, the government should step aside to allow private sector competition (like SpaceX) to drive down costs and accelerate progress.

The Pentagon's research arm, DARPA, used a million-dollar prize for a driverless car race to catalyze innovation. This contest model successfully attracted and identified the diverse engineering talent who would later lead the entire autonomous vehicle industry.

Frame moonshot projects like Google's Waymo not as singular bets, but as platforms for innovation. Even if the primary goal fails, the project should be structured to spin off valuable 'side effects'—advances in component technologies like AI, mapping, or hardware that benefit the core business.

OpenAI runs numerous parallel research projects (expansion), knowing most will fail. When a few show promise, it consolidates talent and resources onto those winners (contraction) to scale them up, before spreading out again to explore the next frontier. This cycle is applied to product as well.

Unlike weak-link problems (e.g., food safety) where you fix the worst part, science is a strong-link problem where progress depends entirely on the best outcomes. The optimal strategy is therefore to increase variance by funding more weird, high-risk ideas.