Loonshots

1. Loonshots are fragile breakthrough ideas that look crazy until they win. Bahcall coins the term for projects dismissed as foolish by most insiders — radar before WWII, statins before they cured heart disease, search before Google. They share a pattern: long incubation, repeated near-death, and a champion who refuses to give up before the world catches up. 2. The behavior of an organization is governed by phase transitions, not by culture. Drawing on his background in physics, Bahcall argues that the same people can be innovative in one structure and bureaucratic in another. Crossing a size or incentive threshold flips an organization from nurturing wild ideas to crushing them, just as water freezes at zero degrees regardless of the molecules involved. 3. Separate the artists from the soldiers. Healthy innovative organizations split into two groups: an artists' wing that develops loonshots and a soldiers' wing that scales and ships proven franchises. Each needs different incentives, metrics, and managers, and treating them the same kills both. 4. The two groups must love each other. Separation without exchange produces a research lab that ships nothing and a product line with nothing new to ship. Bahcall stresses bilateral respect — the soldiers must protect the artists' time and budget, and the artists must accept that the soldiers' revenue funds the experiments. 5. Manage the transfer, not the technology. The hardest leadership job is moving an idea from the artists' wing into the soldiers' wing. Bahcall illustrates with Vannevar Bush's wartime science office, Theodore Vail's Bell Labs, and Steve Jobs's second run at Apple — each case study turns on careful translation across the divide. 6. Incentives at moderate sizes destroy loonshots. As an organization grows, the rational career move shifts from championing risky ideas to climbing the ladder. Bahcall derives this mathematically and argues that flattening hierarchies, reducing managerial spans, and increasing project-skill fit can hold off the phase change. 7. Three deaths kill most loonshots. Bahcall identifies recurring failure modes: dismissing the early ugly version, killing the project after the first false signal, and starving the champion of resources just before the breakthrough. Most postmortems mistake these structural failures for bad ideas. 8. The lesson generalizes beyond companies. Nations, militaries, and research agencies are subject to the same phase transitions. Bahcall closes with policy implications: how to design institutions that protect the small group of crazy projects on which long-run progress depends.