6 ideas
Complicated systems, like a jet engine, have many parts but behave predictably, so expertise and planning can master them. Complex systems, like economies, epidemics and organizations, have interacting parts whose small changes produce disproportionate and unforeseeable effects. Forecasting fails when tools built for complicated problems are applied to complex ones.
Predictions extrapolate from past data, so they are least reliable at the discontinuities that matter most. They also make people feel the future is settled, which shrinks the options they consider. Treating forecasts as knowledge surrenders agency that people could otherwise exercise to shape outcomes.
Optimizing for efficiency, as in just-in-time supply chains, strips out the slack, redundancy and spare capacity that absorb shocks. A system tuned perfectly for the expected case becomes brittle when the unexpected arrives. Under uncertainty, a 'just in case' design that looks wasteful is often the rational choice.
Scenario planning builds several plausible, divergent futures and asks what one would do in each, without betting on any single outcome. Its value is mental rehearsal: participants learn to notice early signals and have already thought through responses when conditions shift. Organizations that practice it adapt faster because the unexpected has already been imagined.
CERN runs decades-long, multinational projects whose results cannot be specified in advance. Progress rests on shared purpose, trust and continuous experimentation rather than on predicted deliverables. The case shows how institutions can commit to long time horizons and still stay open to discovering what they did not expect.
Effective containment depended on preparedness: trained people, existing networks, local trust and the ability to improvise. Capacity and relationships built in advance mattered more than precise predictions of the threat.