10. Build the Radar
Why prediction fails
In a complicated system, careful analysis can find cause and effect. In a complex one, as Dave Snowden puts it, "the relationship between cause and effect is coherent only in retrospect" — you can explain it after the fact, but you cannot extrapolate it forward.1 That single observation invalidates most of what passes for strategic forecasting. Trend extrapolation and time-series models rest on assumptions a changequake breaks: that current trends continue (continuity), that variables move independently, that systems settle toward equilibrium, that change is gradual. Real systems sit far from equilibrium and change in punctuated jumps.2 They get there through emergence and feedback: small effects compound, a threshold tips, and a cascade runs faster than anyone modelled — which is precisely why the curve looks gentle until it doesn't. The design response is counter-intuitive to anyone trained on efficiency. A system tuned to be optimal for today's conditions is brittle against tomorrow's; one built to be adaptive — carrying redundancy and diversity, pushing decisions to where the information is — survives the jump. In a complex world redundancy is not waste; it is the price of staying upright when the regime shifts.