Rule 1: Think of the economy as being more like a cat than a washing machine.
We are victims of the post-Enlightenment view that the world functions like a sophisticated machine, to be understood like a textbook engineering problem and run by wonks. In other words, like a home appliance, not like the human body. If this were so, our institutions would have no self-healing properties and would need someone to run and micromanage them, to protect their safety, because they cannot survive on their own.
By contrast, natural or organic systems are antifragile: They need some dose of disorder in order to develop. Deprive your bones of stress and they become brittle. This denial of the antifragility of living or complex systems is the costliest mistake that we have made in modern times. Stifling natural fluctuations masks real problems, causing the explosions to be both delayed and more intense when they do take place. As with the flammable material accumulating on the forest floor in the absence of forest fires, problems hide in the absence of stressors, and the resulting cumulative harm can take on tragic proportions.
And yet our economic policy makers have often aimed for maximum stability, even for eradicating the business cycle. "No more boom and bust," as voiced by the U.K. Labor leader Gordon Brown, was the policy pursued by Alan Greenspan in order to "smooth" things out, thus micromanaging us into the current chaos. Mr. Greenspan kept trying to iron out economic fluctuations by injecting cheap money into the system, which eventually led to monstrous hidden leverage and real-estate bubbles. On this front there is now at least a glimmer of hope, in the U.K. rather than the U.S., alas: Mervyn King, governor of the Bank of England, has advocated the idea that central banks should intervene only when an economy is truly sick and should otherwise defer action.
Promoting antifragility doesn't mean that government institutions should avoid intervention altogether. In fact, a key problem with overzealous intervention is that, by depleting resources, it often results in a failure to intervene in more urgent situations, like natural disasters. So in complex systems, we should limit government (and other) interventions to important matters: The state should be there for emergency-room surgery, not nanny-style maintenance and overmedication of the patient—and it should get better at the former.
In social policy, when we provide a safety net, it should be designed to help people take more entrepreneurial risks, not to turn them into dependents. This doesn't mean that we should be callous to the underprivileged. In the long run, bailing out people is less harmful to the system than bailing out firms; we should have policies now that minimize the possibility of being forced to bail out firms in the future, with the moral hazard this entails.