My discussion of failure in the context of the Iraq discussion is part of my broader beliefs about innovation. I saw a great speech a little while back by the guy who's in charge of designing new products at Palm. He talked about an excercise that he does with various groups, where he gives them pieces of spaghetti and some tape and tells them to build the tallest structure they can.

Engineers do all right; MBAs do the worst, because they waste time arguing about who will be in charge. But the best performing group? Kindergarteners. Little kids don't try to design a structure. They just keep trying things, and stick with anything that works. Their structures certainly didn't look as elegant as the neat frames designed by the engineers. But they did the job, which is to be as tall as possible.

To succeed quickly, he said, what you want to do is fail. A lot. Failing eliminates wrong answers faster than any possible analysis. I was reminded of the famous Thomas Edison quote: asked how it felt to have failed to invent an electric lightbulb, Edison said "I haven't failed! I've discovered 10,000 filaments that don't work."

That, in my opinion, is the secret strength of the American economy. We don't try to plan anything. We just see what works, and get rid of what doesn't. We make it easy to try, and hard to keep going once you've failed. Once people have failed, we make it easy to try again--even after our "draconian" 2005 reform, our bankruptcy policy remains the most generous in the world towards both individuals and corporations. But we do not keep moribund industries limping along in the hopes that things will change.

There is a credible argument that the reason Japan's recession lasted ten years is that neither the government nor the banks could let enterprises fail. They carried bad businesses and their loans for years rather than write off the loans. Those ailing firms sucked up all the capital that should have been used to rebuild the economy.

I think this determines how we should handle the financial meltdown. In many sorts of repeat games, "trust, but punish" is the value maximizing strategy for all participants. That's my approach to financial regulation. We should aim to provide maximum transparency, and keep the moral hazard problem down by establishing capital requirements that reduce the government's downside risk. But we should be very suspicious of prospective regulations that aim to prevent failure; the best way to do that is usually to close off potential avenues of success.

As I support easy bankruptcy, I also support having the government be insurer of last resort; letting those who have failed drown in their failure may be morally satisfying, but it is rarely good for society. When we shun failure, we lose all the information they learned.

That doesn't mean that I am in favor of bailing out bankers with taxpayer money. Failure should hurt; that's the "punish" part. Bankruptcy should be unpleasant and destroy your credit; running a company into the ground should be an anguishing experience for bankers.

How, exactly, we should punish them I am still not sure. A lot of commentators I see are underestimating how much Bear's managers are suffering. No matter what your level of wealth, losing your job and being publicly branded a failure isn't pleasant, even if you have millions of dollars. Still, I'd like to see policies aimed more explicitly at making it hurt. A good place to start might be to make the assets of company officers part of the bankruptcy pool--but I haven't thought that through, and I'm sure that there are consequences to doing so that I am not anticipating.

Whatever the punishment, it should be severe but brief. Hit them once, hard, and then get on with life.