Whatever else this teaches us, it may well be that the major message is that failures that seem unlikely are not, in fact, inconceivable, except in the deeply wrong sense that Vizzini uses it. In a warming world, for example, it becomes even more urgent to acknowledge that outer limits may be crossed, (I do not, of course, mean to imply any connection between earthquakes and global warming, merely noting the dramatic increase in natural disasters that is attributable to climate change) and that betting against the probability of events is a risky proposition.

At the same time, the costs of building levees to withstand category 5 hurricanes (and no, Katrina was not a category 5 hurricane) and deepwater drilling platforms with automatic shutoffs, and nuclear plants in safe zones to withstand higher earthquakes are enormous - at a time when the US alone has 3 trillion dollars worth of delayed infrastructure work that it has been putting off on everything from bridges to sewers to nuclear power plants to the electrical grid to everything else. That 3 trillion does not include the costs of bringing things up to code for more extreme events. We seem to be reaching a point where Joseph Tainter's observations on the diminishing returns of complexity become strikingly obvious.

My own argument, which I've been making for some years, and which has come to have some little currency in at least some areas of agricultural planning, is that we should turn it around and presume failure. That is, we should ask ourselves "what strategies are most effective and least risky in failure situations...given that systems failures happen all the time." In this model, distributed systems are less dangerous than centralized ones, and those that give partial return even if projects aren't completed or if models break down are more valuable than those that give more but only if we front-load a huge investment to them. It creates, in the end a different way of looking at our world, and one, I would argue, we desperately need.