There's a parallel between the riots and the stock market sell-off. Both are examples of how social behaviour can shift from stability to instability.
There's always some doom-monger trader who's keen to sell. Usually, though, his selling readily finds a buyer, so prices don't move much.
Analogously, there's always some nutter who wants to rob and trash places. But normally, his friends don't go along with him, so he suppresses his tendencies.
In other words, when behaviour is uncorrelated - when sellers find buyers or when the lawless are surrounded by the lawful - we get stability. What we're seeing now, though, is a higher correlation of behaviour. When sellers have tried to sell, they've run into other sellers - so prices have tanked low enough to lure out the buyers. And the lawless have found themselves accompanied by other criminals, so their behaviour has been encouraged rather than suppressed.
This is what Left Outside describes as a shift in equilibrium, from law-abiding to criminal behaviour and from stable-ish prices to unstable ones.
But what explains this? A big part of the story must be information cascades. These happen when people base their behaviour not upon their private information, but rather upon what others are doing. So traders think 'others are selling, so this is no time to go long'. And some youth think 'hey, those guys are getting away with a good ruck, I'll pile in too.' As Ian says, rioting is imitative behaviour.
This sort of story is a complement to attempts to explain the riots by reference to anger, deprivation, powerless or alienation. Such explanations run into a problem - that people have been angry, poor, powerless and alienated for years. So why are rioting now when they weren't for a long time? The answer is that an information cascade has led them to do so, whereas in the past such a cascade was absent. In similar fashion, financial markets can shift from having low volatility to high volatility because a cascade arises.
This story also contradicts a claim implicit in something Will says. He suggests that nudges and choice architectures are alternatives to neoclassical rational choice theory. In this story, though, they are complements. For the poor and desperate, rational choice gives (some of them) a disposition towards violence and looting; they have nothing to lose but their chains and a pair of trainers to gain. But whether this disposition leads to actual criminality depends upon nudges. And just recently, they have been nudged by their peers into thinking that they can actually get away with making trouble. So they do so.
Illuminating as the theory of information cascades can be, there is a problem with it. We cannot forecast when such cascades will emerge. We can only identify them in hindsight. They allow us to explain behaviour, but not predict it.
And perhaps this will always remain the case. I say so because of a recent paper by Klaus Adam and Albert Marcet, who show how only a very small information cascade can generate significant asset price volatility. This suggests that small changes in the extent to which we use public signals versus private information might lead to quite large changes in behaviour. Analogously, it implies that only small cascades can make the difference between peaceable behaviour and lawlessness.
Perhaps, then, social orders are more brittle than we like to think.