I don't take any credit for the title of this blog. It was used by Kurt Vonnegut to describe the sea freezing solid in Cat's Cradle, which was used by Philip Ball in Critical Mass to illustrate his argument for sudden shifts in mass behaviour, which in turn came to mind when I was reading Paul Ormerod's enlightening RSA pamphlet about the implications of networks for public policy. Vonnegut to Ormerod via Ball and me – that's an example of networks in action!
Paul Ormerod, an economist, uses behavioural economics and network theory; Philip Ball, a science writer, uses what he calls the 'physics of society'. They come at the same phenomenon – the fact that small nudges can have slight or large effects depending on the conditions – from totally different angles. Can they both be right?
Ball's idea – one of many in his engaging and thought-provoking book – is that large populations display the same characteristics as materials in their solid, liquid or gaseous states. He describes 'phase transitions', which is what happens when a solid melts into a liquid, or a gas condenses into a liquid, and so on. These transitions are abrupt, and require only a tiny 'nudge' to set them off if the conditions are right. For example, under normal conditions water freezes at exactly 0 degrees – it only needs a tiny change from 0.1 to -0.1 for the transition to occur. But phase transitions are also almost always predictable – the change from water to ice will almost always occur at that critical temperature point.
The interesting thing when thinking about Paul Ormerod's work is that occasionally, the critical point can be passed without the transition occurring. The material is then 'metastable', and will revert to its stable state if it is disturbed. Ice, for example, has to form around some particle in the water; if water is absolutely pure, it can be 'supercooled' to below 0 degrees without freezing, but as soon as a particle is introduced it freezes immediately. This makes the phase transition unpredictable. It also means that a system's behaviour is determined by its history – if it was water to start with, it stays as water for longer than it should.
Ball uses this theory to explain why phantom traffic jams appear on motorways for no apparent reason. Everyone can be driving along smoothly, but without knowing it in a metastable state – faster than the density of traffic would normally allow. This is fine, until the system is disturbed, perhaps by someone braking unexpectedly. The system then abruptly and unpredictably reverts to its stable state of congestion. But because the system is determined by its history, it then stays congested for longer than it should, creating apparently causeless traffic jams.
Ball gives other examples of unpredictable social behaviour that can be explained in this way, such as sudden changes in crime levels. So I wondered whether Paul Ormerod's English football fans in Sardinia might fit his model as well. Ormerod explains the unpredictable change in mass behaviour from order to anarchy in network terms: it was only sparked when a particularly influential crowd member was nudged in the right way. But perhaps it was Ball's kind of unpredictability at work instead: the crowd was under pressure and had entered a metastable state; people ought to have been rioting for some time, but because they started out placid they remained so until the system was disturbed by the police officer's gunshot. Ormerod's unpredictability is factor of targeting; exactly when the nudge is applied is less important than who is nudged. For Ball, it is a factor of timing: anyone can be nudged to set off the transition, but only if the pressures on the whole system are such that it has entered a metastable state.
If this is true for traffic, crime and football fans, what else? Could it be that public engagement with the Big Society, for example, will suddenly shoot up because it becomes more and more credible and attractive but is resisted for 'too long' because of its uncertain start? Or could smoking rates suddenly go through the floor because interventions make quitting more and more attractive, but ingrained reasons to stay a smoker mean prevalence stays at the current level for 'too long'? In both cases, it would only take an individual or small group to be nudged to move to the stable state for the rest of the population to follow suit abruptly. Who knows – it's hard to predict. But just as with Paul Ormerod's network argument, Ball's theory makes it difficult to know what will work spectacularly well, and what will flop for no apparent reason.
You might not agree with all of Ball’s arguments, and some of his points do seem to stretch the evidence, but his ideas about the behaviour of crowds and populations make sense to me. So there you go. The Grand Ah-Whoom – something else for policy makers to worry about?