Readers of Duncan Watts’ Everything Is Obvious know that most attempts to analyze human events are shot through with fallacy and error. This is true of ordinary persons watching the evening news, but also of the professional analysts in CIA and academia. The great experts in world politics have been wrong often enough to put in doubt the whole concept of expertise, as Philip Tetlock’s remarkable studies have shown. Analyzing events is, in principle, problematic. We know much less than we think.
I won’t dwell on the reasons produced by Watts to explain this delusion. In brief, we love to stretch common sense and Newtonian (or billiard-ball) causation beyond the breaking point. When we fail, we take it for granted it was because of insufficient information. This too is a failure of understanding. It’s not that we lack enough information, it’s that no amount of information can ever be enough.
Human events unfold within complex systems governed by weird, nonlinear dynamics. Prediction by means of billiard-ball mechanics is impossible, in principle. Because each complex system develops in unique ways, events are also rarely susceptible to probabilistic analysis. Rightly considered, a question like “Who will win the 2012 presidential elections?” refers to a single token. There have been no previous 2012 presidential elections to average out with this one.
Of course, analysts persist in making predictions. They are addicted to prophecy. Tetlock proved that they guess right about as often as the flip of a coin, but it doesn’t matter. This is what analysts do: who they are. The robe of the magus fits strangely on the scientist’s lab coat, but the point is clear. These are the people who see into the future. Unfortunately, to keep up the pose – to validate their expertise – they must insist that the future resemble the past. They freeze yesterday, and imagine it’s tomorrow. With an election, they point to polls. They say things like, “Since FDR, no president has been re-elected with an unemployment rate of 7.2 percent or higher.” Prediction, explicit or implied, rests on arbitrary statistics and the assumption that nothing in the future will perturb the infinite number of variables pushing and pulling at the data.
Hence the high rate of failure. If we take seriously what Watts and other deep thinkers like Alicia Juarrero have shown, we must parse human affairs by very different methods than this.
An individual person, Juarrero has demonstrated, behaves like a complex system. In a somewhat analogous manner, we can interpret complex events much like we interpret the actions of a good friend. First, we need familiarity with the dynamics of the system. Expertise does count. Second, we assign a character to certain features of the system, based on a combination of factors but giving priority to the narrative by which the system explains and justifies itself.
Narratives capture much of the human intention in the event: the fact that this particular system contains individual agents trying to impose their will. Assigning a character is thus an act of imaginative interpretation. I can say “John is a devout Christian, he won’t come with me to watch the X-rated movie” or “The Cuban regime holds the US responsible for all political opposition, so protesters will be treated like traitors.” On the surface, such statements resemble mechanical predictions, but they are in fact accounts of intention. They recapitulate the logic of the system, against which objective factors, like actual behavior, must be measured.
In fact, John may be false to his principles and watch the X-rated movie. The Cuban regime, from weakness or calculation, may ignore or tolerate opposition. The analyst must abandon the pretense that the future is a tableau vivant reenactment of the past, and embark on an exploration of the dynamics tugging at an event.
All complex systems are inherently unstable and tend toward disequilibrium. Small perturbations can lead to great turbulence; sustained turbulence can result in phase change, a large-scale reconfiguration of the system and its dynamics. The analyst’s job should be to call out potential drivers of disequilibrium – that is, of change – at every stage.
Perturbations are objective shocks which the system must account for. A perturbation can be a “newsworthy” event like a military defeat and a sharp economic downturn, or a hidden development like a demographic imbalance and the slow decline of wealth. Because the flow of causation in complex systems is nonlinear, a seemingly insignificant blip on the radar can be magnified into a transformative force. Five years ago, a few dozen opposition bloggers in Hosni Mubarak’s Egypt appeared to be little more than political background noise. By January of this year, Facebook groups with hundreds of thousands of members coordinated the mass protests which overthrew Mubarak. The rise of Egypt’s online opposition was not predictable, yet in hindsight assumed greater significance than many more attention-grabbing geopolitical developments. This suggests that the analyst, instead of pursuing the obvious, must become a scout of the improbable, exploring change on the margins of the system.
Turbulence is perturbation gone pandemic and approaching disequilibrium. As with all things complex, the causes are mysterious. Some factors are objective, but among the most decisive, I suspect, is the perceived inadequacy of a master narrative to account for perturbing events. Whether a regime can explain and justify its actions in the face of events, more than the events themselves, will determine the extent of the impact. The horrendous March 2011 earthquake in Japan, for example, caused minor political aftershocks; while the less severe (though still devastating) 1972 earthquake in Nicaragua, because it reinforced popular perceptions of the Somoza regime’s callousness, helped drive that political system past turbulence into phase change. Analytically, the fate of the ruling narrative looms even larger than usual at this stage.
On rare occasions, turbulence crosses the line to disequilibrium, ending in phase change. The dynamics of the system are radically reconfigured. An old regime dies. A new one arises on its ashes. This process is beyond mysterious: it’s inexplicable. The classic example is the collapse of the Soviet Union in 1991, which caught most sovietologists by surprise – and the causes of which, proximate and ultimate, are still hotly debated.
Such disputes, Watts would argue, are sterile. Phase change shows complex systems at their most nonlinear: for all we know, the beating of a butterfly’s wing in Siberia caused the fall of the Soviet empire. The analyst should turn the page and focus on the objective and narrative elements – including accounts of the transformation – competing for primacy in the still-turbulent environment. His aim must be to gain sufficient familiarity with the emerging dynamics of the new system that he can begin to assign a character to some of them.
To sum up: analysis of events, as conducted by professionals in places like CIA and the news media, has been theoretically confused and, on the evidence, a failure in practice. If we accept the world described by Watts, a new model of analysis must replace the old: one that is at once more modest and more adventurous. Modesty pertains to prediction and probability. We should give up the illusion that human events are like the orbit of Halley’s comet, and accept them as complex, historical, and brimming with group and individual intentions: understandable, if at all, from within their own internal logic, their narratives of themselves, their character. Adventure pertains to the nature of complex systems, which force the analyst to abandon tableaux vivant prophetic productions and become a rider on the open range of improbability, tracking the sources of change.
Implicit in this new model of analysis is a narrative form of communication. Quantitative data must fit within a qualitative interpretation. The analyst can explain an event only by imaginatively recapitulating its dynamics: the eccentricities of the system, the “strange attractors,” constitute the plot of the tale.
This raises an interesting problem. If the arguments put forward by Christian Smith are correct – and I am convinced they are – shared narratives must reflect the “moral order” embedded in every culture, institution, group, and person. Nothing human is value free. For the analyst of events, this means that familiarity with a system will entangle him in the moral assumptions of his subject matter about what is good, true, important, etc.
Of course, the analyst’s own insights and narratives will be entangled in a different set of assumptions. Analyst and event, explanation and the thing explained – both are beset and constrained by value judgments intrinsic to every human system. The ideal of an Olympian “objectivity,” we must finally agree, is puffery and self-delusion. The typically Western belief that one can stand, like God, above events, is false and thus destructive of sound analysis. The only way out of the dilemma requires large doses of honesty and humility. To the extent moral and ideological constraints are made manifest in his description of events, the analyst will approach the only possible ideal of intellectual integrity.