I have a review in Nature of Albert-László Barabási’s new book Bursts. The book is nice, but the review was necessarily truncated, and here is how I really wanted to put it.
Bursts: The Hidden Pattern Behind Everything We Do
Dutton, New York, 2010
Is human behaviour deterministic or random? Psychoanalysts, economists and behavioural geneticists, however unlikely as bedfellows, all tend to assume cause and effect: we do what we do for a discernible reason, whether obeying the dictates of the unconscious, rational self-interest or our genetic predisposition. But those assumptions have not produced anything like a predictive model of human actions, and we are daily presented with reason to suspect that our actions owe more to sheer caprice than to any formula. Given the disparity of individual decisions, perhaps our behaviour shows no more pattern than coin-tossing: maybe collectively it is dominated by the randomness encoded by the gaussian distribution, the familiar bell-curve statistics of a series of independent events whose outcomes are a matter of chance.
Albert-László Barabási’s Bursts explains how this notion of randomness has been undermined by recent research, much of it conducted by him and his collaborators, that has revealed a hitherto unexpected pattern in human activities ranging from the sending of emails (and before that, postal letters) to our movements through the world. We conduct our affairs in bursts, for example sending out several emails in a short space of time and then none for hours. Even our everyday wrist movements, when monitored with accelerometers, show this bursty behaviour, with spells of motion interspersed with periods of repose. Because the distribution of bursts differs for people who are clinically depressed, these seemingly irrelevant statistics might offer a simple diagnostic tool.
Burstiness could seem so intuitively obvious as to be trivial. That we find a moment to catch up with email responses, rather than attending to them one by one at random intervals, is scarcely puzzling or surprising. But such rationalizing narratives don’t fully account for everything: why, then, does it take us a few minutes to respond to some messages but weeks to get to others? Barabási and his coworkers explained that on the assumption that we prioritize, adding new priorities to our ‘must-do’ lists each time others are cleared.
Barabási renders observations like this, which could seem dry or frivolous, both engaging and illuminating through human stories: Einstein unwittingly stalling the career of Theodor Kaluza by taking two years to reply to a letter, or the hapless artist Hasan Elahi being taken for questioning by US Homeland Security because of his ‘suspicious movement’. Barabási shows that Elahi’s globetrotting really was anomalous – whereas the algorithm he developed in his lab to predict people’s whereabouts based on their personal bursty signature forecast everyone else’s movements with more than 80 percent accuracy, Elahi foiled the program with his genuine randomness.
Burstiness is not confined to human activity, and so is not somehow a by-product of cognition. It is seen in the foraging patterns of several animals (though not, as once claimed in this journal, albatrosses). It even fits the transcriptional activity of genes and evolutionary speciation. But Barabási cannot yet say whether this ubiquity stems from the same basic cause, or whether burstiness happens to be a statistical signature that many mechanisms can generate. The same question has been raised of the power-law statistics found for many natural and social phenomena (in fact bursts also produce power laws), and of fractal structures. To put it another way, is burstiness a discriminating and informative character, or just a common epiphenomenon of several distinct processes? We don’t yet know.
Moreover, the burstiness of human behaviour doesn’t obviously warrant the air of determinism that hangs over the book. ‘Prediction at the individual level is growing increasingly feasible’, Barabási asserts. But bursts per se don’t obviously help with the sort of detailed, moment-by-moment prediction he is discussing here – like the avalanches of self-organized criticality, they remain unpredictable as individual events, differing from gaussian randomness only because they are correlated. They simply help us get the overall statistics right.
While popular science books written by researchers presenting new ideas typically have an ex cathedra quality, Bursts shows the influence of the journalistic approach of professional writers, exemplified by James Gleick and Malcolm Gladwell, narrative-driven and replete with personality sketches. Barabási is rather good at these story-telling tricks, and his opening paragraph is a masterful example of the genre, drawing us in with a puzzle we know will be resolved only much later.
Whether his daring device of punctuating the exposition with the tale of how his Transylvanian compatriot György Székely led a peasant revolt in Hungary in 1514 works is less clear. Barabási implies that this tale illustrates some of the conclusions about burstiness and unpredictability, but that’s far from obvious. Because I am apparently Barabási’s personal Hasan Elahi, a vanishingly rare outlier who happens to have an interest both in Székely Transylvania and the peasant uprisings of the early sixteenth century, I was happy to indulge him. I suspect not everyone will do so. But they should try, because Bursts reveals Barabási to be not just an inventive and profoundly interdisciplinary scientist but an unusually talented communicator.