Predicting human activity
We usually assume that we do things for a reason, whether we are obeying the dictates of the unconscious, rational self-interest or our genetic predisposition. Yet such determinism cannot predict the diverse range of human behaviour. We are left to suspect that our actions may be no more patterned than coin-tossing.
In Bursts, physicist Albert-László Barabási explains how this notion of randomness has been undermined by recent research, including his own. We conduct our affairs in bursts, he says: for example, sending out several e-mails in a short space of time and then none for hours, or pottering around our neighbourhood and then travelling 1,000 miles. Barabási explains that we organize tasks in bursts because we prioritize them, attending to each on a timescale that is appropriate to its urgency.
Even our everyday wrist movements, when monitored with accelerometers, show bursts of motion that are interspersed with periods of repose. As a team at the University of Tokyo has found, the distribution of these bursts differs for people who are clinically depressed, suggesting that such statistics might offer a diagnostic tool.
The book is replete with human stories that animate what might otherwise seem a dry account of statistical patterns. Barabási describes Albert Einstein unwittingly stalling the career of fellow physicist Theodor Kaluza by taking two years to reply to a letter, and the US-based artist Hasan Elahi being questioned by the FBI because of his ‘suspicious’ travel patterns. After his interview, Elahi set up a public website to record his movements as an art project. It revealed that Elahi’s globetrotting was genuinely anomalous. An algorithm developed by Barabási was able to forecast the movements of anonymous individuals by combining information about their locations, as revealed by mobile-phone use, with their personal pattern of ‘bursty’ movements between locations. It was more than 80% accurate for everyone except Elahi, whose movements were unique in that they were truly random.
Barabási’s success in predicting human mobility patterns from mobile-phone data leads to his plausible, if ominous, suggestion that individuals could be constantly tracked using such techniques coupled with widespread surveillance technologies. Yet his assertion that the prediction of most things we do at the individual level “is growing increasingly feasible” is not persuasive. Our predictability, to the extent that our choices and movements form a pattern, relies more on extrapolation of past behaviour — as exploited by web-based ‘recommender systems’ that draw on our purchase or browsing history — than on burst characteristics. Similar to avalanches and earthquakes, bursts have statistical orderliness but remain unpredictable as individual events.
The underlying origin of ‘burstiness’ is unknown. These intense periods of activity are not a by-product of advanced cognition because they apply beyond human behaviour — to the different foraging patterns of animals, the transcriptional activity of genes, and evolutionary speciation, for example. But Barabási cannot say whether their ubiquity stems from the same cause or whether bursts are merely a statistical signature, like power laws or fractals, that many different mechanisms can generate.
Barabási punctuates his exposition with the tale of the 1514 peasant revolt in Hungary, led by his compatriot György Székely. This story supposedly illustrates the difficulty of predicting human affairs, but one could make that point using most episodes in history, and the link to bursts is tenuous. I was happy to indulge him in this digression; others might not, but I encourage them to try, because Bursts reveals Barabási to be both an inventive, interdisciplinary scientist and a talented communicator.