Posted by: michaeldaybath | July 15, 2010

Bursts

Albert-László Barabási, Bursts: The hidden pattern behind everything we do (New York: Dutton, 2010). 310 pp. ISBN 978-0-525-95160-5

I have been reading a recent book on human dynamics by Albert-László Barabási of Northeastern University. Bursts is an attempt to produce a popular account of recent developments in network science, tracing its relevance to patterns of human behaviour.

Barabási interleaves his account of the science with an imaginative account of the Hungarian Peasant Uprising of 1514 led by György Dózsa [1]. The relevance of this to network science is not immediately clear, but the book attempts to contrast the supposed intuition of István Telegdi – who (at least according to the chronicles) accurately predicted the unintended consequences of recruiting thousands of peasants to the 1514 crusade – with the philosopher Karl Popper’s critique of historicism, the opposition to the once popular idea that history progresses on predictable (or deterministic) lines that can be derived from fundamental laws.

The rest of the book attempts to demonstrate that at least some human behaviour can be accurately predicted. Barabási  mines the history of science and society for examples of predictive models that apparently fail to deliver (e.g., long-term meteorology) and a few case studies from network science where it has been demonstrated that human behavior (e.g. travel or Website visits) does not obviously conform to well-known physical laws like Einstein’s theory of diffusion or Rutherford’s exponential model. However, there was one statistical pattern that seemed to reoccur regularly in the data analysis undertaken by Barabási and his colleagues. They were able to characterise both of these  human activities (and many others) as “bursty,” repeated statistical analysis demonstrating power law distribution patterns. The same power law distributions were found in cellphone calls, library loans, printer jobs and much else  (p. 105).

No matter what human activity we examined, the same bursty pattern greeted us: long periods of rest followed by short periods of intense activity … Bursts were everywhere in nature, from the edits individuals published at Wikipedia.org, to the trades made by currency brokers; from the sleeping patterns of humans and animals, to the tiny moves the juggler makes to keep his sticks in the air.

The same kinds of behaviour can be identified in many other different contexts, even in the fossil record. Barabási argues that this indicates that “burstiness is not something we invented but was in use well before intelligent life ever emerged on earth” (p. 240). The point about “bursty” behaviour is that it is not random. Paradoxically, “burstiness” enables predictions to be made about future  likely behaviour. These are the kinds of patterns that can be analysed to help forecast things like shopping preferences, creditworthiness, and even national security risks.

There are a few things of interest in this book for those interested in the longer-term curation of research data. Like other popular scientific books, Bursts contains several examples of where data collected for one purpose has been re-used to explore new research questions. Indeed, this is very much a feature of the emerging science of networks [2]. Much of this data comes from the commercial sector – e.g. records from cell phone companies or Web search portals – but there are some good examples in the book of the re-use of data from other research disciplines.  Sometimes this has lead to problems when the exact nature of the source data has not been understood properly. One interesting example in this book is the re-use of ecological tracking data from albatross (Diomedea exulans) in the South Atlantic. This was used in the mid-1990s to hypothetise that foraging albatross follow Lévy flight patterns, a mathematical concept characterised by Mark Buchanan as  “many-legged journeys in which most of the legs are short, but a few are much longer … found in some sorts of diffusion, in fluid turbulence, even in astrophysics” [3]. Further analysis of albatross movement data in 2004, however, revealed that the Lévy flight patterns were most likely an artifact of the data collection process, suggesting that the longer “journeys” actually represented time that the birds spent sitting on nests. More recent work has reasserted that foraging animals of all kinds do most likely follow “bursty” Lévy flight patterns, but the lesson about the need to be careful when reusing data generated by others remains a good one.

This is a very interesting and persuasive book. For me, the weakest parts are where Barabási tries to undermine Popper’s critique of historicism by referring to Telegdi’s accurate predictions about the consequences of the 1514 crusade. Even if we ignore any potential problems with chronology or bias in the historical sources about the Hungarian Peasant Uprising, it does not seem to me that any example of accurate  foresight from history  could invalidate Popper’s wider philosophical points about the wisdom of using “scientific” methods for determining the future. As Frederic Raphael puts it [4]:

It requires no experiment to conclude that what has already happened could not have happened. The past is not, however, a test-bed for the future. No historian’s hindsight, however shrewd in its observations, cannot [sic?] generate sufficient kudos to warrant putting our faith in his foresight […] Unfortunately or not, no conclusions can be drawn about the predictability of the future from the fact that historians have uncovered the causes of what happened in the past.

In his attack on historicism, Popper was primarily arguing against those – like Social Darwinists or Marxists – that claim that there are deterministic patterns to history, i.e. that so called “scientific” laws determine what will happen in the future. It is no accident that he wrote The Poverty of Historicism in the middle of a century where this kind of thinking seemed to lead inexorably to totalitarianism. The point is not that human behaviour is random, but that the ultimate outcomes of even predictable behaviour will be very difficult to foretell. As Popper himself wrote [5]:

… we might say that the human factor is the ultimately uncertain and wayward element in social life and in all social institutions. Indeed this is the element which ultimately cannot be completely controlled by institutions (as Spinoza first saw); for every attempt at controlling it completely must lead to tyranny; which means, to the omnipotence of the human factor – the whims of a few men, or even one.

In the end, Telegdi’s prophesy may have turned out to have been completely mistaken. Even as Barabási tells the story in Bursts, there were many points at which the 1514 crusade could have had a different outcome.

Barabási may be correct to forecast that predictive tools will continue to improve, and that they will “move from focusing on individuals to focusing on the groups to which they belong to” (p. 257). As he says elsewhere, however,  predictions like those made by Telegdi are essentially based on informed opinion – “if we are wrong, it is forgotten; if we are right, it does not make us a prophet” (p. 255). To me, that still sounds just about right.

References

[1] Norman Housley, “Crusading as social revolt: The Hungarian Peasant Uprising of 1514,” Journal of Ecclesiastical History 49 (1998): 1-28. doi:10.1017/S0022046997005605

[2] M. E. J. Newman, Networks: An introduction (Oxford: Oxford University Press,2010). ISBN 978-0-19-920665-0

[3] Mark Buchanan, “Ecological modelling: The mathematical mirror to animal nature,” Nature 453 (2008): 714-716. doi:10.1038/453714a

[4] Frederic Raphael, Popper (New York: Routledge, 1999), p. 32.

[5] Karl Popper, The Poverty of Historicism (London: Routledge, 2002), pp. 146-147.

Book cover for "Bursts" (2010)

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Categories

%d bloggers like this: