Upload
mark
View
213
Download
0
Embed Size (px)
Citation preview
LAN
DSA
T/N
ASA
Mark Buchanan
When someone makes a ‘prediction’, heor she usually contends that a specificevent will take place at some point in
the future: it will rain tomorrow morning; anearthquake will shake the San Francisco areanext Thursday. This is ‘prediction’ as we usu-ally define it. But in modern science, the ideaof prediction has evolved subtler interpreta-tions, and the concept continues to developtoday, as scientists grapple with phenomenaof ever-greater complexity.
By 700 BC, the Babylonians had devel-oped skill in predicting lunar eclipses; theyunderstood their world better and feared itless, presumably, than earlier peoples did.Over the ensuing 2,500 years, science honedits ability to make accurate predictions,with Galileo, Kepler and finally Newtonbringing the world’s mechanistic pre-dictability to centre stage. In the newtonianUniverse, as Pierre-Simon Laplace pointedout, a being of sufficient intelligence couldpredict the future in detail by knowing thepositions and velocities of all particles atany moment.
Today, we casually use the term ‘predic-tion’ in a slightly more general sense. Onemight predict that a new material will be
superconducting below 40 K, or that a mousethat lacks a certain gene will show a particu-lar trait. Such predictions always precedetheir experimental test, yet aim less to foretellthe future than to explore scientific under-standing. Prediction in this sense is theengine of science: we design the present (theexperiment) and observe the future (theresult) so as to compare our theories toempirical reality.
A more interesting twist to the idea ariseswhen forecasting in the newtonian sense isimpossible. No one can solve the equationsof motion for the 1024 particles in a bucket ofwater, nor would such a solution be useful.For this reason, Boltzmann, Maxwell andother physicists of the nineteenth centuryexecuted a strategic retreat to ‘prediction’ in aweaker sense. One can predict the statisticaldistribution of molecular velocities in thebucket. Foregoing exact knowledge, statisti-cal mechanics also predicts the likelihood ofthe 1024 molecules being in any particularmicroscopic state. Strikingly, acquiescenceto probabilities at this level yields definitepredictions at the higher collective level —we can predict that the water will be solidbelow 0 7C, for example.
More recently, quantum physics hastaken statistical prediction as a fundamentalconcept, and the idea has also entered thestudy of chaotic systems, for which detailedprediction is impossible over long periods.Climate scientists refrain from making pre-cise predictions, instead exploring ‘scenar-ios’, which effectively project a probableincrease in global mean temperature of1.4–5.8 7C by the year 2100. Earth scientistssimilarly find it useful to communicate theirknowledge to the public by predicting theprobability of an earthquake in a given window of space, time and magnitude.Today we are so used to statistical predictionsthat it is difficult to appreciate the profundityof this retreat into probabilities.
Yet a retreat into statistics does not implyan abandonment of the search for meaning-ful regularities. Indeed, this search lies at theheart of recent studies of ‘complex systems’.Such systems frequently involve many inter-acting parts that achieved their current orga-nization through a process of evolution. Wemight think, for example, of a food web, orthe intricate tree-like network of streams andtributaries that make up the drainage basinof a great river. In each case, accidental events— such as the chance extinction of a species— have lasting consequences for the system.As with biological evolution, one mightreplay the ‘tape of history’ many times andnever twice obtain the same result, as theoutcome depends on innumerable fine
details that lie beyond scientific scrutiny. For such systems, one might sensibly
wonder whether there is anything to pre-dict. Yet for river drainage basins, studieshave identified a well-defined fractal struc-ture — in a statistical sense, river networksexhibit a degree of self-similarity, withsmall portions of the network being roughcopies of larger portions. All drainagebasins appear to be alike in this respect,regardless of differences in geography andthe many accidents of history that deter-mined their precise structure. Moreover,this ‘law-like’ feature seems to emerge as theinevitable result of a dynamic process thatminimizes the dissipation of energy.
Given this theoretical understanding, it ispossible to predict the expected fractal char-acter of any newly discovered basin. An ero-sion basin recently identified on the surface ofMars appears to validate this prediction. Forcomplex systems, many macroscopic detailsmay indeed be unpredictable; yet predictablestatistical regularities may lie hidden behindthese details. Recent work on complex net-works underlines this point, illuminatingdeep similarities between the structure of sys-tems as diverse as social networks, the Inter-net and the biochemistry of the cell.
The idea of prediction may have still fur-ther implications for the study of evolution.As the physicist Giorgio Parisi has argued, itmay be that for some systems of biologicalinterest — cells, or even large proteins — itwill never be possible to predict the proper-ties of individual systems. Science mayinstead retreat further, making predictionsthat can be confirmed only by studying thestatistics of many cells or proteins, for exam-ple. Such an approach may seem to under-mine the search for detailed understanding,yet freedom in interpreting the concept of‘prediction’ will continue to broaden thescope and power of science. ■
Mark Buchanan is a science writer based in theUnited Kingdom.
FURTHER READINGRodríguez-Iturbe, I. & Rinaldo, A. Fractal River Basins(Cambridge Univ. Press, 1999).Albert, R. & Barabási, A.-L. Rev. Mod. Phys. 74,47 (2002).Caldarelli, G., De Los Rios, P. & Montuori, M. Publishedonline at http://xxx.arxiv.cornell.edu/abs/cond-mat/0107228Parisi, G. Physica A 263, 557–564 (1999).
A game of chanceconcepts
NATURE | VOL 419 | 24 OCTOBER 2002 | www.nature.com/nature 787
PredictionToday we are so used to statisticalpredictions that it is difficult toappreciate the profundity of thisretreat into probabilities.
Twisting tail: the delta of the Lena River in Russiareveals its intricate and convoluted structure.
© 2002 Nature Publishing Group