The methods of physics, applied cautiously, can really help.

The flow of ideas from physics to elsewhere has a long and sometimes chequered history. Niels Bohr famously speculated that the uncertainty principle of quantum physics might be useful in psychology, where observation unavoidably influences an individual's behaviour. Some artists drew on the concepts of Einstein's theory of relativity and applied them — with terrific distortion, obviously — to explain the relativity of moral values. Einstein thought there was something 'psychopathological' about such efforts. “The present fashion of applying the axioms of physical science to human life”, he said, “is not only a mistake but has something reprehensible to it.”

Fortunately, things are rather different today. We've lost the naive belief that physical laws might have direct analogues in the human world, yet have gained confidence that the methods of physics, applied cautiously, can really help. Physics-based modelling of financial markets, to take one example, grows more impressive day by day.

In the early 1990s, economist Brian Arthur introduced a model to explore situations in which people have to make decisions, yet cannot do so on the basis of calculation. He imagined a popular restaurant with live music every Thursday evening. Although many people would like to go, if too many go, the place is overcrowded and everyone is miserable. Hence the problem: it's good to go when most other people don't go, and to stay home when they do go. There's no rational way to decide, as the wisdom of a choice depends ultimately on what other people do.

Drawing on earlier psychological studies, Arthur supposed that real people handle such problems in an adaptive way. We keep a handful of 'strategies' in our heads about how past bar attendance might predict future attendance, and make decisions by using whichever has been most accurate in the recent past. Arthur programmed a computer to simulate a group of people facing this problem, and found that as individuals adapted, attendance ended up fluctuating about the crowding threshold.

A number of physicists have since used this model to explore more realistic social systems, including financial markets. A financial market is clearly not a restaurant, yet there are similarities. The wisdom of a decision now, to buy or sell stock, depends not only on the real facts about the company in question, but on the unpredictable decisions of many others. Also, many investors look to the record of past price movements to predict what will happen next.

Work in this area has led to a variety of increasingly sophisticated models, that give realistic results when compared with empirical reality. They reproduce the statistical character of real market fluctuations, for example, and have even been used to predict market movements with some success (for a recent review, see http://arxiv.org/abs/physics/0606107). Intriguingly, researchers have found surprising links to physics. These models show phase transitions in the collective market organization, depending on the number of participating people, and such transitions can be described analytically using the replica trick of spin-glass theory.

This excursion from physics into the social sciences shows real promise — not just hope and hype — for building a fundamental understanding of collective human systems. I don't think Einstein would find anything 'psychopathological' in that.