There may well be laws governing the occurrence of upheavals in human affairs; but they are unlikely to tell us when they will occur

Winston Churchill said that a young person entering politics needs “the ability to foretell what is going to happen, tomorrow, next week, next month and next year. And to have the ability afterwards to explain why it didn’t.” Alas, only the last part can be done.

Ten or a dozen years ago many informed observers would have agreed on most of the following. Inflation is endemic to western economies; Labour is unelectable; the UK will integrate more closely with the European Union; full employment will never again be achieved; the US is a sclerotic economy but Japan has the secret of rapid growth; British beef is the best in the world and organic farming will remain the preserve of a minority of cranks. This list is taken from a paper by an “expert on social trends“, Bob Tyrrell, just published by the Centre for Policy Studies (Things Can Only Get...Different, £7.50). It must however be said that Mr Tyrrell’s own scenarios for the years ahead do not look any more convincing than the ones he is mocking.

The more interesting question is why it is so difficult to predict, and what we can nevertheless say in the absence in of a crystal ball. Karl Popper, the philosopher of science, had a knock-out argument against long range historical prophecy. This is that the course of human affairs is affected by new knowledge, which is by its nature unforeseeable.

It is, however, unlikely that new knowledge will determine whether or not there is a recession in the US this year. Another Popper argument is relevant here concerning a misconception about scientific method. Some of the physical sciences have predictive power. But such predictions are conditional. They assert that certain changes, such as an increase to a certain points in the temperature of water in a kettle will, granted certain other conditions - for example a given atmospheric pressure - lead to boiling. But they cannot tell us whether the required conditions will be fulfilled or not.

The historical prophecies that practical men demand are unconditional scientific predictions. They can be derived from valid scientific theories, if and only if, they can be combined with the correct assertion that the required conditions are in fact fulfilled. These requirements can only be fulfilled only for systems that are “well isolated, stationary and recurrent.“ This happens to be approximately true of the solar system, which is why predictions of events such as eclipses of the sun are possible many years ahead. But contrary to popular belief such systems are not typical even of the physical world, let alone the rapidly changing society of human beings.

The problems of forecasting are rooted in the complexity of the phenomena under examination. Popper’s friend, Friedrich Hayek was virtually read out of economics for doubting whether it would be possible to discover simple, quantitative regularities between economic variables - which is what mainstream macroeconomists have been trying to do in the last half century.

What then is there left to discover? Hayek’s view was that the economist, like the biologist, was concerned with rules and patterns. A student of evolution can say something about the conditions in which new species evolve; but he cannot predict what these will be or when they will occur. Economic theory can tell us that it is impossible to maintain a fixed rate of exchange and at the same time a national inflation objective. Nevertheless we cannot predict when the exchange rate will give way or where it will go when the chosen parity collapses.

Thanks to recent developments in mathematical physics it is possible to say more about such patterns, both in nature and in human society, than Hayek was able to do. A fascinating guide has been provided by a book entitled Ubiquity by Mark Buchanan, just published by Wiedenfeld and Nicolson. This book is not free of some of the faults of popular science books. The short table of contents consists of the usual cute headings which give little real guide to the structure of the book. A typical first chapter begins “Hell Creek slips quietly out of one end of the Fort Peck reservoir in easternmost Montana and from there cuts a lonely meandering path into the hills”. Nevertheless it is not too difficult to find ones’s way into the arguments; and at least the book lacks the “gee whizz” and “wow” element so common to this genre.

We have already heard something of chaos and catastrophe theories. Buchanan discusses a third "c", complexity. He uses the analogy of a sand pile. If it is sufficiently steep a few extra grains can cause an upheaval; but without knowing the position of every grain it is quite impossible to say where or when the upheaval will occur. The sand pile is however “in a critical state“ - that is one vulnerable to disturbance - which he believes typical of much of social life as well, for instance, as the earth’s crust in regions subject to earthquake.

Nevertheless there is a “power law“ that can be given for earthquakes and similar phenomena and that seems to fit many human events. This says that the frequency of disturbances is inversely proportional to their size. Crudely: the bigger the rarer.

Take a completely different event: the extinction of dinosaurs 65m years ago. The conventional wisdom is that this was due to a giant meteor hitting the Yucatan peninsula of Mexico. An alternative explanation is that such geological cataclysm are simply rare events liable to happen without any obvious “cause“;and indeed the cataclysm of 65m years ago was only the latest, and not the biggest, of five such upheavals in the last 300m years.

Consider a single grain of sand falling on the western portion of a sand pile triggering a series of upheavals in the whole pile. Should the western authorities have been able to remove some of the sand from the initial spot and thus reduce its vulnerability? Alas no. To have foreseen the disaster the authorities “the would have needed near perfect knowledge of the position of grains over the entire pile coupled with unlimited computing power to work out the consequences of a grain falling.“ Thus there is no way of knowing beforehand which grains should be removed and to where.

The simplest power law is based on the power 2. Earthquake B with double the force of Earthquake A is four times less likely to happen. For every city such as Atlanta with a population of 4m, there are four cities having populations half that size e.g. Cincinnati; and for every Cincinnati there are four cities half as large again. Thus there is no typical size for a city in the US or elsewhere and no reason to find any special cause behind the emergence of the very biggest.

A different example relates to wealth. Find out how many people in the US have a net worth of a billion dollars. You will then find that four times as many have a net worth of about half a billion, and so on.

The inverse square law is the simplest of the power laws. There are others. Wars have been studied in relation to the deaths involved. The power law at work here is slightly more complicated. Each time you double the number of deaths wars become - not four times but 2.62 times - less likely. A similar magnitude also occurs in the power law for the frequency for forest fires of a specified size.

There are no deterministic laws of history relating to the causes of the war: only the power laws which suggests that big wars are, fortunately less frequent than small ones. Wars and revolutions “don’t come in simple cycles, and they don’t telegraph their arrival in advance” any more than earthquakes do. Even the rough magnitude of the next stock market shift is unforeseeable. Great stock market crashes are simply ordinary events which occur infrequently.

Buchanan has the modesty to admit that we are only at the beginning of research. There are numerous problems of measurement. Even if we agreed how to measure a recession, would a boom be measured simply as an upward deviation from trend? Or would it require a new power law? And would prices as well as quantities come into the picture. This is apart from the need for further reflection on why such power laws apply.

Are there areas not not in a critical state where we can plan ahead more confidently? I suspect there are at least two in macroeconomics. There is the very short term outlook for say six to 12 months ahead, where something can be said by the economic statistician on the basis of typical lags between events that have already happened, such as oil price increases or wage explosions and their path through to rest of the economy.

There are also trends which tend to reassert themselves over five to ten years or more. For instance British economic officials have found that almost whatever the period they look at in over a century the annual trend rate of growth has been 2¼ per cent. Even this trend is only probable. It is likely that there will be rare upheavals leading to decades of above and below trend growth, the reasons for which will be explained after the event. Missing from the list, unfortunately, are the short to medium term forecasts, covering, say, 18 months to five years, which are so much desired by central bankers for monetary policy or governments planning their spending over a parliament.

The serious study of complexity is in its infancy. My intention is only to suggest that we can retain the hope of applying scientific method to the study of human affairs without expecting ever to find that magic crystal ball for which there is such insistent demand.