The business of science, it might be said, is to distinguish what is knowable from what is not knowable. The first great flowering of modern scientific thinking, in the days of Newton, Leibniz and Descartes, established a revolutionary perspective of certainty and predictability on a world previously dominated by a largely religious or superstitious belief in nature’s untameable randomness. Instead of being at the mercy of fate, humanity through science could be its master. Everything, in theory, was knowable. If the position, mass, velocity and direction of every particle could be known, so it was thought, then in principle the entire future of the cosmos could be extrapolated from this knowledge.
This faith in the power of science to unlock any secret seems touchingly naïve today, after the cold showers of quantum physics and chaos theory. But the war continues, between the certainty and uncertainty principles, between what science can do and what it can’t. And inevitably, with possibly the biggest financial crash since the 1930’s on the world’s doorstep, some scientists are looking at the economy and asking the same big questions.
Do financial booms and busts have causes, and are those causes identifiable, and more crucially, predictable? Or is the economy essentially a chaos system, whose workings a computer the size of Jupiter could still not reliably forecast?
Sumit Paul-Choudhury argues (New Scientist, 21 June) that financial bubbles are not only unpredictable and unstoppable, but even useful and desirable. According to this theory, bubbles generate an enormous incentive to take reckless risks in developing new technologies or systems with important social benefits but low financial returns. When the bust comes, the reckless lose their shirts, but the social benefits remain for the rest of us. Thus, for example, the dot-com bubble and bust ruined investors but laid the foundations of the modern internet. The recent housing bubble stimulated the building of lots of houses, which will still be there when prices have crashed, and much more affordable in the future.
There is a lot one could say to this. Firstly, a financial bubble is by definition an inflation in credit out of all proportion to any parallel increase in production, and is in consequence the most inefficient and wasteful method of stimulating development. To say that some good comes out of such catastrophic events is not to say anything at all. Development would have happened anyway, and regularly does, without any inflationary cycle to push it along. Secondly, it is an ivory-tower argument which takes no account of the terrible toll such busts have, not on fatcat investors who can afford it, but on millions of workers who already live on the breadline and have no resources with which to withstand the depredations of global recession. Third, it is an example of ‘spin’, where an admission of lack of control is packaged with a sales-pitch, to make a virtue out of a necessity. It is like arguing that bubonic plague serves a useful purpose, because it stimulates change in society.
When divorced from this preposterous spin, the admission that humans cannot control the economy walks a very dangerous edge. It is only a short step to the Marxian conclusion that the economy – capitalism – is an irrational system and should be abolished in favour of a more rational one. Aware of this, some scientists pursue the neo-Newtonian ideal of being able to predict the market. To this end, they offer us computer models.
What one has to say about computer models from the outset is that they can be a very powerful tool for understanding complex systems, provided that the parameters fed into the models are correct in the first place. The more complex the system, the more complex the parameters, and the less certainty over the initial algorithms. Climate modelling is a case in point. The best computers in the world can only predict the weather with any confidence up to three days in advance, after which the variables spiral exponentially out of control. Thus, attempts to predict the consequences of global warming vary widely.
The established way to test a model is to see how well its predictions accord with past documented events, in this case economic crises. Older models, which presupposed standard economic theories of rational trading and the law of value, that is, prices tending to gravitate towards their proper values, have had no success in predicting inflationary bubbles. Some success is now being claimed for models which recognise irrational elements such as trader fear and the herd instinct, and which are designed around artificially intelligent buyers and sellers who interact among themselves, just like real traders (New Scientist, 19 July). But these new models only deal in probabilities. They estimate that the probability of a bubble and bust event is a good deal more likely than older ‘equilibrium economics’ models suggested. But of course they can’t say when. Worse, while the weakness of fixed parameter models is that the parameters may be wrong, the weakness of artificially intelligent models, computer models which can ‘learn’ and modify their own parameters, is that they may rapidly become as complex and opaque as the system they are trying to emulate. One may end up with a computer model which becomes as incomprehensible as its real life counterpart.
The observation has been made in this column before that a computer model of socialist production and distribution, while complex, could be a useful contribution to socialist thinking and would not have to factor in such unquantifiable elements as trader fear or speculator frenzy. Indeed the strength of the socialist model would be in its relative simplicity. Once total demand and total supply are known, a small standard deviation would suffice because in the real world, based not on floating prices but on fixed use-values and known energy costs, production would proceed in a steady state. Only large scale catastrophic natural events, such as droughts, earthquakes, tsunamis or severe storms would cause any blip in the production process, but unless an event was so catastrophic that it affected global production, such as an unstoppable plague or an asteroid impact, the essentially steady and predictable production of socialist society would be able to absorb it. There’s a Nobel prize waiting for the computer scientist who comes up with the first working model of socialist non-market economics. But of course, they’d only get their prize in socialism. And, one need hardly add, there wouldn’t be any money attached.
Paddy Shannon
