Drawings by János HÁY.


When I was a young student in Paris - the city with the most movie-houses in the world - I saw many films, a great many in fact: big blockbusters as well as films which had passed almost un-noticed. But the reasons for the success of certain films and the failure of others largely eluded me: if one took two films of the same genre, and of a more or less equivalent quality, one was a box-office smash and the other quickly disappeared from the scene. Why was this?
Years later I spoke of these impressions of mine with a famous cinema expert: he too had asked himself the same thing, and he told me that the reasons for the success or the failure of a film are often incomprehensible. The cinema distribution companies are well aware that it is not enough to have for the film a cast of stars or a famous director: there is always the risk of it being a flop.
Arthur De Vany and W. David Wallis (1) analysed the box-office takings of the 50 most popular films in the United States during a nine-month period in 1995. They found that the top four films on the list had taken over 20% of all the tickets sold, while the last four on the list took only 0,001%. The first film on the list had takings of 49 million dollars and the last on the list under 5.000 dollars a relationship of 10.000:1!

But this enormous disparity in viewing figures can be observed in almost all cultural sectors, philosophy and religion included. In other words, cultural processes tend to function like fashions: some traits or products become popular while others despite their true value or qualities remain ignored. But how is this avalanche effect produced, according to which we all go to see the same few films, read the same few books, and always profess the same few philosophies or religions?
Thanks to the research inspired by the theories of chaos and complexity, we have finally become aware of the fact that that this “iniquity” is not specific to human beings. It is the effect of processes which mathematicians call non-linear and chaotic (2).

The ants and the marbles

In the ‘80s various entomologists repeated the following experiment: they put two deposits of food for ants at the same distance from the anthill (3) . When the ants had discovered the existence of these deposits and had started taking the food, the experimenters kept supplying the two sites with food so that they were always rigorously equivalent. This then was an application of the famous dilemma of Buridano’s donkey to the world of ants.

1. Economic Journal, November 1996.

2. Not all non-linear processes are chaotic. As we will see, chaos does not mean pure chance.

3. See A. Kirman, chapter in Arthur, Durlauf & Lane, The Economy as an Evolving Complex System II, Santa Fe Institute, Addison-Wesley 1997. Cf. P. Ormerod, Butterfly Economics, Pantheon Books 1998.

But the results of this experiment surprised the researchers. They had predicted that, following an initial period of oscillation, with time the choices of the ants would have become stabilized in an equal way: 50% of the ants would have gone to site A, and the other 50% to site B. But that did not happen. In fact the choices of the ants were subject to chaotic oscillations, which never became stabilized.

On the Y-axis we have the percentage of the ants which visited site A, while on the X-axis we have the time which passed. Over time we have continuous variations, not a stable equilibrium. We can say that in certain periods it is “fashionable” to go to site A, in other periods it is “fashionable” to go to site B, and in still other periods the popularity of the two sites is equal. It is said that the system of the anthill-site relationship is far from equilibrium: it undergoes continual variations. We notice that a chaotic series of the same type describes processes as different and varied as, for example, the annual variations of the GNP of an industrialised country or the annual variations of the numbers of sun spots.

And yet after a sufficiently long period of observation, it turns out that on average the ants pass more or less the same amount of time in A as they do in B: or - as the “chaos-ologists” say - the series has an attractor. In the long term then, order appears; it is said that this system is locally unpredictable and globally stable. But how can we explain this chaotic unpredictability?
Ants, like human beings, are imitative animals: if an ant goes first to site A, others will follow it to the same site, which will attract still more ants and so on. It is probable that some ants are more able to lead others, more “charismatic” than others: this would explain why at a certain moment most of the ants go to A and not to B, or vice versa. This avalanche effect is called a self-reinforcing process and it was formulated mathematically in 1983 by Arthur, Ermoliev and Kaniovski (4) as the “urn problem”.

Let us imagine that there are some black and white marbles, equal numbers of each, in an urn and imagine that someone takes out a certain number of marbles in a random way. The rule for putting them back into the urn is “if by chance more than half of the marbles taken out are white, then you must put them all back in, but with an extra white marble; the same thing applies if more than half of the marbles taken out are black”. If this operation is repeated, following the rule every time, it can happen that a distinctive tendency towards white marbles, for example, develops. Since the possibility of taking more white marbles increases each time a handful is taken out (because a white marble is added every time, there are more white marbles in a handful), after a while the urn could become filled with an overwhelmingly greater number of white marbles. In other words, at each “turn” the probability of getting a majority of white or black marbles does not stay the same, because at every turn the probabilities of the following turn are modified.
This mathematical game represents what happens with the ants: if by chance, at the beginning, more ants go to a certain site, this behaviour will influence that of the other ants, and therefore an “avalanche” effect is created in favour of one of the two sites. The future evolution of a process therefore depends on the initial processes, even if they have a minuscule effect. The scientists and experts say that the system is notably dependent on the initial conditions.

4. B. Arthur, Y. Ermoliev & Y. Kaniovski, “A Generalised Urn Problem and Its Applications”, Kibernetica, January-February 1983.

The market, blindfolded goddess

Some economists have observed that the same dynamics happens in commodities markets. In many cases, when two products are in competition, the objectively better one does not prevail: instead the winner is the one which, for whatever reason, is chosen by an initially superior number of people, often for totally casual reasons, or in any case inscrutable ones. It is by now obligatory to cite the example of the QWERTY keyboard, the one which we all use to this day. This is a particularly irrational system, and yet it is the one which has prevailed. The industries of the sector have proposed more easy to use and efficient types of keyboard many times, but every time this has been a failure.

When the first video recorders came onto the market two companies were in competition, each of which offered a different model: Betamax and VHS. In the end VHS won and Betamax disappeared. And yet, from the technical point of view the Betamax video system was, in many ways, superior. The point is that at the time, since they were both new products, the consumers were not experts. Buyers - also of books, political ideas, artistic tendencies, religions, etc. - are interactive agents; they influence each other reciprocally, and are not isolated atoms. Like the ants, also the consumers of video recorders have ended up imitating each other and fortune has rewarded the VHS. All those who preach the thaumaturgic virtues of the market, according to which the mechanisms of the market always reward the best products (also cultural and political ones) should meditate on cases of this kind.

In general, unjustifiable disparities very often arise between two or more cultural objects of analogous quality. In fact very few products (today they are usually American) take up alone the greater part of our cultural consumption, and all the other products have to be content with the crumbs that are left. This happens because of the urn effect described above: a “virtuous circle” of success is created starting from an initial positive impact. In other words, the market hyperbolises objective differences - or creates them wherever they do not exist or almost exist. In other words, a large proportion of cultural processes have the characteristics of what we could define as fashions.
So do the dynamics of reciprocal imitation alone explain the production of fashions in human cultures?

Simmel’s paradigm

The only truly important general theory of fashion is an essay by Georg Simmel of 1895, Die Mode (5). Simmel says that every fashion is a process which is always unstable and which depends on two contrasting and interacting forces. One force is the impulse of every human being to imitate someone else - usually a person who is considered, for some reason, as superior or in any case worthy of being imitated. The second force is the impulse of everyone to distinguish him/herself from his/her fellows - above all from those persons who are perceived, for some reason, as inferior. The relation between the tendency to imitate and the tendency to distinguish oneself varies from one human being to another, but it is rare for one of the two to be totally absent in an individual.

Let us consider a feature of recent fashions, for example the wearing of short shirts which reveal the female navel. This was certainly first done by young women of the higher classes and the most influential cultural sectors, who live in the great western metropolises. Then, gradually, women of an increasingly inferior social condition and those who live in less and less central areas imitated this exhibition of the navel. But as this trait became propagated and as it imposed itself - becoming ever more fashionable - it became less and less distinctive: this is why the ladies of a higher and more influential world or the trend-setters who have launched a fashion tend for this very reason to abandon it and pass on to something else. This explains the perennial instability of fashion: its triumph with the masses is tantamount to the digging of its grave. If one says of anything that it is “obligatory fashion”, this means that it is already on the wane and due to disappear.

Raymond Boudon has applied this model also to the field of sophisticated intellectual fashions (6) . A new philosophical idea, let us, for the sake of argument, call it heraclitism, springs up in one of the prestigious cultural breeding grounds - in the American campuses of the ivy league, in Paris, Oxford or Cambridge, in some Germany university, or any of the few other truly influential centres of intellectual production in the world. Then, gradually, heraclitism spreads to the outer provinces of Western culture, followed by Oriental culture. But as it becomes diffused, heraclitism is appropriated by intellectuals and professors of lesser calibre, less and less brilliant, ever more pedestrian and conformist - and so it becomes the obligatory paradigm taught in even the universities of marginal importance. Thus after a few decades the new philosophical elite, instructed in the rules of heraclitan obedience in one of the above-mentioned great philosophical centres, precisely in order to distinguish themselves from the mass of their colleagues, opt for a rival but less successful theory. Since by now heraclitism has become commonplace, a way of thinking which is taken for granted and therefore lazy, it is not too difficult for these young lions of the intellect to upset the status quo and promote the alternative philosophy. And so the cycle begins again.

But a point remains which the theory of Simmel and Boudon does not deal with: why is it precisely that trait - why exactly the navel en plein air - which is imitated and not another? Why is heraclitism adopted and not another philosophy having its own justificatory arguments and supporters? In effect the foremost avant-garde stylists usually base their work on a stock of ideas: they know that at the most only one proposal will be imitated and become fashionable. Every ambitious philosopher tries to launch a new and original way of thinking, but in the end only very few of these become hegemonic schools. So, what qualities must a cultural trait possess in order to be a success, even if only an ephemeral one?

5. G. Simmel, “Fashion”, International Quarterly, 10, n. 1, October 1904, pp. 130-155; “Fashion”, American Journal of Sociology, n. 62, 1957 [Philosophie der Mode, Pan-Verlag, Berlin 1905].
6. R. Boudon, La logique du social, Hachette, Paris 1979.

Cultural evolution has no meaning

The theory of chaos suggests the following idea to us: it is not necessary for a fashionable trait to have any particular qualities; instead it is enough for it to obey “the dynamics of the ants”. Of course it is necessary for some facilitating conditions to be satisfied: that the trait should first of all distinguish the persons who “make the fashion”, in other words that they should be in the prestigious position which makes them elegantiarum arbitri - or cogitationum arbitri (arbiters of concepts). It is also necessary for certain influential media to start acting as a sufficient sounding board. Given these conditions, a cultural product will be able to impose itself, while another analogous one will disappear immediately. We may consider the recent case of the book Empire (7) by Negri and Hardt, which seems to give a new new lymph to the radical and alternative social tendencies of the sixties: it became a best seller in Italy because it was already a blockbuster in the guiding market of America. This previous success abroad convinced the Italian reviewers to take it seriously, and thus “the process of the urn” was triggered off.

The unpredictability of the fortunes of a product is due to the fact that various negligible and minimal differences at the moment the process begins - the fact that at the beginning a book or a film has had a good review, for example - can lead to spectacular differences when it becomes fully developed. We have already seen that the system shows a sensitive dependence on the initial conditions, better known as the butterfly effect: “the beating of a butterfly’s wings in Beijing can cause a hurricane in California” . Also modern cultures, like the weather, are particularly unstable systems: minimal variations can produce dramatic results; while in more stable systems (such as certain archaic societies) even enormous impact do not manage to disturb the basic equilibrium.
The classical sociologists of culture usually maintain that cultural fashions have deeply rooted sociological motivations. For example, it is said that young people today tattoo themselves in order to overturn a dominant conception which exalts the reversibility of any choice, and the unlimited ability and tendency to change; thus in this way they are polemically affirming their preference for irreversible acts. Take as an example the great vogue for the thought of Popper in Italy in recent years: it is seen as indicating the decline of the totalising theories (such as Marxism) and the growing power of science. By following this line of argument it has even been declared that the periods in which women’s dresses become shorter coincide with periods of irrepressible female emancipation! But this is simply not true.
It is all very well to look for the deeper meanings of fashions, both in futile as well as serious fields. But one should ask oneself whether these fashions prevail because they express certain deep tendencies in the social way of being, or if they only seem to express deep tendencies because they happened to prevail in a certain specific period. The relationship between signifiers (the fashionable trait) and signified is much more complex than the classic sociology of culture would lead us to believe.

7. M. Hardt, A Negri, Empire, Harvard Univ. Press, 2000

Intelligible unpredictability

And so cultural processes are often non-linear and chaotic. A linear process is a classic chain of causes and effects: if I heat some water I know that it will boil at 100° centigrade. But human culture is a non-linear system and the single changes are thus unpredictable. An order can instead be mapped out over a long period. In other words, the socio-cultural facts show an aleatory and unpredictable tendency when seen in detail: no one knows what women will be wearing in a couple of years. But with time stability is revealed which, far from denying chaotic processes, is like their precipitate. An order disguised as disorder emerges.

However the claim of being able to predict cultural phenomena in a precise way comes up against some decisive limits. Today various fashion industries devote substantial financing to sophisticated research by social psychologists, hoping to discover a theory which will make fashion in some way predictable. It would be like manna from heaven: these companies could launch a sure-fire winning product every year. But they are just throwing away their money, because the increase of intelligibility that a theory of fashion can give us does not necessarily, ipso facto, lead to more predictability. Luckily that which human beings tend to prefer from one moment to another - whether it be a type of trousers or a philosophical or aesthetical conception - remains mostly unpredictable. Non-linearity is the ontological expression of the liberty of nature and therefore also of those natural beings who are human beings.

In fact no sociologist has ever truly been able to predict any social macro-phenomenon - and when he has actually got it right, it has almost always been by chance. 30 years ago who could have predicted the re-awakening of Islam and the jihad, which worries us so much today? Who could have foreseen even in 1965 the explosion of the radical protest movements of only two or three years later? And even as recently as 4 years ago who would have predicted the no-global vogue? Which American intellectual of the ‘60s would ever have predicted the hegemony of the deconstructionist tendencies in the American humanistic faculties? And we could continue this list indefinitely.

There is no need to speak at length of economic predictions, which almost always turn out to be wrong. I have the good habit of not reading even the economic predictions for the coming year, even if they come from the most prestigious sources. As Paul Ormerod (8) underlines , even the immense sums paid to the famous financial consultants employed by private clients and companies are simply thrown away: the oscillations of the financial markets are not even chaotic, they are often simply random. This is because the development of human culture (and also the economy) is indeterminate: at any given moment there is not just one track that a culture (or an economy) could follow, but many.
It would however be a mistake to conclude that this failure of the social sciences to predict the future evolution of such developments is the effect of the scientific backwardness of these sciences. In reality other much “harder” sciences are not able to do much better.
For example, the Darwinian theory of evolution - which is the dominant paradigm in biology - does not allow us to predict the future changes and developments of animal species. 80 million years ago, in the Cretaceous era, no biologist would have been able to predict the advent of a mammal of medium size called homo sapiens, around 78 million years later. Like the evolution of culture, also the evolution of life is to large extent unpredictable. And as Edward Lorenz has demonstrated, not even meteorology is ever able to predict with much precision. This should be enough to make us very mistrustful of futurologists of whatever tendency or school they may belong to, even if they are paid vast amounts: there is something rotten in the pretension of being able to predict the future, in many fields.
Chaos theory in fact tells us that certain unpredictability is not the effect of our inability to accumulate a mass of information which would allow us to make the prediction, but is an integral part of the non-linear structure of many natural processes.
But fortunately not everything is unpredictable. One should not confuse chaos with pure randomness. Not even the best meteorologist can say if next summer will be hotter or cooler than this summer, but we can all safely bet that the next summer will be hotter than the next winter. The theory of chaos also shows how order emerges within natural and cultural processes and that there is therefore something predictable: but it underlines the fact that order is basically a form of stable chaos. The fact that for centuries only women have worn dresses, for example, is probably the effect of stable chaos in the field of clothing habits, and perhaps even the fact that in western societies the Christian faith continues to prevail is a form of stable chaos.

For example, certain fashions in clothing turn out to be more or less periodical, and therefore they manifest a sort of order. It can be noticed that there is a cyclic variation in the length and the breadth of skirts (the anthropologist Kroeber even calculated the period involved (9) ). But even if we have periods of vertiginously short skirts and periods of extraordinarily long skirts, we can recognise an attractor, which is to say a sort of medium length of skirts in the West, for example in the last 200 years. Moreover, we can suppose with a certain margin of certainty that for the whole 21st century in the West women will continue to wear skirts, while - except for a few Scots - men will not wear them (this is however only a fairly safe bet, not a certainty). This stable order is always provisional and threatened by complexity. We should finally start thinking that we all live on the edge of chaos.

For this reason, if they were truly digested, the theories of complexity and chaos could change our way of seeing what happens in our cultures. They lead us to mistrust all the totalising and totalitarian conceptions which have the pretension of telling us with certainty what the world will be like and which therefore supply us with the instruments to dominate as we may please – or to help us submit to those who, in their opinion, will dominate us. Living on the edge of chaos is also an aesthetic choice: the acceptance of living joyously with the unpredictable, the new and the unknown. Rather than being simply the humiliation of our arrogance, it is the renunciation of the imaginary "regular income" of determinism and the transformation of our uncertainties into a genuine wealth to help us to survive.

8. P. Ormerod, The Death of Economics, Faber and Faber, London 1994.
9. A.L. Kroeber e J. Richardson, Three Centuries of Women's Dress Fashion, Univ. of California, Berkeley & Los Angeles 1940. They noticed that the rhythm of change of women’s evening gowns “not only is regular (the amplitude was of around half a century, the complete oscillation of a century), but it also tends towards the alternation of the forms according to a rational order: for example the width of the skirt and the width of the waist are always in an inverse relationship: when one is narrow the other is wide” (R. Barthes, Système de la Mode, Einaudi, 1970, pp. 299-300).

The need for linearity

Today many intellectuals avail themselves of every opportunity to refer to the theories of complexity and chaos. Also the chaotic theory of fashions will soon become, one can safely suppose, a fashion itself. And yet this approach has hitherto been and still remains - also among open-minded and well-informed intellectuals - a dead letter. For example, the overwhelming majority of people, even those with a certain degree of culture, continue to think of political processes in linear terms. I cannot exclude the idea that modern democracies function only on the basis of a linearist illusion according to which it is necessary for us to think that policies are either good or bad in absolute terms, and that certain actions of the government or the Central Bank are the real causes of a certain economic disaster or on the contrary of an economic boom.

The supposition that the relationship between the input (political or economical measures) and the output is linear seems to us a necessary condition for being able to judge political actions. But the theories of chaos teach us that the causal relationships in societies are not linear - that a wise and far-sighted policy in certain contexts can lead to dreadful results, or vice versa. Who could have foreseen, for example, that the application of the free-market doctrine of opening the markets fostered by the International Monetary Fund would have led to excellent results in some countries, while it is leading other countries, especially in Latin-America, towards bankruptcy?
In political and social life we always need to identify someone to be held responsible - or indeed a scapegoat - while in reality there is not a linear relationship between certain inputs and the final output. If the chaotic conception of society and politics truly became a part of our mentality, a significant portion of the old categories which still condition our thinking – for example, the fundamental political opposition between the left and the right, or between innovation and conservation - would lose the greater part of their meaning. But the irrepressible need to simplify reality will surely prevail over the disenchanted acceptance of complexity.


  © All rights belong to the authors or their heirs. 2004.