------- www.t0.or.at/delanda/meshwork.htm MESHWORKS, HIERARCHIES AND INTERFACES this is one of 7 Delanda items at: t0.or.at/delanda/ ------- Economics, computers and the war machine ---------- MARKETS AND ANTIMARKETS IN THE WORLD ECONOMY --------------- THE GEOLOGY OF MORALS ---------------- artnode.se/artorbit/issue1 (the fourth dates from feb 99) Deleuze and the genesis of form -------- Cybernetic culture research unit e-mail abstract culture syzygy archive id(entity) links occultures The Decimal Labyrinth (everything up to zone 4) -------- ccru.info/id(entity)/communiquetwo.htm ------------- altx.com/wordbombs/fuller.html  ==============  www.t0.or.at/delanda/meshwork.htm MESHWORKS, HIERARCHIES AND INTERFACES by Manuel De Landa The world of interface design is today undergoing dramatic changes which in their impact promise to rival those brought about by the use of the point-and-click graphical interfaces popularized by the Macintosh in the early 1980's. The new concepts and metaphors which are aiming to replace the familiar desk-top metaphor all revolve around the notion of semi-autonomous, semi-intelligent software agents. To be sure, different researchers and commercial companies have divergent conceptions of what these agents should be capable of, and how they should interact with computer users. But whether one aims to give these software creatures the ability to learn about the users habits, as in the non-commercial research performed at MIT autonomous agents group, or to endow them with the ability to perform transactions in the users name, as in the commercial products pioneered by General Magic, the basic thrust seems to be in the direction of giving software programs more autonomy in their decision-making capabilities. For a philosopher there are several interesting issues involved in this new interface paradigm. The first one has to do with the history of the software infrastructure that has made this proliferation of agents possible. From the point of view of the conceptual history of software, the creation of worlds populated by semi-autonomous virtual creatures, as well as the more familiar world of mice, windows and pull-down menus, have been made possible by certain advances in programming language design. Specifically, programming languages needed to be transformed from the rigid hierarchies which they were for many years, to the more flexible and decentralized structure which they gradually adopted as they became more "object-oriented". One useful way to picture this transformation is as a migration of control from a master program (which contains the general task to be performed) to the software modules which perform all the individual tasks. Indeed, to grasp just what is at stake in this dispersal of control, I find it useful to view this change as a part of a larger migration of control from the human body, to the hardware of the machine, then to the software, then to the data and finally to the world outside the machine. Since this is a crucial part of my argument let me develop it in some detail. The first part of this migration, when control of machine-aided processes moved from the human body to the hardware, may be said to have taken place in the eighteenth century when a series of inventors and builders of automata created the elements which later came together in the famous Jacquard loom, a machine which automated some of the tasks involved in weaving patterns in textiles. Jacquards loom used a primitive form of software, in which holes punched into cards coded for some of the operations behind the creation of patterned designs. {1} This software, however, contained only data and not control structures. In other words, all that was coded in the punched cards was the patterns to be weaved and not any directions to alter the reading of the cards or the performance of the operations, such as the lifting of the warp threads. Therefore, it was the machine's hardware component that "read" the cards and translated the data into motion, in which control of the process resided. Textile workers at the time were fully aware that they had lost some control to Jacquards loom, and they manifested their outrage by destroying the machines in several occasions. The idea of coding data into punched cards spread slowly during the 1800's, and by the beginning of our century it had found its way into computing machinery, first the tabulators used by Hollerith to process the 1890 United States census, then into other tabulators and calculators. In all these cases control remained embodied in the machine's hardware. One may go as far as saying that even the first modern computer, the imaginary computer created by Alan Turing in the 1930's still kept control in the hardware, the scanning head of the Turing machine. The tape that his machine scanned held nothing but data. But this abstract computer already had the seed of the next step, since as Turing himself understood, the actions of the scanning head could themselves be represented by a table of behavior, and the table itself could now be coded into the tape. Even though people may not have realized this at the time, coding both numbers and operations on numbers side by side on the tape, was the beginning of computer software as we know it. {2} When in the 1950's Turing created the notion of a subroutine, that is, the notion that the tasks that a computer must perform can be embodied into separate sub-programs all controlled by a master program residing in the tape, the migration of control from hardware to software became fully realized. From then on, computer hardware became an abstract mesh of logical gates, its operations fully controlled by the software. The next step in this migration took place when control of a given computational process moved from the software to the very data that the software operates on. For as long as computer languages such as FORTRAN or Pascal dominated the computer industry, control remained hierarchically embedded in the software. A master program would surrender control to a subroutine whenever that sub-task was needed to be performed, and the subroutine itself may pass control to an even more basic subroutine. But the moment the specific task was completed, control would move up the hierarchy until it reached the master program again. Although this arrangement remained satisfactory for many years, and indeed, many computer programs are still written that way, more flexible schemes were needed for some specific, and at the time, esoteric applications of computers, mostly in Artificial Intelligence. Trying to build a robot using a hierarchy of subroutines meant that researchers  had to completely foresee all the tasks that a robot would need to do and to centralize all decision-making into a master program. But this, of course, would strongly limit the responsiveness of the robot to events occurring in its surroundings, particularly if those events diverged from the predictions made by the programmers. One solution to this was to decentralize control. The basic tasks that a robot had to perform were still coded into programs, but unlike subroutines these programs were not commanded into action by a master program. Instead, these programs were given some autonomy and the ability to scan the data base on their own. Whenever the found a specific pattern in the data they would perform whatever task they were supposed to do. In a very real sense, it was now the data itself that controlled the process. And, more importantly, if the data base was connected to the outside world via sensors, so that patterns of data reflected patterns of events outside the robot, then the world itself was now controlling the computational process, and it was this that gave the robot a degree of responsiveness to its surroundings. Thus, machines went from being hardware-driven, to being software-driven, then data-driven and finally event-driven. Your typical Macintosh computer is indeed an event-driven machine even if the class of real world events that it is responsive to is very limited, including only events happening to the mouse (such as position changes and clicking) as well as to other input devices. But regardless of the narrow class of events that personal computers are responsive to, it is in these events that much of the control of the processes now resides. Hence, behind the innovative use of windows, icons, menus and the other familiar elements of graphical interfaces, there is this deep conceptual shift in the location of control which is embodied in object-oriented languages. Even the new interface designs based on semi-autonomous agents were made possible by this decentralization of control. Indeed, simplifying a little, we may say that the new worlds of agents, whether those that inhabit computer screens or more generally, those that inhabit any kind of virtual environment (such as those used in Artificial Life), have been the result of pushing the trend away from software command hierarchies ever further. The distinction between centralized and decentralized control of given process has come to occupy center-stage in many different contemporary philosophies. It will be useful to summarize some of this philosophical currents before I continue my description of agent-based interfaces, since this will reveal that the paradigm-shift is by no means confined to the area of software design. Economist and Artificial Intelligence guru Herbert Simon views bureaucracies and markets as the human institutions which best embody these two conceptions of control.{3} Hierarchical institutions are the easiest ones to analyze, since much of what happens within a bureaucracy in planned by someone of higher rank, and the hierarchy as a whole has goals and behaves in ways that are consistent with those goals. Markets, on the other hand, are tricky. Indeed, the term "market" needs to be used with care because it has been greatly abused over the last century by theorists on the left and the right. As Simon remarks, the term does not refer to the world of corporations, whether monopolies or oligopolies, since in these commercial institutions decision-making is highly centralized, and prices are set by command. I would indeed limit the sense of the term even more to refer exclusively to those weakly gatherings of people at a predefined place in town, and not to a dispersed set of consumers catered by a system of middleman (as when one speaks of the "market" for personal computers). The reason is that, as historian Fernand Braudel has made it clear, it is only in markets in the first sense that we have any idea of what the dynamics of price formation are. In other words, it is only in peasant and small town markets that decentralized decision-making leads to prices setting themselves up in a way that we can understand. In any other type of market economists simply assume that supply and demand connect to each other in a functional way, but they do not give us any specific dynamics through which this connection is effected. {4} Moreover, unlike the idealized version of markets guided by an "invisible hand" to achieve an optimal allocation of resources, real markets are not in any sense optimal. Indeed, like most decentralized, self-organized structures, they are only viable, and since they are not hierarchical they have no goals, and grow and develop mostly by drift. {5} Herbert Simon's distinction between command hierarchies and markets may turn out to be a special case of a more general dichotomy. In the view of philosophers Gilles Deleuze and Felix Guattari, this more abstract classes, which they call strata and self-consistent aggregates (or trees and rhizomes), are defined not so much by the locus of control, as by the nature of elements that are connected together. Strata are composed of homogenous elements, whereas self-consistent aggregates articulate heterogeneous elements as such. {6} For example, a military hierarchy sorts people into internally homogenous ranks before joining them together through a chain of command. Markets, on the other hand, allow for a set of heterogeneous needs and offers to become articulated through the price mechanism, without reducing this diversity. In biology, species are an example of strata, particularly if selection pressures have operated unobstructedly for long periods of time allowing the homogenization of the species gene pool. On the other hand, ecosystems are examples of self-consistent aggregates, since they link together into complex food webs a wide variety of animals and plants, without reducing their heterogeneity. I have developed this theory in more detail elsewhere, but for our purposes here let's simply keep the idea that besides centralization and decentralization of control, what defines these two types of structure is the homogeneity or heterogeneity of its composing elements. Before returning to our discussion of agent-based interfaces, there is one more point that needs to be stressed. As both Simon and Deleuze and Guattari emphasize, the dichotomy between bureaucracies and markets, or to use the terms that I prefer, between hierarchies and meshworks, should be understood in purely relative terms. In the first place, in reality it is hard to find pure cases of these two structures: even the most goal-oriented organization will still show some drift in its growth and development, and most markets even in small towns contain some hierarchical elements, even if it is just the local wholesaler which manipulates prices by dumping (or withdrawing) large amounts of a product on (or from) the market. Moreover, hierarchies give rise to meshworks and meshworks to hierarchies. Thus, when several bureaucracies coexist (governmental, academic, ecclesiastic), and in the absence of a super-hierarchy to coordinate their interactions, the whole set of institutions will tend to form a meshwork of hierarchies, articulated mostly through local and temporary links. Similarly, as local markets grow in size, as in those gigantic fairs which have taken place periodically since the Middle Ages, they give rise to commercial hierarchies, with a money market on top, a luxury goods market underneath and, after several layers, a grain market at the bottom. A real society, then, is made of complex and changing mixtures of these two types of structure, and only in a few cases it will be easy to decide to what type a given institution belongs. A similar point may be made about the worlds inhabited by software agents. The Internet, to take the clearest example first, is a meshwork which grew mostly by drift. No one planned either the extent or the direction of its development, and indeed, no one is in charge of it even today. The Internet, or rather its predecessor, the Arpanet, acquired its decentralized structure because of the needs of U.S. military hierarchies for a command and communications infrastructure which would be capable of surviving a nuclear attack. As analysts from the Rand Corporation made it clear, only if the routing of the messages was performed without the need for a central computer could bottlenecks and delays be avoided, and more importantly, could the meshwork put itself back together once a portion of it had been nuclearly vaporized. But in the Internet only the decision-making behind routing is of the meshwork type. Decision-making regarding its two main resources, computer (or CPU) time and memory, is still hierarchical. Schemes to decentralize this aspect do exist, as in Drexler's Agoric Systems, where the messages which flow through the meshwork have become autonomous agents capable of trading among themselves both memory and CPU time. {7} The creation by General Magic of its Teletext operating system, and of agents able to perform transactions on behalf of users, is one of the first real-life steps in the direction of a true decentralization of resources. But in the meanwhile, the Internet will remain a hybrid of meshwork and hierarchy components, and the imminent entry of big corporations into the network business may in fact increase the amount of command components in its mix. These ideas are today being hotly debated in the field of interface design. The general consensus is that interfaces must become more intelligent to be able to guide users in the tapping of computer resources, both the informational wealth of the Internet, as well as the resources of ever more elaborate software applications. But if the debaters agree that interfaces must become smarter, and even that this intelligence will be embodied in agents, they disagree on how the agents should acquire their new capabilities. The debate pits two different traditions of Artificial Intelligence against each other: Symbolic AI, in which hierarchical components predominate, against Behavioral AI, where the meshwork elements are dominant. Basically, while in the former discipline one attempts to endow machines with intelligence by depositing a homogenous set of rules and symbols into a robot's brain, in the latter one attempts to get intelligent behavior to emerge from the interactions of a few simple task-specific modules in the robot's head, and the heterogeneous affordances of its environment. Thus, to build a robot that walks around a room, the first approach would give the robot a map of the room, together with the ability to reason about possible walking scenarios in that model of the room. The second approach, on the other hand, endows the robot with a much simpler set of abilities, embodied in modules that perform simple tasks such as collision-avoidance, and walking-around-the-room behavior emerges from the interactions of these modules and the obstacles and openings that the real room affords the robot as it moves.{8} Translated to the case of interface agents, for instance, personal assistants in charge of aiding the user to understand the complexities of particular software applications, Symbolic AI would attempt to create a model of the application as well as a model of the working environment, including a model of an idealized user, and make these models available in the form of rules or other symbols to the agent. Behavioral AI, on the other hand, gives the agent only the ability to detect patterns of behavior in the actual user, and to interact with the user in different ways so as to learn not only from his or her actual behavior but also from feedback that the user gives it. For example, the agent in question would be constantly looking over the user's shoulder keeping track on whatever regular or repetitive patterns it observes. It then attempts to establish statistical correlations between certain pairs of actions that tend to occur together. At some point the agent suggests to the user the possibility of automating these actions, that is, that whenever the first occurs, the second should be automatically performed. Whether the user accepts or refuses, this gives feedback to the agent. The agent may also solicit feedback directly, and the user may also teach the agent by giving some hypothetical examples. {9} In terms of the location of control, there is very little difference between the agents that would result, and in this sense, both approaches are equally decentralized. The rules that Symbolic AI would put in the agents head, most likely derived from interviews of users and programmers by a Knowledge Engineer, are independent software objects. Indeed, in one of the most widely used programming languages in this kind of approach (called a "production system") the individual rules have even more of a meshwork structure that many object-oriented systems, which still cling to a hierarchy of objects. But in terms of the overall human-machine system, the approach of Symbolic AI is much more hierarchical. In particular, by assuming the existence of an ideal user, with homogenous and unchanging habits, and of a workplace where all users are similar, agents created by this approach are not only less adaptive and more commanding, they themselves promote homogeneity in their environment. The second class of agents, on the other hand, are not only sensitive to heterogeneities, since they adapt to individual users and change as the habits of this users change, they promote heterogeneity in the work place by not subordinating every user to the demands of an idealized model. One drawback of the approach of Behavioral AI is that, given that the agent has very little knowledge at the beginning of a relationship with a user, it will be of little assistance for a while until it learns about his or her habits. Also, since the agent can only learn about situations that have recurred in the past, it will be of little help when the user encounters new problems. One possible solution, is to increase the amount of meshwork in the mix and allow agents from different users to interact with each other in a decentralized way. {10} Thus, when a new agent begins a relation with a user, it can consult with other agents and speed up the learning process, assuming that is, that what other agents have learned is applicable to the new user. This, of course, will depend on the existence of some homogeneity of habits, but at least it does not assume a complete homogenous situation from the outset, an assumption which in turn promotes further uniformization. Besides, endowing agents with a static model of the users makes them unable to cope with novel situations. This is also a problem in the Behavioral AI approach but here agents may aid one another in coping with novelty. Knowledge gained in one part of the workplace can be shared with the rest, and new knowledge may be generated out of the interactions among agents. In effect, a dynamic model of the workplace would be constantly generated and improved by the collective of agents in a decentralized way, instead of each one being a replica of each other operating on the basis of a static model centrally created by a knowledge engineer. I would like to conclude this brief analysis of the issues raised by agent-based interfaces with some general remarks. First of all, from the previous comments it should be clear that the degree of hierarchical and homogenizing components in a given interface is a question which affects more than just events taking place in the computer's screen. In particular, the very structure of the workplace, and the relative status of humans and machines is what is at stake here. Western societies have undergone at least two centuries of homogenization, of which the most visible element is the assembly-line and related mass-production techniques, in which the overall thrust was to let machines discipline and control humans. In this circumstances, the arrival of the personal computer was a welcome antidote to the development of increasingly more centralized computer machinery, such as systems of Numerical Control in factories. But this is hardly a victory. After two hundred years of constant homogenization, working skills have been homogenized via routinization and Taylorization, building materials have been given constant properties, the gene pools of our domestic species homogenized through cloning, and our languages made uniform through standardization. To make things worse, the solution to this is not simply to begin adding meshwork components to the mix. Indeed, one must resist the temptation to make hierarchies into villains and meshworks into heroes, not only because, as I said, they are constantly turning into one another, but because in real life we find only mixtures and hybrids, and the properties of these cannot be established through theory alone but demand concrete experimentation. Certain standardizations, say, of electric outlet designs or of data-structures traveling through the Internet, may actually turn out to promote heterogenization at another level, in terms of the appliances that may be designed around the standard outlet, or of the services that a common data-structure may make possible. On the other hand, the mere presence of increased heterogeneity is no guarantee that a better state for society has been achieved. After all, the territory occupied by former Yugoslavia is more heterogeneous now than it was ten years ago, but the lack of uniformity at one level simply hides an increase of homogeneity at the level of the warring ethnic communities. But even if we managed to promote not only heterogeneity, but diversity articulated into a meshwork, that still would not be a perfect solution. After all, meshworks grow by drift and they may drift to places where we do not want to go. The goal-directedness of hierarchies is the kind of property that we may desire to keep at least for certain institutions. Hence, demonizing centralization and glorifying decentralization as the solution to all our problems would be wrong. An open and experimental attitude towards the question of different hybrids and mixtures is what the complexity of reality itself seems to call for. To paraphrase Deleuze and Guattari, never believe that a meshwork will suffice to save us. {11} Footnotes: {1} Abbot Payson Usher. The Textile Industry, 1750-1830. In Technology in Western Civilization. Vol. 1. Melvin Kranzberg and Carrol W. Pursell eds. (Oxford University Press, New York 1967). p. 243 {2} Andrew Hodges. Alan Turing: The Enigma. (Simon & Schuster, New York 1983). Ch. 2 {3} Herbert Simon. The Sciences of the Artificial. (MIT Press, 1994). p.43 {4} Fernand Braudel. The Wheels of Commerce. (Harper and Row, New York, 1986). Ch. I {5} Humberto R. Maturana and Francisco J. Varela. The Tree of Knowledge. The Biological Roots of Human Understanding. (Shambhala, Boston 1992). p. 47 and 115. {6} Gilles Deleuze and Felix Guattari. A Thousand Plateaus. (University of Minnesota Press, Minneapolis, 1987). p. 335 {7} M.S. Miller and K.E. Drexler. Markets and Computation: Agoric Open Systems. In The Ecology of Computation. Bernardo Huberman ed. (North-Holland, Amsterdam 1988). {8} Pattie Maes. Behaviour-Based Artificial Intelligence. In From Animals to Animats. Vol. 2. Jean-Arcady Meyer, Herbert L. Roitblat and Stewart W. Wilson. (MIT Press, Cambridge Mass, 1993). p. 3 {9} Pattie Maes and Robyn Kozierok. Learning Interface Agents. In Proceedings of AAAI È93 Conference. (AAAI Press, Seattle WA. 1993). p. 459-465 {10} Yezdi Lashari, Max Metral and Pattie Maes. Collaborative Interface Agents. In Proceedings of 12th National Conference on AI. (AAAI Press, Seattle WA. 1994). p. 444-449 {11} Deleuze and Guattari. op. cit. p. 500. (Their remark is framed in terms of "smooth spaces" but it may be argued that this is just another term for meshworks). =============== Economics, Computers and the War Machine. by Manuel De Landa. When we "civilians" think about military questions we tend to view the subject as encompassing a rather specialized subject matter, dealing exclusively with war and its terrible consequences. It seems fair to say that, in the absence of war (or at least the threat of war, as in the case of government defense budget debates) civilians hardly ever think about military matters. The problem is that, from a more objective historical perspective, the most important effects of the military establishment on the civilian world in the last four hundred years have been during peacetime, and have had very little to do with specifically military subjects, such as tactics or strategy. I would like to suggest that, starting in the 1 500's, Western history has witnessed the slow militarisation of civilian society, a process in which schools, hospitals and prisons slowly came to adopt a form first pioneered in military camps and barracks, and factories came to share a common destiny with arsenals and armories. I should immediately add, however, that the influence was hardly unidirectional, and that what needs to be considered in detail are the dynamics of complex "institutional ecologies", in which a variety of organizations exert mutual influences on one another. Nevertheless, much of the momentum of this process was maintained by military institutions and so we may be justified in using the term "militarisation". On one hand, there is nothing too surprising about this. Ever since Napoleon changed warfare from the dynastic duels of the eighteenth century to the total warfare with which we are familiar in this century, war itself has come to rely on the complete mobilization of a society's industrial and human resources. While the armies of Frederick the Great were composed mostly of expensive mercenaries, who had to be carefully used in the battlefield, the Napoleonic armies benefited from the invention of new institutional means of converting the entire population of a country into a vast reservoir of human resources. Although technically speaking the French revolution did not invent compulsory military service, its institutional innovations did allow its leaders to perform the first modern mass conscription, involving the conversion of all men into soldiers, and of all women into cheap laborers. As the famous proclamation of 1793 reads: "...all Frenchmen are permanently requisitioned for service into the armies. Young men will go forth to battle; married men will forge weapons and transport munitions; women will make tents and clothing and serve in hospitals; children will make lint from old linen; and old men will be brought to the public squares to arouse the courage of the soldiers, while preaching the unity of the Republic and hatred against Kings." {1} This proclamation, and the vast bureaucratic machinery needed to enforce it, effectively transformed the civilian population of France into a resource (for war, production, motivation) to be tapped into at will by the military high command. A similar point applies to the industrial, mineral and agricultural resources of France and many other nation states. Given the complete mobilization of society's resources involved in total war it is therefore not surprising that there has been a deepening of military involvement in civilian society in the last two centuries. However, I would want to argue that, in addition to the links between economic, political and military institutions brought about by war time mobilizations, there are other links, which are older, subtler but for the same reason more insidious, which represent a true militarisation of society during peace time. To retire to the French example, some of the weapons that the Napoleonic armies used were the product of a revolution in manufacturing techniques which took place in French armories in the late eighteenth century. In French armories, the core concepts and techniques of what later would become assembly­line, mass production techniques, were for the first time developed. The ideal of creating weapons with perfectly interchangeable parts, and ideal which could not be fulfilled without standardization and routinization of production, was taken even further in American arsenals in the early 19th century. And it was there that military engineers first realized that in practice, standardization went hand in hand with replacement of flexible individual skills with rigid collective routines, enforced through constant discipline and monitoring. Even before that, in the Dutch armies of the sixteenth century, this process had already begun. Civilians tend to think of Frederick Taylor, the late nineteenth century creator of so­called "scientific management" techniques, as the pioneer of labor process analysis, that is, the breaking down of a given factory practice into micro­movements and the streamlining of these movements for greater efficiency and centralized management control. But Dutch commander Maurice of Nassau had already applied these methods to the training of his soldiers beginning in the 1560's. Maurice analyzed the motion needed to load, aim and fire a weapon into its micro­movements, redesigned them for maximum efficiency and then imposed them on his soldiers via continuous drill and discipline. {2} Yet, while the soldiers increased their efficiency tremendously as a collective whole, each individual soldier completely lost control of his actions in the battlefield. And a similar point applies to the application of this idea to factory workers, before and after Taylorism. Collectively they became more productive, generating the economies of scale so characteristic of twenty­century big business, while simultaneously completely losing control of their individual actions. This is but one example of the idea of militarisation of society. Recent historians have rediscovered several other cases of the military origins of what was once thought to be civilian innovations. In recent times it has been Michel Foucault who has most forcefully articulated this view. For him this intertwining of military and civilian institutions is constitutive of the modern European nation­state. On one hand, the project of nationbuilding was an integrative movement, forging bonds that went beyond the primordial ties of family and locality, linking urban and rural populations under a new social contract. On the other, and complementing this process of unification, there was the less conscious project of uniformation, of submitting the new population of free citizens to intense and continuous training, testing and exercise to yield a more or less uniform mass of obedient individuals. In Foucault's own words: "Historians of ideas usually attribute the dream of a perfect society to the philosophers and jurists of the eighteenth century; but there was also a military dream of society; its fundamental reference was not to the state of nature, but to the meticulously subordinated cogs of a machine, not to the primal social contract, but to permanent coercions, not to fundamental rights, but to indefinitely progressive forms of training, not to the general will but to automatic docility... The Napoleonic regime was not far off and with it the form of state that was to survive it and, we must not forget, the foundations of which were laid not only by jurists, but also by soldiers, not only counselors of state, but also junior officers, not only the men of the courts, but also the men of the camps. The Roman reference that accompanied this formation certainly bears with it this double index: citizens and legionnaires, law and maneuvers. While jurists or philosophers were seeking in the pact a primal model for the construction or reconstruction of the social body, the soldiers and with them the technicians of discipline were elaborating procedures for the individual and collective coercion of bodies." {3} Given that modern technology has evolved in such a world of interacting economic, political and military institutions, it should not come as a surprise that the history of computers, computer networks, Artificial Intelligence and other components of contemporary technology, is so thoroughly intertwined with military history. Here, as before, we must carefully distinguish those influences which occurred during war­time from those that took place in peace­time, since the former can be easily dismissed as involving the military simply as a catalyst or stimulant, that is, an accelerator of a process that would have occurred more slowly without its direct influence. The computer itself may be an example of indirect influence. The basic concept, as everyone knows, originated in a most esoteric area of the civilian world. In the 1 930's British mathematician Alan Turing created the basic concept of the computer in an attempt to solve some highly abstract questions in metamathematics. But for that reason, the Turing Machine, as his conceptual machine was called, was a long way from an actual, working prototype. It was during World War 11, when Turing was mobilized as part of the war effort to crack the Nazi's Enigma code, that, in the course of his intense participation in that operation, he was exposed to some of the practical obstacles blocking the way towards the creation of a real Turing Machine. On the other side of the Atlantic, John Von Neuman also developed his own practical insights as to how to bring the Turing Machine to life, in the course of his participation in the Manhattan Project and other war related operations. In this case we may easily dismiss the role that the military played, arguing that without the intensification and concentration of effort brought about by the war, the computer would have developed on its own, perhaps at a slower pace. And I agree that this is correct. On the other hand, many of the uses to which computers were put after the war illustrate the other side of the story: a direct participation of military institutions in the development of technology, a participation which actually shaped this technology in the direction of uniformization, routinization and concentration of control. Perhaps the best example of this other relation between the military and technology is the systems of machine­part production known as Numerical Control methods. While the methods developed in 19th. century arsenals, and later transferred to civilian enterprises, had already increased uniformity and centralized control in the production of large quantities of the same object (that is, mass production), this had left untouched those areas of production which create relatively small batches of complex machine parts. Here the skills of the machinist were still indispensable as late as World War II. During the 1950's, the Air Force underwrote not only the research and development of a new system to get rid of the machinist's skills, but also the development of software, the actual purchase of machinery by contractors, and the training of operators and programmers. In a contemporary Numerical Control system, after the engineer draws the parts that need to be produced, the drawings themselves are converted into data and stored in cards or electronically. From then on, all the operations needed to be performed, drilling, milling, lathing, boring, and so on, are performed automatically by computer­controlled machines. Unlike mass­production techniques, where this automatism was achieved at the expense of flexibility, in Numerical Control systems a relatively simple change in software (not hardware) is all that is needed to adapt the system for the production of a new set of parts. Yet, the effects on the population of workers were very similar in both cases: the replacement of flexible skills by rigid commands embodied in hardware or software, and over time, the loss of those skills leading to a general process of worker de­skilling, and consequently, to the loss of individual control of the production process. The question in both cases is not the influence that the objects produced in militarized factories may have on the civilian world. One could, for instance, argue that the support of the canned food industry by Napoleon had a beneficial effect on society, and a similar argument may be made for many objects developed under military influence. The question, however, is not the transfer of objects, but the transfer of the production processes behind those objects that matters, since these processes bring with them the entire control and command structure of the military with them. To quote historian David Noble: "The command imperative entailed direct control of production operations not just with a single machine or within a single plant, but worldwide, via data links. The vision of the architects of the [Numerical Control] revolution entailed much more than the automatic machining of complex parts; it meant the elimination of human intervention ­a shortening of the chain of command ­ and the reduction of remaining people to unskilled, routine, and closely regulated tasks." And he adds that Numerical Control is a "giant step in the same direction [as the 19th. century drive for uniformity]; here management has the capacity to bypass the worker and communicate directly to the machine via tapes or direct computer link. The machine itself can thereafter pace and discipline the worker." {4} Let's pause for a moment and consider a possible objection to this analysis. One may argue that the goal of withdrawing control from workers and transferring it to machines is the essence of the capitalist system and that, if military institutions happened to be involved, they did so by playing the role assigned to them by the capitalist system. The problem with this reply is that, although it may satisfy a convinced Marxist, it is at odds with much historical data gathered by this century's best economic historians. This data shows that European societies, far from having evolved through a unilinear progression of "modes of production" (feudalism, capitalism, socialism), actually exhibited a much more complex, more heterogeneous coexistence of processes. In other words, as historian Ferdinand Braudel has shown, as far back as the fourteenth and fifteenth centuries, institutions with the capability of exercising economic power (large banks, wholesalers, long­distance trade companies) were already in operation. and fully coexisted with feudal institutions as well as with economic institutions that did no have economic power, such as retailers and producers of humble goods. Indeed, Braudel shows that these complex coexistances of institutions of different types existed before and after the Industrial Revolution, and suggests that the concept of a "capitalist system" (where every aspect of society is connected into a functional whole) gives a misleading picture of the real processes. What I am suggesting here is that we take Braudel seriously, forget about our picture of history as divided into neat, internally homogeneous eras or ages, and tackle the complex combinations of institutions involved in real historical processes. The models we create of these complex "institutional ecologies" should include military organizations playing a large, relatively independent role, to reflect the historical data we now have on several important cases, like fifteenth century Venice, whose famous Arsenal was at the time the largest industrial complex in Europe, or at eighteenth century France and nineteenth century United States, and their military standardization of weapon production. Another important example, involves the development of the modern corporation, particularly as it happened in the United States in the last century. The first American big business was the railroad industry, which developed the management techniques which many other large enterprises would adopt later on. This much is well known. What is not so well known is that military engineers were deeply involved in the creation of the first railroads and that they developed many of the features of management which later on came to characterize just about every large commercial enterprise in the United States, Europe and elsewhere. In the words of historian Charles O'Connell: "As the railroads evolved and expanded, they began to exhibit structural and procedural characteristics that bore a remarkable resemblance to those of the Army. Both organizations erected complicated management hierarchies to coordinate and control a variety of functionally diverse, geographically separated corporate activities. Both created specialized staff bureaus to provide a range of technical and logistical support services. Both divided corporate authority and responsibility between line and staff agencies and officers and then adopted elaborate written regulations that codified the relationship between them. Both established formal guidelines to govern routine activities and instituted standardized reporting and accounting procedures and forms to provide corporate headquarters with detailed financial and operational information which flowed along carefully defined lines of communication. As the railroads assumed these characteristics, they became America's first 'big business'." {5} Thus, the transfer of military practices to the civilian world influenced the lives not only of workers, but of the managers themselves. And the influence did not stop with the development of railroads. The "management science" which is today taught in business schools is a development of military "operations research", a discipline created during World War 11 to tackle a variety of tactical, strategic and logistic problems. And it was the combination of this "science of centralization" and the availability of large computers that, in turn, allowed the proliferation of transnational corporations and the consequent internationalization of the standardization and routinization of production processes. Much as skills were replaced by commands in the shop floor, so were prices replaced by commands at the management level. (This is one reason not to use the term "markets" when theorizing big business. Not only they rely on commands instead of prices, they manipulate demand and supply rather than being governed by them. Hence, Braudel has suggested calling big business "anti­markets"). {6} Keeping in mind the actual complexity of historical processes, as opposed to explaining everything by the "laws of capitalist development", is crucial not only to understand the past, but also to intervene in the present and speculate about the future. This is particularly clear when analyzing the role which computers and computer networks may play in the shaping of the economic world in the coming century. It is easy to attribute many of the problems we have today, particularly those related to centralized surveillance and control, to computer technology. But to do this would not only artificially homogenize the history of computers (there are large differences between the development of mainframes and minicomputers, on one hand, and the personal computer, on the other) but it would obscure the fact that, if computers have come to play the "disciplinarian" roles they play today it is as part of a historical processes which is several centuries old, a process which computers have only intensified. Another advantage of confronting the actual heterogeneity of historical processes, and of throwing to the garbage the concept of "the capitalist system", is that we free ourselves to look around for combinations of economic institutions which coexist with disciplinarian anti­markets but do not play by the same rules. Historically, as Braudel has shown, economic power since the 14th century has always been associated with large size enterprises and their associated "economies of scale". Although technically this term only applies to mass­produced objects, economies of scale meaning the spreading of production costs among many identical products, we may use it in an extended way to define any economic benefits to managers, merchants and financiers stemming from the scale of any economic resource. Coexisting with economies of scale there are what is called "economies of agglomeration". These are economic benefits which small businesses enjoy from the concentration of many of them in a large city. These economies stem from the benefits of shop­talk, from unplanned connections and mutual enhancements, as well as for the services which grow around these concentrations, services which small business could not afford on their own. To conclude this talk I would like to give one example, from the world of computers, of two American industrial hinterlands which illustrate the difference between economies of scale and of agglomeration: Silicon Valley in Northern California, and Route 128 near Boston: "Silicon Valley has a decentralized industrial system that is organized around regional networks. Like firms in Japan, and parts of Germany and Italy, Silicon Valley companies tend to draw on local knowledge and relationships to create new markets, products, and applications. These specialist firms compete intensely while at the same time learning from one another about changing markets and technologies. The region's dense social networks and open labor markets encourage experimentation and entrepreneurship. The boundaries within firms are porous, as are those between firms themselves and between firms and local institutions such as trade associations and universities." {7} The growth of this region owed very little to large financial flows from governmental and military institutions. Silicon Valley did not develop so much by economies of scale, as by the benefits derived from an agglomeration of visionary engineers, specialist consultants and financial entrepreneurs. Engineers moved often from one firm to another, developing loyalties to the craft and region's networks, not to the corporation. This constant migration, plus an unusual practice of information sharing among the local producers, insured that new formal and informal knowledge diffused rapidly through the entire region. Business associations fostered collaboration between small and medium­sized companies. Risk­taking and innovation were preferred to stability and routinization. This, of course, does not mean that there were not large, routinized firms in Silicon Valley, only that they did not dominate the mix. Not so in Route 128: "While Silicon Valley producers of the 1970's were embedded in, and inseparable from, intricate social and technical networks, the Route 128 region came to be dominated by a small number of highly self­sufficient corporations. Consonant with New England's two century old manufacturing tradition, Route 128 firms sought to preserve their independence by internalizing a wide range of activities. As a result, secrecy and corporate loyalty govern relations between firms and their customers, suppliers, and competitors, reinforcing a regional culture of stability and self­reliance. Corporate hierarchies insured that authority remains centralized and information flows vertically. The boundaries between and within firms and between firms and local institutions thus remain far more distinct." {8} While before the recession of the 1980's both regions had been continuously expanding, one on economies of scale and the other on economies of agglomeration (or rather, mixtures dominated by one or the other), they both felt the full impact of the downturn. At that point some large Silicon Valley firms, unaware of the dynamics behind the region's success, began to switch to economies of scale, sending parts of their production to other areas, and internalizing activities previously performed by smaller firms. Yet, unlike Route 128, the intensification of routinization and internalization in Silicon Valley was not a constitutive part of the region, which meant that the old meshwork system could be revived. And this is, in fact, what happened. Silicon Valley's regional networks were re­energized, through the birth of new firms in the old pattern, and the region has now returned to its former dynamic state, unlike the command­heavy Route 128 which continues to stagnate. What this shows is that, while both scale and agglomeration economies, as forms of positive feedback, promote growth, only the latter endows firms with the flexibility needed to cope with adverse economic conditions. In conclusion I would like to repeat my call for more realistic models of economic history, models involving the full complexity of the institutional ecologies involved, including markets, anti­markets, military and bureaucratic institutions, and if we are to believe Michel Foucault, schools, hospitals, prisons and many others. It is only through an honest philosophical confrontation with our complex past that we can expect to understand it and derive the lessons we may use when intervening in the present and speculating about the future. References: {1} Excerpt from the text of the levee en mass of 1793, quoted in William H. McNeill. The Pursuit of Power. Technology, Armed Force and Society since A.D. 1000. (University of Chicago Press, 1982). p. 192 {2} ibid. p. 129 {3} Michel Foucault. Discipline and Punish. The Birth of Prison. (Vintage Books, New York, 1979) p. 169 {4} David Noble. Command Performance: A Perspective on Military Enterprise and Technological Change. In Merrit Roe Smith ed. Military Enterprise. (MIT Press, 1987). p. 341 and 342. {5} Charles F. O'Connell, Jr. The Corps of Engineers and the Rise of Modern Management. In ibid. p. 88 {6} Fernand Braudel. The Wheels of Commerce. (Harper and Row, New York, 1986). p.379 {7} Annalee Saxenian. Lessons from Silicon Valley. In Technology Review, Vol. 97, no. 5. page. 44 {8} ibid. p. 47 =================== MARKETS AND ANTIMARKETS IN THE WORLD ECONOMY by Manuel De Landa. One of the most significant epistemological events in recent years is the growing importance of historical questions in the ongoing reconceptualization of the hard sciences. I believe it is not an exaggeration to say that in the last two or three decades, history has almost completely infiltrated physics, chemistry and biology. It is true that nineteenth century thermodynamics had already introduced an arrow of time into physics, and hence the idea of irreversible historical processes. It is also true that the theory of evolution had already shown that animals and plants were not embodiments of eternal essences but piecemeal historical constructions, slow accumulations of adaptive traits cemented together via reproductive isolation. However, the classical versions of these two theories incorporated a rather weak notion of history into their conceptual machinery: both thermodynamics and Darwinism admitted only one possible historical outcome, the reaching of thermal equilibrium or of the fittest design. In both cases, once this point was reached, historical processes ceased to count. For these theories, optimal design or optimal distribution of energy represented, in a sense, an end of history. Hence, it should come as no surprise that the current penetration of science by history has been the result of advances in these two disciplines. Ilya Prigogine revolutionized thermodynamics in the 1960's by showing that the classical results were only valid for closed systems where the overall amounts of energy are always conserved. If one allows energy to flow in and out of a system, the number and type of possible historical outcomes greatly increases. Instead of a unique and simple equilibrium, we now have multiple ones of varying complexity (static, periodic and chaotic attractors); and moreover, when a system switches from one to another form of stability (at a so-called bifurcation), minor fluctuations can be crucial in deciding the actual form of the outcome. Hence, when we study a given physical system, we need to know the specific nature of the fluctuations that have been present at each of its bifurcations, in other words, we need to know its exact history to understand its current dynamical form. {1} And what is true of physical systems is all the more so for biological ones. Attractors and bifurcations are features of any system in which the dynamics are nonlinear, that is, in which there are strong interactions between variables. As biology begins to include these nonlinear dynamical phenomena in its models, for example, in the case of evolutionary arms-races between predators and prey, the notion of a "fittest design" loses its meaning. In an arms-race there is no optimal solution fixed once and for all, since the criterion of fitness itself changes with the dynamics. This is also true for any adaptive trait which value depends on how frequent it occurs in a given population, as well as in cases like migration, where animal behavior interacts nonlinearly with selection pressures. As the belief in a fixed criterion of optimality disappears from biology, real historical processes come to reassert themselves once more. {2} Computers have played a crucial role in this process of infiltration. The nonlinear equations that go into these new historical models cannot be solved by analytical methods alone, and so scientists need computers to perform numerical simulations and discover the behavior of the solutions. But perhaps the most crucial role of digital technology has been to allow a switch from a purely analytic, top-down style of modeling, to a more synthetic, bottom-up approach. In the growing discipline of Artificial Life, for instance, an ecosystem is not modeled starting from the whole and dissecting it into its component parts, but the other way around: one begins at the bottom, with a population of virtual animals and plants and their local interactions, and the ecosystem needs to emerge spontaneously from these local dynamics. The basic idea is that the systematic properties of an ecosystem arise from the interactions between its animal and plant components, so that when one dissects the whole into parts the first thing we lose is any property due to these interactions. Analytical techniques, by their very nature, tend to kill emergent properties, that is, properties of the whole that are more than the sum of its parts. Hence the need for a more synthetic approach, in which everything systematic about a given whole is modeled as a historically emergent result of local interactions. {3} These new ideas are all the more important when we move on to the social sciences, particularly economics. In this discipline, we tend to uncritically assume systematicity, as when one talks of the "capitalist system", instead of showing exactly how such systematic properties of the whole emerge from concrete historical processes. Worse yet, we then tend to reify such unaccounted-for systematicity, ascribing all kinds of causal powers to capitalism, to the extent that a clever writer can make it seem as if anything at all (from nonlinear dynamics itself to postmodernism or cyberculture) is the product of late capitalism. This basic mistake, which is, I believe, a major obstacle to a correct understanding of the nature of economic power, is partly the result of the purely top-down, analytical style that has dominated economic modeling from the eighteenth century. Both macroeconomics, which begins at the top with concepts like gross national product, as well as microeconomics, in which a system of preferences guides individual choice, are purely analytical in approach. Neither the properties of a national economy nor the ranked preferences of consumers are shown to emerge from historical dynamics. Marxism, is true, added to these models intermediate scale phenomena, like class struggle, and with it conflictive dynamics. But the specific way in which it introduced conflict, via the labor theory of value, has now been shown by Shraffa to be redundant, added from the top, so to speak, and not emerging from the bottom, from real struggles over wages, or the length of the working day, or for control over the production process. {4} Besides a switch to a synthetic approach, as it is happening, for instance, in the evolutionary economics of Nelson and Winter in which the emphasis is on populations of organizations interacting nonlinearly, what we need here is a return to the actual details of economic history. Much has been learned in recent decades about these details, thanks to the work of materialist historians like Fernand Braudel, and it is to this historical data that we must turn to know what we need to model synthetically. Nowhere is this need for real history more evident that in the subject of the dynamics of economic power, defined as the capability to manipulate the prices of inputs and outputs of the production process as well as their supply and demand. In a peasant market, or even in a small town local market, everybody involved is a price taker: one shows up with merchandise, and sells it at the going prices which reflect demand and supply. But monopolies and oligopolies are price setters: the prices of their products need not reflect demand/supply dynamics, but rather their own power to control a given market share. {5} When approaching the subject of economic power, one can safely ignore the entire field of linear mathematical economics (so-called competitive equilibrium economics), since there monopolies and oligopolies are basically ignored. Yet, even those thinkers who make economic power the center of their models, introduce it in a way that ignores historical facts. Authors writing in the Marxist tradition, place real history in a straight-jacket by subordinating it to a model of a progressive succession of modes of production. Capitalism itself is seen as maturing through a series of stages, the latest one of which is the monopolistic stage in this century. Even non-Marxists economists like Galbraith, agree that capitalism began as a competitive pursuit and stayed that way till the end of the nineteenth century, and only then it reached the monopolistic stage, at which point a planning system replaced market dynamics. However, Fernand Braudel has recently shown, with a wealth of historical data, that this picture is inherently wrong. Capitalism was, from its beginnings in the Italy of the thirteenth century, always monopolistic and oligopolistic. That is to say, the power of capitalism has always been associated with large enterprises, large that is, relative to the size of the markets where they operate. {6} Also, it has always been associated with the ability to plan economic strategies and to control market dynamics, and therefore, with a certain degree of centralization and hierarchy. Within the limits of this presentation, I will not be able to review the historical evidence that supports this extremely important hypothesis, but allow me at least to extract some of the consequences that would follow if it turns out to be true. First of all, if capitalism has always relied on non-competitive practices, if the prices for its commodities have never been objectively set by demand/supply dynamics, but imposed from above by powerful economic decision-makers, then capitalism and the market have always been different entities. To use a term introduced by Braudel, capitalism has always been an "antimarket". This, of course, would seem to go against the very meaning of the word "capitalism", regardless of whether the word is used by Karl Marx or Ronald Reagan. For both nineteenth century radicals and twentieth century conservatives, capitalism is identified with an economy driven by market forces, whether one finds this desirable or not. Today, for example, one speaks of the former Soviet Union's "transition to a market economy", even though what was really supposed to happen was a transition to an antimarket: to large scale enterprises, with several layers of managerial strata, in which prices are set not taken. This conceptual confusion is so entrenched that I believe the only solution is to abandon the term "capitalism" completely, and to begin speaking of markets and antimarkets and their dynamics. This would have the added advantage that it would allow us to get rid of historical theories framed in terms of stages of progress, and to recognize the fact that antimarkets could have arisen anywhere, not just Europe, the moment the flows of goods through markets reach a certain critical level of intensity, so that organizations bent on manipulating these flows can emerge. Hence, the birth of antimarkets in Europe has absolutely nothing to do with a peculiarly European trait, such as rationality or a religious ethic of thrift. As is well known today, Europe borrowed most of its economic and accounting techniques, those techniques that are supposed to distinguish her as uniquely rational, from Islam. {8} Finally, and before we take a look at what a synthetic, bottom-up approach to the study of economic dynamics would be like, let me meet a possible objection to these remarks: the idea that "real" capitalism did not emerge till the nineteenth century industrial revolution, and hence that it could not have arisen anywhere else where these specific conditions did not exist. To criticize this position, Fernand Braudel has also shown that the idea that capitalism goes through stages, first commercial, then industrial and finally financial, is not supported by the available historical evidence. Venice in the fourteenth century and Amsterdam in the seventeenth, to cite only two examples, already show the coexistance of the three modes of capital in interaction. Moreover, other historians have recently shown that that specific form of industrial production which we tend to identify as "truly capitalist", that is, assembly-line mass production, was not born in economic organizations, but in military ones, beginning in France in the eighteenth century, and then in the United States in the nineteenth. It was military arsenals and armories that gave birth to these particularly oppressive control techniques of the production process, at least a hundred years before Henry Ford and his Model-T cars {10} Hence, the large firms that make up the antimarket, can be seen as replicators, much as animals and plants are. And in populations of such replicators we should be able to observe the emergence of the different commercial forms, from the family firm, to the limited liability partnership to the joint stock company. These three forms, which had already emerged by the fifteenth century, must be seen as arising, like those of animals and plants, from slow accumulations of traits which later become consolidated into more or less permanent structures, and not, of course, as the manifestation of some pre-existing essence. In short, both animal and plant species as well as "institutional species" are historical constructions, the emergence of which bottom-up models can help us study. It is important to emphasize that we are not dealing with biological metaphors here. Any kind of replicating system which produces variable copies of itself, coupled with any kind of sorting device, is capable of evolving new forms. This basic insight is now exploited technologically in the so-called "genetic algorithm", which allows programmers to breed computer software instead of painstakingly coding it by hand. A population of computer programs is allowed to reproduce with some variation, and the programmer plays the role of sorting device, steering the population towards the desired form. The same idea is what makes Artificial Life projects work. Hence, when we say that the forms the antimarket has taken are evolved historical constructions we do not mean to say that they are metaphorically like organic forms, but that they are produced by a process which embodies the same engineering diagram as the one which generates organic forms. Another example may help clarify this. When one says, as leftists used to say, that "class-struggle is the motor of history", one is using the word "motor" in a metaphorical way. On the other hand, to say that a hurricane is a steam motor is not to use the term metaphorically, but literally: one is saying that the hurricane embodies the same engineering diagram as a steam motor: it uses a reservoir of heat and operates via differences of temperature circulated through a Carnot cycle. The same is true of the genetic algorithm. Anything that replicates, such as patterns of behavior transmitted by imitation, or rules and norms transmitted by enforced repetition can give rise to novel forms, when populations of them are subjected to selection pressures. And the traits that are thus accumulated can become consolidated into a permanent structure by codification, as when informal routines become written rules. {11} In this case, we have the diagram of a process which generates hierarchical structures, whether large institutions rigidly controlled by their rules or organic structures rigidly controlled by their genes. There are, however, other structure-generating processes which result in decentralized assemblages of heterogeneous components. Unlike a species, an ecosystem is not controlled by a genetic program: it integrates a variety of animals and plants in a food web, interlocking them together into what has been called a "meshwork structure". The dynamics of such meshworks are currently under intense investigation and something like their abstract diagram is beginning to emerge. {12} From this research, it is becoming increasingly clear that small markets, that is, local markets without too many middlemen, embody this diagram: they allow the assemblage of human beings by interlocking complementary demands. These markets are indeed, self-organized decentralized structures: they arise spontaneously without the need for central planning. As dynamic entities they have absolutely nothing to do with an "invisible hand", since models based on Adam Smith's concept operate in a frictionless environment in which agents have perfect rationality and all information flows freely. Yet, by eliminating nonlinearities, these models preclude the spontaneous emergence of order, which depends crucially on friction: delays, bottlenecks, imperfect decision-making and so on. The concept of a meshwork can be applied not only to the area of exchange, but also to that of industrial production. Jane Jacobs has created a theory of the dynamics of networks of small producers meshed together by their interdependent functions, and has collected some historical evidence to support her claims. The basic idea is that certain relatively backward cities in the past, Venice when it was still subordinated to Byzantium, or the network New York-Boston-Philadelphia when still a supply zone for the British empire, engage in what she calls, import-substitution dynamics. Because of their subordinated position, they must import most manufactured products, and export raw materials. Yet, meshworks of small producers within the city, by interlocking their skills can begin to replace those imports with local production, which can then be exchanged with other backward cities. In the process, new skills and new knowledge is generated, new products begin to be imported, which in turn, become the raw materials for a new round of import-substitution. Nonlinear computer simulations have been created of this process, and they confirm Jacobs' intuition: a growing meshwork of skills is a necessary condition for urban morphodynamics. The meshwork as a whole is decentralized, and it does not grow by planning, but by a kind of creative drift. {13} Of course, this dichotomy between command hierarchies and meshworks should not be taken too rigidly: in reality, once a market grows beyond a certain size, it spontaneously generates a hierarchy of exchange, with prestige goods at the top and elementary goods, like food, at the bottom. Command structures, in turn, generate meshworks, as when hierarchical organizations created the automobile and then a meshwork of services (repair shops, gas stations, motels and so on), grew around it. {14} More importantly, one should not romantically identify meshworks with that which is "desirable" or "revolutionary", since there are situations when they increase the power of hierarchies. For instance, oligopolistic competition between large firms is sometimes kept away from price wars by the system of interlocking directorates, in which representatives of large banks or insurance companies sit in the boards of directors of these oligopolies. In this case, a meshwork of hierarchies is almost equivalent to a monopoly. {15} And yet, however complex the interaction between hierarchies and meshworks, the distinction is real: the former create structures out of elements sorted out into homogenous ranks, the latter articulates heterogeneous elements as such, without homogenization. A bottom-up approach to economic modeling should represent institutions as varying mixtures of command and market components, perhaps in the form of combinations of negative feedback loops, which are homogenizing, and positive feedback, which generates heterogeneity. What would one expect to emerge from such populations of more or less centralized organizations and more or less decentralized markets? The answer is, a world-economy, or a large zone of economic coherence. The term, which should not be confused with that of a global economy, was coined by Immanuel Wallerstein, and later adapted by Braudel so as not to depend on a conception of history in terms of a unilineal progression of modes of production. From Wallerstein Braudel takes the spatial definition of a world-economy: an economically autonomous portion of the planet, perhaps coexisting with other such regions, with a definite geographical structure: a core of cities which dominate it, surrounded by yet other economically active cities subordinated to the core and forming a middle zone, and finally a periphery of completely exploited supply zones. The role of core of the European world-economy has been historically played by several cities: first Venice in the fourteenth century, followed by Antwerp and Genoa in the fifteenth and sixteenth. Amsterdam then dominated it for the next two centuries, followed by London and then New York. Today, we may be witnessing the end of American supremacy and the role of core seems to be moving to Tokyo. {16} Interestingly, those cities which play the role of core, seem to generate in their populations of firms, very few large ones. For instance, when Venice played this role, no large organizations emerged in it, even though they already existed in nearby Florence. Does this contradict the thesis that capitalism has always been monopolistic? I think not. What happens is that, in this case, Venice as a whole played the role of a monopoly: it completely controlled access to the spice and luxury markets in the Levant. Within Venice, everything seemed like "free competition", and yet its rich merchants enjoyed tremendous advantages over any foreign rival, whatever its size. Perhaps this can help explain the impression classical economists had of a competitive stage of capitalism: when the Dutch or the British advocated "free competition" internally is precisely when their cities as a whole held a virtual monopoly on world trade. World-economies, then, present a pattern of concentric circles around a center, defined by relations of subordination. Besides this spatial structure, Wallerstein and Braudel add a temporal one: a world-economy expands and contracts in a variety of rhythms of different lengths: from short term business cycles to longer term Kondratiev cycles which last approximately fifty years. While the domination by core cities gives a world-economy its spatial unity, these cycles give it a temporal coherence: prices and wages move in unison over the entire area. Prices are, of course, much higher at the center than at the periphery, and this makes everything flow towards the core: Venice, Amsterdam, London and New York, as they took their turn as dominant centers, became "universal warehouses" where one could find any product from anywhere in the world. And yet, while respecting these differences, all prices moved up and down following these nonlinear rhythms, affecting even those firms belonging to the antimarket, which needed to consider those fluctuations when setting their own prices. These self-organized patterns in time and space which define world-economies were first discovered in analytical studies of historical data. The next step is to use synthetic techniques and create the conditions under which they can emerge in our models. In fact, bottom-up computer simulations of urban economics where spatial and temporal patterns spontaneously emerge already exist. For example, Peter Allen has created simulations of nonlinear urban dynamics as meshworks of interdependent economic functions. Unlike earlier mathematical models of the distribution of urban centers, which assumed perfect rationality on the part of economic agents, and where spatial patterns resulted from the optimal use of some resource such as transportation, here patterns emerge from a dynamic of conflict and cooperation. As the flows of goods, services and people in and out of these cities change, some urban centers grow while others decay. Stable patterns of coexisting centers arise as bifurcations occur in the growing city networks taking them from attractor to attractor. {17} Something like Allen's approach would be useful to model one of the two things that stitch world-economies together, according to Braudel: trade circuits. However, to generate the actual spatial patterns that we observe in the history of Europe, we need to include the creation of chains of subordination among these cities, of hierarchies of dependencies besides the meshworks of interdependencies. This would need the inclusion of monopolies and oligopolies, growing out of each cities meshworks of small producers and traders. We would also need to model the extensive networks of merchants and bankers with which dominant cities invaded their surrounding urban centers, converting them into a middle zone at the service of the core. A dynamical system of trade circuits, animated by import-substitution dynamics within each city, and networks of merchants extending the reach of large firms of each city, may be able to give us some insight into the real historical dynamics of the European economy. {18} Bottom-up economic models which generate temporal patterns have also been created. One of the most complex simulations in this area is the Systems Dynamics National Model at MIT. Unlike econometric simulations, where one begins at the macroeconomic level, this one is built up from the operating structure within corporations. Production processes within each industrial sector are modeled in detail. The decision-making behind price setting, for instance, is modeled using the know-how from real managers. The model includes many nonlinearities normally dismissed in classical economic models, like delays, bottlenecks and the inevitable friction due to bounded rationality. The simulation was not created with the purpose of confirming the existence of the Kondratiev wave, the fifty-two year cycle that can be observed in the history of wholesale prices for at least two centuries. In fact, the designers of the model were unaware of the literature on the subject. Yet, when the simulation began to unfold, it reached a bifurcation and a periodic attractor emerged in the system, which began pulsing to a fifty year beat. The crucial element in this dynamics seems to be the capital goods sector, the part of the industry that creates the machines that the rest of the economy uses. Whenever an intense rise in global demand occurs, firms need to expand and so need to order new machines. But when the capital goods sector in turn expands to meet this demand it needs to order from itself. This creates a positive feedback loop that pushes the system towards a bifurcation. {19} Insights coming from running simulations like these can, in turn, be used to build other simulations and to suggest directions for historical research to follow. We can imagine parallel computers in the near future running simulations combining all the insights from the ones we just discussed: spatial networks of cities, breathing at different rhythms, and housing evolving populations of organizations and meshworks of interdependent skills. If power relations are included, monopolies and oligopolies will emerge and we will be able to explore the genesis and evolution of the antimarket. If we include the interactions between different forms of organizations, then the relationships between economic and military institutions may be studied. As Galbraith has pointed out, in today's economy nothing goes against the market, nothing is a better representative of the planning system, as he calls it, than the military-industrial complex. {20} But we would be wrong in thinking that this is a modern phenomenon, something caused by "late capitalism". In the first core of the European world-economy, thirteenth century Venice, the alliance between monopoly power and military might was already in evidence. The Venetian arsenal, where all the merchant ships were built, was the largest industrial complex of its time. We can think of these ships as the fixed capital, the productive machinery of Venice, since they were used to do all the trade that kept her powerful; but at the same time, they were military machines used to enforce her monopolistic practices. {21} When the turn of Amsterdam and London came to be the core, the famous Companies of Indias with which they conquered the Asian world-economy, transforming it into a periphery of Europe, were also hybrid military-economic institutions. We have already mentioned the role that French armories and arsenals in the eighteenth century, and American ones in the nineteenth, played in the birth of mass production techniques. Frederick Taylor, the creator of the modern system for the control the labor process, learned his craft in military arsenals. That nineteenth century radical economists did not understand this hybrid nature of the antimarket can be seen from the fact that Lenin himself welcomed Taylorism into revolutionary Russia as a progressive force, instead of seeing for what it was: the imposition of a rigid command-hierarchy on the workplace. {22} Unlike these thinkers, we should include in our simulations all the institutional interactions that historians have uncovered, to correctly model the hybrid economic-military structure of the antimarket. Perhaps by using these synthetic models as tools of exploration, as intuition synthesizers, so to speak, we will also be able to study the feasibility of counteracting the growth of the antimarket by a proliferation of meshworks of small producers. Multinational corporations, according to the influential theory of "transaction-costs", grow by swallowing up meshworks, by internalizing markets either through vertical or horizontal integration. {23} They can do this thanks to their enormous economic power (most of them are oligopolies), and to their having access to intense economies of scale. However, meshworks of small producers interconnected via computer networks could have access to different, yet as intense economies of scale. A well studied example is the symbiotic collection of small textile firms that has emerged in an Italian region between Bologna and Venice. The operation of a few centralized textile corporations was broken down into a decentralized network of firms, in which entrepreneurs replace managers and short runs of specialized products replace large run of mass produced ones. Computer networks allow these small firms to react flexibly to sudden shifts in demand, so that no firm becomes overloaded while others sit idly with spare capacity. {24} But more importantly, a growing pool of skills is thereby created, and because this pool has not been internalized by a large corporation, it can not be taken away. Hence this region will not suffer the fate of so many American company towns, which die after the corporation that feeds them moves elsewhere. This self-organized reservoirs of skills also explain why economic development cannot be exported to the third world via large transfers of capital invested in dams or other large structures. Economic development must emerge from within as meshworks of skills grow and proliferate. {25} Computer networks are an important element here, since the savings in coordination costs that multinational corporations achieve by internalizing markets, can be enjoyed by small firms through the use of decentralizing technology. Computers may also help us to create a new approach to control within these small firms. The management approach used by large corporations was in fact developed during World War II under the name of Operations Research. Much as mass production techniques effected a transfer of a command hierarchy from military arsenals to civilian factories, management practices based on linear analysis carry with them the centralizing tendencies of the military institutions where they were born. Fresh approaches to these questions are now under development by nonlinear scientists, in which the role of managers is not to impose preconceived plans on workers, but to catalyze the emergence of meshworks of decision-making processes among them. {26} Computers, in the form of embedded intelligence in the buildings that house small firms, can aid this catalytic process, allowing the firm's members to reach some measure of self-organization. Although these efforts are in their infancy, they may one day play a crucial role in adding some heterogeneity to a world-economy that's becoming increasingly homogenized. FOOTNOTES: {1} Ilya Prigogine and Isabelle Stengers. Order out of Chaos. (Bantam Books, New York 1984). p.169. {2} Stuart A. Kauffman. The Origins of Order. Self Organization and Selection in Evolution. (Oxford Univ. Press, New York 1993) p.280 {3} Christopher G. Langton. Artificial Life. In C.G. Langton ed. Artificial Life. (Addison-Wesley, 1989) p.2 {4} Geoff Hodgson. Critique of Wright 1: Labour and Profits. In Ian Steedman ed. The Value Controversy. (Verso, London 1981). p.93 {5} John Keneth Galbraith. The New Industrial State. (Houghton Mifflin, Boston 1978) p.24 {6} Fernand Braudel. Civilization and Capitalism, 15th-18th Century. Vol 2. (Harper and Row, New York 1982) p.229 {7} ibid. p.559-561 {8} William H. McNeill. The Pursuit of Power. (University of Chicago Press, 1982) p.49 {9} Merrit Roe Smith. Army Ordnance and the "American system" of Manufacturing, 1815-1861. In M.R.Smith ed. Military Enterprise and Technological Change. (MIT Press, 1987) p.47 {10} Richard Nelson and Sidney Winter. An Evolutionary Theory of Economic Change. (Belknap Press, Cambridge Mass 1982) p.98 {11} Richard Dawkins. The Selfish Gene. (Oxford University Press, New York 1989) ch.11 {12} Stuart Kauffman. The Evolution of Economic Webs. In Philip Anderson, Kenneth Arrow and David Pines eds. The Economy as an Evolving Complex System. (Addison-Wesley, 1988) {13} Jane Jacobs. Cities and the Wealth of Nations. (Random House, New York 1984) p.133 {14} The dichotomy Meshwork/Hierarchy is a special case of what Deleuze and Guattari call Smooth/Striated or Rhizome/Tree. Gilles Deleuze and Felix Guattari. 1440: The Smooth and the Striated. In A Thousand Plateaus. (University of Minnesota Press, Minneapolis 1987) ch.14 {15} John R. Munkirs and James I. Sturgeon. Oligopolistic Cooperation: Conceptual and Empirical Evidence of Market Structure Evolution. In Marc. R. Tool and Warren J. Samuels eds. The Economy as a System of Power. (Transaction Press, New Brunswick 1989). p.343 {16} Fernand Braudel. op. cit. Vol 3. p.25-38 {17} Peter M. Allen. Self-Organization in the Urban System. In William C. Schieve and P.M.Allen eds. Self-Organization and Dissipative Structures: Applications in the Physical and the Social Sciences. (University of Texas, Austin 1982) p.136 {18} Fernand Braudel. op. cit. Vol 3. p.140-167 {19} J.D. Sterman. Nonlinear Dynamics in the World Economy: the Economic Long Wave. In Peter Christiansen and R.D. Parmentier eds. Structure, Coherence and Chaos in Dynamical Systems. (Manchester Univ. Press, Manchester 1989) {20} John Galbraith. op. cit. p. 321 {21} Fernand Braudel. op. cit. Vol 2 p. 444 {22} Vladimir Lenin. The Immediate Tests of the Soviet Goverment. Collected Works, Vol 27 (Moskow 1965). {23} Jean-Francois Hennart. The Transaction Cost Theory of the Multinational Enterprise. In Christos Pitelis and Roger Sudgen eds. The Nature of the Transnational Firm. (Rutledge, London 1991). {24} Thomas W. Malone and John F. Rockart. Computers, Networks and the Corporation. In Scientific American Vol 265 Number 3 p.131 Also: Jane Jacobs, op. cit. p.40 Fernand Braudel, op cit Vol 3 p. 630 {25} Jane Jacobs. op. cit. p.148 {26} F. Malik and G. Probst. Evolutionary Management. In H.Ulrich and G. Probst eds. Self-Organization and Management of Social Systems. (Springer Verlag, Berlin 1984) p. 113 =============== Manuel DeLanda, writer and artist, has published, among other works, War in the Age of Intelligent Machines, and A Thousand Years of Nonlinear History. by Manuel DeLanda One constant in the history of Western philosophy seems to be a certain conception of matter as an inert receptacle for forms that come from the outside. In other words, the genesis of form and structure seems to always involve resources that go beyond the capabilities of the material substratum of these forms and structures. In some cases, these resources are explicitly transcendental, eternal essences defining forms which are imposed on infertile materials. The clearest example of this theory of form is, of course, religious Creationism, in which form begins as an idea in God's mind, and is then imposed by a command on an obedient and docile matter. But more serious examples also exist. In ancient philosophies Aristotle's essences seem to fit this pattern, as do those that inhabit Platonist heavens. And although classical physics began with a clean break with Aristotelian philosophy, and did endow matter with some spontaneous behavior (e.g. inertia), it reduced the variability and richness of material expression to the concept of mass, and studied only the simplest material systems (frictionless planetary dynamics, ideal gases) where spontaneous self-generation of form does not ocurr, thus always keeping some transcendental agency hidden in the background. Yet, as Gilles Deleuze has shown in his work on Spinoza, not every Western philosopher has taken this stance. In Spinoza, Deleuze discovers another possibility: that the resources involved in the genesis of form are not transcendental but immanent to matter itself. A simple example should suffice to illustrate this point. The simplest type of immanent resource for morphogenesis seems to be endogenously-generated stable states. Historically, the first such states to be discovered by scientists studying the behavior of matter (gases) were energy minima (or correspondingly, entropy maxima). The spherical form of a soap bubble, for instance, emerges out of the interactions among its constituent molecules as these are constrained energetically to "seek" the point at which surface tension is minimized. In this case, there is no question of an essence of "soap-bubbleness" somehow imposing itself from the outside, an ideal geometric form (a sphere) shaping an inert collection of molecules. Rather, an endogenous topological form (a point in the space of energetic possibilities for this molecular assemblage) governs the collective behavior of the individual soap molecules, and results in the emergence of a spherical shape. Moreover, the same topological form, the same minimal point, can guide the processes that generates many other geometrical forms. For example, if instead of molecules of soap we have the atomic components of an ordinary salt crystal, the form that emerges from minimizing energy (bonding energy in this case) is a cube. In other words, one and the same topological form can guide the morphogenesis of a variety of geometrical forms. A similar point applies to other topological forms which inhabit these spaces of energetic possibilities. For example, these spaces may contain closed loops (technically called "limit cycles" or "periodic attractors"). In this case the several possible physical instantiations of this space will all display isomorphic behavior: an endogenously generated tendency to oscillate in a stable way. Whether one is dealing with a socio-technological structure (such as a radio transmitter or a radar machine), a biological one (a cyclic metabolism), or a physical one (a convection cell in the atmosphere), it is one and the same immanent resource that is involved in their different oscillating behavior. Since this is a crucial issue in Deleuze's philosophy let me explain this point in a little more detail. Deleuze calls this ability of topological forms to give rise to many different physical instantiations, a process of "divergent actualization", taking the idea from French philosopher Henri Bergson who, at the turn of the century, wrote a series of texts where he criticized the inability of the science of his time to think the new, the truly novel. The first obstacle was, according to Bergson, a mechanical and linear view of causality and the rigid determinism that it implied. Clearly, if all the future is already given in the past, if the future is merely that modality of time where previously determined possibilities become realized, then true innovation is impossible. To avoid this mistake, he thought, we must struggle to model the future as truly open ended, and the past and the present as pregnant not only with possibilities which become real, but with virtualities which become actual. The distinction between the possible and the real, assumes a set of predefined forms (or essences) which aquire physical reality as material forms that resemble them. From the morphogenetic point of view, realizing a possibility does not add anything to a predefined form, except reality. The distinction between the virtual and the actual, on the other hand, does not involve resemblance of any kind (e.g. our example above, in which a topological point becomes a geometrical sphere) and far from constituting the essential identity of a form, it subverts identity, since now forms as different as spheres and cubes emerge from the same topological point. To quote from what is probably his most important book, "Difference and Repetition": "Actualization breaks with resemblance as a process no less than it does with identity as a principle. In this sense, actualization or differenciation is always a genuine creation." And Deleuze goes on to discuss processes of actualization more complex than bubbles or crystals, processes such as embryogenesis, the development of a fully differenciated organism starting from a single cell. In this case, the space of energetic possibilities is more elaborate, involving many topological forms governing complex spatio-temporal dynamisms: "How does actualization ocurr in things themselves?...Beneath the actual qualities and extensities [of things themselves] there are spatio-temporal dynamisms. They must be surveyed in every domain, even though they are ordinarly hidden by the constituted qualities and extensities. Embryology shows that the division of the egg is secondary in relation to more significant morphogenetic movements: the augmentation of free surfaces, stretching of cellullar layers, invagination by folding, regional displacement of groups. A whole kinimatics of the egg appears which implies a dynamic". In "Difference and Repetition", Deleuze repeatedly makes use of these "spaces of energetic possibilities" (technically refered to as "state spaces" or "phase spaces"), and of the topological forms (or "singularities") that shape these spaces. Since these ideas reappear in his later work, and since both the concept of "phase space" and that of "singularity" belong to mathematics, it is safe to say that a crucial component of Deleuzian thought comes from the philosophy of mathematics. And, indeed, chapter four of "Difference and Repetition" is a meditation on the metaphysics of the differential and integral calculus. On the other hand, given that "phase spaces" and "singularities" become physically significant only in relation to material systems which are traversed by a strong flow of energy, Deleuze philosophy is also intimately related to that branch of physics which deals with material and energetic flows, that is, with thermodynamics. And, indeed, chapter five of "Difference and Repetition" is a philosophical critique of nineteenth century thermodynamics, an attempt to recover from that discipline some of the key concepts needed for a theory of immanent morphogenesis. At the beginning of that chapter, Deleuze introduces some key distinctions that will figure prominently in his later work, specifically the concept of "intensity", but more importantly, he reveals in the very first page his ontological commitments. It is traditional since Kant to distinguish between the world as it appears to us humans, that is, the world of phenomena or appereances, and the world as it exists by itself, regardless of whether there is a human observer to interact with it. This world "in itself" is refered to as "nuoumena". A large number of contemporary thinkers, particularly those that call themselves "postmodernists", do not believe in nuomena. For them the world is socially constructed, hence, all it contains is linguistically-defined phenomena. Notice that even though many of these thinkers declare themselves "anti-essentialist", they share with essentialism a view of matter as an inert material, only in their case form does not come from a Platonic heaven, or from the mind of God, but from the minds of humans (or from cultural conventions expressed linguistically). The world is amorphous, and we cut it out into forms using language. Nothing could be further from Deleuzian thought than this postmodern linguistic relativism. Deleuze is indeed a realist philosopher, who not only believes in the autonomous existance of actual forms (the forms of rocks, plants, animals and so on) but in the existance of virtual forms. In the first few lines of chapter five of "Difference and Repetition", where Deleuze introduces the notion of "intensity" as a key to understand the actualization of virtual forms, he writes: "Difference is not diversity. Diversity is given, but difference is that by which the given is given...Difference is not phenomenon but the nuoumenon closest to the phenomenon...Every phenomenon refers to an inequality by which it is conditioned...Everything which happens and everything which appears is correlated with orders of differences: differences of level, temperature, pressure, tension, potential, difference of intensity". {2} Let me illustrate this idea with a familiar example from thermodynamics. If one creates a container separated into two compartments, and one fills one compartment with cold air and the other with hot air, one thereby creates a system embodying a difference in intensity, the intensity in this case being temperature. If one then opens a small hole in the wall dividing the compartments, the intensity difference causes the onset of a spontaneous flow of air from one side to the other. It is in this sense that intensity differences are morphogenetic, even if in this case the form that emerges is too simple. The examples above of the soap bubble and the salt crystal, as well as the more complex foldings and stretchings undergone by an embryo, are generated by similar principles. However, in the page following the quote above, Deleuze argues that, despite this important insight, nineteenth century thermodynamics cannot provide the foundation he needs for a philosophy of matter. Why? Because that branch of physics became obsessed with the final equilibrium forms, at the expense of the difference-driven morphogenetic process which gives rise to those forms. But as Deleuze argues, the role of virtual singularities can only be grasped during the process of morphogenesis, that is, before the final form is actualized, before the difference dissapears. This shortcoming of nineteenth century thermodynamics, to overlook the role of intensity differences in morphogenesis, to concentrate on the equlibrium form that emerges only once the original difference has been cancelled, has today been repaired in the latest version of this branch of physics, appropriatedly labeled "far-from-equilibrium thermodynamics". Although Deleuze does not explicitly refer to this new branch of science, it is clear that far-from-equilibrium thermodynamics meets all the objections which he raises against its nineteenth century counterpart. In particular, the systems studied in

this new discipline are continuously traversed by a strong flow of energy and matter, a flow which does not allow the differences in intensity to be cancelled, that is, maintains these differences and keeps them from cancelling themselves. It is only in these far-from-equilibrium conditions that the full variety of immanent topological forms appears (steady state, cyclic or chaotic attractors). It is only in this zone of intensity that difference-driven morphogenesis comes into its own, and that matter becomes an active material agent, one which does not need form to come and impose itself from the outside. To return once more to the example of the developing embryo, the DNA that governs the process does not contain, as it was once believed, a blueprint for the generation of the final form of the organism, an idea that implies an inert matter to which genes give form from the outside. The modern understanding of the procesess, on the other hand, pictures genes as teasing out a form out of an active matter, that is, the function of genes and their products as now seen as merely constraining and channeling a variety of material processes, ocurring in that far-from-equlibrium zone, in which form emerges spontaneously. To complete my characterization of Deleuze theory of the genesis of form, I would like to explore the way in which his more recent work (in collaboration with Felix Guattari) has extended these basic ideas, greatly increasing the kind of immanent resources that are available to matter for the creation of form. In particular, in their joint book "A Thousand Plateaus", they develop theories of the genesis of two very important types of structures, to which they refer with the terms "strata" and "self-consistent aggregates" (or alternatively "trees" and "rhizomes"). Basically, strata emerge from the articulation of homogeneous elements, whereas self-consistent aggregates emerge from the articulation of heterogeneous elements as such. {3} Both processes display the same "divergent actualization" which characterized the simpler processes behind the formation of soap bubbles and salt crystals. In other words, in both processes we have a virtual form (or abstract machine, as they now call it) underlying the isomorphism of the resultant actual forms. Let's begin by briefly describing the process behind the genesis of geological strata, or more specifically, of sedimentary rock, such as sandstone or limestone. When one looks closely at the layers of rock in an exposed mountain side, one striking characteristic is that each layer contains further layers, each composed of small pebbles which are nearly homogeneous with respect to size,shape and chemical composition. It is these layers that are referred to as "strata". Now, given that pebbles in nature do not come in standard sizes and shapes, some kind of sorting mechanism seems to be needed to explain this highly improbable distribution, some specific device which takes a multiplicity of pebbles of heterogeneous qualities and distributes them into more or less uniform layers. One possibility uncovered by geologists involves rivers acting as sorting machines. Rivers transport rocky materials from their point of origin to the place in the ocean where these materials will accumulate. In this process, pebbles of variable size, weight and shape tend to react differently to the water transporting them. These different reactions to moving water are what sorts out the pebbles, with the small ones reaching the ocean sooner than the large ones. This process is called "sedimentation". Besides sedimentation, a second operation is necessary to transform these loose collections of pebbles into a larger scale entity: a sedimentary rock. This operation consists in cementing the sorted components, an operation carried out by certain substances dissolved in water which penetrate the sediment through the pores between pebbles. As this percolating solution crystallizes, it consolidates the pebble's temporary spatial relations into a more or less permanent "architectonic" structure. These double articulation, sorting and consolidation, can also be found in biological species. Species form through the slow accumulation of genetic materials. Genes, of course, do not merely deposit at random but are sorted out by a variety of selection pressures which include climate, the action of predators and parasites and the effects of male or female choice during mating. Thus, in a very real sense, genetic materials "sediment" just as pebbles do. Furthermore, these loose collections of genes can (like sedimented pebbles) be lost under some drastically changed conditions (such as the onset of an Ice age) unless they become consolidated together. This second operation is performed by "reproductive isolation", that is, by the closure of a gene pool which occurs when a given subset of a reproductive community, becomes incapable of mating with the rest. Through selective accumulation and isolative consolidation, a population of individual organisms comes to form a larger scale entity: a new individual species. We can also find these two operations (and hence, this virtual diagram) in the formation of social classes. Roughly, we speak of "social strata" whenever a given society presents a variety of differentiated roles to which not everyone has equal access, and when a subset of those roles (i.e. those to which a ruling elite alone has access) involves the control of key energetic and material resources. In most societies roles tend to "sediment" through a variety of sorting or ranking mechanisms, yet not in all of them ranks become an autonomous dimension of social organization. In many societies differentiation of the elites is not extensive (they do not form a center while the rest of the population forms an excluded periphery), surpluses do not accumulate (they may be destroyed in ritual feasts), and primordial relations (of kin and local alliances) tend to prevail. Hence a second operation is necessary: the informal sorting criteria need to be given a theological interpretation and a legal definition. In short, to transform a loose ranked accumulation of traditional roles into a social class, the social sedimement needs to become consolidated via theological and legal codification. {8} Is there also a virtual diagram behind the genesis of meshworks? In the model proposed by Deleuze and Guattari, there are three elements in this other virtual diagram, of which two are particularly important. First, a set of heterogeneous elements is brought together via an articulation of superpositions , that is, an interconnection of diverse but overlapping elements. And second, a special class of operators, or intercallary elements, is needed to effect this interlock via local connections. Is it possible to find instances of this diagram in geology, biology and sociology? Perhaps the clearest example is that of an ecosystem . While a species may be a very homogeneous structure, an ecosystem links together a wide variety of heterogeneous elements (animals and plants of different species) which are articulated through interlock, that is, by their functional complementarities. Since one of the main features of ecosystems is the circulation of energy and matter in the form of food, the complementarities in question are alimentary: prey-predator or parasite-host being two of the most common. In this situation, symbiotic relations can act as intercallary elements aiding the process of building food webs by establishing local couplings. Examples include the bacteria that live in the guts of many animals allowing them to digest their food, or the fungi and other microorganisms which form the "rhizosphere", the underground food chains which interconnect plant roots and soil. The world of geology also has actualizations of these virtual operations, a good example being that of igneous rocks. Unlike sandstone, igneous rocks such as granite are not the result of sedimentation and cementation, but the product of a very different construction process forming directly out of cooling magma. As magma cools down its different elements begin to separate as they crystallize in sequence, those that solidify earlier serving as containers for those which acquire a crystal form later. In these circumstances the result is a complex set of heterogeneous crystals which interlock with one another, and this is what gives granite its superior strength. Here the intercallary elements include anything which brings about local articulations from within the crystals, including nucleation centers and certain line defects called dislocations, as well as local articulation between crystals, such as events ocurring at the interface between liquids and solids. Thus, granite may be said to be an instance of a meshwork. In the socio-economic sphere, pre-capitalist markets may be considered examples of cultural meshworks. In many cultures weekly markets have traditionally been the meeting place for people with heterogeneous needs and offers. Markets connect people by matching complementary demands, that is, by interlocking them on the basis of their needs and offers. Money (even primitive money such as salt blocks or cowry shells) may be said to perform the function of intercallary element: while with pure barter the possibility of two exactly matching demands meeting by chance is very low, when money is present those chance encounters become unnecessary, and complementary demands may find each other at a distance, so to speak. Thus, much as sandstone, animal species and social classes may be said to be divergent actualizations of a virtual process of "double articulation" which brings homogenous components together, granite, ecosystems and markets are actualizations of a virtual process which links heterogenous elements through interlock and intercalation. These virtual processes are, according to Deleuze, perfectly real, a real virtuality which has nothing to do with what we call virtual reality. And yet, because this real virtuality constitutes the nuomenal machinery behind the phenomena, that is, behind reality as it appears to us humans, because this real virtuality governs the genesis of all real forms, it cannot help but be related to virtual realities, not only those created by computer simulations, but also by novelists, filmmakers, painters and musicians. Deleuze's work is, from the beginning, concerned as much with physics and mathematics, as it is with art. But it seems to me, only when we understand the Deleuzian world of material and energetic flows, and the forms that emerge spontaneously in these flows, can we begin to ask "what is a novel or a painting or a piece of music" in this world? In other words, the movement should be from a rich material world pregnant with virtualities, to literature or art, and not from literature (and texts, discourses, metaphors) to a socially constructed world where matter has once again, become an inert receptacle for external forms. It is in this sense, that Deleuze's work constitutes a true challenge to language-obsessed postmodernism, a neomaterialism which promises to enrich the conceptual reservoirs of both science and art and that one day could lead to a complete reconceptualization of our history as well as of our alternatives for the future.~ ========== and ../popefuller.htm WARNING: This Computer Has Multiple Personality Disorder by Simon Pope and Matthew Fuller Pandemonium is the complete system of Lemurian demonism and time sorcery. It consists of two principal components: Numogram (time-map) and Matrix (listing the names, numbers and attributes of the demons). The system is constructed according to immanent criteria latent in decimal numeracy, and involves only basic arithmetical operations (assembled from additions and subtractions). The Numogram, or Decimal Labyrinth, is composed of ten zones (numbered 0-9) and their interconnections. These zones are grouped into five pairs (syzygies) by nine-sum twinning [zygonovism]. The arithmetical difference of each syzygy defines a current (or connection to a tractor zone). Currents constitute the primary flows of the numogram. Each zone number when digitally cumulated defines the value of a gate, whose reduction sets the course of a corresponding channel. Channels constitute the secondary flows, time-holes, or secret interconnections of the numogram. The arrangement of currents divides the Maze into three basic time-systems. Firstly, the currents of the three central syzygies mutually compose a cycle, rotating in anticlockwise steps. Lemurian sorcery calls this inner loop the Time-Circuit. Secondly, and thirdly, in both the Upper and the Lower syzygies the currents produced fold back into (a half of) themselves, constituting autonomous loops: the Warp (upper), and Plex (lower). Warp and Plex circuitries are of an intrinsically cryptic nature, which is compounded by the enigmas of their interconnection. They are variously considered to be Outside- or Outer-time. The gates and their channels knit the Maze together, providing connections between otherwise incompatible time-systems. They open and close the ways of sorcerous traffic. Although each gate deranges time in its own way, their operations vary with a certain regional consistency. 1. Numogram and Otz Chaiim. To those familiar with the Western Magical Tradition, it is likely that the Numogram will initially evoke the Qabbalistic Tree of Life. Both are constructed as decimal diagrams, involving webs of connectivity between ten basic zones, mysteriously twisted into a cryptic ultra-cycle (that links upper and lower regions). Both treat names as numbers, and numerize by digital reduction and cumulation. Both include passages across abysmal waters and through infernal regions. Both map zones onto spinal levels. Despite these manifold interlinkages, there are compelling reasons to consider the Tree of Life a scrambled variant of the Numogram, rather than a parallel system. During its long passage through Atlantean and post-Atlantean hermetic traditions the systematic distortions of the Numogram (introduced to confuse the uninitiated) gradually hardened into erroneous doctrines, and a dogmatic image of the Tree. Most evidently, a vulgar distribution of the numbers - in their exoteric counting-order - was substituted (redundantly) for the now esoteric numogrammatical distribution, which proceeds in accordance with immanent criteria (the web emerging qabbalisitically from the zone-numbers themselves). More devastatingly, the orginal consistency of numeracy and language seems to have been fractured at an early stage, introducing a division between the number of the Sephiroth (10) and that of the Hebrew alphabet (22). The result was a break between the nodes of the tree and the interconnecting paths, ruining all prospect of decipherment. The Sephiroth -segmented over-aganist their connections - become static and structural, whilst the paths lose any rigorous principle of allocation. A strictly analogous outcome is evident in the segmentation of the Tarot into Major and Minor Arcana. Increasingly desperate, arbitrary, and mystifying attempts to re-unite the numbers and their linkages seems to have bedevilled all succeeding occult traditions. 2. Numogram and I Ching. There is considerable evidence, both immanent and historical, that the chinese I Ching and the Nma numogram share a hypercultural matrix. Both are associated with intricate zygonomies, or double-numbering systems, and process abstract problematics involving subdivisions of decimal arrays (as suggested by the Ten Wings of traditional I Ching commentary). Digital reduction of binary powers stabilizes in a six-step cycle (with the values 1, 2, 4, 8, 7, 5). These steps correspond to the lines of the hexagram, and to the time-circuit zones of the Numogram, producing a binodecimal 6-Cycle (which is also generated in reverse by quintuplicative numbering). In both cases a supplementary rule of pairing is followed, according to a zygonovic criterion (9-twinning of reduced values: 8:1, 7:2, 5:4, mapping the hexagram line pairs). The numogram time-circuit, or I Ching hexagam, implictly associates zero with the set of excluded triadic values. It is intriguing in this respect that numerous indications point to an early struggle between triadic and binary numbering practices in ancient chinese culture, suggesting that the binary domination of decimal numeracy systematically produces a triadic residue consistent with nullity. The hexagram itself exhibits obvious tension in this respect, since it reinserts a triadic hyperfactor into the reduced binodigital set (compounded by its summation to twenty-seven, or the third power of three). An ancient binotriadic parallel to the I Ching, called the T'ai Hsuan Ching (or Book of the Great Dark) consisted of eighty-one tetragrams, reversing the relation of foregrounded and implicit numerical values. The division of Lao Tse's Tao Te Ching into eighty-one sections suggests that this numerical conflict was an animating factor in the early history of Taoism. 3. Ethnography of the Nma. Nma culture cannot be decoded without the key provided by the Lemurian Time-Maze. The influence of a hyper triadic criterion of time is evident in the relics of Nma kinship organization, calendrics, and associated rituals. Prior to the calamity of 1883, the Nma consisted of true tribes (tripartite macrosocial divisions). They were distributed in a basic tridentity (interlocking large-scale groupings into Tak- Mu- and Dib-Nma), supported by a triangular patrilocal marriage-cycle. Each marriage identified a woman with a numogram current, or time-passage. (Tak-Nma women marrying into the Mu-Nma, Mu-Nma ditto Dib-Nma, Dib-Nma ditto Tak-Nma). The common calendar of all three tribes was based upon a zygotriadic system (using 6 digits to divide a double-year period of 729 days into fractional powers of three). The Mu-Nma still employ such a calendar today. (The current Mu-Nma calendar is adjusted by regular intercalations of three additional days every second cycle, or four years. The earlier practice of intercalations is not easily recoverable). In the rituals of the Nma the time-circuit is concretized as a hydro-cycle: a division and recombination of the waters. The three stages of this recurrent transmutation are, 1) the undivided waters (oceanic), 2) cloud-building (evaporation), and 3) down-pour (precipitation, river-flow). These are associated with the great sea-beast (Mur Mur), the lurker of steaming swamps (Oddubb), and that which hunts amongst the raging storms (Katak). The cycle is closed by a return to the abysmal waters, intrinsically linking the order of time, and its recurrence, to an ultimate cataclysm (prior to any opposition of cyclic and apocalyptic time). It is in this context that the transcultural deluge-mythos can be restored to its aboriginal sense (which also corresponds to the Hindu Trimurti, with its three stages of creation, preservation and destruction). 4. The Numogram Zones The Zones. Zone Zero. Zone One. Zone Two. Zone Three. Zone Four. Zone Five. Zone Six. Zone Seven. Zone Eight. Zone Nine. Ccru is committed to an ongoing research program into the numeracy of the 'lost lemurian polyculture' apparently terminated by the KT missile of BCE 65 000 000. During the last century, various aspects of this primordially ancient 'digital hyperstition,' 'mechanomics,' 'schizonumerics,' or 'numbo-jumbo' have been painstakingly re-assembled through certain cryptic investigations, pre-eminently those associated with the names Echidna Stillwell, Chaim Horovitz, and Daniel Barker. From the Mu-Archive in Tibet Horovitz unearths an 'ultimate decimal qabbala' oriented to the cultic exploration of the numerals zero-to-nine as cosmic zones. In contradistinction to the late-Babylonian (or Judeo-Christian) qabbala, the 'method of Mu' involves a rigorous collapse of transcendent symbolism into intrinsic or immanent features, excavating the latent consistency between the numerical figures, their arithmetic functions, and their cultural associations. Horovitz describes these procedures as a diagonal path between esoteric numerology and exoteric mathematics, and also defines them negatively as a 'non-numerology' or 'ulterior-arithmetic.' Atlanto-Babyonian State-societies preserved some of the most fully degraded late-Muvian conclusions, but only by assimilating them to a 'Gnostic Arithmetic,' fossilizing the numbers into spiritual beings, ideal individuals, and general concepts. Within these familiar traditions the sense of the numbers as raw functions of cosmic distribution was systematically subordinated to magical and religious principles, whilst their intensive potentials as transmutational triggers was drained-off into geometrical structures and logical representations. The productive synthesis of Stillwell's numogrammatic researches with Barker's 'tic-systemic' approach provides the requisite cutting-tools for re-opening the virtual-numeric labyrinth. This involves the re-activation of those 'lemurian' cultural practices which traffick with numbers as techno-sorcerous entities: the diagrammatic tokens, intensive thresholds, cosmic coincidences and hyperstitional influences that populate the plane of Unlife. Ccru has collated material from a series of occultural investigations that demonstrate the virtual existence of a lost lemurian art of interplanetary communication, or ‘planetwork.' This system maps the major bodies of the Solar-system onto the ten digital labyrinth Zones (beginning from Sol = 0). The numerals one to nine function as astronomical ordinals, designating the terms of the planetary sequence in ascending order of orbital span (mean distance from the sun), orbital period (local year length), and gravitational attenuation (einsteinean spatial flatness). This heliocentrism (with its implicit repudiation of terrestrial phenomenology) does not contradict the broad counter-solar trend in lemurian culture, with its repulsion of centralization and gravitational capture. There has never been a lemurian solar cult. Lemurian Planetwork communicates with the substellar bodies as distributed hyper-intelligences exerting singular influences (or ‘Barker-traces'). These planetary forces of suggestion are propagated through contemporary mythologies, systematic coincidences, and accidental scientific fictions (whether lunar seas, martian canals, jovian monoliths, or life on Europa). Various cryptic records indicate the existence of considerable calendrical lore based upon the Planetwork system, yet little of this has been definitively reconstructed. What is certain is that it takes the mercurian year for its basic unit, and uses this regular beat in the calendrical interweaving of (nonmetric) speeds and slownesses. The Zone-Sequence -by mercurian periods- with planetary attributions: Zn-0 [0000.00] Sun Zn-1 [0001.00] Mercury Zn-2 [0002.55] Venus Zn-3 [0004.15] Earth Zn-4 [0007.95] Mars Zn-5 [0049.24] Jupiter Zn-6 [0122.32] Saturn Zn-7 [0348.78] Uranus Zn-8 [0684.27] Neptune Zn-9 [1028.48] Pluto Many tales tell of a lemurian hyperstition composed of numbers that function as interconnected zones, zone-fragments, and particles. With Stillwell's epoch-switching discovery of the Numogram - and subsequent mapping of this 'digital labyrinth' - it became possible to compile cartographies of these zones, in which numbers distribute themselves throughout tropics, clusters, and regions. The zones thus function as diagrammatic components of flat cosmic maps (variously charting systems of coincidence, nebular circulations, spinal nestings, and the folds of inner/outer time). Amongst numerous systematizations of occult cartography that of Chaim Horovitz (direct descendant of the infamous 'mad rabbi of Kiev') is especially remarkable. Based upon lemurian digital relics extracted from the Mu-Archive, it enables the conversion of numogram-zones (and sub-zones) into cascade-phases, accessed through numerical 'doors.' The Horovitzean phases constitute qabbalistic groupings or cross-sections of the pandemonium population (simultaneously numbering the impulse-entities and defining their collective 'tone'). Those critics who seek to reduce Horovitz's work to an 'immensely indirect rediscovery of Pascal's triangle' fail to appreciate either the true antiquity of 'Pascal's' system or the machinic novelty of it's Horovitzean reanimation. Systematic issues concerning the Numogram Gates have been separated out from the other interconnective features of the zones. It has been known since the dawn of occult cartography that every Zone supports a Gate, and that their corresponding channels spin the web of esoteric fibres. All sorcerous cultures involve themselves in the exploration of these paths. A Sarkonian mesh-tag is provided for each zone as a key to Axsys-format and Crypt-compatibility. The modern figure zero (0) is a closed circle, oval, or ellipse (sometimes differentiated from the letter ‘O' by a diagonal slash). Its archaic Hindu form was the ‘bindu' (or dot, retained in the modern system as the ‘decimal point'). Both of these ciphers are of such abstraction that no rapid summary can be other than misleading. The figure ‘0' designates the number zero, anterior to the distinction odd/even, and also to the determination of primes (zeroth prime = 1). Zero is the only natural number that is indivisible by one. The division of any number by zero produces an infinity (multiplication of any number by zero = 0), in this respect zero treats itself as any other number. Zero digitally cumulates to zero. Numeric Keypad direction: anomalous. As an arithmetical function zero is strongly affined to place value - or ‘positional' - systems in which it typically operates as the designator of empty magnitudes. The modern decimal and binary systems are the most familiar examples of such modular numeracies. (The widespread assumption that such a zero-function is indispensable to any possible place-value numeracy is, however, a fallacious one). On the number line zero marks the transition from negative to positive numbers. In modern binary code zero is instantiated by electronic ‘off' (see One). In set theory zero corresponds to the null (or empty) set. In coordinate geometry zero marks the ‘origin' or intersection point of all dimensional axes, and is marked uniquely across all dimensions. In game theory a zero-sum game is one in which all gains and losses are mere redistributions (‘I win, you lose' or inversely). In Boolean algebra zero symbolizes logical negation. Absol