Learning from “A New Kind of Science”:

Can  we decode the program for world system evolution?





In the first chapter of his impressive volume, a chapter entitled “Foundations for a New Kind of Science” Stephen Wolfram reviews a number of fields

to which his work might be relevant but he is opines that  “no doubt there will also quite quickly be made all sorts of claims about applications of my ideas to the social sciences” and warns that some fundamental limits may confront such an endeavor  (2002:9).   In fact in the main body of his work he has very little to say about the social world except for some brief remarks on the randomness of fluctuations in stock market prices.


                        What follows is a brief attempt to bring out the relevance of Wolfram’s approach to the understanding of the social organization of the human species (the world system), and   the way that organization has evolved over long historical time (la longue duree) in a process that produces globalization.   It is now a matter of record that the paper just published (Devezas and Modelski 2003), and written before Wolfram’s book had appeared, shows a degree of convergence of the two approaches.  “World System Evolution” deals, of course, with a subject of more limited scope than Wolfram’s sweeping picture of a new kind of science (large parts of which remain to be filled out) but the analogies are worth pointing out, and exploring, because the subject matter is obviously important.


                        Table 1 summarizes the main lines of analogy, and contrast, between the two approaches, that one that is based on an abstract system, that of Cellular Automata(CAs), and the other, that models a natural system, the evolution of social structuring on a global scale (globalization).  


Both systems concern the behavior of populations of discrete entities and in that sense both have a common form;   but the substrate differs in

the two cases, one being an abstract, and the other, a natural system.  We take the world system to be comprised of a population of acting units, the interactors, and a set of replicators that encode, and transmit, information between the generations of interactors thus forming the lineages that link these generations.


The focus of analysis in both types of system is on change, that is on the process that transforms the system over time.   In the case of CAs,

Wolfram regularly uses the term “evolution” in reference to the pattern of change that a particular system is undergoing (as e.g. rule 30 at 2002:28-30) but he does so in a non-technical sense, synonymous with change, and. elsewhere, as starting on p.14, he tends to minimize the role of Darwinian theory, doubting that natural selection might be responsible for much of the complexity observed in biology or elsewhere.  In the case of world system, the process is explicitly and technically) evolutionary (that is involving at the minimum both variation and selection).   In fact we postulate the existence of a “cascade” of such self-similar processes, each undergoing phased transformation but at different albeit synchronized, rates.




Table 1:  Comparing Wolfram’s Cellular Automata with World System Evolution


Cellular Automata                                                               World System Evolution


CA systems are populations of discrete elements                     World system is a population

                                                                                            of interactors and  replicators in lineages


System behavior is process of change                                     System passes through phases in a

via discrete steps or stages                                                    cascade of evolutionary processes


Rules governing system behavior constitute                              Each period of a

a program implementing laws of nature                                    (self-similar) evolutionary process implements

        the learning algorithm, given specific initial conditions


Simple programs show repetition and nesting                            World system processes exhibit repetition, nesting, and branching

(including branching)                                             


Simple programs can produce great complexity of behavior        World system exhibits complex     behavior                                                            


System behavior corresponds to a computation    



                        Each system undergoes a process of change, and that process is rule-governed.  The rules constitute a program, and each system changes in a programmed fashion.   According to Wolfram, (2002:715) “the rules such processes followed … are defined by the basic laws of nature”.   World system evolution proceeds according to the rules of evolutionary theory and, more specifically, those of “universal Darwinism” that specifies the elements of a learning algorithm.   Each period of the several (and self-similar) processes in the “cascade” implements that algorithm, given appropriate initial conditions.   Power-law behavior of world system evolution (Devezas and Modelski 2003) indicates the presence of self-similarity.


                        Wolfram defines as simple systems those that show behavior that is either repetitive, or nesting.   Branching, too, is a form of nesting.   Similarly, repetition,  and nesting, are central features of world system evolution  (Devezas and Modelski 2003).    Each period of the evolutionary learning process repeats once its phase structure is completed.   The longer-period evolutionary processes moreover, nest within shorter-period ones.   For example, the economic processes nest within political processes that are twice the length of economic processes.  Moreover, world system evolution as a whole might be seen as a branching process where in for instance, at a point in time, world system processes develop global and other branches.  That would suggest that the basic make-up of world system evolution is, in Wolfram’s terms, “simple”.


The central insight of Wolfram’s work is the frequently reiterated assertion  that ‘simple programs can produce great complexity ‘.   If, then, world

system evolution, and globalization, arguably show repetitive, and nesting, behavior, and therefore also represent a “simple” system, working to simple rules, then it is conceivable that the complexity of the world system might indeed be produced by a set of simple programs.    That would tend to lend encouragement to those who seek to uncover such a program, and support the concept of a “learning algorithm”.


                        Finally, Wolfram maintains that the behavior of the systems he analyzes is “a computation”:  “all processes, whether they are produced by human effort or occur spontaneously in nature, can be viewed as computations”, producing “a fundamental equivalence between many different kinds of processes” (2002:715,716).   The question might then arise:  is world systemevolution computationally irreducible (that is, capable to making complete predictions by making a sufficient number of computations)?   The answer, on Wolfram’s terms, might be that “if the behavior of the system is obviously simple – and is say either repetitive or nested – then it will always be computationally reducible” (2002:741).   In other words, since the world system shows repetition and nesting, and is therefore “simple”, its behavior might be subject to prediction.


                        To sum up, two important suggestions might be culled from Wolfram’s work::  that world system evolution, despite the apparent complexity of its results, might be powered by a simple program, and that world system behavior might in principle (and in some respects) be predictable, once its program is decoded.  The world of possibilities is enlarged but the hard work remains to be accomplished.   In other words, this work empowers researchers to search for simple programs but it does not supply ready-made solutions for students of  i.a. globalization.  And, of course, it does not guarantee that such programs will be simple, or predictive.   In relation to social problems  the tone of Wolfram’s conclusions is unusually restrained::  “there can be no abstract basic science of the human condition;  only something that involves all sorts of specific details of humans and their history” (2002:846).


                        For as Ray Kurzweil (2002:9,5) has pointed out  “…cellular automata on their own do not evolve sufficiently.  They quickly reach a limited asymptote in their order of complexity.  An evolutionary process, involving conflict and competition” [and, we might add, cooperation]  is needed”.   More specifically , by adding “an evolutionary algorithm, we start to get far more interesting, and more intelligent results”.


                        The question remains:  can we decode (and write) a program for globalization (world system evolution) that produces both repetition, and nesting and that, in favourable conditions, evinces evolutionary behavior matching long-run historical processes? 



Devezas, T. and G. Modelski  (2003)   “Power Law behavior and World System Evolution”   Tech. Forecasting and Soc. Change,  October.

Kurzweil,  Ray   (2002)   “Reflections on  Stephen Wolfram’s “A New Kind of Science”   at www.kurzweilai.net/articles/art0464.html

Wolfram,   S.   (2002)   A New Kind of Science,   Wolfram Media.       



December 28, 2003