Customer Reviews

Complexity and Postmodernism: Understanding Complex Systems

5 star:		(7)
4 star:		(5)
3 star:		(1)
2 star:		(2)
1 star:		(1)

 

 

Cilliers has undertaken an important job - exploring the linkage between complexity thinking and postmodernism. He has made excellent use of some main writers on postmodernism and shown some important relation to studies of representation and self-organizing systems. He works hard to help us escape the locked-in positions of positivistic and foundationalist science, but his major conceptual base in connectionism displays an unabashed modernist view. While connectionism is an important tool in exploring the ideas about how the mind/brain works, it ignores other important ideas arising from the work of Maturana/Varela and Niklas Luhmann on auto-poiesis and John Holland on complex adaptive systems. More significantly, Cilliers is locked into the ideas of networks. It is a valuable tool for the technological advances, but for a full philosophical exploration he undertakes, we needs also to look at field thinking, particularly that arising in quantum fields discussion such as in Sunny Auyang' work.

What I find most difficult in Cillier's retention of the modernist view of competition. Our cultures may be agonistic but is competition fundamental to the development of human life?

I read this book primarily through an interest in the philosophy of language. Of particular relevance in this respect is the emphasis on a characterisation of complexity as being opposed to traditional notions of representation. Cilliers draws parallels between the philosophy of Saussure and Derrida and scientific developments in distributed representation, particularly with respect to connectionist approaches as implemented in neural networks. Cilliers argues that a classical representational theory of language that posits syntax as an instantiation of semantics does not sufficiently allow for the complexity evident in language, but rather that meaning is constituted by the dynamic relationships between both the components of language and the environment in which it is embedded. Cilliers explicitly rejects rule-based symbol systems as being adequete for modelling language, referring to recent scientific research using neural networks to simulate language learning indicating that "though rules may be useful to describe linguistic phenomena, explicit rules need not be employed when language is acquired or when it is used" (p. 32). In Chapter 4 (pp. 48-57), Cilliers considers the Chinese Room Gedankenexperiment from the perspective of his thesis. He suggests that the debate has unquestionably assumed that the formal model of language represented by the argument is correct, that is, that a rule-book such as the one supposed is even possible. Cilliers suggests that this assumes certain features of language: that a formal grammar for a natural language can be constructed and represented in a lookup table; that there is a clean split between syntax and semantics; and that language represents rather than constitutes meaning (p. 53).

The overall picture of language that Cilliers develops has important parallels with the views of Wittgenstein, though, somewhat surprisingly, Wittgenstein is never explicitly mentioned (except with regard to his family concepts). Firstly, meaning is construed as occuring through dynamic processes (use) rather than static representations (the conception that Wittgenstein's private language argument criticises). Secondly, the idea that there is some fact of the matter (whether inside or outside human agents) that determines meaning is explicitly rejected. Finally, a straightforward split between syntax and semantics is denied (a distinction that the sceptical interpretation of Wittgenstein, offered by Kripke, takes advantage of).

In summary, I would recommend this book to anyone interested in making connections between dynamic systems theory and philosophy of mind or language -- Cilliers proves an effective communicator in both of the fields he wishes to connect.

This work is essential for a cutting-edge understanding of how two independently cultivated lines of investigation - complexity and postmodernism - have fortuitously dovetailed, providing us with a new level of perspective upon the character and evolution of contemporary technology. I highly recommend reading this work in tandem with Don Ihde's groundbreaking study EXPANDING HERMENEUTICS: VISUALISM IN SCIENCE, itself a phenomenologically well-grounded yet visionary exposition of where the computer-inspired "visual turn" in hermeneutics is leading us in the 21st century.

Frankly, I'm astonished by some of the favorable reviews this book has received. First of all, I still haven't figured out if this really is a book or if it's a collection of essays, due to the amount of repetition of content between chapters.

Cilliers attempts to demonstrate the mutual relevance of complexity science (CS) and postmodern philosophy, but his knowledge of CS and thermodynamics seems to go no deeper than what he's read on the dustjackets of pop-sci books. The number of claims he makes that are either blatantly false or not necessarily true are outnumbered only by the number of uninsightful comments and statements that appear to have been gleaned directly from more technical sources. Here are a few to make one's skin crawl:

On p. 6, as an example of a non-linear relationship: "money can receive compounded interest". In fact, this is a classic *linear* relationship (so common it's often used as an introductory problem the first day of a course in linear differential equations). The equation representing it is simply: dM/dt = n*M, where M is the amount of money in an account, and n is the interest rate. The solution is Mo * e^(nt), where Mo is the initial amount of money in the account and 'e' represents 'exponential'. (Simply because compounded interest generates an exponential curve over time does not make the relationship non-linear; the underlying equation is linear.)

On p. 4: "Any analysis of a complex system that ignores the dimension of time is incomplete, or at most a synchronic snapshot of a diachronic process." This is completely false - One of the very purposes of 'phase space' analysis is to *completely* represent a system without considering time. The elliptical relationship between velocity and momentum in a simple harmonic oscillator is a common example that many might remember from high school physics.

On p. 8: "In classical mechanics, time was reversible, and therefore not part of the equation. In thermodynamics time plays a vital role." This quote still makes me tear at my hair. The *exact opposite* is true: almost every equation in classical mechanics (projectile motion, harmonic oscillation, planetary motion) explicitly involve time as a dimension, while, because thermodynamics is only concerned with initial and final (equilibrium) states, few thermo equations do so.

On p. 3, Cilliers says: "The grains of sand on a beach do not interest us as a complex system." but includes later in the book a quote from complexity scientist Per Bak, who has achieved his fame specifically for the study of the 'self-organized criticality' of sand grains.

And this is just the first few pages! The list goes on and on: He repeatedly confuses the thermodynamic concepts of 'closed' and 'isolated' systems; He seems to think that 'non-linear' equations are all somehow phenomenally complex and unsolvable and that the phrase 'non-linear' is therefore a synonym for being non-reductionist, non-rational, and, in short, 'postmodern'. (In doing so, he falls into many of the traps Alan Sokal identified in Fashionable Nonsense.)

I think that the basic concept behind the book could have been interesting, but due to Cilliers elementary-level grasp of half the subject matter with which he deals, the statement Cilliers himself makes on p. 133 (in reference to a recent book by Rouse) applies equally well to this text: "For me, reading this book was about as pleasant as it would be to eat it."

Prof. Cilliers's elucidation of the key elements of complexity theory is not only informative but fascinating reading. He has taken two subjects (complexity and post-modernism), each of which can be frustrating and confusing to the average reader, clearly explained them, and then convincingly related them to each other. By describing each of these subjects in the context of the other (in true post-structural style), Prof. Cilliers makes each of them more understandable. Highly recommended!

As a professional engineer with a strong interest in postmodern philosophy I identify closely with the author and I am very amazed at how he could relate the extremely abstract concepts of post-structuralism with the more concrete example of neural networks. His unmasking of the metaphysics of representation that underlies current research in artificial intelligence was a great insight for me. At only 142 pages, this book seemed very inviting and thus I bought it. But don't be misled, what this book lacks in length is more than made up by its density. For me, who prior to this had only read introductory books on postmodernism and had only vague notions about connectionism and neural networks, it turned out to be extremely challenging and demanding to read, and completing it gave me a sense of achievement similar to being done with a hard project. I think some parts were unnecessarily abstract, which, knowing the author's talent for making analogies and examples, felt like a disappointment. Other parts, such as his comments on postmodern ethics, simply begged for further elaboration or at least to references on the works of others in this field. I think I will return to this book once I read more on Derrida and Lyotard for a better understanding. I really hope that by then the author will have come out with a sequel to this very interesting and groundbreaking line of work.

Though it helps to have a better understanding of the modern philosophy of science, it is not necessary in order understand the idea the book presents to the reader. It is very well thought out, cogent and I would recommend it to anyone who is interested in cross-disciplined approaches to understanding concepts. I appreciate the process oriented nature of his thesis and how the classic Newtonian physics are inadequate to defining our rapidly changing universe. Be prepared to think, but prepare yourself for a good adventure.

Because they are structures of relationships and patterns, non-linear, dynamically self-organizing processes such as those described by Stuart Kauffman and Ilya Prigogine are characterized by distributed control and non-linear feedback, adapting and complexifying in response to interaction with their environment. Unlike Newtonian processes, self-organizing structures are essentially historical and contextual. More importantly, such complex dynamical systems are describable only at the holistic level, in terms of their higher-order relationships and patterns. For all these reasons, they are not reducible to the sum of their component parts.

Cilliers argues brilliantly that the existence and importance of complex systems suggest that Derrida and Lyotard's post-structuralism "is not merely a subversive form of discourse analysis but a style of thinking that is sensitive to the complexity of the phenomena under consideration." If human beings and their societies exhibit dynamics homologous to those of complex biological and chemical processes, postmodernism's rejection of all metanarratives --including ethical metaprescriptions -- is simply an echo of Prigogine's phrase, "Nature is too rich to be described in a single language." However, Cilliers points out, "the fact that there are manay narrative paths... doesnot imply that anything goes. All narratives are subject to some form of constraint, and some paths are ruled out" by the historical and contextual framework which they describe. Knowledge and ethics are always local and tentative.

When read in conjunction with Andy Clark's Being There, and Dynamics in Action, by Alicia Juarrero,a new conceptual philosophical framework based on complexity theory seems to be in the making! An exciting and revolutionary triad!

An excellent well-rounded introduction to complexity, presented in a clear and well-referenced manner. Cilliers outlines the philosophical underpinnings for many of the central issues in complex systems. The introductory chapter includes the most straightforward explanation of the properties of complex systems that I've seen (far better than Lewin, Waldrop, or Gell-Mann). A must-own for anyone interested in complexity or cognitive and neural systems, or philosophy of science for that matter.

POINT OF VIEW

This book has got my attention by the recommendation of a friend, from South-Africa, pointing to a certain author living in South-Africa. And because my subjects are matching in various ways with the topics of complexity and post-modernism, I started reading this book.

PERSONAL SUMMARY

For me was the book -- at one hand -- extremely stimulating because Cilliers presents in his book not only the classical AI topics but tries to reflect these positions in the light of many modern philosophical 'schools' of analytical philosophy, cognitive science as well as in the light of structuralism, post-structuralism, and post-modernism. This is a really strong point. What was confusing and even disappointing is the sloppy argumentation in theoretical issues. From this I have inferred the hypothesis that his main thesis is probably completely wrong. Despite this strong critic I esteem the book a lot and would recommend reading it.

Here a few more comments. The book is too rich to talk about all interesting topics.

MAIN THESIS

Cilliers thinks that 'complexity' is a fundamental property of the world we are living in and that traditional methods of science and analytical philosophy are not sensitive enough to complexity. He talks explicitly and in a critical manner about methods like rule-based (symbolic) Artificial Intelligence, analytical method, deductive logic, atomism, formal rule-based grammars , closed algorithms and symbolic representation. As only alternative he proposes a connectionist approach which he uses as an example concept which is understood to realize the main ideas of 'post-structuralism', of 'post-modernism' as well as of complexity in general.

COMPLEXITY?

The weak point of the book starts right in the beginning where the term 'complexity' has to be introduced. Stating, that the phenomenon of 'complexity' is not merely determined by the point of view of the observer (3), this assumes, that there is some 'objective reality' 'outside' and 'independent' of the observer. This represents a philosophical position which contradicts nearly everything which the author wants to bring forward in the rest of the book (post-structuralism, post-modernism). In the following the reader will not find a real working definition (2), but only circumscribing characterizations of different properties (3-5) which are loosely exemplified by taking phenomena from the field of economy (5). Other examples of complex systems to which the book points are brain and language (5). Thus, the author describes 'complex systems' as something which exists outside and independent of an observer in an 'informal language'. This informal text can perhaps be understood as a post-structural 'narrative', which allows different interpretations which makes a detailed examination difficult, if not impossible. 'Shallow meaning' as new kind of science?

There are much more details to this topic which I leave out here (e.g. the author mentions the concept of 'entropy' in the context of thermodynamics (Boltzmann as well as information theory (Shannon) or the concept of 'randomness' as 'densenes' in the context of algorithmic information theory (Chaitin), but without giving a clear theoretical account how these concepts are connected and what they have to do with complexity. He only states that complexity has not be confused with 'randomness' and 'chance' (109), and, that complexity is much more than 'chaos' (98).

From all the before mentioned characterizing properties of complex systems two are especially important for the author: 'presentation' (storage of information) and 'self-organization' (10).

RULE-BASED OR CONNECTIONIST SYSTEMS?

Raising the question, how one could 'model' complex systems the author mentions so-called 'rule-based' systems (also called 'symbolic AI' 13-15) as illustrated by formal systems in general, as well as expert systems, turing machines, and formal languages. These are than compared to so-called connectionis models (16-18) which are inspired by the biological brain as a network of neurons. In such networks are the 'important informations' represented as 'weights' attached to the connections. The 'learning' of such networks follows e.g. Hebb's learning rule, later improved by the generalized delta-rule. Looking to a single element reveals no reasonable behaviour, only observing the 'whole' network.
This kind of argumentation can be found in many publications, but this argumentation is nevertheless misleading and even wrong. Taking here only one point of many: The turing machine is mathematical concept allowing the formal presentation of every kind of computable process, which is known until today. Every kind of a concrete neural network can be simulated within the formal concept of the turing machine. Thus the formal concept of the turing machine can be used as a 'meta-concept' where neural networks are possible examples, 'instances' of this concept. Moreover, it is possible to simulate an instance of a turing machine within the formal (! concept of a neural network. This shows the formal equivalence of of both concepts. In an analogue manner it is possible to show that an implementation of a certain classical rule-based system for some learning tasks which appears to be 'cumbersome' or 'inflexible' is no argument against formal models and theories in general. One can represent any kind of connectionist system within a formal system including all the know dynamics. Thus the contradiction is not formal vs. connectionist, but between different 'implementations' of formal systems compared to neural networks as another kind of an implementation.

REPRESENTATION AND POST-STRUCTURALISM

There are many places where Cilliers talks about representation in his book (e.g. 11, 30, 32f, 58ff), one important section seems to be the analysis of the position of Saussure (38ff) as well as the following discussion by Derrida (42ff).
Saussure introduces within his 'Cours de linguistique général' the concept of 'sign' as a relation between an acoustic expression -- internally represented as an acoustic 'image' -- which signifies something else -- called a 'concept' --, which is 'signified' by the signifier and which represents the 'meaning' of the signifier within the sign-relation. While the individual signifier-event is local to the sign-user, the 'language as object of science' is only given in the set of relationships between all signifier. This implies 'differences' between the individual signifiers.
While Saussure in his writings keeps the signified as a kind of 'counterpart' to the systems of signifiers does Derrida switch from the priority of the spoken language to the written language and does cut the dependency of the signifier of the signified. Meaning is now declared 'independent' from the subject, the consciousness loses it's control about meaning and language, language becomes subject-free, a system of 'differences' and 'traces' (where 'traces is not really 'defined'), and as a consequence of this, not the subject controls now language but the language controls the subject (e.g. 43).

CONCLUDING REMARKS

As stated above I have left out many exciting topics. Only one concluding remark:
From the point of view of the reviewer is the deconstruction of Saussure by Derrida a nice 'game of concepts', but finally it doesn't work. It is in 'contradiction' with it's own assumptions. Within the modern research of language-learning artificial agents one can see that Saussure 'works', Derrida not. Nevertheless seems Derrida's position very stimulating and instead of bringing metaphysics to an end -- as Derrida wanted to show -- it is not unlikely now, that we will get Metaphysics back, much more brilliant and forceful than ever....