Fake Banner
    A Pattern-oriented Approach to Complexity (PAC): Patterns.
    By Kees Pieters | March 24th 2011 10:21 AM | Print | E-mail | Track Comments
    About Kees

    With a Ph.D.from the University for Humanistics on complexity, complex systems and technology and society, Kees holds degrees in electrotechnics...

    View Kees's Profile
    In the first post I introduced a view on 'complexity thinking' which presupposes a life-world where ambiguity, uncertainty, indeterminacy and other hard to grasp concepts play an important role, both in a creative as in a destructive way.

    1. Introduction
    2: Patterns

    This life-world also allows for the formation (organisation) of  order, and a methodology of complexity should offer some assistance in the recognition of these orderly phenomena. PAC uses a pattern-orientation to 'make sense' of this life-world, and this sense is one of many other scientific methodologies that take this approach.

    Patterns were pioneerd in the 1970's by the Austrian-British architect Christopher Alexander, who saw certain regularities in the various (medeaval) towns he visited and realised that many of regularities 'made sense'. A town square, for instance, had a number of functions, and a good designed town square facilitated these functions optimally. As a result, (well-designed) town squares of different cities would tend to have certain similarities, which Alexander dubbed 'patterns'. But these patterns were not restricted to town squares alone, for any well-designed public meeting place would share patterns with this aspect of a town square. Hence, the underlying pattern is 'A Place to Wait'. Alexander collected well over 250 patterns in the pattern library that he described in his seminal book A Pattern Language: Towns, Buildings, Construction, which is still considered a 'must-read' in building architecture.
    In 1987 this idea was taken over for software architecture by Kent Beck and Ward Cunningham, and was hyped by the extremely popular book
    by the 'Gang of Four' (GoF). Currently it is the predominant means of designing software architecture.
    The success of patterns has resulted in many attempts to apply these ideas in other areas, but as a methodological construct, it is a bit problematic that the notion of pattern is usually only used in a colloquial sense. There is no real 'metaphysics of patterns' so to speak. With the help of a, for me, very important article by Volker Grimm and his colleagues  in Science Magazine, I tried to fill in some of the gaps, and came to the following conclusions:
    • You can speak of 'patterns' when they can be observed in various situations (duh)
    • They are substrate -neutral, which means that they express themselves in different media
    • Patterns have a structure.
    • Patterns are building blocks; new patterns can be assembled from simpler ones
    • Simple patterns express themselves more widely than complex ones, and in a larger amount of different media. Mathematics can be seen as a collection of patterns that have a wide applicability.
    So far, this seems rather intuitive. But Patterns also have some other qualities that are hard to reconcile with the above list:
    • They are vaguely familiar phenomenon
    • They can express themselves differently in different media
    One one side, they seemed to be quite clear, and on the other they were vague. Then I realised that maybe there are two different forms of patterns, of which the first -contextual patterns- are related to what we already know (or have seen before). Every moment we are bombarded with sensory input (qualia) and our brains try to make sense of all this by making selections and mapping the qualia with that what is stored in our brain. So new qualia -conceptual patterns- are matched with existing, contextual ones, and through these interferences you get in-formation. In other words, mutually interacting patterns take up a certain form, and what we usually call 'information' is the result of this proces. With this, the self-refernetial metaphysics I needed for a methodology of complexity became possible, because everything we observe or experience becomes the result of intrerferences of patterns; matter, heat, gamma radiation, but also knowledge is the result. Our brains are capable of mimicking and processing these interferences, which allows us to make sense of our life-world. In the end, both matter and information are the result of patterns that 'have taken up form'.

    But these forms take place in interaction with a lot of other conceptual patterns, which deforms them, stretches and tears them, so 'formation' always takes place in a contingent environment. However, some patterns are sufficiently robust that they create stable forms. Especially patterns that create their own contexts -basically they feed back onto themselves- can create very stable forms. So a pattern library of complexity should include quite a number of such robust self-referential patterns.