PatternTheory

MeatballWiki | RecentChanges | Random Page | Indices | Categories

PatternTheory is the abstract form of ChristopherAlexander's philosophy of living, unfolding systems. It's the holistic theory of everything, especially of complex systems, e. g. like cities or societies, in a simple form it is used for software systems. It is the theory needed to cope with the problems of the 3rd millenium. It' cybernetics taken to the next level, not only self-regulation and self-organization, but also the creation and especially the development/unfolding of systems is taken into account. The general solution to many pressing issues of or time is there, but it is not yet seen seen by the public. I try to give an overview that can be read in 10 minutes. -- HelmutLeitner

At the core of Alexanders Theories are about 40-50 concepts that go into a few main groups. The application of these concepts might be compared to playing a game like chess: The theory is not closed in the sense that it deterministic laws, it is a theory of options that suggest to unfold a system according to certain principles in a creative design process. It's comparable to music: the result should not be seen in the one perfect piece of music, but in the unlimited variety of wonderful music that touches us.

There are general concepts:

e.g. system, structure, process, patterns, ...
e.g. whole, center, field of centers, resonance, ...
e.g. life, process increasing the quality of living systems (maybe "livingness"), unfolding, ...

There are fifteen properties of structures that - if cooperating - increase the quality of structures:

e.g. contrast/difference, symmetry, functional boundaries, proportions, similarity, gradient, alternating repetiton (rhythm), simplicity, individuality, adaptated form, free space, connectedness, ambiguity, functional center, ...
(one can also understand these as "categories of form" that endow us with a language to describe general systems and their dimensions of development/unfolding)

There are about a dozen principles that increase the quality of processes:

e.g. simplest steps (efficiency), step-by-step, reversible action (testing and learning from errors), structure-preserving-transformation, importance of sequence, ...
e.g. win-all solutions, open participation, judgement by resonance, ...

There are concepts regarding patterns:

e.g. patterns as solutions to problems in a context
e.g. patterns as a formal structure: e. g. name, context, problem, solution, ins and outs, pros and cons, examples, know-how needed, connections, alternatives, ...
e.g. pattern languages as collections of patterns and their application know-how, sharing of knowledge in a modular form, ...
e.g. pattern mining methodology: pattern thumbs, pattern clouds, pattern gestalt, ....

These concepts lead to an understanding of systems. This means that the systems can be described in general terms, that are outside of the typical application context. The same is then true for the process of developing a system. Strength and weaknesses become visible, and options for unfolding the system.

One should understand that pattern theory is not just a way an idiosyncratic way of thinking, it is a logical and even categorial structuring of how one can think. E.g. what we can reasonably understand can be brought into a problem-solution-formalism. It's not only that we can find patterns but that - if we accept the formalism - we must find them and everthing we know can be brought into this form. The same is true for the dichotomy of structure (properties) and process (principles). Whatever exists in the universe (system) can be perceived under the aspect of change in time. Things that are unchanged on a certain timescale are "rigid structure", those that change are "process flow". Where we draw the boundary is our decision, but the ( system - structure - process - pattern ) categories will always be applicable. We have structured our thinking, like setting up an x-y-z coodinate system (Descartes). In itself it is debatable and useless (tautology, we get out of it what we have put into it) but if we use it and fill it with content, it may show its strength as a tool or methodology.

There are background views that are result and basis of the theory at the same time:

the unfolding of life is the fundamental process in the universe (also: nature as our teacher)
living processes are not predictable and not predetermined (oversimplified: quantum fluctuations)
the quality comes from participation, from those directly involved
the designer must take the role of a "coach" not that of the "remote planner and god-architect"

This last point also creates the main acceptance problem of pattern theory and Alexander. Pattern theory is about the transfer of power from the top to the bottom of the creative process. Our modern society is so successful as a whole because the normal member has much more options and autonomy (kind of freedom) than in former societies. Nevertheless, there are always powers (politics, industries, media) that would profit from taking this autonomy away from the people. They want to suggest what to vote, buy or think, in their own interest. There is a constant competition regarding the social status and influence, whether in a family, a church or a NGO. Power is a negative term, its too simple. At the basis is the wish to dominate the creative process of what should happen or not. Alexander says: if we want our processes to work those that are affected by the process must be enabled to understand the process and to participate in the process. This is revolutionary thinking that is unloved.

Discussions & questions

A few (incoherent?) points...

I quite appreciate the current effort to create a 'summary' since it is clarifying much of the material I have reviewed previously.
Coming at the subject of Patterns from an entry point that starts with ChristopherAlexander, one gets a very 'universal' and 'holistic' (top-down) view of patterns. A different (bottom-up ?) view emerges when the entry points are closer to the REGEXP insights that are fundamentally important in more advanced programming methods. Does anyone else see these two extremes merging in the middle?

-- HansWobbe

Hans, are you referring (by "Regexp") to what you've called "strings" theory? (Not to be confused with microphysics.) (Rather, computer strings.)

I am reminded of a saying at Damanhur: "The crystal of truth is singular and whole, but it can only be penetrated through a facet."

-- LionKimbro

Lion, thank you for your reference to Damanhur, they were not on my radar screen, although their centre in the Piemont, Italy is not very far from my home, maybe 8 hours by car. They even seem to have started an organisation in Austria. The quotation is more than valid in many ways. It resonates with what wie discussed on http://www.communitywiki.org/en/InformationKnowledgeAndWisdom in the way that facets correspond to knowledge and truth corresponds to wisdom. One problem of Pattern theory / Alexander is that it has so many facets, maybe 30 or more. Each could be a valid access point to the whole of the theory. But in using one facet (e. g. architecture, cybernetics, philosophical holism, system theory, software patterns, methodology of education/therapy as self unfolding, science theory, the artistical creative process, ...) as an access point, the typical reaction is that of reduction ("ok, that's it, about X") and of statistical refusal ("You know, X is none of my main interests"), which is understandable.

Hans, please tell us more about what you name Google:REGEXP insights. Neither wikipedia nor google did help me to access what you mean. As far as I see, there is no contradiction between top-down and bottom-up approaches with Alexander. I see a kind of "modular realism" compared to idealistic holistic approaches ("we are influenced by the hole universe"). The "whole" of Alexander is never an abstract whole but a concrete one: the organism (man, tree, artifacts), the building (house, bridge), and up the hierarchy (village or neighbourhood, city, region). The whole is a unit that has individuality and a history of development over time (the seed that grows into a large tree as the same whole, the city of New York that maybe once started with the first wooden building). Unfolding always starts with a whole or center, somewhere in the hierarchy of wholes, taking the upward/downward/sideward effects into account. Don't know whether this fits to your question, so I'll stop at this point. Please clarify.

-- HelmutLeitner

Material removed to PatternTypes to help this page stay focused on being the "Summary" proposed by Helmut.

Hans, could you at least name/create a page to move this side-discussion there? -- HelmutLeitner
Helmut, I am happy to do so. I am also thinking that PatternTypes might be needed to distinguish CategoryPattern which is currently specific to ChristopherAlexander and other possible pattern forms. -- HansWobbe

Fridemar, thank you for explaining the the possible background of what Hans talks about. I agree that both, Alexandrian patterns (AP) and "regular patterns in language or visual systems", have a common aspect of "recognition". But, I think this is a superficial similarity. Everything is at some point subject to a recognition process. Regularity and language may or may not be connected to APs. There are many contextual layers to look at regular patterns and their problem solution capacity. Alexander looks at architecture but typically not at structural building materials like bricks (the brick is outside of his contextual focus, as long it is not visible as a design element), although a brick obviously also solves a problem.

Language is a very Alexandrian field, because WORD and GRAMMATICAL_STRUCTURES (as a pattern group) serve a purpose and obviously solve problems in the context of communication. Each and every word is a pattern instance in the context of language. It's helpful to ignore this for our pattern work most of the time, when working in non-language contexts. Chomsky seems non-Alexandrian to me, because he suggests strict grammers. The Alexandrian picture of the world is different, creative, always allowing alternative. Natural language is something living. I do not see this with Chomsky.

Highly "organic" patterns (as regular patterns) can be generated even by simple userdefined rule systems as rendered by this freely available artwork tool: http://www.contextfreeart.org/gallery/.

E.g. take as examples: [ContextFreeArt:Flower] or even this shorter one: [ContextFreeArt:FoggyTree] It would be nice, if someone could find/create such CF patterns that emulate typical Alexander Patterns. Voilà, here we have a generated plan for [ContextFreeArt:buildings and roads] -- FridemarPache

I'm obviously very pro-Patterns, but I will admit that the whole PatternTheory exhibits an inherent weakness. The way that Alexander has approached Patterns resembles Descartes' [Vortex] [Theory]. An elegant, internally consistent description of reality that was not empirically predictive. While you may argue that Alexander grounded his theory in example after example, that was only on known data, not new data. -- SunirShah

Sunir, basically I agree with you. But I wouldn't see the PatternTheory as primarily a predictive theory, but as a methodical concept in creative contexts where predictability can not be attained. Descartes moves on the space of beginning natural sciences, while Alexander goes beyond. The question is: what is science? Simplified, I'd say: science is about abstract knowledge that is useful and can be shared. This included natural sciences and Alexander, but doesn't reduce science to predicting theories. -- HelmutLeitner