Coevolving Innovations

… in Business Organizations and Information Technologies

Systems generating systems — architectural design theory by Christopher Alexander (1968)

The systems thinking roots from architect Christopher Alexander aren’t completely obvious in his work on pattern language.  A republished version of an 1968 article resurfaces some clarification on a perspective on systems thinking originating from practices in architecture.  This article introduced ways in which systems thinking could be most directly applied to built environments.  The cross-appropriation of pattern languages across a variety of domain types — object-oriented programmers were the earliest motivating adopters — could be enlightened by revisiting the foundations.  Alexander concisely presented 4 points, and then provided detailed reasoning for each:

1. There are two ideas hidden in the word system: the idea of a system as a whole and the idea of a generating system.

2. A system as a whole is not an object but a way of looking at an object. It focuses on some holistic property which can only be understood as a product of interaction among parts.

3. A generating system is not a view of a single thing. It is a kit of parts, with rules  about the way these parts may be combined.

4. Almost every ‘system as a whole’ is generated by a ‘generating system’. If we wish to make things which function as ‘wholes’ we shall have to invent generating systems to create them.  [Alexander 2011, p. 59; Alexander 1968, p. 605]

In a properly functioning building, the building and the people in it together form a whole: a social, human whole. The building systems which have so far been created do not in this sense generate wholes at all.  [Alexander 2011, p. 58; Alexander 1968, p. 605]

Let’s leave analytical explications of the original 1968 text as secondary, to first appreciate the idea of “systems generating systems” through sensemaking done some decades after 1968, and in the broader context of Alexander’s other writings and interviews.

Molly Wright Steenson, as part of her 2014 dissertation, has a 66-page digest of Alexander’s work between 1962 and 1968.  Her deep reading was reflected in a 2009 recorded presentation on “Loving and Hating Christopher Alexander“.  Generally speaking, interaction designers love Christopher Alexander’s approach, while architects hate Christopher Alexander’s approach.

SVA Dot Dot Dot Lectures: Molly Wright Steenson from MFA Interaction Design.

Amongst the lovers and haters of Christopher Alexander is a predisposition towards interaction compatible with systems thinking.  For built environments, architecture can be described through a language of patterns, where those patterns may or may not be generative.  In her 2014 dissertation, Steenson fleshes out Alexander’s 1968 “Systems Generating Systems” with the broader context of the 1979 The Timeless Way of Building, and 1983 publication by Stephen Grabow of interviews with Alexander.

Generating Systems

Alexander describes pattern languages as “generative,” referring to the quality of multiplicity, of a system that operates both as a whole and as a set of rules.  A system, like a language, works on multiple levels.  The system presents itself on the surface, he writes, when “we are confronted with an object which displays some kind of behaviour which can only be understood as a product of interaction among parts within the object.  We call this kind of behaviour, holistic behaviour.”262 It also incorporates the rule set for the manipulation of the elements that it composes. This dualistic system is analogous to the functions of the pattern language. Just as a generating system is a kit of parts, “Each pattern is a rule which describes what you have to do to generate the entity which it defines.”263 [Steenson 2014, pp. 90-91]

262Christopher Alexander, “Systems Generating Systems,” AD 38(1968): 606.
263Alexander, The Timeless Way of Building, 182.

Alexander intends to literally mesh genetics and linguistics and apply them to architecture with his notion of generativity. He was interested in the idea that a rule set — the syntax of a language — can generate a building “not as a mechanical technique (as might perhaps be naively understood in the automobile industry) but as a structural principle of natural creation as it is understood in modern science,” writes Alexander’s biographer Stephen Grabow.266  “The idea that a set of known rules could actually generate a building is as disturbing as the idea that a human being is generated by a few genetic rules operating on chromosomes or that a poem is generated by a few grammatical rules operating on language. And yet that is precisely what Alexander is claiming.”267  Alexander refers to Noam Chomsky’s generative grammar in an interview in his biography, stating that it is his intention to apply such a grammar to architecture.  The “structure of the underlying language…is doing most of the hard work.”  The pattern language provides the syntax, where the patterns when combined and executed by the user reflect the language’s semantics. Alexander seems to channel Chomsky’s notion of deep and surface structures: while a pattern might have many potential instantiations on its surface, its sum of underlying possibilities are what make it generative.  In essence, if a generative grammar can produce sentences regardless of the language, if genetic code can produce a bird, then a generative system can produce architecture. Alexander says, “… I’m making the statement that I can actually set up those rules so that if you follow a sequence of them in the order prescribed you will have a building.”268  The user of the language makes choices about what and how to use the language, based on context: It is the structure that makes this possible.269  Indeed, genetics, languages, and architecture are inextricable in Alexander’s view. Designing buildings or a town is “fundamentally a genetic process.”270  Moreover, he writes, “patterns always come from languages”; these languages are analogous to the genetic code that shapes a living being.271 [Steenson 2014, pp. 91-92]

266Stephen Grabow, Christopher Alexander: The Search for a New Paradigm in Architecture (Boston: Oriel Press, 1983), 9.
268Ibid., 48.
269Ibid., 49.
270Alexander, The Timeless Way of Building, 240.
271Ibid., 199.

From a systems perspective, just having a collection of patterns isn’t sufficient.  A generative pattern language has to meet a higher standard, in three ways.

There are three elements of pattern languages that make them generative. First, pattern languages contain an inherent rule set that determines their logic. Alexander writes, “Thus, as in the case of natural languages, the pattern language is generative. It not only tells us the rules of arrangement, but shows us how to construct arrangements — as many as we want — which satisfy the rules.”272  In an interview with his biographer, Alexander noted, “We give names to things but we don’t give many names to relationships.”273  The pattern language was an attempt to address these relationships.  “So it not only defines the sentences which make sense in a given situation; it also gives us the apparatus we need to create these sentences. It is, in other words, a generative system, which allows us to generate sentences that are appropriate to any given situation.”274  Following the operations in order, suggested by the system, creates a generative, coherent whole out of the parts the system is organizing. [Steenson 2014,  pp. 92-93]

272Ibid., 186.
273Grabow, Christopher Alexander, 46.
274Alexander, The Timeless Way of Building, 186.

Second, pattern languages and other generating systems produce effects greater than the sum of their parts. Alexander writes in Timeless Way, “This quality in buildings and in towns cannot be made, but only generated, indirectly, by the ordinary actions of the people, just as a flower cannot be made, but only generated from the seed.”275 These systems may come to necessitate their own propagation, he suggests, when we use them. He writes, “The patterns in the world merely exist. But the same patterns in our minds are dynamic. They have force. They are generative. They tell us what to do; they tell us how we shall, or may, generate them; and they tell us too, that under certain circumstances, we must create them.”276 [Steenson 2014,  p. 93]

275Ibid., xi.
276Alexander, The Timeless Way of Building, 186.

Third, in addition to their self-perpetuating properties, generating systems contain the mechanism for their own propagation. The pattern language, then, “like a seed, is the genetic system which gives our millions of small acts the power to form a whole.”277 A language, then, can foster “a process of unfolding, like the evolution of an embryo, in which the whole precedes the parts, and actually gives birth to them, by splitting.”278 At the same time, the language grows through accretion. “Next, several acts of building, each one done to repair and magnify the product of the previous acts, will slowly generate a larger and more complex whole than any single act can generate,” he writes.279 Alexander sees it as a genetic allegory: in his description of “the timeless way of building,” an unnamable quality that his systems are intended to elicit, he writes, “In this sense, then, we have found an example of the kind of code which does, at certain times play just the role in buildings and in towns that the genetic code plays in a living organism.”280 [Steenson 2014,  pp. 93-94]

277Ibid., xiii.
279Ibid., xiv.
280Grabow, Christopher Alexander, 49. Italics Grabow’s

A pattern language can be more than grammar.  As a generative language about design, it can approach the sophistication of a semantic network.

Despite the fact that Alexander employs the notion of generative grammar, he argues that Chomsky’s generative grammars are too basic and that pattern languages surpass them because of their engagement with semantic networks. Alexander says in an interview with Stephen Grabow,

Chomsky’s work on generative grammar will soon be considered very limited… It does not deal with the interesting structure of language because the real structure of language lies in the relationships between words — the semantic connections. The semantic network — which connects the word “fire” with “burn,” red,” and “passion” — is the real stuff of language. Chomsky makes no attempt to deal with that and therefore, in a few years, his work will be considered primitive.281 [Steenson 2014,  pp. 94]

281Grabow, Christopher Alexander, 49. Italics Grabow’s.

The “real stuff” that interested Alexander had to do with the architectural equivalent of the semantic networks: the interrelations of the words, their meanings and their evocations with each other. He continues,

In that sense, pattern languages are not like generative grammars. What they are like is the semantic structure, the really interesting part of language and which only a few people have begun to study. The structure which connects words together…is much more like the structure which connects patterns together in a pattern language. So pattern languages are not so much analogous to generative grammars as they are to the real heart structure of language which has hardly been described yet.282

This places Alexander in a strange situation. On one hand, the notion of generativity provides the pattern language with the means of its propagation.  The formatting and compression of the patterns and the sequencing provided by the language provides the framework for using the patterns as a program to create form.  But he aims for semantics, allegory, and poetics, as well as the aspects of language that generate feelings, emotions, a sense of order — all of which extend beyond the structural, topological and syntactic aspects of his program. These semantic conceptions equate the patterns with timeless cycles of life, with a program for the built environment, for the order of existence.  [Steenson 2014,  pp. 94-95]

282While it is true that Chomsky’s interest in generative grammar shifted over the decades, eventually moving toward the Minimalist Program in the early 1990s, Alexander would also step away from the notion of a semantic network and more toward the pursuit of the geometrics of order. Ibid.

Thus, from a perspective of 2014, the idea of systems generating systems may have first emerged in 1968, but shows up in the 1973 publication of The Timeless Way of Building.  The basic systems thinking foundations help in understanding the 2003-2004 four volumes of The Nature of Order … but we’ll leave that for another discussion.

Turning from a retrospective view to an earlier point in Alexander’s development, Steenson traces back to the writings of Ross Ashby through to the 1964 publishing of Notes on the Synthesis of Form.

In order to develop a model for stability in design problems, Alexander looked to cybernetics for models of homeostasis and ultrastability. Such systems could stabilize themselves regardless of what disturbed them, including variables that weren’t considered when the system was designed.121  Alexander was particularly inspired by W. Ross Ashby’s 1952 book, Design for a Brain, which he cites numerous times in Notes, especially in Chapter 3, “The Source of Good Fit.” [Steenson 2014, p. 46]

121“ultrastable, adj.” OED Online. September 2013. Oxford University Press, accessed November 12, 2013, Ultrastability places two sets of environmental and reactive variables in a primary feedback loop. A slower, second feedback affects the reactive variables by acting on the step-mechanisms and setting parameters for the environmental variables. Ashby, Design for a Brain: The Origin of Adaptive Behavior, 98.

In a 2011 edited volume, Sean Ahlquist and Achim Mengers describe how the foundations of General Systems Theory by Ludwig von Bertalanffy were adapted to design thinking and architectural theory by Christopher Alexander..

Systems Thinking

[….]  In the mid-20th century, Ludwig von Bertalanffy introduced general system theory as a response to the long evolution of the conception of the ‘whole’ of functioning systems. His theory is based on a fundamental critique on classic physics and its deductive methods and focus on isolated phenomena. Bertalanffy considered such methods as unsuitable for biology, reasoning that nothing in nature exists in isolation or simple dependencies, but rather needs to be understood as complex systems of interactions and reciprocities.   [Ahlquist and Mengers 2011, pp. 13]

[….] General system theory was formulated as a universal scientific discipline to decipher the laws which rule the definition of ‘organised wholes’. Folded into architecture, as exemplified by Christopher Alexander’s writings on design process and physical organisations, form and functionality in the 1960s, the theory implied profound changes to design thinking. Contrary to established design approaches, Alexander argued that an overall design problem cannot be divided into sub-problems, and consequently, that it is impossible to arrive at a novel design solution as a summary process of solving individual problems one after the other.

Alexander describes a system as that which focuses on an overall behaviour accomplished through the ‘interaction among parts‘. The use of the words among parts is critical in this definition as it states that the knowledge of the components has to be complemented with the knowledge of how the components interact, whether it be in competition or in contiguity. As defined initially by Aristotle and later expanded upon by Bertalanffy, systems do not function simply as the summation of the whole. The holistic nature cannot be seen in the individual part, nor can it be seen with the addition of its parts. The system behaviour emerges only in the dynamics of the interactions of the parts. This is not a cumulative linear effect but rather a cyclical causal effect in which the complexity of the level and amount of interaction cannot be directly deciphered.

The introduction of system theory in architecture and the consequential focus on systems thinking in design applied a twofold shift. First, the shift signified a dismissal of the view of architecture as comprised of entities in static isolation for one which defines form as the culmination of systems which interact with its context in matter, physicality and personal engagement. Second, this shift in architectural thinking introduced fundamental concepts for how the computation of such interrelational, complex behaviour-based systems could be achieved.  [Ahlquist and Mengers 2011, p. 15]

This brings us to a 2011 perspective by Ahlquist and Mengers on the republication of the 1968 article by Alexander.  An introductory preface is provided.

In his earlier writings, Christopher Alexander established the notion of the unselfconscious process. Initially, this refers to the notion of process, in a social context, as being interrelational by which culture, building and environment are concurrently formed. In this selected text, published in an issue of Architectural Design (1968), it Is extended into the notion of how architectural problems may be solved through an analogous process where design forms through the iterative readings and responses to interrelational conditions, with the intention of producing environments synchronous with their cultural settings. Inherent in this position regarding process and form is the substantial step that Alexander makes to transfer the conception of form as an observed object to one as an externalised operating system. In this article, he argues how such a system is born, itself, of a generative system, establishing the duality between the object as a computing agent and the method as a computational process. Alexander lays out particular aspects of such a process, describing three conditions: the global behaviour, the components that. form such behaviour and the types of local relationships among those components. While these three aspects characterise a system, this does not extensively describe the process by which a system can be achieved. Alexander explains, through defining four characteristics of a system, the distinction between the behaviour, as a collection of actions, and the system which generates that behaviour, as a series of interactions, and at what level of abstraction such complexity can be understood.  [Ahlquist and Mengers 2011, p. 58]

Let’s view selected passages from the full 1968 article.

1. There are two ideas hidden in the word system: the idea of a system as a whole and the idea of a generating system.

For systems thinkers, emergence is a property of a whole that is not a property in its parts.  The classical example is that wetness is a property of water, and not a property of the hydrogen and oxygen that make up water.  Alexander doesn’t use the word “emergence”, but does capture that idea in the “generating system”.

The word system, like any technical word borrowed from common use, has many meanings and is imprecise. This lack of precision in a technical word might seem dangerous at first; in fact it is often helpful. It allows new ideas to flourish while still vague, it allows connections between these ideas to be explored, and it allows the ideas to be extended, instead of having them cut short by premature definition and precision.

The word ‘system’ is just such a word. It still has many meanings hidden in it. Among these meanings there are two central ones: the idea of a system as a whole, and the idea of a generating system.

These two views, though superficially similar, are logically quite different. in the first case the word ’system’ refers to a particular holistic view of a single thing. In the second case, the word ’system’ does not refer to a single thing at all, but to a kit of parts and combinatory rules capable of generating many things. [Alexander 2011, p. 58; Alexander 1968, p. 605]

Why use the word “generation” and/or “generative?  In 1995, on the C2 wiki entry on “Generative Pattern” Jim Coplien wrote:

A generative pattern is one of the KindsOfPatterns.  It is first a pattern; a solution to a problem in a context. In the early days of patterns, we used the term generative to mean creational. But a closer reading of Alexander shows that by generative, he means something that leads to emergent behavior.

Generative patterns work indirectly; they work on the underlying structure of a problem (which may not be manifest in the problem) rather than attacking the problem directly. Good design patterns are like that: they encode the deep structure (in the Senge sense) of a solution and its associated forces, rather than cataloging a solution.

We can contrast a Generative Pattern with a GammaPattern, which is not generative. (That doesn’t make them bad, just different. Much of the software visualization work going on in the industry is all about Gamma patterns.)
— JimCoplien 1995/05/28

This distinction between non-generative and generative is fleshed out to a greater extent in writing by Coplien in 1996.


In many problem-solving strategies, we try to attack problems directly. In doing so, we often attack only symptoms, leaving the underlying problem unresolved. Alexander understood that good solutions to architectural problems go at least one level deeper. The structures of a pattern are not themselves solutions, but they generate solutions. Patterns that work this way are called generative patterns. A generative pattern is a means of letting the problem resolve itself over time, just as a flower unfolds from its seed:

9. This quality in buildings and in towns cannot be made, but only generated indirectly by the ordinary actions of the people, just as a flower cannot be made, but only generated from the seed (Alexander, 1979. p.xi)

And later:

An ordinary language like English is a system which allows us to create an infinite variety of one dimensional combinations of words, called sentences…. A pattern language is a system which allows its users to create an infinite variety of those three dimensional combinations of patterns which we call buildings, gardens, towns.

Thus, as in the case of natural languages, the pattern language is generative. It not only tells us the rules of arrangement, but shows us how to construct arrangements as many as we want which satisfy the rules. (Alexander, 1979: pp. 185 186)

Like many other facets of Alexander’s philosophy, this philosophy can be traced back to Eastern schools of thought (Lao Tsu principles of nonaction, part 3). This generativity is an important aspect of the Quality Alexander seeks. It is an elusive quality, so elusive he calls it the quality without a name; … but we need not turn to esoteric sources for insights on the importance of generativity in problem-
solving; other contemporary sources will do. In Senge we find:

What, exactly, does it mean to say that structures generate particular patterns of behavior? (Senge, 1990: p. 45) [Coplien 1996, p. 32]
… a fundamental characteristic of complex human systems … [is that] cause and effect are not close in time and space. By effects, I mean the obvious symptoms that indicate that there are problems drug abuse, unemployment, starving children, falling orders, and sagging profits. By cause I mean the interaction of the underlying system that is most responsible for generating the symptoms, and which, if recognized, could lead to changes producing lasting improvement. Why is this a problem? Because most of us assume they are most of us assume, most of the time, that cause and effect are close in time and space. (Senge, 1990: p. 63) [Coplien 1996, pp. 32-33]

Why is generativity important? First, as Senge says, most real problems go deeper than their surface symptoms, and we need to address most interesting problems with emergent behavior. Second, a good pattern is the fruit of hard work and intense review and refinement. Simple problems can be addressed through simple rules, since the solutions are more direct or obvious than we find in generative solutions. The pattern form excels an engaging the reader in generative solutions: to understand the principles and values of lasting solutions and long-term emergent behavior. Good patterns go beyond the quick fix. [Coplien 1996, pp. 33-34]

For Alexander, then, a system that is described as a whole but does not have any emergent properties misses a hidden idea.  For a systems thinker, the lack of a whole suggests a collection of parts in a network, rather than a systemic whole.

2. A system as a whole is not an object but a way of looking at an object. It focuses on some holistic phenomenon which can only be understood as a product of interaction among parts.

Alexander moves the ideas of system and whole away from the tangibleness of building construction, towards a phenomenological view of architecture.

Let us consider some examples of holistic phenomena which need to be viewed as systems.  [Alexander 2011, p. 59; Alexander 1968, p. 605]

The great depression is an obvious example of a holistic phenomenon. We cannot understand the depression, except as a result of interaction among rates of consumption, capital investment and savings: the interactions can be specified in the form of equations; if we follow these equations through to their conclusion, we see that under certain conditions they must always lead to a depression.  [Alexander 2011, p. 59; Alexander 1968, p. 605-606]

The stability of a candle flame is another example of a holistic phenomenon. Why does it maintain approximately the same size and shape throughout its flickering?  In this case, the ‘parts’ are flows of vapourised wax, oxygen and burnt gases — the processes of combustion and diffusion give the interaction between these flows — and these interactions show us at what size and shape the flame will be approximately stable.  [Alexander 2011, p. 59; Alexander 1968, p. 606]

Alexander describes more examples of a rope and a computer.  He then focuses on stability as an essential character that is a property of the whole.

Another kind of holistic behaviour is that instability which occurs in objects that are very vulnerable to a change in one part: when one part changes, the other parts change also. We see this in the case of erosion: cutting down trees robs the soil of the roots which hold it together, so that wind and water can strip the soil of all remaining plants, and make a desert. We see it again in the death of the traditional farm: when the combine harvester replaced traditional harvesting, the entire balance of scale economies was destroyed, the little farms collapsed, and gave way to giant farms.

Let us summarise the content of these examples. In every case we are confronted with an object which displays some kind of behaviour which can only be understood as a product of interaction among parts within the object. We call this kind of behaviour, holistic behaviour.

The central point of the whole argument can be stated very simply. The most important properties which anything can have are those properties that deal with its stability. It is stability which gives a thing its essential character. The strength of an arch, the even burning of a flame, the growth of an animal, the balance of a forest ecology, the steady flow of a river, the economic security of a nation, the sanity of a human individual, the health of a society: these are all, in one way or another, concerned with stability.  [Alexander 2011, p. 60; Alexander 1968, p. 606]

Stability, no matter in which of its many forms, is a holistic property. It can only be understood as a product of interaction among parts. The essential character of anything whatever, since it must at heart be based on some kind of stability, must be understood as a product of interactions within the whole. When we view a thing in such a way as to reveal its character in holistic terms, we speak of it as a system.  [Alexander 2011, p. 60; Alexander 1968, p. 606-607]

On the second point on “holistic phenomenon” and “interaction between parts”, Alexander is rather exhaustive in his distinctions.  The extended content can be excised with little impact for systems thinkers (who won’t learn much).

In order to speak of something as a system, we must be able to state clearly: (1) the holistic behaviour which we are focusing on; (2) the parts within the thing, and the interactions among these parts, which cause the holistic behaviour we have defined; (3) the way in which this interaction, among these parts, causes the holistic behaviour defined.

If we can do these three, it means we have an abstract working model of the holistic behaviour in the thing. In this case, we may properly call the thing a system, If we cannot do these three, we have no model, and it is meaningless to call the thing a system. The idea of a system is synonymous with the idea of an abstract model of some specific holistic behaviour. We may speak of the economic system in a country, because we can construct a system of equations which reproduce important holistic phenomena like depressions or inflation. If we couldn’t do this, it would be meaningless to speak of economic systems.

We must not use the word system, then, to refer to an object. A system is an abstraction. It is not a special kind of thing, but a special way of looking at a thing. It is a way of focusing attention on some particular holistic behaviour in a thing, which can only be understood as a product of interaction among the parts. [Alexander 2011, p. 60; Alexander 1968, p. 607]

This second section focused on clarifying the understanding of systems and wholes, and didn’t mention the idea of generativity.  This becomes the focus in the third section.

3. A generating system is not a view of a single thing. It is a kit of parts, with rules about the way these parts may be combined.

On this third point, Alexander moves on from the discussion on wholes (i.e. “single thing”), to focus down on part-part relations.  The description of “rules” would seem to lead to an appreciation that some part-part relations will not (or should not) work.

This is a different use of the word system from the first one. In colloquial English we often use the word system to mean ‘a way to do something”: that’s what a betting system is; that’s what the Montessori system is; that’s what the democratic system is.

Each of these systems is, at heart, a system of rules. A betting system tells you how to place your bets, the Montessori system lays down rules to be followed by children and teachers in nursery school„ the democratic system of government lays down certain rules about the nature of representation, the choice of representatives and the conduct of elections. In all these cases, the rules are designed to generate things. A betting system supposedly generates winning bets, an educational system generates well-educated pupils, the democratic system supposedly generates freedom and good government.  [Alexander 2011, p. 64; Alexander 1968, p. 609]

We may generalise the notion of a generative system. Such a system will usually consist of a kit of parts {or elements) together with rules for combining these parts to form allowable ‘things’. The formal systems of mathematics are systems in this sense. The parts numbers, variables, and signs like + and =. The rules specify ways of combining three parts to form expressions, and ways of forming expressions from other expressions, and ways of forming true sentences from expressions, and ways of forming true sentences from other true sentences. The combinations of parts, generated by such a system, are the true sentences, hence theorems, of mathematics. Any combination of parts which is not formed according to the rules is either meaningless or false.  [Alexander 2011, pp. 64-65; Alexander 1968, p. 609]

A generating system, in this sense, may have a very simple kit of parts, and very  simple rules. Thus the system of triangles which may be put together to form a square, is a generating system. Its rules generate all the ways of putting these triangles together  to form a square. It is typical of a system that the rules rule out many combinations of  the parts. Thus these triangles could be put together in an infinite variety of ways — but  most of these ways are ruled out, because the outside perimeter is not a square and this thing is not connected.  [Alexander 2011, p. 65; Alexander 1968, p. 609-610]

Another example of a generating system, is the system of language. [….]

Perhaps the most interesting and important generating system in the world is the genetic system.  [….]

A building system is a generating system in this sense. It provides a kit of parts — columns, beams, panels, windows, doors — which must be put together according  to certain rules.  [Alexander 2011, p. 65; Alexander 1968, p. 610]

From a structural perspective, not every combination of parts fits together.  From a process perspective, even if the correct parts are available, not every sequence of assembly will produce the generating system.

4. Almost every ‘system as a whole’ is generated by a generating system. If we wish to make things which function as ‘wholes’ we shall have to invent generating systems to create them.

Alexander doesn’t rule out spontaneous order, but sees that as a rare event.  For a system as a whole to have the properties desired, the builders will most probably have to have a generating system to create the system as a whole.  Examples of physical systems and biological systems are cited.

There is a relationship between the two ideas of system which have been defined. Almost every object with behaviour that depends on some ‘system as a whole’ within the object, is itself created by a generating system.

Take an obvious and simple case: a hi-fi system. its purity of performance can only be understood as a product of the combined effect of all the various components, working as a whole. The same hi-fi system is also generated by a generating system: the kit of all the parts on the market, and the rules governing the electrical connections and impedance matching between these parts.  [Alexander 2011, p. 65; Alexander 1968, p. 610]

To take a more complicated case: the railroad switch-yard. [….]

The most complicated case of all, and the clearest, is that of an animal. A landing seagull certainly needs to be seen as a system: so does almost everything else that seagulls do. At the same time, this seagull is created by a generating system: the genetic system. An animal is both something which needs to be seen holistically, and generated by a generating system.

The relationship between holistic systems and generating systems is easy to understand. If an object has some holistic property caused by interaction among parts — then it is clear that these particular parts and these particular interactions will only come into being if the parts have very constrained relationships to one another. The object then, must be generated by some process which assembles parts according to certain constraints, chosen to ensure the proper interaction of these parts, when the system operates. This is exactly what a generating system is.

The generating system need not be conscious (as in the case of the switch-yard), nor even always explicit (as in the genetic case). Sometimes the processes which make up the generating systems are integral with the object being formed — thus the candle flame is generated by chemical processes which are the same as those processes which then maintain the system’s equilibrium and make up the interacting parts, when we view the flame as a holistic system.   [Alexander 2011, p. 66; Alexander 1968, p. 610]

Alexander turns to making a criticism and challenge for designers, as professionals who are focused on the parts and not the whole.

[….]  It is true then, that almost every ‘system as a whole’ is generated by a generating system. This axiom contains a remarkable lesson for designers. Man as a designer is concerned with the design and construction of objects which function as wholes. Most of the important properties a city needs to support life, for instance, are holistic properties.

Our axiom means this: to ensure the holistic system properties of buildings and cities, we must invent generating systems, whose parts and rules will create the necessary holistic system properties of their own accord.

This is a radical step in the conception of design. Most designers today think of themselves as the designers of objects. If we follow the argument presented here, we reach a very different conclusion. To make objects with complex holistic properties, it is necessary to invent generating systems which will generate objects with the required holistic properties. The designer becomes a designer of generating systems — each capable of generating many objects — rather than a designer of individual objects.  [Alexander 2011, p. 66; Alexander 1968, p. 610]

Closing the article, Alexander makes it clear that the whole is not delimited to just a building, but also the social group that occupies it.

A final word of caution. As we have already seen, a building system is an example of a generating system. It is a kit of parts with rules of combination. But not every generating system necessarily creates objects with valuable holistic properties. The generating system which makes squares out of triangles is an example. It is a perfectly good generating system; yet the objects it produces do nothing: they have no holistic system properties whatever.  In the same sense, those building systems which have so far been conceived make buildings, but they do not make buildings with any really important holistic system properties. In a properly functioning building, the building and the people in it together form a whole: a social, human whole. The building systems which have so far been created do not in this sense generate wholes at all. While it is inherent in the generating system of an animal that the finished animal will work as a whole, it is not inherent in any of today’ s building systems that the buildings they produce will work as social or human wholes. Creating building systems in the present sense is not enough. We need a new, more subtle kind of building system, which doesn’t merely generate buildings, but generates buildings guaranteed to function as holistic systems in the social, human sense. [Alexander 2011, p. 66-67; Alexander 1968, p. 610]

This 1968 article predates the pattern language publications starting in 1977.  In Alexander’s patterns on towns, building and construction, human beings are parts of the whole as much as built environments.  A critical eye on pattern languages in other domains would check for similar inclusion of human beings in social interactions.


Ahlquist, Sean, and Achim Mengers. 2011. “Introduction — Computational Design Thinking.” In Computational Design Thinking, edited by Achim Mengers and Sean Ahlquist, 10–29. Chichester, England: John Wiley & Sons.

Alexander, Christopher. 2011. “Systems Generating Systems.” In Computational Design Thinking, edited by Achim Mengers and Sean Ahlquist, 58–67. Chichester, England: John Wiley & Sons. Reproduced by permission of Christopher Alexander, from “Systems Generating Systems”, Architectural Design, volume 38 (December), John Wiley & Sons Ltd (London), 1968, pp. 605-610. Originally published in Systemat, a journal of the Inland Steel Products Company.

Coplien, James O. 1996. Software Patterns. SIGS Books.

Steenson, Molly Wright. 2014. Architectures of Information:  Christopher Alexander, Cedric Price, and Nicholas Negroponte and MIT’s Architecture Machine Group. Doctoral dissertation, Princeton, NJ: Princeton University.

Steenson, Molly Wright. 2009. “Loving and Hating Christopher Alexander” presented at the Dot Dot Dot Lecture, April 15, School of Visual Arts, New York City.

Thee are kits of parts (which make) this kit of parts
Semaphores, signs, codes and dictionaries are all types of systems that generate systems
(Title image in Alexander “Systems Generating Systems”, Architectural Digest 38 (1968): 605.


Leave a Reply

Your email address will not be published. Required fields are marked *

  • RSS (Mastodon)

    • daviding: “Pre-announcing April 30 Dialogic Drinks session I'm leading …” April 23, 2024
      Pre-announcing April 30 Dialogic Drinks session I'm leading on "#Yinyang and Daojia into #SystemsThinking through Changes", online 18:30 Singapore, 11:30 London, 6:30am Toronto. Repeating May 2, 8:00pm ET. Official #EQLab notifications
    • daviding: “Diachrony (or diachronic shifts) resurrects a word from 1857…” April 10, 2024
      Diachrony (or diachronic shifts) resurrects a word from 1857, better expressing *changes through time*. A social practice publication in 1998 contrasts synchronic with diachronic.
    • daviding: “Web video introduction of 15 minutes for 1-hour Lunch and Le…” March 22, 2024
      Web video introduction of 15 minutes for 1-hour Lunch and Learn #CentreForSocialInnovationToronto on "Systems Changes Dialogues for Social Innovation" invites practitioners for upcoming monthly meetings. Evocative animated images, details deferred to conversations with mentors.
    • daviding: “Web video of slides from "From Unfreezing-Refreezing, to Sys…” March 21, 2024
      Web video of slides from "From Unfreezing-Refreezing, to Systems Changes Learning" for Dialogic Drinks of #EQLab represents only 1/5 of the time compared to peer-led discussions. Concise hosting called for brevity, and richer presentations. #SystemsThinking
    • daviding: “Hosting multiple Dialogic Drinks on "From Unfreezing-Refreez…” March 8, 2024
      Hosting multiple Dialogic Drinks on "From Unfreezing-Refreezing, to Systems Changes Learning" online, March 12 (Europe), March 14 (Americas), March 15 (Australia). #Leadership meets #SystemsThinking . Short presentations, longer discussions
  • RSS on IngBrief

    • The Nature and Application of the Daodejing | Ames and Hall (2003)
      Ames and Hall (2003) provide some tips for those studyng the DaoDeJing.
    • Diachronic, diachrony
      Finding proper words to express system(s) change(s) can be a challenge. One alternative could be diachrony. The Oxford English dictionary provides two definitions for diachronic, the first one most generally related to time. (The second is linguistic method) diachronic ADJECTIVE Oxford English Dictionary, s.v. “diachronic (adj.), sense 1,” July 2023, For completeness, prochronic relates “to […]
    • Introduction, “Systems Thinking: Selected Readings, volume 2”, edited by F. E. Emery (1981)
      The selection of readings in the “Introduction” to Systems Thinking: Selected Readings, volume 2, Penguin (1981), edited by Fred E. Emery, reflects a turn from 1969 when a general systems theory was more fully entertained, towards an urgency towards changes in the world that were present in 1981. Systems thinking was again emphasized in contrast […]
    • Introduction, “Systems Thinking: Selected Readings”, edited by F. E. Emery (1969)
      In reviewing the original introduction for Systems Thinking: Selected Readings in the 1969 Penguin paperback, there’s a few threads that I only recognize, many years later. The tables of contents (disambiguating various editions) were previously listed as 1969, 1981 Emery, System Thinking: Selected Readings. — begin paste — Introduction In the selection of papers for this […]
    • Concerns with the way systems thinking is used in evaluation | Michael C. Jackson, OBE | 2023-02-27
      In a recording of the debate between Michael Quinn Patton and Michael C. Jackson on “Systems Concepts in Evaluation”, Patton referenced four concepts published in the “Principles for effective use of systems thinking in evaluation” (2018) by the Systems in Evaluation Topical Interest Group (SETIG) of the American Evaluation Society. The four concepts are: (i) […]
    • Quality Criteria for Action Research | Herr, Anderson (2015)
      How might the quality of an action research initiative be evaluated? — begin paste — We have linked our five validity criteria (outcome, process, democratic, catalytic, and dialogic) to the goals of action research. Most traditions of action research agree on the following goals: (a) the generation of new knowledge, (b) the achievement of action-oriented […]
  • Recent Posts

  • Archives

  • RSS on

  • RSS on Media Queue

    • What to Do When It’s Too Late | David L. Hawk | 2024
      David L. Hawk (American management theorist, architect, and systems scientist) has been hosting a weekly television show broadcast on Bold Brave Tv from the New York area on Wednesdays 6pm ET, remotely from his home in Iowa. Live, callers can join…Read more ›
    • 2021/06/17 Keekok Lee | Philosophy of Chinese Medicine 2
      Following the first day lecture on Philosophy of Chinese Medicine 1 for the Global University for Sustainability, Keekok Lee continued on a second day on some topics: * Anatomy as structure; physiology as function (and process); * Process ontology, and thing ontology; * Qi ju as qi-in-concentrating mode, and qi san as qi-in-dissipsating mode; and […]
    • 2021/06/16 Keekok Lee | Philosophy of Chinese Medicine 1
      The philosophy of science underlying Classical Chinese Medicine, in this lecture by Keekok Lee, provides insights into ways in which systems change may be approached, in a process ontology in contrast to the thing ontology underlying Western BioMedicine. Read more ›
    • 2021/02/02 To Understand This Era, You Need to Think in Systems | Zeynep Tufekci with Ezra Klein | New York Times
      In conversation, @zeynep with @ezraklein reveal authentic #SystemsThinking in (i) appreciating that “science” is constructed by human collectives, (ii) the west orients towards individual outcomes rather than population levels; and (iii) there’s an over-emphasis on problems of the moment, and…Read more ›
    • 2019/04/09 Art as a discipline of inquiry | Tim Ingold (web video)
      In the question-answer period after the lecture, #TimIngold proposes art as a discipline of inquiry, rather than ethnography. This refers to his thinking On Human Correspondence. — begin paste — [75m26s question] I am curious to know what art, or…Read more ›
    • 2019/10/16 | “Bubbles, Golden Ages, and Tech Revolutions” | Carlota Perez
      How might our society show value for the long term, over the short term? Could we think about taxation over time, asks @carlotaprzperez in an interview: 92% for 1 day; 80% within 1 month; 50%-60% tax for 1 year; zero tax for 10 years.Read more ›
  • Meta

  • Creative Commons License
    This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License
    Theme modified from DevDmBootstrap4 by Danny Machal