From CasGroup

Revision as of 21:08, 11 February 2011 by Admin (Talk | contribs)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

Self-organization is a process where the internal organization of a system increases, and the organizational change takes place without being controlled, guided and managed completely by the environment (i.e. the system is involved in the increase of organization itself). The name already implies that feedback is an important factor in many self-organizing systems, and in fact feedback and emergence are the two most fundamental terms related to self-organization. As the name suggests, Self-Organization occurs in a system if, left to itself, it maintains its organization or even tends to become more organized. "Left to itself" means without being controlled and guided explicitly by external orders or commands. A true self-organizing system increases and maintains organization without organizer, central control and explicit management through managers, only through context dependent local interactions. Such a system can not be divided into a passive, managed sub-system and an active, managing sub-system.



Various Attemps and Views

Basic Principles
Natural Selection
Red Queen Effect
Swarm Intelligence
Self-Organized Criticality
Butterfly Effect
Control Loop
Lever Point
Frozen Accidents
Path Dependence
Complexity, Simplicity
Edge of Chaos
edit this box

Self-Organization is problematic. The first problem is that it is like intelligence, complexity or life not easy to define. According to John Skår (2003), one way is "to present concepts, definitions, principles and experiments of self-organization. At present there are several concepts, numerous definitions and some principles and experiments". The other would be "to place a number of experience-based examples, samples and observations before the reader". Carl Anderson lists 10 different definitions (2002): Self-Organization is considered to be..

  1. ..indicative of a machine that is "determinate and yet able to undergo spontaneous changes of internal organisation"
  2. ..a set of dynamical mechanisms whereby structures appear at the global level of a system from interactions among its lower level components
  3. ..associated with the spontaneous emergence of long-range spatial and/or temporal coherence among the variables of the (organized) system
  4. ..the spontaneous emergence of coherence or structure without externally applied coercion or control
  5. ..a system is self-organizing if it acquires a spatial, temporal or functional structure without specific interference from the outside. By ‘specific’ we mean that the structure or functioning is not impressed on the system, but that the system is acted upon from the outside in a nonspecific fashion
  6. ..the ability of systems comprising many units and subject to constraints, to organize themselves in various spatial, temporal or spatiotemporal activities. These emerging properties are pertinent to the system as a whole and cannot be seen in units which comprise the system
  7. ..the creation of macroscopical patterns by the action of forces distributed in a much more homogeneous way than the structures that arise. Hence, this kind of transformation implies a spontaneous breaking of symmetry
  8. ..the spontaneous emergence of nonequilibrium structural organization on a macroscopic level due to collective interactions between a large number of simple, usually microscopic, objects
  9. ..a process where the organization (constraint, redundancy) of a system spontaneously increases, i.e. without this increase being controlled by the environment or an encompassing or otherwise external system
  10. ..a process in which pattern at the global level of a system emerges solely from numerous interactions among the lower-level components of the system. Moreover, the rules specifying interactions among the system’s components are executed using only local information, without reference to the global pattern

So what is 'self-organization' really ? It depends on your point of view. Self-Organization means different things to different people, and it is a vague and ambiguous term. There are many different meanings of self-organization. The computer scientist is used to organize everything himself up to the last detail. For large-scale system which are very complex or very distributed he would of course like to have self-organizing systems which are fault tolerant, scaleable, robust and adaptive. If there is no central organizer, then the failure or loss of single elements obviously does not affect the behavior of the whole system. Unfortunately, these systems in nature have also drawbacks, they are more or less unpredictable, unreliable and incomprehensible as well. The biologist thinks that self-organization is related to emergence, swarm intelligence and stigmergy. Camazine et al. define in their book "Self-Organization in Biological Systems" that "Self-organization is a process in which pattern at the global level of a system emerges solely from numerous interactions among the lower-level components of the system. Moreover, the rules specifying interactions among the system's components are executed using only local information, without reference to the global pattern." This 'emergence' viewpoint is useful to explain the behavior of simple swarms, flocks and colonies. The shortcoming of this view is the fact that the units which form swarms, flocks and colonies are themselves often far more complex than the groups which they form, a beehive or an ant colony, for example, is a far simpler system than any individual insect. Another weakness that this definition might appear confusing and irritating, because the environment of the system is not mentioned explicitly, although it plays an essential role in this kind of organization. The chemist is certain that self-organization is some form of autocatalysis, a pattern formation process in a reaction-diffusion system or a kind of self-assembly. Unfortunately inorganic systems that are not alive do not assemble themselves and are far less complex than organic systems. The physicist finally argues that self-organization is something related to self-organized criticality, phase transitions and power-laws. The disadvantage of this view is that these systems are very simple and primitive, for example a pile or heap of sand. This form of self-organization is possible in open systems, where you can import energy and export entropy.

Computer Science Definition Biological Definition Chemical Definition Physical Definition
Applicable to Multi-Agent Systems, Software Systems Swarms, Flocks, Colonies Reactions, Inorganic Substances Ensembles of particles, physical systems
Advantage Useful for Fault-Tolerance, Scalability, Robustness Useful to explain Swarm Intelligence Useful to explain Inorganic Reactions Useful to explain Power-Laws and Critical States
Drawback Unpredictability, Unreliableness, Incomprehensibility Individuals far more complex than the groups they form Organic substances far more complex than inorganic Not useful to explain more complex structures


For the computer scientist, self-organizing systems promise to be fault tolerant, scaleable, robust, for the biologist and the chemist, they promise to explain the extraordinary complexity and high organization found in living and organic systems (maybe offering clues to the origin of life), for they physicist, they promise to help understanding phase transitions, critical states and power-laws. Whether the concept of self-organization is able to keep all of these promiseses or not remains at least partially an open question.


Self-organization is a buzzword. There's no strict agreement on exactly what self-organization is, it remains a very broad, comprehensive and general term with many meanings. The term self-organization in the literal sense (organization without organizer) only says that the system is not organized by someone else. It does not say who organizes the system, and therefore the concept is like other terms which are defined by negation (for instance Non-Linearity) inherently ambiguous, broad, and vague. This is one reason why the concept is so popular, but it is also a reason why it is problematic, for example if we try to transfer examples and results from one domain to another. If we try to formalize the concept through a formal definition, the concept becomes more concrete, but also less comprehensive and less powerful.

Like many mysterious things which appear puzzling, it is partially an illusion which depends on our point of view. The common illusion is that it is a ubiquitous or fundamental principle. Yet this is doubtful. Self-organization is in fact like emergence more a rare phenomena which can be observed at margins and boundaries of systems, and depends therefore on the location of the observer. As we all know nothing organizes itself in daily life. This rareness makes the subject very interesting (for the scientist), but also very frustrating (for the engineer). In other words the reason why the concept is interesting - it is rare and normally the exception - is also the reason why it is difficult to find, to construct and to apply.

Self-organization is often more a desired property than a real phenomenon, especially in systems which a very small or very large, very distributed or very complex, for instance in sensor networks and in large distributed systems. It is of course desirable to build systems that organize themselves automatically if we cannot handle the associated problems manually. The difficulty is that systems which can organize themselves do not always organize themselves in the way we want it.

The common way to solve these problems is to use self-management instead of self-organization: to organize the system by introducing organizers, and to manage the systems by managers with obey our objectives and goals. Self-management by autonomous agents or managers is of course always possible, and hierarchies of managers are in fact ubiquitous in social systems.


The idea of self-organization is old, although in ancient times nearly everything was more or less associated with god, because any idea which contraticted creation was easily regarded as blasphemy. According to C.R. Shalizi, the first statement of a vague concept related to self-organization can be found by Descartes, in the fifth part of his Discourse on Method, where he presents it hypothetically, as something God could have arranged to have happen, if He hadn't wanted to create everything Himself:

"Consider what would happen in a new world, if God were now to create somewhere in the imaginary spaces matter sufficient to compose one, and were to agitate variously and confusedly the different parts of this matter, so that there resulted a chaos as disordered as the poets ever feigned, and after that did nothing more than lend his ordinary concurrence to nature, and allow her to act in accordance with the laws which He had established . . . . I showed how the greatest part of the matter of this chaos must, in accordance with these laws, dispose and arrange itself in such a way as to present the appearance of heavens; how in the meantime some of its parts must compose an earth and some planets and comets, and others a sun and fixed stars. . . . I came next to speak of the earth in particular, and to show how . . . the mountains, seas, fountains, and rivers might naturally be formed in it, and the metals produced in the mines, and the plants grow in the fields and in general, how all the bodies which are commonly denominated mixed or composite might be generated . . . [S]o that even although He had from the beginning given it no other form than that of chaos, provided only He had established certain laws of nature, and had lent it His concurrence to enable it to act as it is wont to do, it may be believed, without discredit to the miracle of creation, that, in this way alone, things purely material might, in course of time, have become such as we observe them at present; and their nature is much more easily conceived when they are beheld coming in this manner gradually into existence, than when they are only considered as produced at once in a finished and perfect state."
Rene Descartes (1637, part 5 of "Discourse on the Method of Rightly Conducting the Reason, and Seeking Truth in the Sciences"), see [1] or [2]

Amnong the pioneers of self-organization in the 20th century are Heinz Von Foerster, Ludwig von Bertalanffy and Ilya Prigogine. Many works in this area emphasized the existence of self-organization in non-equilibrium and non-linear systems. Both terms, non-equilibrium and non-linear, are quite vague and abstract, because there are many different non-linear systems, but only a few linear. An equilibrium has a unified, simple form, but a non-equilibrium has many varieties.

The term "self-organizing" seems to have been first mentioned around 1947 by the psychiatrist, engineer and cybernetics pioneer W. Ross Ashby. William Ross Ashby distinguished later in his paper "Principles of the Self-Organizing System" between two forms of self-organization:

The first meaning refers to a system that starts with its parts separate and whose parts then act so that they change towards forming connections of some type. According to Ashby, such a system is self-organizing in the sense that it changes from "parts separated" to "parts joined", from unorganized to organized structure. This form is related to complex networks.

The second meaning refers to a system which improves its organization. This implies a change from bad or useless forms to good and useful types of organization. He argued that no machine can be self-organized in this sense, unless it is coupled to another machine. In this case one machine is the organizer or manager, and the other is the managed element. This form is related to adaptation, self-management and other self-* properties.

It is remarkable, that Ashby as a pioneer in the area of self-organization claims that there's no such thing as self-organization in many cases. It is clear that no machine, desktop or room is able to organize itself. The ingredients of a cake do not get together and bake themselves into a cake. Later generations appreciateed and embraced the positive aspects of the concept but forgot the negative aspects and limitations.


How is it possible

The question how self-organization is possible arises because self-organization is rare. Normally things do not organize themselves without organizers in daily life. If an organizer, manager or agent is incorporated into the system in order to organize, manage, observe and control it, a self-organization is without doubt always possible. This comparatively simple form of self-management is useful to construct systems with self-* properties. If an organizer, manager or agent does not exist, the problem becomes more interesting.

Different people have different answers to the question if self-organization is possible or not. For the consultant and the biologist it is clear that self-organizing systems exist, whereas for the chemist and the physicist who mainly deal with non-living systems it is clear that systems are usually not self-organizing. The computer scientist is confused and would like to have self-organizing systems with all the desirable properties (robust, fault-tolerant, scaleable,...), but without the negative ones (unpredictable, unreliable, incomprehensible,...), especially if systems become very large, very distributed and very complex.

Computer Scientist Consultant, Biologist Chemist Physicist
Why is it not possible Software is not fault-tolerant or robust and very brittle, it follows only detailed programmed instructions Why ? It is certainly possible.. the only substances that organize themselves are living substances and organic life-forms the second law of thermodynamics says that entropy (disorder) only increases in the course of time
Why is it possible redundancy and replication can increase fault-tolerance, agents can be used for self-management individuals are controlled by selfish genes, swarms can communicate with invisible pheromones hypercycles and autocatalysis can maintain structures through positive feedback open systems can decrease entropy, and complex structures can also appear temporarily at phase transitions or at 'the edge of chaos'
How rare is it ? Very rare (only in some special systems that are build to be scalable, robust and fault-tolerant) Hard to say, because we understand living systems so little yet.. Very rare (only in exceptional cases) Very rare (only in some open systems)

Order and Disorder

Self-Organization is possible in general by transfer of organization across the system-environment boundary (which means import of order and export of disorder)

Self-organization in general (the literal sense) means the evolution of a system into an organized form in the absence of external pressures. This is not a magic process, and it does not mean there is no external influence at all. A self-organization process appears normally in an open system (contrary to closed systems) with a constant input, throughput and output of information or energy. Self-organization is paradox: it arises through the interaction of its components without direct environmental influence, but depends at the same time on open systems and on the inflow of energy or inforamtion from the environment. This paradox between independence or autonomy (closed system with self-* properties) and dependence or embeddedness (open-system) with respect to the environment lies at the heart of the concept.

Self-Organization can be considered a differentiation between "realization" and "inspiration" or "consumer" and "provider": the driving force which inspires and enables the system to get more organized is nearly always the environment, but the actual process of realization is done by the system itself. The process of self-organization needs a provider and a consumer: an external provider of organization in the environment which can be used, and a self-perpetuating consumer process which is able to exploited this source. The difference between a system which is organized by someone and a system which organizes itself is blurred and fluent. Normally system do not organize themselves. If they do, they are often "selfish" or "egoistic", because they are usually open systems that extract order and organization from the environment in which they are embedded.

Another name for the same principle is self-perpetuating reduction in entropy, because a reduction in entropy (related to disorder) is identical to an increase in order and organization. Self-organizing systems also seem to "produce" entropy: they produce waste and increase disorder in the environment. Since entropy is often explained by state spaces, this definition is identical to a move from a large region of state space to a persistent smaller one, under the control of the system itself. There are many similar sounding self-* names and self-* processes like self-organizing, self-perpetuating, self-sustaining, self-maintaining, self-regenerating, self-reproducing. They are all related to each other, since self-organization is a very general and ambiguous term.

Order for free?

The idea of self-organization challenges the idea of ever-decreasing order based on the second law of thermodynamics. However, at the microscopic or local level, the two need not be in contradiction: it is possible for a system to reduce its entropy by transferring it to its environment. In open systems, it is the flow of matter and energy through the system that allows the system to self-organize, to import order and organization and to export entropy and disorder to the environment. This is the basis of the theory of dissipative structures. Ilya Prigogine noted that self-organization can only occur far away from thermodynamic equilibrium. Therefore open systems can exhibit self-organization, while isolated systems cannot decrease their entropy. Biological systems are open systems feeding from the environment and dumping waste and garbage into it. Paul Erdös said "Mathematicians are machines which turn coffee into theorems." In this sense, self-organizing systems are selfish machines which turn order into disorder, available organization into useless waste, and usable energy into unusable heat.

The second law of thermodynamics corresponds to our intuitive expectation that order does not increase spontaneously. Left to itself, a thing does not become more organized. Your desktop does not organize itself, your bookshelf does not order itself, and your coffee-cup does not extract heat from the room to heat itself to the right temperature. Left to themselves, systems become more messy, orderless and disorganized. Self-Organization is possible in open systems, but it is the exception. Order does not "pop up" for free at every corner. As C.R. Shalizi says "when we encounter a very high degree of order, or an increase in order" we expect that "something, someone, or at least some peculiar thing, is responsible". This expectation is right. As already said, Self-Organization is the exception. It does not occur everyhwere.

How do we recognize it

Because self-organization is like organization, pattern and order a vague and ambiguous idea, the concept of "self-organization" or the phrase "order for free" easily lead on the wrong track. In most cases where systems are amazingly complex there is also some other basic process involved, often evolution or metabolism. Quite frequently such systems are the result of a very long evolutionary process. Metabolism is the self-organization process which maintains the continuous growth and regeneration of biological bodies, which is based on the principles built-up (anabolism) through break-down (katabolism).

Three ways to reach self-organization

Self-organization is about the organization of a system. What a system is depends on the view of the observer. A cell, a organ, an organism, a population, or a ecosystem can all be considered as a system. The important notion for self-organization is the boundary between system and environment. It is this boundary or border that defines where the organization takes place. It can be the border between individual element and group (class, type, role, function, job, niche,.. ), or in general between system and environment (surface, facade, shell, ..).

We can recognize self-organization of a system if we look at the border or boundary between the system and the environment. If the organization of the system increases inside of the border, without any direct guidance, control or management from the outside of the border, then one can speak of a process of self-organization.

How do we reach it

In engineering, e.g. software engineering, there are three basic ways to reach and achieve self-organization in principle:

  1. Superimposed: Add a self-* property like self-optimizing, self-healing or self-protecting to an existing system
  2. Systematic Without Emergence: Top-Down without any feedback and emergence
  3. Systematic With Emergence: Bottom-Up with feedback and emergence, often with the help of a code

The problems and drawbacks of each approach are obvious. Approach No. 1 is only superimposed, not built from scratch into the system. There are countless possible self-* properties, and each self-* property has to be added manually. Approach No. 2 is ubiquitous in social systems, and uses a self-management hierarchy or bureaucracy. The drawback is that the hierarchical structure is not really flexible, robust or fault-tolerant. Approach No. 3 is the most interesting, but also the most problematic and frustrating, because emergence is like self-organization the exception and not the rule.

What prevents it

Bureaucracies, egoistic managers and inflexible hierarchies often prevent self-organization. Many companies seem to run well even without CEO. Similarly, many countries seem to function without government, many teams in sport seem to function without trainer, and many orchestras can play without conductors. If all is running well, no CEO, president, trainer or conductor seems to be necessary. The question is then

  • why do orchestras have a conductor (companies a CEO, teams a trainer, countries a president,..) if they do not really need it ?
  • is an orchestra without a conductor (a company without a CEO, a team without a trainer, a country without a president,..) a form of self-organization ?

The answer to the first question is perhaps that an conductor or organizer is especially needed to setup the organization. Once the organization is clarified and written down somewhere, he is no longer needed constantly, at least he does not need to be present permanently. The answer to the second question is no, an orchestra without a conductor may exist, but it is not always a form of self-organization. The organization can also be written down somewhere, for example in form of laws, instructions, detailed notes or sheets of music.

Why is there no army, government, church, university or company completely without central organizer, without general, president, bishop, dean, director or CEO? Many systems would function without them, they are only needed at the beginnging (to hire the right people for the basic positions, and to determine the basic policies, rules and guidelines) and in times of change again. The answers has to do with power and selfish behavior: once they have the power to rule, the founders, managers, directors and presidents don't want to give off/up power again. Although they are no longer needed permanently, the greedy managers stay around because they are hungry for power.

If a country is newly founded, then the parliament is often quite small. During the years, it grows constantly, and the number of delegates and members gets larger every year (as well as their ample salary and their pensions, which are determined by themselves). If the country is old, it suffers from a large parliament and huge bureaucracy. It should be the opposite way round: if a country is newly founded, then the parliament needs to be large, if it is old, it can be smaller. With companies it is similar: If a country is newly founded, then the executive board and management body is often quite small. During the years, it grows constantly, and the number of directors and managers gets larger every year (as well as their ample salary which is partially determined by themselves). Again it should be the opposite way round: if a company is newly founded, then the executive board and management body needs to be large (to work out business plans, strategies, policies and all that corporate governance stuff), if it is old and the company is running well it can be smaller.

The question if an orchestra needs a central conductor is similar to the question if a company needs a CEO, a country a president or a church a bishop. In organizations (whether countries or companies doesn’t matter) leadership is needed especially if the company or country has just been created (to define a direction) or needs organizational change (to redefine the direction and organization). In the meantime, the CEO or president is often superfluous. The same argument applies to an orchestra. A conductor is certainly need at the beginning, to synchronize the start, and at the end, to synchronize the ending, and perhaps in the middle to help those who have long breaks (during “organizational change”). All the rest is basically a “show”, because all musicians have detailed notes, even for the breaks. The conductor is maybe needed before the actual performance, in order to rehearse the composition, to define the direction: to set the general tempo and speed, to modulate the volume, and to evaluate the overall sound. In team sports, teams usually don’t organize themselves. Every team member has a clear role (for instance goal keeper - defender - midfielder - striker) determined by the coach or trainer, who also sets the overall strategy. The trainer is needed at the beginning in order to form a team, and during times of change if the team is reorganized (for example during a game if a player needs to be exchanged).

The most efficient method to manage a system with a set of manager is a flexible and dynamic hierarchy. As long as no management is need, the hierarchy remains flat or empty, and the management positions are only occupied if there is really a need for management. In real life this is hardly possible, because a manager will defend his position even it is no longer necessary for the whole system: a president will defend its own position even if he not needed currently, a conductor will stress its own position even if he not necessary, and a CEO, CTO or CIO may defend its own position even it is no longer necessary for the company. This is a major drawback of self-management. Although the best way to organize something is probably to find someone who feels personally responsible for it, he may continue to stand up for it even if it the thing becomes obsolete.


Principles and Phenomena

The major principles related to self-organization are:

Although many evolutionary and adaptive systems sometimes seem to organize themselves, the concept of self-organization is not directly related to evolution, adaptation and adjustment to the environment, at least not to adaptation by direct contact with the environment. Many forms of self-organization do not rely on fitness, natural selection, or adaptation through direct contact to the environment. It is more a marginal phenomenon that can appear at any border, boundary or margin of a system, whether it is evolutionary, adaptive or complex does not matter.

Attributes and Features

Any system that takes an organized form that is not imposed from outside (by walls, machines or forces) can be considered as self-organizing. Typical attributes and features of self-organizing systems are the following complementary attributes:

  1. self-determination (autonomy) and context-dependence
  2. robustness/resilience and unpredictability/uncontrollability
  3. self-regulation/self-reference (feedback) and embeddedness/openness

The first point emphasizes the duality autonomy/context-depence. In decentralized and distributed systems there is often no central control, and the agents in a self-organizing system are acting autonomously. At the same time their actions are often context-sensitive or context-dependent. Top-down feedback often works through these context dependent interactions. The context (the current situation, circumstances and conditions of the environment, including the positions and behaviors of the other agents) can constrain the actions of the agents, and assign them a certain role. Self-organizing systems are characterized by emergence of global order from local interactions, absence of central or centralized control (autonomy), and decentralized or distributed control.

The second point highlights the duality robustness/uncontrollability. The decentralized or distributed control is closely associated with redundancy, because it does not depend on a single, central coordinator or organizer. On the one hand, such systems are tolerant and resilient against failures, on the other hand, unpredictable emergent properties make the system uncontrollable. Self-organizing systems have an intrinsic redundancy, robustness, and resilience or tolerance despite pertubations and faults.

The third point is about the duality self-reference/openness. Self-organizing systems contain self-* processes. They are closed because they refer to themselves, but at the same time they are open, because they are usually embedded in an environment. The vague terms far-from-equilibrium or dissipative are used to denote a continuous flow of energy or information through a dynamic system which is characterized by consumation of energy or information and production of entropy, heat, waste or disorder.

In living systems, there is a balance in metabolism between catabolism (breakdown and decomposition of complex structures) anabolism (built up and construction of complex structures), which is responsible for self-maintenance and self-regeneration.

Limits of Self-Organization

Size Limit

The concept of self-organization literally only says that systems *can* organize themselves, it does not explain *how* a system organizes itself

The concept of self-organization is most useful if it is applied to the border or boundary of small and simple systems. Over a certain size limit, to concept loses its explanatory power. Of course you can say that the earth as a whole is a self-organizing system which consumes order and energy (in form of sunlight) and which produces disorder and entropy (in form of heat). Yet the same can also be said about any living organism or any computer. A living organism or a digital computer also consume energy and produce heat, although they all work in an completely different way. Therefore the concept is most useful for a large number of small elements and less useful for small number of very large elements.

Another example where the concept of self-organization becomes less useful is the brain. There is certainly no central master neuron which determines the organization of the other neurons, and if one neuron fails, it does not affect the function of the whole system. Yet there are so many other processes involved - growth, development, evolution, adaptation, learning - that the concept of self-organization alone says nothing. Without constant learning from others a brain while neither learn a language nor an intelligent behavior. Schools with teachers are not for nothing an established institution.

Evolution and Self-Organization

Intricate organization is the hallmark of a a complex system - for example an economy, a weather system, an immune system, a liquid or a brain. Although the elements and elementary components are simple - for example atoms or agents - the organization and organized interaction makes the system complex. Such a system is what it is and does what it does because of the way in which its constituent parts are organized and not because of what they are. It can only be understood in terms of its parts and the interactions between them.

Since in many natural systems no central organizer is visible, the systems often appear to be the result of self-organization. But many of these systems are also subject to and result of evolution, the most basic principle of biology. In fact evolution and natural selection act on nearly all living systems, and in many cases their effect is much stronger than short-lived forms of self-organization.

Self-Organization in isolated systems is often an illusion or linked to evolution

Self-Organization in isolated systems is often an either an illusion (related for example to strong fluctuation or a sudden transfer of organization or complexity from one sub-system to another) or linked to evolution and other basic processes, for instance evolution together with increased division of labor and cooperation.

Limits and Codes

With initiatives and visions as organic computing and autonomic computing, scientists and engineers try to construct biologically inspired systems. The hope is that by learning from organic systems, we can discover and apply new forms of distributed and decentralized organization. Yet there is probably no mysterious unknown principle of self-organization. Many elementary tools such as pheromones in ant colonies and waggle dances in honey bee swarms are well known, and they are a infact a concrete form of language, a code which defines where the food can be found.

Even nature and its universal tool, evolution, seem to avoid systems with very high complexity. If the number of combinatorial possibilities and possible combinations gets to large or explodes (as it can be observed in cases of "strong emergence"), then a combinatorial limit is reached. In this case evolution often forms closed entities like cells or organs, and finally invents a new code (genetic code in form of DNA or memetic code, i.e. normal language). Thus the emergence of a new code marks the limits of self-organization.

A code is a set of rules, instructions and symbols, which can store information and messages. It can be used to control and organize processes. With the genetic and memetic codes, evolution created a model in form of a serial language or linear string, which is a new layer of abstraction and at the same time a tool to describe and control activities and processes. This code can also be a binary code (0/1), a dynamic code as a waggle dance used by honey bees, or a pheromon code in form of a gradient field.

Useful information stored in a code can be used, re-used, altered and changed later on. A code reduces dependency from the current moment and increases independence and autonomy. For evolution, a new code means the gateway to a new evolutionary system.

Human engineers have designed and invented a lot of codes: the binary code independent from electronic devices, assembler code independent from specific binary coding of instructions, high level programming languages independent from specific processor instruction sets, and finally platform independent Java bytecode and the code of the Common Intermediate Language (CIL) from .NET.

Web 2.0 and Wiki

The latest and most "trendy" form of self-organization can be observerd in Web 2.0 systems and Wikis like this DCS-Wiki. This form of self-organization through tagging and blogging based on file-, link- and information-sharing, is more social than technical, because it is based on user-generated and user-organized content. It seems to work best in very large systems like Wikipedia (in fact the more the corresponding community believes that it works, the better it seems to work) and less good in smaller systems, in other words it is often absent in small groups and present in very large groups. Therefore self-organization is the exception here too, because there are of course many small groups but only a few very large groups.

In small Web 2.0 systems, even if they work in the same way as larger successfully working systems (for instance small local Wikis vs. large global Wikipedia), only those people participate who are directly associated with it, i.e. those people who had the idea to setup the Wiki in the first place. Only a few idealists will create an account to contribute his or her ideas voluntarily. If someone contributes something, than usually in order to advertise something: a website, a service, a book, etc.

Internet groups like Newsgroups, Google Groups, Yahoo! Groups, MSN Groups have been around for a long time (since the first days of the Web) and they share a common social phenomenon. One can call it perhaps "group threshold", "self-organization threshold" or "selfish posting". In every internet newsgroup, discussion group or blog usually 10% of the group members do 90% of all postings, and the majority of members will not post many messages. The group-owner or founder is usually among the 10% minority, because he has an inherent incentive to support the group ("his group"). Often he is the only one who posts anything.

The question is why is self-organization absent in small groups (for example small WIKIs) and present in larger groups (for example WIKIPEDIA) ? When does self-organization start and set in ? A small local WIKI works in the same way as the larger WIKIPEDIA, but in the former usually only the owner or founder makes contributions, and in the latter people contribute voluntarily. Is there a kind of self-organization threshold, critical value, or tipping point for this form of social systems? If such a threshold exists, can we quantify it? It is larger than 7, 70, 700 or 7,000 ? The threshold seems to be rather diffuse, blurred and fuzzy, because it is a social and psychological phenomenon related to many causes and factors:

  1. the feedback illusion that only arises if the system is something "big": the user gets only the feeling of being part of something "big" if the system is already "big". The participants think they have a big and lasting effect, gain a higher reputation or a better image, and the benefit of being known to be smart about something. In reality they often have a small and non-permanent effect, gain little reputation, and are not known at all. This illusion is larger in big systems.
  2. the motivation through a big common enemy in form of a global "superpower" exists only in "big systems" (for instance Microsoft in the case of Open Source software or the expensive Encyclopedia Britannica in the case of Wikipedia)
  3. the incentive and reputation gained through ranking lists and statistics becomes bigger in big systems

Articles and References



  • Principles of Self-organizing Systems, Ross Ashby, Originally published in 1962 by H. Von Foerster and G. W. Zopf (Editors), Principles of self-organization, London: Pergamon, 1962. p.225-78, re-published in the E:CO journal with an introduction by Jeffrey Goldstein



G. Di Marzo Serugendo et al., Self-Organisation: Paradigms and Applications, in "Engineering Self-Organising Systems: Nature-inspired approaches to software engineering", G. Di Marzo Serugendo, A. Karageorgos, O. F. Rana, F. Zambonelli (Eds), LNAI 2977, Springer-Verlag, 2004.



Scott Camazine et al., Self-Organization in Biological Systems, Princeton University Press 2003, ISBN 0691116245

Stuart A. Kauffman, The Origins of Order: Self-Organization and Selection in Evolution, Oxford University Press 1993, ISBN 0195079515

F.Eugene Yates (Editor), Self-Organizing Systems: The Emergence of Order, Plenum Press 1988, ISBN 0306421453

Erich Jantsch, Self Organizing Universe: Scientific and Human Implications, Pergamon Press 1980, ISBN 0080243118

Marshall C. Yovits, Scott Cameron, Self-organizing Systems, Proceedings of an Interdisciplinary Conference May 1959, Pergamon Press, 1960

Marshall C. Yovits, George T. Jacobi, and Gordon D. Goldstein (Eds.), Self-Organizing Systems, Spartan Books, 1962

Heinz von Foerster, George W. Zopf (eds.) Principles of Self-Organization, Pergamon Press, 1962

Personal tools