Bas Reus' quest on self-organization and online collaborative spaces

Complex Adaptive Systems, my understanding

Posted in philosophy, self-organization by Bas Reus on October 23, 2009

Some commenters on previous posts on this blog referred to CAS or Complex Adaptive Systems. This term is somewhat fuzzy for me, as I’ve never read about CAS before. So now is the time to do so. A first lookup in Wikipedia is always a good start, so that’s what I did. I must say, the C in CAS already becomes apparent when you look at the definitions. One of the definitions that is mentioned is the following:

A Complex Adaptive System (CAS) is a dynamic network of many agents (which may represent cells, species, individuals, firms, nations) acting in parallel, constantly acting and reacting to what the other agents are doing. The control of a CAS tends to be highly dispersed and decentralized. If there is to be any coherent behavior in the system, it has to arise from competition and cooperation among the agents themselves. The overall behavior of the system is the result of a huge number of decisions made every moment by many individual agents.

Water dripsSo this definition says that a CAS is a network, where many actors act for themselves in a response to their (changing) environment. If I interpret this correctly, human behaviour is a CAS as well. Almost all humans are connected to each other via a number of other humans. Or the Internet is a CAS, where many endpoints are connected to the same network, they determine the network, they are the network. Or maybe the universe and evolution as well.

My interpretation is that we use the term CAS when we do not understand the behaviour of a system or phenomenon or when it can’t be controlled. Examples that are given are ant colonies, stock markets, the ecosystem, or political parties. All are difficult to understand, if they can be understood at all, and even the actors in it probably do not understand their system that they are part of, for example the politicians in a political party or the ants in the colony. These systems or phenomena can’t be controlled, their behaviour can seem unpredictable. And that’s a good thing, the urge to control is overrated very much. Maybe some influence can be desired sometimes, if possible.

The Wikipedia article also states that the principles of self-organization and emergence are very important in these systems. The relation between self-organization and CAS became apparent in the discussion on self-organization as well. But then we come to the differences between human beings with a mind of their own, and other players like ants or cells. Can self-organization occur in an organization where people are involved? Or is it just not possible because we can think for ourselves and can act by reason? However, the latter is a philosophical discussion. Do we act by reason or by drifts for power? The philosophers Immanuel Kant and Friedrich Nietzsche thought about that very differently. So maybe this discussion is always a philosophical one.

If we go back to the definition, the C in CAS is only true when you look at the phenomenon from a birds-eye perspective. All the actors deep down in the system are probably not aware (if they could) that they are part of the system, and just follow simple rules. So from their perspectives, there is not much complexity. They adapt to their environment, like a water drip just follows the easiest path. This drip is not aware of the ecosystem that it is part of, just like the system is not aware of the single drip. However, it is possible to influence the flow of the water, because we understand the characteristics of water. But it is not possible to influence the whole system where water is a part of, it’s just too complex.

Translated to organizations, complexity is there or not depending on the perspective you’re in. The higher in the hierarchy, the more complex the organization as a whole seems to function. If you are high in the organization, you’re aware of the size of the organization, and therefore aware of the variety of actors. How they all interact, is difficult to grasp. The lower in the hierarchy, the less you are aware of all the other players that exist in the organization, and the more focussed you are on your tasks which are relatively not complex at all. Well, that’s my understanding at this point.


20 Responses

Subscribe to comments with RSS.

  1. sourcepov said, on October 23, 2009 at 18:02

    Great summary Bas, thanks for posting. As always, you have a knack for getting us thinking.

    Will leave ants & philosophy for experts to comment, as those are foundations of CAS thinking. They definitely get us in the right mindset. Another good metaphor is “dancing landscapes” which speaks to constantly changing environment variables, vs. a static set of external forces. That is a key “real world” variable that helps explain complex events in business, economics, socio-political systems, etc.

    Peter Senge makes reference to complex factors as the “Fifth Discipline” aka ‘systems thinking’ (1990). I am still working through those correlations, but I think there’s an important link.

    I am especially intrigued by your last paragraph, as you have introduced to me a new concept of “relative” complexity, that is, the degree of complexity is a function of levels in the hierarchy. Using an organization chart as a quick visual example: as you get closer to a specific task-related function, the apparent complexity diminishes.

    That makes sense: fewer actors, fewer interactions, fewer variable interdependencies, less complexity.

    So I would assume that understanding and attempting to harness or influence CAS is also a function of establishing the right ‘complexity context’. And if you get granular or atomic enough in your analysis, the complexity goes away?

    Fascinating. Haven’t seen that one discussed yet.

    Btw my primary source for this material has been Scott Page (U.Mich) who has written on this and published some lectures, which I highly recommend. Another good source extending beyond CAS is M.Waldrop (@mitchwaldrop): Complexity: Emerging Science at the Edge of Order & Chaos”.

    My work in this space is on social ecosystem change. We have built public collaboration models to drive issue and solution framing for complex ecosystems, and we have tried to incorporate “CAS thinking” to our process. It includes focus on the 4-core CAS drivers: a) diversification, b) connection, c) interdependence/engagement, and d) adaptation/learning.

    We chat about it under #ECOSYS w/ framing here:

    Social media, Twitter in particular, creates a significant breeding ground for what I’m calling CAS “learning cells” – in the vernacular, problem-solving communities that self-organize.

    Are you a CAS stakeholder? Let me know, and I’ll add you

    Exciting stuff. Like Bas, would love your feedback.

    Chris (@SourcePOV)

  2. uberVU - social comments said, on October 24, 2009 at 01:21

    Social comments and analytics for this post…

    This post was mentioned on Twitter by bottomup: @ChrisPRodgers @johnt @sourcePOV @rotkapchen @StephenBilling @TomgibbonsTMS: New post on #CAS, read my understanding

  3. Jacques Knight said, on October 26, 2009 at 16:33

    Hello Bas,

    Nice post.

    Notice how we are not aware of most of the processes going on in ourselves, or how management is not aware of most of what is going on in an organization. So, although as you said complexity increases with scale, the intelligence of the actors seems to be constant. Every actor knows what he is doing to a certain point beyond which the consequences become too complex.

    Jacques (@Jacques_Knight)

  4. Bas Reus said, on October 26, 2009 at 17:09

    Chris, nice to elaborate some more on relative complexity or complexity context. I think it’s not only a function of levels in the hierarchy, but if is a function at all, you should account the heterarchical (see Stephenson) levels as well. Formal or informal connections, in terms of complexity it should not make a difference.

    I think it’s time for me to read some material of Page and/or Waldrop…

    Thanks. Interesting point of view you state here. What are/aren’t we aware of? What can/cannot we be aware of? Is awareness predefined by our limited capacity? Or is there a lot of tacit knowledge that we are aware of, but we do not or are not able to share?

  5. John Tropea said, on October 27, 2009 at 03:03

    Pithy post Bas…and thanks for making me think and allowing me to comment to consolidate my thinking.

    Steven Johnson nicely puts a CAS as

    “when the emergence becomes useful to the livelihood of the system it is part of (self organisation)”

    Basically you can’t understand the system by looking at it’s parts…reductionism (which is how science sense-makes)

    Snowden says:

    “A complex problem is not the sum of its parts. It cannot be broken down with each solution aggregated; it must be solved as a whole.”

    Also see systems biology

    You got it right here, when you say:

    “we come to the differences between human beings with a mind of their own, and other players like ants or cells. Can self-organization occur in an organization where people are involved? Or is it just not possible because we can think for ourselves and can act by reason?”

    And I think Stephen Billing made this clear

    Definitely a difference with CAS where agents act based on hormones or gene instructions, compared to humans

    Steven Johnson talks about free will, self consciousness as the defining characteristic

    “…the intelligence of the colony actually relies on the stupidity of its component parts

    “it’s very hard to imagine any human society in which people would go around responding to what happened at that moment without any conception of why they’re doing what they’re doing”

    “Human behaviour works at two comparable scales…driving a car has short-term and long-term consequences.
    The short term influences whether we make it to soccer practice on time; the long term alters the shape of the city itself.
    We interact directly with, take account of-and would seem to control-the former. We are woefully unaware of the latter”

    And I love how you say the same thing, but using the perspective of a birds eye view:

    “If we go back to the definition, the C in CAS is only true when you look at the phenomenon from a birds-eye perspective. All the actors deep down in the system are probably not aware (if they could) that they are part of the system, and just follow simple rules. So from their perspectives, there is not much complexity.”

    I too have come across this way of looking at it when reading Steven Johnson’s book, “Emergence”

    “but while cells lack a bird’s-eye view of the organism that contains them, they can make street-level assessments via the molecular signals transmitted through the cell junctions. This is the secret self-assembly: cell collectives emerge because each cell looks to its neighbours for cues about how to behave.”

    And here is a similar excerpt when looking at what unexpected macro things emerge from local behaviours in cities

    “…residents arent setting out to build bigger settlements; theyre all solving local problems, such as how to make their fields more productive, or what to do with all the human waste of a busy town. And yet those local decisions combine to form the macrobehaviour of urban explosion”

    Like SourcePOV I too like your mention of relative complexity:

    “The lower in the hierarchy, the less you are aware of all the other players that exist in the organization, and the more focused you are on your tasks which are relatively not complex at all. ”

    Isn’t this the whole problem of org units not being aware of each other in order to be a more cohesive org…that’s why I like the idea of ambient awareness.

    In an organisation we have local behaviours as people and our teams/units…and all the people and the teams together result in sometimes really bad emergence, ie our self organisation surfaces uncooperative and unsustainable results/outcomes as we are not aware of each other’s activities and goals, and how our actions effect each other.

    So yes, we humans are complex creatures, we self-organise, display emergence, but we don’t natually adapt to the best interest of the health of the system (ie the organisation), as we are not a CAS.

    Jaques, it’s so true when you say “that complexity increases with scale”
    …try to coordinate a lunch outing for 20 people…

    “and the intelligence remains constant”
    …that’s why we have coordinator…as the attendees are still doing their thing in their bubble…it’s up to the coordinator to juggle the complexity. I think the next step is a facilitator where we create conditions for adaptation, and where we stand at the fringes guiding it.

    I guess in business we define the macro picture up front, and try to harness people in their local environments to be aware of the macro picture (cooperation).
    So as we defined previously, because of volition and self consciousness/awareness humans are not a natural CAS, but we are complex, and we can manually add the adaptiveness part by manipulating or harnessing the environment with rules and feedback.

    Snowden talks about attractors and dampeners acting as feedback and boundaries in order to facilitate the system to adapt.

    An example is a thermometer has a goal of equilibrium by realigning itself when it receives negative feedback…it’s acting to a rule of homoestasis.

    The important thing now is to move on from the philosophical stuff and try to cope with organisations by creating manual conditions for adapting as Snowden is demonstrating. We know our flaws (whoa, I never thought of free-will as a flaw until the big picture), so lets act on sensing and guiding them to a better place.

    My blog is down at the moment, so here’s an excerpt from my post-km post:

    Steven Johnson gives an example of programmed billiard balls that alter their movement when interacting with other balls…he calls this complex behaviour,

    “a system with multiple agents dynamically interacting in multiple ways, following local rules and oblivious to any higher-level instructions”

    “But it wouldn’t truly be considered emergent until those local interactions resulted in some kind of discernable macrobehaviour.” eg. the balls end up on either side of the table in clusters, even on one side and odd on the other.

    “That would mark the beginnings of emergence, a high-level pattern arising out of parallel complex interactions between local agents…the balls aren’t programmed explicitly to cluser in two groups…yet out of those low-level routines, a coherent shape emerges.”

    But he goes on to say that this is not adaptive, until it becomes useful.

    eg. if it was in the interest of our pool hall to attract players, it would be adaptive behaviour for the balls to end up forming one cluster in a triangle shape with the white ball on the other end…as this is useful.

    “The system would use local rules between interacting agents to create higher-level behaviour well suited to its environment. Emergent complexity without adaptation is like the intricate crystals formed by a snowflake: it’s a beautiful pattern, but it has no function”

    He talks about emergent behavior becoming smarter over time and responding to environmental changes.



  6. Stuart G. Hall said, on October 27, 2009 at 11:36

    Interesting point about relative complexity, thanks. I try to act using CAS ideas by knowing that project issues for example will often ‘sort themselves out’ if I let them. IOW there are already some simple folksonomical tools that help ppl lower down the org hierarchy take advantage of CAS as practical tools.

  7. Gordon Rae said, on October 27, 2009 at 12:03

    I want to dissent from John Tropea’s comment “you can’t understand the system by looking at its parts” because we all know that you can. In fact, it’s hard to say anything meaningful about a CAS unless you talk about how the parts interact with one another to create outcomes. For example, doctors can diagnose kidney failure from the build up of waste products in the blood. What you can’t do is solve problems with the system by aggregating the solutions you devise to problems with the components.

    Take for example a rugby team that is unable to score a try. You can only do so much by taking each player one at a time, and improving their throwing and catching skills. At some point, you have to put the players on the pitch together, and let them interact.

    You might still find the problem is a component that needs to be replaced – for example Jones can’t catch the ball when it’s passed from his left side – but you can’t assume that all problems will be solved that way.

  8. Stephen Billing said, on October 27, 2009 at 12:16

    It’s very interesting to see you ‘thinking out loud’. Good learning process.

    I think the relative complexity idea is a red herring. Yes, the numbers of connections in a network grow exponentially as the number of people in the network grows. But another key tenet of complexity is self-similarity, which is to say that the patterns are similar at all levels in the CAS hence, fractals. So in a CAS, all levels are complex.

    You argue that it is simpler for someone at the front line than higher in the hierarchy and this has given birth to the idea that the higher in the hierarchy experience higher levels of complexity. You are referring here to power differentials, but not to different levels of complexity. The idea of a CAS is that the agents are interacting without any individual agent being able to control the pattern of interactions that emerges. This is known as self-organisation. So, by analogy, even the most senior person cannot control the interactions that occur – hence the failure of so many management edicts.

  9. Bas Reus said, on October 27, 2009 at 17:02

    Wow, thanks for your responses. Good learning process indeed. John, thanks for all your examples.

    There are at least two ideas or concepts that are not agreed upon here in the discussion:
    – systems (related are systems thinking and systems theory)
    – relative complexity

    Systems thinking is a subject that I yet have to explore more in depth. I will probably will come back to these later. It is a subject that many people interpret differently, or just can’t seem to get any consensus on.

    Relative complexity is something I refer to in this post. Stephen argues this idea is a red herring. Let me elaborate on this. I do not refer to power differentials, nor do I necessarily refer to more perceived complexity on higher hierarchical levels. It really is a network thing. It can be that people that are lower in the hierarchy have many informal connections, that will increase the complexity. If it is a function of something, it is a function of connections in the network you’re in, how many connections you maintain, formal and informal combined. Chances are that the higher in the hierarchy, the complexity will increase, but not necessarily.

    Still, the question is whether a CAS should be compared to an organization. I think it can not, the A in CAS, and perhaps the S as well are problematic there. And Stephen for example explains that in CAS, all levels (all nodes in the network?) are complex. But the C in CAS can be applied to organizations. An organization can be complex, but not a CAS. However, looking at CAS made we coin the term relative complexity when complexity is translated to organizations.

  10. Bill Free said, on October 27, 2009 at 21:17

    I agree that organizations are not systems, but individuals within organizations use systems to accomplish their work.

    That said, the systems that operate within organizations tend to be programmatic, driven by reactive as opposed to generative thinking and thus not adaptive. What Senge calls the “learning organization” essentially embraces CAS thinking at all levels. From this perspective, Stephen has it right when he calls “relative complexity” a red herring.

  11. Jacques Knight said, on October 28, 2009 at 00:21

    re : relative complexity

    There is a sense in which complexity increases with scale*, but this sense is not relevant for CAS, because of emergent patterns. This fits with the idea of constant intelligence across levels of organization. Management does not know/control all that is going on in the organization, but also does not need to know/control everything.

    * Specifically, for an outside observer. What we are looking at here is the actor’s perspective.

  12. Bas Reus said, on October 28, 2009 at 10:11

    Bill, Jacques, thanks for your comments.

    As I interpret you correctly, a CAS is complex and adaptive on all levels by nature. Therefore we cannot speak of relative complexity. Probably that’s another discussion.

    I’m curious what you think of ‘learning cells’ that Chris (sourcePOV) is referring to in the first comment. Can these be compared to CAS? He uses Twitter as an example.

  13. John Tropea said, on October 29, 2009 at 01:51

    In the last paragraph in this post Snowden says organisations are complex adaptive systems:

    “Common perceptions of the work world as machine-like and ordered, and thus subject to the rules of order, are cultural legacies of the industrial revolution that still blind us to the fact that organisations are in fact complex adaptive systems.”

    …whereas the comments in this blog post would say that orgs are complex, people having conversations (interactions) are self-organising, good or bad emergence occurs, but given all this an org is not a system, let alone adaptive.

    I understand this, but what do you call it when you take an approach as Snowden does to create conditions for an org to become more adaptive.

    Here’s a great excerpt from the same post:

    “In the idealistic approach, the leaders of an organization set out an ideal future state that they wish to achieve, identify the gap between the ideal and their perception of the present, and seek to close it. This is common not only to process-based theory but also to practice that follows the general heading of the “learning organization”. Naturalistic approaches, by contrast, seek to understand a sufficiency of the present in order to act to stimulate evolution of the system. Once such stimulation is made, monitoring of emergent patterns becomes a critical activity so that desired patterns can be supported and undesired patterns disrupted. The organization thus evolves to a future that was unknowable in advance, but is more contextually appropriate when discovered.”

    So what are the defining characteristics that don’t make an org a system?

    Stephen can you elaborate on this post
    When you say a change in the macro causes changes in the parts of the system is not true in an org, as there are people involved…we are not just robots. Management wonder why changes don’t filter down or take effect, as they are not taking into account the people. As mentioned before Snowden says this ideal future state approach is idealistic, or not realistic anyway.

    Can you give an example of a system compared to an org, where a change in one part will not be resisted by another part?

    The systems thinking page on wikipedia says:

    “component parts of a system can best be understood in the context of relationships with each other and with other systems, rather than in isolation.The only way to fully understand why a problem or element occurs and persists is to understand the part in relation to the whole”

    To me this describes organisations (even though it is said here they are not systems), as what one department does may affect another department. So to understand what’s happening in one department and why it’s happening, it’s good to look at other departments as what they are doing is having impact on the department in question.

    And this comes back to silos in orgs and lack of awareness, therefore not as agile.

    I think social computing tools like networks, microblogs, blogs, forums, wikis are enabling more silo busting, transparency, communication…so we can be more aware and therefore adaptive.

    Stephen Billing always says that the key is conversation, closing that gap between management and workers, and I think these social tools amplify this idea.

    Gordon, I’m still thinking about your reply, and totally agree that aggregated solutions to the parts will not solve system problems.

    I guess what I was saying is that to solve a problem with a part may not be effective by just looking at the part, you have to look at it’s relationship with other parts.

    I think this applies to both “complexity” and “complicated”

    A car is not complex, it’s complicated…you can take it apart and put it back together, you can look at a part and fix it. But at the same time this part may break again, as you need to fix another part that is causing it to break. So things that are complicated still have relationships between the parts.
    But what I read from Snowden is that what makes it “not complex” is that all the parts are known.

    Click to access 37_Intranet_as_complex_ecology_final_.pdf

    Whereas a gardeners environment is more complex, she can do everything to make her garden grow effectively, but their are always unknowns like bad weather, etc

    Snowden goes on to say:

    “For a complex system we need to create an ecology; this is achieved by drawing boundaries between spaces to reduce uncertainties and intervening to encourage growth, such interventions are best if they take the form of simple actions that organically evolve into complex and hopefully desirable forms of behaviour.”


    Maybe the gardener can create a permaculture environment to best cope

    Gordon, when you mentioned aggregated parts is not the same as the whole, it reminded me of something Clay Shirky said that describes “more is different:

    “sociology is not just psychology applied to groups; individuals in group settings exhibit behaviours that no one could predict by studying single minds”

    “no one has ever been bashful or extroverted while sitting alone in their room, no one can be a social climber or a man of the people without reference to society, and these characteristics exist because groups are not just simple aggregations of individuals”

  14. sourcepov said, on October 29, 2009 at 02:49

    Getting traction via blog comments can be difficult. I find semantics and the theoretical and/or practical contexts vary widely from person to person. This makes for many interpretations.

    Simply saying ‘organization’ gets us in trouble because it can bring to mind a static ‘organization chart’ which is clearly hierarchical, somewhat fixed (not adaptive) and less of a system than it is an abstraction of relationships. I even made reference to organization charts and hierarchies, which reinforced that view. I shouldn’t have gone there.

    More typically, and certainly relative to CAS, when I say ‘organization’ I mean it as ‘a team of people working together to solve a problem’ which is probably better described as a ‘work group’. In this light, the CAS aspects become more clear. (C)omplexity exists in the emergent value that may be produced, the parts can self organize, and for practical purposes, the actors/agents can’t see the macro effect of what they’re doing. Any assembly of humans that interact can produce (A)daptation (aka learning) which I believe is what produces the emergence, and the flow of inputs, addition of value (useful work) and outputs qualifies it as a (S)ystem. So I think where Snowden and Senge are wanting to take us is in the direction of CAS as a functioning ‘Learning Organization” comprised of work groups .. not a static hierarchical abstraction.

    My interest in making the CAS linkage to Learning Organizations is that Scott Page (U.Mich) outlines 4 control variables that create conditions for emergence to occur, as I outlined above. These are diversification, connection, interdependence, and adaptability. If work groups as Learning Organizations can be tuned to produce what he describes as a “sweet spot” or “interesting in between” (not too much of the 4 variables, and not too little) then we might gain important insight on how to optimize such work groups to maximize innovation.

    If there are issues with these conclusions, let’s revisit definitions first, to make sure we’re talking about the same thing.

    If you know of authoritative sources on CAS, please share them.

    I’m no expert on CAS .. I’m a student of it .. but I truly want to drive a fuller understanding, so that work in collaborative innovation can be both informed and optimized. I’ve been fighting silo-thinking and dysfunctional work groups for a very long time. CAS (in context of ‘systems thinking’) as a way to optimize work groups truly resonates with me .. I think there’s something here ..

    Thanks again, Bas, for teeing this up ..

  15. sourcepov said, on October 29, 2009 at 04:14

    And btw if the ‘relative complexity’ idea is a red herring, that’s fine by me .. it may simply be the ‘complicated’ v. ‘complex’ debate coming back to visit.

  16. John Tropea said, on October 29, 2009 at 07:22

    SourcePOV – I wouldn’t put Senge and Snowden in the same bucket, as Snowden is a naturalistic approach, rather than an idealistic approach “fake intervention: the top down determination of what is right”

    I posted excerpts of this in my last comment, but forgot to include the link

  17. Rotkapchen said, on October 29, 2009 at 17:11

    Gordon: The kidney analogy is actually a classic one for supporting John’s point. Western medicine fails us for anything but total failure: emergency medicine. We’re good at emergency medicine — obvious failure. But because Western medicine lacks a systemic view it cannot help identify larger issues or help with ‘real’ diagnosis.

    A classic example given is one where a series of children all had liver conditions. All would have been prescribed the same medication, but in Eastern medicine each would have been given a different treatment because the conditions were different.

    Western medicine lacks the context necessary for relevant diagnosis and treatment — again, except in the case of abject failure that can be ‘measured’ or ‘seen’. This accounts for about 5% of all medical care needed.

    I’d be willing to jump to an analogy to business here and apply the same.

  18. […] are not (well maybe) Complex Adaptive Systems, so we need to make them open and transparent as much as we can, so people can be ambiently aware, […]

  19. […] intentions or local behaviour involved in the the Just-in-time vs Just-in-case concept actually emerge a macro picture…and that’s a change in the internal dynamics of an organisation from a competitive to […]

  20. ソース said, on December 4, 2013 at 12:26

    →一番不可解な提案。保育所呼び出しは半年経過するとなくなるのですか? マタニティ 私はおなかが大きくなってからも付けていました。

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: