I’m not a systems engineer, nor an expert in complexity science. But I have been a long-time student of a certain kind of complex adaptive system, the Canadian economy and the countless economic agents who make up that system. And all complex adaptive systems share some common features, which means that analyzing the behavior of any complex adaptive system requires some common understandings and is prone to many of the same pitfalls. But let’s back it up a few steps and start by defining what ‘systems’ are, complexity in relation to systems, and then explain what a complex adaptive system is.
According to Wikipedia, a system “is a group of interacting or interrelated elements that act according to a set of rules to form a unified whole”. Systems can be simple, deterministic, complex, stochastic, or some combination thereof. For example, a simple stochastic system might be an old windmill powered textile loom, which depended (at least in part) on the randomness of gusting winds; the stronger the winds, the faster the looms turn, the more yarn produced.
Complex systems have certain features that make those systems, you guessed it, ‘complex’. Let’s list and then briefly summarize what those characteristics are. Complex systems have the following attributes: (1) nonlinearities, (2) feedback effects, (3) spontaneous order, (4) robustness, without a centralized system of control, (5) emergence, (6) hierarchical organization, and lastly (7) numerosity.
1. Nonlinearity means the system is not characterized by linear relationships i.e. it lacks linearity. Linearity means that algebraic operations apply: adding, subtracting, multiplying by scalars and so forth. In a linear system, we can conceptually “add any two solutions to the equations that describe it and obtain another, and multiply any solution by any factor and obtain another”**. In complex adaptative systems these algebraic manipulations don’t work.
2. Feedback means that the way individual parts, or subgroups, within the system interact with adjacent components of the system at a later time, depends on how it interacted with them in preceding times, and the times before that, and so on. This property is closely related to the concept of recursion.
3. Spontaneous order means there is no single coordinator of the system; rather, order springs forth from the aggregation of innumerable “uncoordinated interactions” between all the individual parts making up the system. Another way to understand spontaneity is that the arrangement of complex (adaptive) systems at any given moment is partly random, but also partly deterministic.
4. Robustness and lack of central control is partly the inverse of spontaneous order and lack of planning. The robustness connotes that the system is stable when there are perturbations, disturbances, and minor shocks, and can rapidly absorb these disturbances without any disruption to the system. Another interpretation is the “ability of a system to correct errors in its structure” with a built in “error-correction mechanism”.
5. Hierarchical organization means that there are multiple layers within the system that form “a hierarchy of system and sub-system”.
6. Numerosity means that the system is comprised of numerous, if not a countless number of parts, and these parts “are engaged in many interactions”.
7. Emergence is the final characteristic of complex systems and perhaps the hardest to explain concretely. So loosely speaking, emergence is the ability of a system to react to changes in the lower rungs of the system, leading to consequent changes that ripple upwards to the higher echelons of the (hierarchical) system. In most systems, impacts are thought to flow downwards from the influences wrought by higher-level forces, senior-ranked persons, benevolent planners, and so forth, not the other way around.
Another interesting feature of complex systems is that they are “neither completely ordered nor completely random”. This makes understanding such systems even more difficult because the web of interrelationships combines deterministic and stochastic links in a way that precludes our ability to disentangle complex systems into constituent components. Consequently, these systems cannot be studied with standard, ‘concise’, analytical models that were devised to explain phenomenon that is either solely deterministic, or solely stochastic, but not both.
Now much of the above discussion applies to complex systems more generally, so what does the extra adjective ‘adaptive’ add in terms of how these systems work? Complex adaptive systems have a few extra features, but two of the more important are: (1) the constituent parts within the system mostly interact with their adjacent neighbors, and thus the system becomes ‘modular’, and (2) that each part of the system “is affected by and affects several other elements or sub-systems” and can adapt or learn in response to perturbations in other parts of the system.
Lastly, it is also important to point out that the behavior of complex adaptive systems cannot be predicted based on the behavior of any lower-order constituent components, and the very premise of these systems is that behavior is not governed from above. What this means is that complex adaptive systems are naturally ordered, but uncontrollable. As a corollary, how behavior will ultimately evolve within these systems is very hard to predict with much precision, especially in the long term.
As we can see from these descriptions, complex systems are, as the name indicates, complex in many ways. It behooves observers of complex adaptive systems, both academic and lay alike, to tread cautiously when trying to analyze, and especially when trying to forecast, how complex adaptive systems actually behave and will evolve. That isn’t to say we shouldn’t try to better understand complex adaptive systems, but that we should recognize how difficult it is to pin down our understanding and ability to predict how such systems will shape-shift over time, with anything approaching deterministic certainty.
With that description out of the way, it’s worthwhile to try and think about a few concrete examples of such systems, to really ponder how well we understand their intricacies, and then to think about the potential efficacy of trying to plan i.e. tamper with the organic dynamics of these systems. The most salient complex adaptive system (to most of us currently living humans) is our physical bodies (and for that matter, the physiology of many complex organisms); then there’s financial markets, and the national and global economies; large urban metropolises; big business or government organizations; social networks; seasonal weather, global climate, local ecosystems; and my personal favorites, bee hives, ant and termite colonies, and the magnificent flight of enormous starling flocks.
Indeed, when scholars of complex adaptive systems want to paint a compelling picture of how these systems behave, many have resorted to showing short videos of a massive starling flocks or time lapse videos of ant colonies doing what ants do as a collective: burrow complicated tunnel networks, forage for food; battle neighboring ant colonies. These sorts or pictures are worth many thousands of words when illustrating complex adaptive systems in action.
By this point, you may be wondering what’s my point? Aside from just wanting to introduce you to this riveting topic (surely if you’ve read this far you must be dying to read more!), I do have one take away. If there’s one thing to highlight, it’s that complex adaptive systems cannot be ordered, or planned, or based on deterministic rules, or regulations, or laws – because other parts of the system will respond in stochastic ways. The inevitable reality is that the law of unintended consequences will, with near certainty, prevent the ordination of desired outcomes when the systems being acted upon are complex and adaptive. If the flight of the starlings offers a lucid picture of how complex adaptive systems behave, then the game of Wack-a-Mole offers a similarly lucid visual of what trying to control, or plan, or coordinate the workings of such systems looks like.
**Apart from the first quote from Wikipedia, all direct quotes are taken from Ladyman, Lambert, and Wiesner (2013).
For further reading, see:
– Ladyman, James, Lambert, James, and Wiesner, Karoline (2013). “What is a complex system?” European Journal for Philosophy of Science 3: 33–67.
– The Wikipedia entry on CAS is also short and sweet. https://en.wikipedia.org/wiki/Complex_adaptive_system
Photo by Pete Godfrey on Unsplash
