2
The Evolution of Complex Systems

Companies and all forms of human organizations function as complex adaptive systems (CAS). The description of such CAS is the subject of this chapter as well as Chapter 3. The present chapter aims to describe the ways in which organizations evolve by interpreting them as self-organized systems whose trajectories over time are marked by periods of dynamic flux and adaptation as well as phases of stability. Chapter 3 will be dedicated to the question of steering CAS, as we noted in Chapter 1, adaptive systems cannot be controlled but can be steered over time. Overall, this Chapter presents the evolutionary characteristics of complex systems from the vantage point of an outside observer - a cybernetic, systems dynamic and economic approach. Chapter 3 will assume a more managerial perspective. The management of systems that, due to their "complexity", are likely to organize themselves through radical and largely unexpected changes is obviously a challenging task. We will see what qualities their managers must have in order to overcome this challenge, as well as the types of procedures they must implement in order to intelligently "steer" them within such a context. These two chapters necessarily have a strong connection, which will lead us to deal with the same subjects, but from different perspectives. In particular, the paradigm shifts described in this chapter will be interpreted in Chapter 3 as the emergence of new signals for management, i.e. "weak signals" that the astute manager must know how to decipher.

Any manager strives to make his organization highly adaptable in the face of the external and internal crises that are a regular part of its ongoing evolution. The question is then knowing how a system adapts when it is largely self-organized (here, we can use the term autonomous system). Indeed, in the case of a system that can be controlled in a linear manner, change (rather than adaptation) fully lies in the hands of the hierarchy. If, on the contrary, numerous feedback loops lead to internal structural changes in response to stimuli, then the system becomes difficult to control. It is, in a manner of speaking, overly autonomous. To what extent should the steerage of a strongly nonlinear system depend on its ability to automatically adapt? Can instructions be given that respect the system's self-organizing logic (its autonomy) to encourage its progress in the desired direction? These are some of the questions to be dealt with in a managerial approach to CAS. In order to begin responding to these questions, it is necessary to study the evolution of complex systems and to understand what exactly the term "adaptive" means based on the type of system.

2.1. Adaptation, learning and flexibility

In the case of a highly elementary system, a structure is adaptive based on the environment if it learns and responds to the feedback loops in the environment. This is a complicated system and does not adapt - the tires and suspension do not learn from the surface they interact with, for example. This sub-system presents good adaptability if it absorbs a large part of the shocks caused by irregularities in the road. This makes the vehicle not only more comfortable, but also more stable and less fragile. This form of adaptability, in comparison with that of a more complex system, may be classified as static.

In contrast to this, CAS, particularly human ones (CAHS), adapt through interacting with their environment, as the rewards and feedback loops in the system cause the CAS to alter their behavior. For an autonomous system, adaptation takes place through sustainable self-organized modifications in the structures - as if the vehicle had not had to be modified from external intervention, but instead redeveloped itself to better overcome the challenges of the road after having experienced a section of it. Let us note that this miracle is simply business as usual for a living being or a company! If managers consider their organization to be like a vehicle, then they will consider it their mission to reorganize the structure based on modifications required by the environment. If instead they are aware of the possibilities of self-organization in the system they are modifying, then they will understand and allow adaptive structural mechanisms to steer the organization to a certain extent. The design of CAS is thus partly a matter of managers' perceiving the system. If their mental representations lead them down the right route, then they perceive the organization as a complex system (including themselves as part of the system), which is adaptive in a dynamic sense, i.e. it forms itself over time through evolutionary changes.

A more precise term than "dynamic" would likely be "evolving". We will come back to the question of evolution later in this chapter, but first, we must define the various models of evolution that are possible. For the moment, let it suffice to say that dynamic flexibility is connected to an ability to learn. What happens in a self-organized system faced with new, unexpected information is adaptation through reorganization. The system does not only react in that moment, like the tire, but also "learns" from the event to change in a sustainable way. This is the beginning of path dependency in the sense that future reactions on the vehicle's part will not be the same.

Static or dynamic adaptation

A simple adaptive system has static flexibility. This means reacting to a new situation with an adapted tactic, but one that leaves no trace on the system. A complex adaptive system (CAS), because it is complex, learns from this event. By adapting (level 1 learning), it learns to adapt (level 2 learning) to use Gregory Bateson's terminology (1972). In the terms of self-organized system theory, we could say that it reacts to an external shock by reorganizing at a higher level of complexity. It profits from the event to increase its repertoire of programmed responses, which will make it even more reactive and efficient in the future. It has increased its strategic intelligence.

Up to this point, we have hypothesized that the complex system is capable of adaptation. In fact, this is not always the case, or this at least depends on the nature of the external shock - as well as on what we consider a reasonable adaptation. Strong qualitative changes may endanger the system. Self-organization at a higher level of complexity is the favorable scenario. Another evolution can be foreseen, i.e. system destruction. In addition, spontaneous reorganization, if it has taken place, may not satisfy the organization's management for one reason or another, hence the slightly problematic nature of the concept of CAS, as useful as it may be. The very nature of adaptation is a complex matter. Does the complex system adapt completely independently? Due to a push from its management? Guided critically and vigilantly by its managers? Despite them? When we speak of strategic intelligence, are we speaking primarily of a quality of the management team or one belonging to the entire system (distributed intelligence)?

In any case, we must keep in mind that the way in which self-organized systems learn is largely an intrinsic characteristic. Its method of reaction through transformation is part of the system's identity, like the mechanisms governed by the DNA in living systems. Two similar self-organized systems (e.g. two companies in the same field and of a comparable size) faced with the same external shock (change of market prices or the emergence of new technology) will not react in the exact same way as there are also internal drivers and feedback loops which will differ within each company. The difference in dynamics is not a simple question of inertia, but rather of path dependency; for the history of each system, its trajectory up to the point being observed, is not neutral in relation to the possible adaptations and future trajectories.

2.2. The nonlinear behavior of "imbalanced" systems

If it is important to understand how CAS work, it is because the self-organization processes that characterize them and that often manifest themselves when changes in the environment (external shocks) come about are nonlinear mechanisms that make them difficult to steer. In physical-chemical systems, these qualitative leaps are typical of imbalanced states. Here, we could mention Ilya Prigogine's "dissipative structures" (see Prigogine and Stengers 1984), when a macroscopic system is penetrated by a flow of matter and energy (which dissipates) and produces unforeseeable self-organized forms, going through profound state changes. To analyze these mutations, physicist Pierre-Gilles de Gennes proposed a general theory of what he calls percolation thresholds - rapid switches from one macroscopic state to another based on an accumulation of microscopic modifications. What is observed and studied as a sudden change in quality - fascinating and complicated for researchers to model - is experienced by the head of an organization as a sizeable managerial challenge.

Kerr (2014) examined the behavior of managers and distinguished those who think in a "linear" manner (linear thinking leaders) from those who have mental representations and professional experience better adapted to steering complex systems with nonlinear behavior. Linear thinking...