Complex Adaptive Systems: a primer for ITSM (i)

The increasing complexity of digital systems necessitates new perspectives and practice.

As an airline customer in the modern era, it is possible to open a mobile app, and swap to a different flight with a few simple clicks. For the end-user, this operation has never been easier than now. However, the reality of this apparently simple action is that a huge number events are triggered in complex web of interrelated digital events. The events occur immediately, and each has an intricate impact on many other components of the system.

It’s interesting to explore this complexity. It needs to be established, for instance, whether I can make this change free of charge, or whether it is billable. If the latter is the case, a payment needs to be handled, which might involve interactions between a payment service run by the OS provider of my phone, a central credit card routing service, and my bank. These entities interact to perform a validation which establishes that I am who I say I am, and whether I have sufficient funds to cover the cost of the purchase. Of course, I might not choose to pay with currency: I may alternatively opt to use loyalty points which have previously been accrued and associated with my customer profile.

Meanwhile, operational systems need to be updated, and these will have multiple owners. Airport systems need to ensure that I am able to proceed through check-in gates into the departure terminal. Border agencies, at both my starting point and destination, may need to be informed of the change. Cabin crew systems and baggage handling data will need to be updated.

The dependencies from such a simple change are not just local to the airline, and not restricted to the airports through which I travel. I may change my flight from a desk in London, but a travel agency in Sydney must immediately be prevented from any simultaneous attempt to sell my newly allocated seat to another person.

Airline ticketing is an example of the kind of huge global system which has arisen in the digital era, comprising an intricate web of interconnected parts, each with its own origin, nature, and behaviour. It is a system in a constant state of operational flux, with events in one part of it impacting multiple other systems in ways that were not designed up-front by any single architect. The system I am interacting with grew over many years through a regular series of independent expansions and changes.

The further we explore this system, the more difficult it is to see its edges. Is the system limited to the systems of the travel industry, or do we need to account for the fact that the change can only take place if the user’s phone is connected to a data service, and charged with electrons supplied by an electricity seller? And how did those electrons get there? They were supplied by the power grid, a similarly intricate and diverse system of interacting independent components.

Airline ticketing, in summary, is just one example of a Complex Adaptive System.

Defining a Complex Adaptive System?

The concept of complex adaptive systems was defined relatively recently (in 1996, by Kevin J Dooley). Complex adaptive systems are a key component of the young science of complexity, and there are well-understood examples of complex adaptive systems in multiple realms such as biology, economics, transport, and urbanisation.

To define them, we probably need to establish a basic definition for the general concept of a “system”. A useful description can be found in the Merriam-Webster dictionary:

System (noun): A regularly interacting or interdependent group of items forming a unified whole.

Systems thinking typically considers systems in terms of the relationships between their inputs and outputs. Linear systems are straightforward, demonstrating a simple, proportional relationship (for example). Nonlinear systems have an output which is not proportional to the inputs, but are nonetheless often predictable and understandable (for example, y=sin x).

Complex adaptive systems, however, do not exhibit such deterministic relationships, nor follow the continuation in Merriam-Webster’s definition which states a system will be “in or tend(ing) to equilibrium”. Instead, their properties are a product of their nature as a set of interrelated entities, each with its own behaviours. The overall structure is not in a state of equilibrium, but instead displays a more dynamic, adaptive state of existence.

So, while there is not a single, definitive definition of complex adaptive systems, they do have a number of well-recognised characteristics which set them apart from simpler systems. These characteristics are fascinating, and there’s plenty of scope to explore their implications in more depth (which is something I plan to do in subsequent articles on this blog). They include (but aren’t limited to):

• Complexity: No single formal model is sufficient to capture all of the properties of the system, and the descriptive models which can represent parts of the system are not derivative of each other.
• Self-Organisation: The system operates not just on the basis of its individual parts, but on the interactions between them. The overall order of the system is not a state of equilibrium, but one of constantly evolving fluctuations and feedback loops in these interactions.
• Emergence: Because the result of combining parts of the system is not simply a sum of the properties of those parts, new system characteristics are continually formed by the establishment and removal of such combinations.
• Resilience: the system exhibits an ability to react to internal failures unforeseen negative events by absorbing the disturbance, and/or reorganising to maintain its functions.
• Observer Dependency: The nature of the system will appear differently to different people, depending on their specific viewpoint.
• Path Dependency: The present nature of the system is heavily influenced by decisions and events in the past, in ways that are difficult to overcome.
• Chaos: Small variations in input can lead to dramatic changes in output, while large variations in input may not create a correspondingly big change in outputs.
• Irreducibility: Transformations to the system can not always be reverted back to the previous state.

Why does this matter to ITSM?

ITSM is built on a somewhat deterministic philosophical legacy of defining and mapping services, and maintaining some degree of central control. Forrester analyst Charles T. Betz, for example, wrote in 2017 in a blog for my own employer:

Re-reading as sympathetically as I can, I find that the overall ITIL narrative is still sequential, plan-centric, and deterministic.

This mindset has recently moved forward, not least because of the influence of DevOps, and ITIL 4 is a significant step forward in many ways. Complex Adaptive Systems themselves are examined for several pages of the ITIL 4 framework (in the High Velocity IT book first published in early 2020). This is welcome and useful, but it is only a beginning.

My view is that it is critical for ITSM professionals to understand that complex adaptive systems exist, and that many of the services they seek to maintain, manage and support are actually parts of broader complex systems. ITSM needs to adapt its ways of working to compex adaptive systems, but to do so we need to drive wider awareness of the concept’s existence, and build a community ability to recognise such systems in practice. This is necessary for a number of ways some of which are as follows:

• IT Systems are increasingly complex. As organisations transform digitally, for example, an ever-broader, deeper, and more interconnected technology structure evolves. Where previously, systems operated on a more simple, linear basis, in relative isolation, the increasing connection and dependency between components, and their continual rate of evolution, tends to push established enterprise digital services into the realm of complex adaptive systems.
• Large scale digital systems can not be thought of in terms of their technology alone. Instead, most are examples of socio-technical systems, in which the system is a product of both the technology underpinning it, and the behaviours and actions of the people interacting with it ITSM itself has some solid socio-technical underpinnings, having long been built, at least partly, on the concept of delivering business value to customers. However, as argued by Gordon Baxter and Ian Sommerville in a 2009 paper for St Andrews University, “The rationale for adopting socio-technical approaches to systems design is that failure to do so can increase the risks that systems will not make their expected contribution to the goals of the organisation.
• Complex systems fail differently. In fact, as I’ve explored before, complexity theory generally holds that a complex system is in a constant state of multiple failures, mitigated by the fundamental resilience of the system (which can be derived from both the structure of the system and through the actions of humans operating and working with it). A seminal paper on this topic is Richard Cook’s How Complex Systems Fail. In short, we need to be able to recognise when a system is complex, and manage it in very different ways to a simpler, linear system.

What next?

This is the first article of a set of more detailed explorations of critical characteristics of complex adaptive systems, and their impact on ways of working in service management. Part two is here: “Complex Adaptive Systems (ii): thinking about emergence”

There is a lot of good material about the emerging science of complexity, and the concept of complex adaptive systems. One good recommendation is a series of lectures shared by TU Delft from its undergraduate syllabus, entitled “Agent Based Modelling of Complex Adaptive Systems”. The introductory lecture by Dr Igor Nicolik (@complexevo) is an excellent primer on the concept and characteristics of complex adaptive systems.

Featured photo by David Berry on Flickr, used under Creative Commons license.