Our approach is based on the concept of 'steering' complex adaptive systems. 'Steering' refers to an action, or series of actions, we make in an attempt to alter the outcomes of the system we are focussing on. The diagram below gives an overview of the steering process. Click on the individual tabs below for more detail on each stage.
Many important problems for society involve the management of interlinked complex adaptive systems. Such systems have well known properties which make understanding and controlling them challenging. These include non-linear responses to interventions, so-called ‚Äúemergent‚Äù phenomena produced by the interaction and feedback between low-level components of the system, the importance of network structure and crucially, the ability to adapt and evolve to changes in their environment. All of these properties present new challenges for policy intervention or engineering as they may give rise to behaviours which run counter to our intuition and experience and may change their responses as we intervene. Additionally many of the complex systems which we would most like to influence have significant social components and may require the integration of participatory or political processes with tools from complexity science.
In order to manage complex adaptive systems, we suggest a ‚Äústeering‚Äù approach; an action or series of actions applied to a complex system for achieving a specific purpose. Steering is a continuous process which involves interacting with, monitoring and learning from the system in question.
The techniques required for effective steering fall into two categories. Firstly we wish to understand, and indeed exploit, the systems‚Äô structure and dynamics in order to intervene effectively with them. Hence we need techniques to uncover this structure and to choose points of intervention. Secondly we frame those techniques within a participatory ‚Äúadaptive management‚Äù structure, which explicitly takes into account the adaptive nature of these systems and our limited capacity to fully model real world complex systems, by building in monitoring and feedback processes with which to modify our interventions as systems respond.
Steering is necessarily a continuous and interactive process with discrete actions as outcomes and feedback from the system itself informing subsequent actions. When we apply an action a) we are unlikely to fully understand the system‚Äôs dynamics or have knowledge of all system states and b) the system may change or adapt in response to our intervention. Taking these considerations together means that the system may not respond (or continue to respond) to our action in the way in which we have predicted. We may need to adapt our actions according to an improved understanding of the system‚Äôs dynamics revealed by the results of the action, or in response to a change in the structure or dynamics of the system caused by its own adaptation. We thus either have to design our systems so as to minimise the effects of such change or to treat our interventions as experiments in an ongoing interaction in which we learn the system.
This first stage of the process focusses on uncovering the structure and dynamics of the system in question.
Empirical investigation of the system is central to this process; however, data may be unavailable and/or difficult to collect. Public data is often limited as much that is relevant to the system is commercially sensitive information. Information can be obtained via qualitative research methods such as interviews, but difficult economic circumstances, a lack of obvious immediate benefit of research and potentially strategic actors with vested interests mean that this information may be sparse and biased.
Modelling work is a useful tool to help us formalise our beliefs about a system's structure and dynamics. However, to build and trust our models, we often need a lot of data, e.g., to help us decide how to design the model, or to test if its results match the real world.
Participatory work. In this context using qualitative participatory techniques to uncover information and increase engagement simultaneously is vital. Participatory techniques are centred on the inclusion and engagement of all stakeholders in the system. Stakeholders should be involved throughout the steering process, to gather both information on the system, and feedback on the models and data already developed.
This stage involves bringing together our newly developed knowledge of the system, and possible external events or contexts that may influence it. With these we can seek to identify plausible scenarios for the development of the system.
The aim here is to set a solid foundation on which to design our interventions. We must have a clear understanding of how we believe a system will change without intervention. These scenarios are in effect, our baseline assumptions of how the system may change.
Next, we should begin to identify the goals behind the 'steering' or intervention we wish to undertake. We may find that one of the scenarios previously identified, is our goal, or we may believe our goal lies outside these scenarios.
In systems with few stakeholders, identifying goals may be relatively easy, with few intersecting interests and beliefs. However in systems with many stakeholders, it will be difficult both to ascertain all those stakeholders' choice of goals, and to approach anything resembling a consensus on goals.
Once we have a solid understanding of the system and agreement on goals, we should determine what might be effective ‚Äúlevers‚Äù or points of intervention by which we might attempt to steer the system. The knowledge we have previously developed should help us to identify which are the key points - components, connections or processes - at which we might look for levers, and what these levers may actually be.
These levers might be a subset of lower-level components of the system which we can identify as being able to drive the system to any given state, or self-organising or adaptive processes which we can influence by providing appropriate ‚Äúcontexts‚Äù or selective pressures.
In this stage, we should decide on appropriate system variables to monitor and put in place an organisational infrastructure with which to monitor them. These may be variables which have been found from the earlier ‚Äúpoints of intervention‚Äù analysis to have a strong bearing on system state and/or indicator variables which characterise the achievement or maintenance of the desired system scenarios.
As the metrics are monitored and measured, we must record and arrange appropriate systems to analyse and present the metrics. This will involve developing reproducible methods for processing raw data, carrying out any calculations required on the data, and presenting it in a form usable by a range of stakeholders and users.
Finally, on-going adaptive management or steering requires that we refine our understanding of the system and redesign our interventions as the system reacts or adapts. This may involve starting again at the beginning of the steering cycle, or jumping to another stage, whichever is deemed most appropriate. It is this iterative looping between stages that is as important as any individual stage.