Complexity Science and wicked problems
Features16 December 2009
Yasmin Merali, Associate Professor of Information Systems at Warwick Business School and Co-Director of the Complexity Science Doctoral Training Centre, defines Complexity Science and explains its growing importance and developments.
Turning to Complexity Science because...
The Internet and related technologies have accelerated the expansion of the globalised economy and the evolution of a network society. The resulting interconnected world is characterised by increased complexity as more people, devices, institutions and countries interact in a multitude of ways, exploiting diverse media and communication capabilities. This presents exciting possibilities for innovation, and allows new ways of organising people and resources across traditional boundaries.
However, these possibilities are accompanied by challenges of dealing with a more dynamic and uncertain environment. The non-linear dynamics of the networked world means that small changes in one locality can have large consequences for the global system. The networked society makes it harder for people, institutions and countries to insulate themselves from the impact of events in distant locations, or to predict how situations may change due to interventions and actions of the diverse stakeholders.
Policy makers must bridge the gap between higher (global, regional or national) level policy formulation and its local implementation. This entails the difficulty of accommodating diverse stakeholders' needs, contributions and agendas, whilst maintaining equity and engaging their best efforts in a wide variety of overlapping and interacting implementation contexts.
Even more demanding is the challenge of understanding the intended and unintended impacts of policy interventions in practice. Policy is concerned with a mix of 'tame' and 'wicked' problems. Tame problems are generally amenable to traditional methods of prediction, analysis and solution. Wicked problems (e.g. climate change, providing health care for an aging population etc.) are often highly non-linear (in time and space), uncertain (as opposed to risky) and characterised by emergent phenomena arising from the complex networking among key actors and the entanglement of social, economic and political systems. "One size fits all" strategies that are good "on average" are likely to fail if they ignore the disproportionate and disruptive influence of outliers.
Consequently, wicked problems demand methods to define their complexity in a meaningful way, analyse its consequences appropriately and develop suitable policy interventions. Wicked problems can be difficult to identify prior to the implementation of a 'solution': legendary failures are attributable to the imposition of tame solutions on wicked problems, for example, widely-publicised failures in government IT projects.
Complex problems do not necessarily require complicated solutions: the difficulty is in identifying the critical dimensions of the problem for policy interventions to operate on. Complexity science offers directly relevant concepts, methods and tools for identifying wicked policy problems, defining relevant interventions and exploring possible outcomes in diverse and heterogeneous implementation contexts.
Modelling Complexity
Complexity Science has emerged as a label applied to the collection of concepts, tools and methods developed over time to understand how the overall behaviour of a system arises from the non-linear interactions of its components with each other and with the environment. Modelling methods have evolved for exploring the non-linear dynamics of complex systems in the natural sciences. Over the past decades there has been an increasing interest in deploying complexity modelling approaches to look at socio-economic systems and challenge traditional management paradigms.
Approaches developed for modelling Complex Adaptive Systems (CAS) are especially relevant. CAS adapt and evolve in the process of interacting with dynamic environments. The adaptive capacity of CAS is predicated on their characteristic network structure and non-linear dynamics and on the heterogeneity of their components. Studies of large-scale CAS (e.g. ecologies) have shown that micro-diversity plays an important role in ensuring their sustainability: components or traits in a population that appear to be insignificant outliers under one set of conditions can become critically important for the survival of the population under a different set of conditions.
The global financial system as a CAS illustrates the importance of network topology and diversity in system robustness and resilience. The density and complexity of the financial network led to profound structural vulnerabilities and amplified uncertainties in the pricing of assets, causing seizures in certain financial markets. Network feedback effects under stress (hoarding of liabilities and fire-sales of assets) coupled with the dominant positions of big players and the erosion of diversity in institutions' business and risk management strategies resulted in the current crisis.
CAS display organisation and adaptation at multiple scales. Adaptation at the macro level is an emergent phenomenon: it arises from the local adaptive behaviour of the system's networked constituents interacting with each other and with the environment (e.g. the aggressive spread of bracken, attributed to climate change and to local land management practices, has a major influence on moorland ecology and its current management). As evidenced by current debates on climate change, there is a reflexive relationship between system and environment: changes in the system both shape and are shaped by changes in the environment. The CAS paradigm imposes a need to consider the dynamics and mutually defining consequences of the relationship between the system and its environment, taking us from issues of adaptation (of the system to the environment) to issues of co-adaptation and co-evolution (of system and environment) in dynamic contexts.
Complexity science-based approaches are designed to deal with emergence and (co)evolution of CAS, and can be used in conjunction with more conventional methods to deliver guidance for policy makers. Most widely used are agent-based models, comprising individual 'agents' - individuals, organisations, countries - implemented as software objects. Agents can be endowed with requisite resources, traits, behaviours and rules for interacting with, and adapting to, each other. Running such a model is an exercise of representing an instance of an agent population, letting the agents interact and monitoring what happens.
Typically, agent-based models deploy a diversity of agents to represent the constituents of the focal system and the modeller defines the environmental parameters that are of interest as the starting conditions for the particular study.
Starting with some initial condition, the simulation consists of applying the rules through several iterations. Repeated runs of the model reveal collective states or patterns of behaviour as they emerge from the interactions of entities over time. Often such models reveal counter-intuitive outcomes. For example it was revealed by modelling and verified through experimentation that placing a pillar slightly off-centre just in front of an emergency exit results in more efficient evacuation.
Complex systems have many degrees of freedom (many elements are partially but not completely independent), they have ambiguous system-environment relationships, and there is a greater diversity of local behaviours than there is of global outcomes. The network dynamics linking micro-level behaviours (which may change in response to changes in local conditions) with macro-level global system properties cannot be completely specified a priori. In order to achieve an effective representation of the dynamics of the processes connecting the local (micro-level) and global (macro-level) characteristics, we need to develop a multi-scale description of complex systems. Agent-based models allow us to study the diversity of (local) behaviours at fine scales and to observe the emergence of the global characteristics at the large scale. For policy makers such models provide an environment for testing assumptions underpinning policy decisions and for exploring how policy interventions could play out over time.
For socio-economic systems the modelling challenge lies in defining characteristics of the system at an appropriate level of abstraction to yield insights about emergence of possible future states under a variety of conditions. This is a transdisciplinary endeavour, requiring the combined expertise of scholars in the social sciences, practitioners from the policy domains being explored, and experts in dynamical systems from mathematics and the physical and biological sciences.
Invitation to engage in Complexity Research
The University of Warwick has invested significantly in developing the transdisciplinary complexity community. It hosted the 2009 European Conference on Complex Systems , and is home to the EPSRC Doctoral Training Centre for Complexity Science with which WBS has strong ties (Professor Mark Salmon and Yasmin Merali are Co-directors). Current complexity network projects at WBS include the EPSRC-funded EmergeNet project and the EU FP7-funded ASSYST programme. You are invited to participate in the network activities and engage with the research programme at the Doctoral Training Centre.
The ASSYST programme will introduce complexity science to business and public sector organisations: we will organise workshops and seminars to identify and develop areas of interest for practitioners. For MBA alumni of the module Business Transformation there will be a Business Transformation Reunion in the New Year with a complexity-based programme. The students at the Doctoral Training Centre for Complexity Science are keen to work on real-world problems and we would very much like to hear from practitioners who have complex problems for the students to work on.
To learn more about the ASSYST and Business Transformation events or to engage with research at the Doctoral Training Centre for Complexity Science please contact Yasmin Merali, details below.
Turning to Complexity Science because...
The Internet and related technologies have accelerated the expansion of the globalised economy and the evolution of a network society. The resulting interconnected world is characterised by increased complexity as more people, devices, institutions and countries interact in a multitude of ways, exploiting diverse media and communication capabilities. This presents exciting possibilities for innovation, and allows new ways of organising people and resources across traditional boundaries.
However, these possibilities are accompanied by challenges of dealing with a more dynamic and uncertain environment. The non-linear dynamics of the networked world means that small changes in one locality can have large consequences for the global system. The networked society makes it harder for people, institutions and countries to insulate themselves from the impact of events in distant locations, or to predict how situations may change due to interventions and actions of the diverse stakeholders.
Policy makers must bridge the gap between higher (global, regional or national) level policy formulation and its local implementation. This entails the difficulty of accommodating diverse stakeholders' needs, contributions and agendas, whilst maintaining equity and engaging their best efforts in a wide variety of overlapping and interacting implementation contexts.
Even more demanding is the challenge of understanding the intended and unintended impacts of policy interventions in practice. Policy is concerned with a mix of 'tame' and 'wicked' problems. Tame problems are generally amenable to traditional methods of prediction, analysis and solution. Wicked problems (e.g. climate change, providing health care for an aging population etc.) are often highly non-linear (in time and space), uncertain (as opposed to risky) and characterised by emergent phenomena arising from the complex networking among key actors and the entanglement of social, economic and political systems. "One size fits all" strategies that are good "on average" are likely to fail if they ignore the disproportionate and disruptive influence of outliers.
Consequently, wicked problems demand methods to define their complexity in a meaningful way, analyse its consequences appropriately and develop suitable policy interventions. Wicked problems can be difficult to identify prior to the implementation of a 'solution': legendary failures are attributable to the imposition of tame solutions on wicked problems, for example, widely-publicised failures in government IT projects.
Complex problems do not necessarily require complicated solutions: the difficulty is in identifying the critical dimensions of the problem for policy interventions to operate on. Complexity science offers directly relevant concepts, methods and tools for identifying wicked policy problems, defining relevant interventions and exploring possible outcomes in diverse and heterogeneous implementation contexts.
Modelling Complexity
Complexity Science has emerged as a label applied to the collection of concepts, tools and methods developed over time to understand how the overall behaviour of a system arises from the non-linear interactions of its components with each other and with the environment. Modelling methods have evolved for exploring the non-linear dynamics of complex systems in the natural sciences. Over the past decades there has been an increasing interest in deploying complexity modelling approaches to look at socio-economic systems and challenge traditional management paradigms.
Approaches developed for modelling Complex Adaptive Systems (CAS) are especially relevant. CAS adapt and evolve in the process of interacting with dynamic environments. The adaptive capacity of CAS is predicated on their characteristic network structure and non-linear dynamics and on the heterogeneity of their components. Studies of large-scale CAS (e.g. ecologies) have shown that micro-diversity plays an important role in ensuring their sustainability: components or traits in a population that appear to be insignificant outliers under one set of conditions can become critically important for the survival of the population under a different set of conditions.
The global financial system as a CAS illustrates the importance of network topology and diversity in system robustness and resilience. The density and complexity of the financial network led to profound structural vulnerabilities and amplified uncertainties in the pricing of assets, causing seizures in certain financial markets. Network feedback effects under stress (hoarding of liabilities and fire-sales of assets) coupled with the dominant positions of big players and the erosion of diversity in institutions' business and risk management strategies resulted in the current crisis.
CAS display organisation and adaptation at multiple scales. Adaptation at the macro level is an emergent phenomenon: it arises from the local adaptive behaviour of the system's networked constituents interacting with each other and with the environment (e.g. the aggressive spread of bracken, attributed to climate change and to local land management practices, has a major influence on moorland ecology and its current management). As evidenced by current debates on climate change, there is a reflexive relationship between system and environment: changes in the system both shape and are shaped by changes in the environment. The CAS paradigm imposes a need to consider the dynamics and mutually defining consequences of the relationship between the system and its environment, taking us from issues of adaptation (of the system to the environment) to issues of co-adaptation and co-evolution (of system and environment) in dynamic contexts.
Complexity science-based approaches are designed to deal with emergence and (co)evolution of CAS, and can be used in conjunction with more conventional methods to deliver guidance for policy makers. Most widely used are agent-based models, comprising individual 'agents' - individuals, organisations, countries - implemented as software objects. Agents can be endowed with requisite resources, traits, behaviours and rules for interacting with, and adapting to, each other. Running such a model is an exercise of representing an instance of an agent population, letting the agents interact and monitoring what happens.
Typically, agent-based models deploy a diversity of agents to represent the constituents of the focal system and the modeller defines the environmental parameters that are of interest as the starting conditions for the particular study.
Starting with some initial condition, the simulation consists of applying the rules through several iterations. Repeated runs of the model reveal collective states or patterns of behaviour as they emerge from the interactions of entities over time. Often such models reveal counter-intuitive outcomes. For example it was revealed by modelling and verified through experimentation that placing a pillar slightly off-centre just in front of an emergency exit results in more efficient evacuation.
Complex systems have many degrees of freedom (many elements are partially but not completely independent), they have ambiguous system-environment relationships, and there is a greater diversity of local behaviours than there is of global outcomes. The network dynamics linking micro-level behaviours (which may change in response to changes in local conditions) with macro-level global system properties cannot be completely specified a priori. In order to achieve an effective representation of the dynamics of the processes connecting the local (micro-level) and global (macro-level) characteristics, we need to develop a multi-scale description of complex systems. Agent-based models allow us to study the diversity of (local) behaviours at fine scales and to observe the emergence of the global characteristics at the large scale. For policy makers such models provide an environment for testing assumptions underpinning policy decisions and for exploring how policy interventions could play out over time.
For socio-economic systems the modelling challenge lies in defining characteristics of the system at an appropriate level of abstraction to yield insights about emergence of possible future states under a variety of conditions. This is a transdisciplinary endeavour, requiring the combined expertise of scholars in the social sciences, practitioners from the policy domains being explored, and experts in dynamical systems from mathematics and the physical and biological sciences.
Invitation to engage in Complexity Research
The University of Warwick has invested significantly in developing the transdisciplinary complexity community. It hosted the 2009 European Conference on Complex Systems , and is home to the EPSRC Doctoral Training Centre for Complexity Science with which WBS has strong ties (Professor Mark Salmon and Yasmin Merali are Co-directors). Current complexity network projects at WBS include the EPSRC-funded EmergeNet project and the EU FP7-funded ASSYST programme. You are invited to participate in the network activities and engage with the research programme at the Doctoral Training Centre.
The ASSYST programme will introduce complexity science to business and public sector organisations: we will organise workshops and seminars to identify and develop areas of interest for practitioners. For MBA alumni of the module Business Transformation there will be a Business Transformation Reunion in the New Year with a complexity-based programme. The students at the Doctoral Training Centre for Complexity Science are keen to work on real-world problems and we would very much like to hear from practitioners who have complex problems for the students to work on.
To learn more about the ASSYST and Business Transformation events or to engage with research at the Doctoral Training Centre for Complexity Science please contact Yasmin Merali, details below.
