The world is complex and unstable. Our approach to policy-making must reflect that fact.

Social scientists and policy-makes have long built their models around a theory of 'rational action' which bears little relation to how people respond to the complexities of the world around them.

Graham Room
31 August 2016
 Susannah George / AP/Press Association Images

"It is easy to focus on the stability that may be evident at home, while overlooking developments elsewhere." Photo: Burning oil fields, Iraq. Susannah George / AP/Press Association Images. All rights reserved.Recent decades have seen a dramatic growth in the literature on complex systems.  Much of this has developed in mathematics, and in the natural and computational sciences.  Increasingly however, this approach is gaining traction in the social sciences. This has in many ways been a breath of fresh air.  Drawing upon dynamic models from the natural sciences that can be extremely fertile in suggesting new insights for the social sciences, introducing models such as tipping points, bifurcations and co-evolving systems. It has created much greater interest in the non-linear dynamics of the social world - and in the macro-patterns that emerge unexpectedly from the micro-interactions in which we are all involved.   

Nevertheless, in applying these perspectives to the social world, we need to take care. The ways we interact involve us exercising and contesting power.  All this also involves uncertainty – and efforts by the powerful to offload that uncertainty onto others.  Yet these have been rather neglected by much of the complexity writing within the social sciences.  


Evidence-based policy-making brings together robust evidence about the outcome or impact of a particular intervention.  Evidence is collected, evaluated and aggregated – ‘systematically reviewed’ - across as wide a range of contexts as possible. The gold standard remains the randomised controlled trial. The prominence given to this gold standard can be explained in part by its high status within medical science and clinical practice. What may also be relevant is the ascendancy of the 'contract culture' for the commissioning of public services from multiple providers. Contracts need to specify the impacts against which the service providers will be judged, and therefore depend on the establishment of an evidence base for such impact (and a policy research community that endorses such an approach). This may also explain why impacts which can be couched in terms of ‘behaviour change’ are especially attractive, if they seem readily identifiable and measurable.

The gold standard remains the randomised controlled trial.  

This tends to assume that any particular intervention can be assessed in isolation from others.  The complexity literature however tells us that we live in a connected world - and that these connections require us to pay attention to the dynamic interactions among policy interventions. Any new intervention is an intrusion into a dynamic policy ‘ecosystem’ and must be evaluated as such.

Government leaders warn that conventional methods of policy intervention no longer seem to work. New models of a dynamically interconnected world are needed, if they are to anticipate, steer and control its turbulent behaviour. The literature on complex systems affords us just such models.  One of the simplest and most elegant is Schelling’s model of residential segregation in cities. His agents have different addresses on a grid or lattice. Schelling posits a ‘tolerance schedule’, a preference not to have within one’s immediately adjacent neighbourhood more than a certain proportion of another race: the general racial composition of the city at large is however of no concern. Starting with an initially random distribution of households across the city, Schelling conducts repeated simulations, as households walk step-wise to an adjacent address, whenever their immediate neighbours exceed this racial threshold. He shows that even a rather mild level of racial antipathy – and concerned only with immediate neighbours - will quickly generate zones of racial segregation across the city.  

Schelling’s actors react to the racial composition of their immediate neighbourhood. As they do so, their actions affect the composition of adjacent neighbourhoods, and the society as a whole moves progressively to a more segregated state.  Any particular individual may find that, whatever the level of their own tolerance, the ethnic composition of their neighbourhood is increasingly likely to exceed that threshold, obliging them to move. Even the most tolerant finds that their neighbourhood steadily loses all diversity. In this way the rate of segregation accelerates, through processes of positive feedback.   What this means is that the level of social segregation that results is quite other than the individuals in question themselves anticipated or sought.  

This is a simple model, but one which illustrates well how ‘macro’ emerges from ‘micro’ in non-linear ways and with counter-intuitive results. Policy interventions launched with insufficient thought, as to their interactions with the wider policy ecosystem, are unlikely to be sufficient to the task.  No policy is launched onto a greenfield site: it is an intervention in a tangled web of institutions that have developed incrementally over extended periods of time and through a succession of political struggles. Complexity analysis can play a central role in the analysis of such policy dynamics.   

Stability and freedom of manoeuvre do not descend equally on all

Where much of the complexity literature is weak, however, is in recognising the exercise of power and interests in human societies and social interactions.  What works in the interests of who?  By reference to which interests do we assess the significance of the effects of any intervention?  In what ways do power inequalities ‘emerge’ from the complex social interactions to which complexity writers refer?   Of course, there is much that remains stable: otherwise we would hardly be able to live our everyday lives. Nevertheless, it is easy to take this stability for granted and not to notice the small changes that are continuously underway.  It is also easy to focus on the stability that may be evident at home, while overlooking developments elsewhere. Stability and freedom of manoeuvre do not descend equally on all, as manna from heaven: many are more cosseted than others from the effects of this instability.    


Policy makers – and their academic policy advisers – adopt a variety of assumptions about the citizens they serve and what motivates those citizens to behave as they do.  These assumptions are not immune to successive intellectual fashions.  Rational action theories remain dominant across much of social science: here social actors assess the menu of choices available to them in terms of their costs, benefits and consequences.  The rational citizen deserves to be provided with clear information about the public services that are available and those services should embody variety and choice.    

However, there is an increasing recognition of the bounds to such rationality, because of the partial information that actors have at their disposal and their limited ability to assess risk. This is true not just of the person in the street: it was also on too evident in the 2008 financial crisis, when the ‘rationality’ of the financiers produced risky behaviours that almost led to systemic collapse. The complexity literature reinforces that recognition of the bounds to rational choice.  The lesson of Schelling’s model, discussed above, is that aggregating individual choices can produce counterintuitive results for the individuals in question, as well as being non-optimal socially. This fundamentally challenges the central place that has been given to individual choice in the rhetoric of policy reform over recent decades.  

Rational action theories remain dominant

Meanwhile, recent years have seen growing interest – especially on the part of economists – in behaviourist perspectives. These in some degree abandon orthodox models of rational economic behaviour, especially as far as ‘ordinary’ people are concerned, instead pointing to the biases and blunders that they display in many of their decisions. These behaviourist perspectives – based on empirical studies of individual behaviour - have increased the attention given in policy debate not only to the contribution of psychologists, but also that of neuroscientists, claiming to measure well-being in a far more rigorous manner than social scientists could ever offer.    

This is the realm of ‘nudge’:  with policy makers seeking to structure the ‘choice architectures’ that face citizens, so that even with their innate biases and short-sightedness, they will make choices that align with their own best interests.  ‘Nudge’ has been well-described by economist Richard Thaler and law professor Cass Sunstein as “libertarian paternalism” in their book of the same name, which endorses this tactic as an enlightened approach to modern policymaking. This might seem a very modest form of paternalism, compared with that which traditional welfare states are often accused of embodying, where the citizen is said to be left with few if any options.  Nudge leaves the citizen to make the final choice: government can structure the ‘choice architecture’; but if, despite this, citizens still make the wrong choices, that is their responsibility.  

But what if the citizen, far from being lazy or short-sighted, is simply but profoundly unsure what to do, given the insecurity and instability of the world around them? After all, Thaler and Sunstein are not equally concerned with all citizens; their particular focus is on ‘the least sophisticated’; those who they reckon are in greatest need of such guidance.  These are precisely those who are most exposed to such insecurity and instability. The citizen may want not more choices, but a guarantee of well-being and security instead. 

The citizen may want not more choices, but a guarantee of well-being and security instead.

Models might be able to better account for these differences if they assumed people not to be purely rational actors, but rather agile actors. These are social actors who probe the complex and turbulent landscapes on which they find themselves, but only if they can do so from the vantage point of some stable and settled ground.  In contrast to rational action theories, it sees social actors not as menu-takers but as menu-shapers, re-working and contesting the social and economic world in which they find themselves. In contrast to behaviourist perspectives, it understands any biases and blunders by reference to the settled ground from which social actors draw their routines and practices: a ground which may be extensive or narrow: and which must be understood in terms not of psychology but of sociology and political economy. The result is a perspective on social action that sharply distances itself from the abstract individualism of much contemporary writing.  

This gives a central place to uncertainty and instability and to the efforts by agile actors to offload its costs onto others. The social distribution of uncertainty thus moves centre-stage.  But of course, this is a struggle played on unequal terms. Those already at a positional disadvantage are likely to carry a disproportionate share.  They may end up with hardly any stable and settled ground from which to shape their world: and with little option therefore but to hunker down and cling to a precarious existence.  

A new social contract

When we identify the factors that hinder ‘agile agents’ from choosing in their best interest, we can gain insight into the social and political conditions under which agile citizenship may be possible for all.  Citizens need to be secure, resilient and adaptable, if they are to survive and thrive in a complex world. No less than the earlier leitmotifs of rational consumer choice and nudge, our notion of agile actors suggests an agenda of policy reform. This involves public policies to provide a stable and secure ground for all - and to invest in the creative energies of everyone.  

“in the twentieth century, citizenship and the capitalist class system have been at war”

The recent direction of UK social policies has been to expose people as much as possible to the movements of the market, narrowing public generosity towards those who remain on state ‘support’. The burden of austerity has fallen on the most disadvantaged, multiplying the uncertainties to which they are exposed.  This is the politics of fear and surrender to the global market: of insecurity and hunkering down.     

In contrast, the post-war social contract between State and citizen, across the western world, involved a pooling of risks and uncertainties through systems of social security. It set the fraternity and mutual interdependence of citizenship against the divisions and inequalities of class and against the turbulence and insecurity of an urban-industrial society. The same period saw governments confronting the economic instability of capitalist society. This has sometimes been characterised as a consensual process, the benign fruit of economic progress. Nevertheless, as T H Marshall warned: “in the twentieth century, citizenship and the capitalist class system have been at war”.   

If the social changes of the 21st Century are to be managed successfully and with public consent, they need a new social contract to underpin them.  We need to mobilise the energies and talents of all sections of society: and we are more likely to pull together if the distribution of rewards is less unequal.  Such a contract would need to include several interrelated elements, going well beyond traditional welfare systems:  

  1. Individual security against risks of income interruption: the heartland of traditional welfare states, albeit in the last half century on the defensive, across much of the industrialised world, in face of neo-liberal hostility to State welfare. Investment in everyone’s capabilities, not just in those with parental wealth: what many have referred to as the ‘social investment state’. There is good evidence that for a given financial outlay, it is investment in the lowest-skilled that can produce the greatest benefit for national productivity.

  2. The rebalancing of our economies to provide ‘decent jobs’ which make use of everyone’s capabilities.

  3. Investment in vibrant local communities, as places of education, learning and creativity for all: in particular for disadvantaged communities, which are often poorly connected to the society at large.

  4. Involvement of all in the governance of social, political and economic institutions, with active citizenship and scrutiny of public policies, and of the corporate interests which might otherwise detract from such a contract.  

Such a contract would involve a broad range of policies of relevance to all citizens, rather than focussing just on society’s casualties. It would need to go far beyond the notion of a basic income, which in various guises has again reared its head across the political spectrum. It would limit the risks of poverty but also promote economic growth; promote individual security but also collective resilience and adaptability. It would also go far beyond the extension of choice in public services, with the citizen seen primarily as a consumer. It would involve rebuilding local and national communities, as points where these different policies can be connected up.  It would leave the market where it belongs, as the servant of the community not its master.  

It would leave the market where it belongs, as the servant of the community not its master

This would also re-shape the debate on immigration. First, by investing in the skills and creativity of our own population, we reduce the need for employers to look elsewhere – for nurses, for IT specialists and others - in ways that denude poorer countries of those in whom they have invested their slender national resources.  Second, by taking collective responsibility for the infrastructures of those communities to which large numbers of immigrants come, rather than ‘devolving’ this burden to the local areas in question, we reduce the risk that those communities will see immigrants as a threat.               

It is not however enough for government to provide stability and security and to invest in agile and creative citizens: they must also be able to hold government to account. This means government being placed under critical scrutiny by citizens, rather than vice versa. This gives a fundamental role to citizens in policy making as well as in policy implementation.  ‘Nudge’ in contrast makes little or no attempt to engage citizens as active, critical and responsible partners: they are deemed hardly up to that. 

A self-organising democracy?

By the start of the new century, New Labour had in some measure joined the Conservatives in their celebration of market capitalism, with philosophical underpinnings that owed much to Hayek.  The economy was a decentralised system, best left to self-organise and self-steer.  Government could not expect to second-guess the market and had therefore better stay out.   

Government would nevertheless maintain its overall grip, albeit at arm’s length

In the public services, both parties sought to expand individual choice: in many cases by expanding the role of the market and its self-organising virtues.  Government would nevertheless maintain its overall grip, albeit at arm’s length; and even under conditions of austerity, big data and smart government would enhance its effectiveness.  Public services would become leaner and fitter and more tightly managed, so as to be better-tuned to consumer choice.   

Hayek’s vision likewise resonates with many of those who have been attracted by the literature on complex systems. He describes a complex and interconnected economy peopled by a myriad of micro-decision-makers – but one which has generated wealth on an unprecedented scale. This would seem an obvious example of the processes of ‘self-organisation’ with which the complexity literature deals: and one with socially benign consequences.  In contrast, the complexity literature might well seem poorly aligned with the sort of social democratic vision that has been outlined above.

There are however several important caveats: First, the complexity literature gives little attention to uncertainty (as distinct from risk) and its implications for social actors. As argued above, uncertainty makes people hunker down. On the economic and on the social front, government has a strong role to play in building stability and confidence – on the part of businesses, communities and individual citizens.  This was of course a major tenet of Keynes’ analysis of economic recession, differentiating him from Hayek. In conditions of great uncertainty, businesses would not invest: strong government action was needed, to boost confidence in the future buoyancy of the economy.  Otherwise the economy would ‘self-organise’ at sub-optimal levels of activity.

Second, Hayek pays insufficient attention to power differentials among social and economic decision-makers. The same goes for much of the complexity literature: including writers such as Schelling, concerned with a mass of micro-decision-makers, each following a simple algorithm as to whether to move house. The real world of contemporary capitalism gives scope for large corporations to shape the rules of those micro-interactions in their own interests. Not least, the more powerful are able to offload the costs of uncertainty onto others.  Complexity analysis must be conjoined with political economy.   

Third, there is little reason to confine the vision of a complex and interconnected society to the economy and market agents alone. The complexity literature can also inform our thinking about citizens and communities, their networks of interaction and the public services on which they draw. These too we can think of as decentralised systems with distributed intelligence, connecting a myriad of agile actors, with their varied projects and visions of change.  Here too we might examine the counter-intuitive outcomes that can emerge and how these vary, depending on the institutional conditions.

Finally, while the complexity literature can help us to understand the non-linear emergence of macro-patterns from micro-interactions, it helps little in evaluating those macro-patterns in terms of how socially benign or otherwise they are.   

What next for complexity theory and social policy?

Complex dynamics can produce a tangled knot of institutions and policy interventions, with path dependencies and lock-ins which are hard to escape, and trade-offs which allow of no easy resolution. ‘Wicked’ problems of this sort are unlikely to be solved by leaving them to the market, or by improving the technical skills of those in government.  What is needed instead is the distributed intelligence and wisdom of all those involved and cooperative solutions: including citizens, their communities and the public services on which they depend.

The literature on complex systems demonstrates that quite small changes in initial conditions can result in dramatically different dynamics and directions of travel.  It underlines that the future is open: there is always an alternative. This chimes well with longstanding critiques of ‘futurology’.    Nevertheless, faced with an open future, the more powerful do not hesitate to intervene, to ensure as far as possible that their interests are protected, their position consolidated and the costs of uncertainty displaced onto others.  To expose how these powerful groups narrow the range of possible futures is one of the tasks of the social scientist.

A second is to illuminate other possible paths of development, thereby enriching democratic debate.  This is no merely technical matter.  These alternative futures span the different values by which citizens live their lives and express their hopes and their fears.   

A third is to fix responsibility and enable powerful actors to be held to account. But is this even in principle possible, given the unexpected and unintended dynamics that can emerge from simple individual choices?  Think for example of Schelling’s interconnected city residents who collectively produce a segregated city: and which in the real world may mean the dramatic narrowing of choices and life chances for some of those ethnic minorities.  Complexity analysis can enrich our analytical toolkit and illuminate these tangled processes.  

Nevertheless, in a complex social system it is never wholly possible to allocate clear responsibility for such effects: it is necessary for the State to take ultimate responsibility for solidarity and compensation across the society as a whole.  Civility does not ‘self-organise’; it must be politically constructed; and we cannot escape the social and political choices of our time.   



The piece is was originally posted on the blog of the Institute for Policy Research, University of Bath.

Who is bankrolling Britain's democracy? Which groups shape the stories we see in the press; which voices are silenced, and why? Sign up here to find out.


We encourage anyone to comment, please consult the oD commenting guidelines if you have any questions.
Audio available Bookmark Check Language Close Comments Download Facebook Link Email Newsletter Newsletter Play Print Share Twitter Youtube Search Instagram WhatsApp yourData