3.1 Predictions - uncertain or meaningless?

All reflexive research can be interpreted from many perspectives, but most reflexive research is not plesionic. It operates on statistically stable populations, where laws of large numbers obtain, dynamics are presumed linear and prediction is at least locally possible. Many social science disciplines, for example, study and are funded by institutions that have little to do with plesionic complexity. To qualify as plesionic, a research programme must deal with patterns of interaction among purposeful neighbours in a physical neighbourhood. This usually involves studying space-time patterns on two or more scales and finding a dynamic balance between synergetic and constraint-based modes of explanation. Archaeology, anthropology, sociology, economics, geography, environmental science and political science have each produced many texts that could be described as 'plesionic' because they deal with interactions between neighbours in populated neighbourhoods. In practice, however, it is easy to evade plesionic complexity by stepping up a level of aggregation and working with statistically large populations of agents coping with institutional or ecological constraints in a stable attractor.

Herbert Spencer's (1864) First Principles, for example, deals primarily with institutional constraints, conflict and the heroic model. Political Marxism (Marx 1867) works in a similar way, ignoring agency and focusing on constraint and conflict in large populations. By downplaying the role of symmetry-breaking events, small-scale synergies and multiplier effects, Spencer and the political Marxists could argue that the ontology of nations and power-relations was stable and the past was a good guide to the future. Human affairs seemed time-symmetric and prediction was possible, subject to some statistical uncertainties that could be ignored at a first level of approximation.

It is relatively easy, in a reflexive study domain, to translate familiar patterns into an ex post narrative and then re-present that narrative as a causal structure that can be used to make predictions ex ante. Those predictions can be strengthened by synergetic multipliers that re-shape human activity systems - perhaps by changing policies or being used to justify the use of institutional power to enforce compliance. The result can be a dangerous self-fulfilling prophecy. This is not a theoretical risk. Archaeologists, biologists, sociologists and economists have, on occasion, been complicit in genocide. The social science wars and revolutions of the 20th century, for example, were often justified with reference to scientific authorities who had backgrounded plesionic complexity by up-scaling their own research.

The consensuses that crystallised around Spencerian liberalism and Marxian dialectic, for example, pitched the world into a series of heroic struggles that killed on an industrial scale. Ernst Schumacher (1973, 31) famously wrote that: 'The greatest danger invariably arises from the ruthless application, on a vast scale, of partial knowledge'. To this we would add that re-framing a plausible ex post narrative as a causal mechanism is a particularly risky strategy.

The contrast between Marx, Spencer and Darwin in this respect is striking. Darwin's Descent of Man (1871b) was an ex post explanation of human evolution in which agency was a significant and pervasive feature. Marx and Spencer, on the other hand, were predicting the future, downplaying the role of agency and emphasising the generality of their insights. Plesionic texts tend to be very sensitive to time-asymmetry and the unreliability of ex ante prediction. The difference is not that between certainty and uncertainty - all predictions are uncertain - it is that between meaningful and meaningless predictions.

A prediction of the mean annual temperature of Britain in 2050, for example, is uncertain but not meaningless because the land mass and the concept of temperature are unlikely to be changed much by human agency. To predict the GDP of Britain in 2050, however, would be meaningless because political and economic institutions can be re-shaped and even destroyed by socially constructed emergents and multipliers.

The dynamics of any human activity system are critically dependent on patterns of belief, habit and socially constructed consensus. If beliefs about virtual agents change, then institutional structures can become vulnerable very quickly. The result is often a power struggle between conservative and reforming factions that can destabilise whole ecosystems. We humans can innovate, changing the course of human evolution by changing our minds (van der Leeuw and Torrence 1989; van der Leeuw and Redman 2002). The ability to do this accentuates the time-asymmetry of human affairs, often generating cascades of innovations, each with emergent side-effects that cause collateral damage and trigger new responses. We have become 'apes in skyscrapers' (Dupré 2014, 310) whose co-evolutionary ecology cannot be understood without reference to agency, habit, purposeful action and belief (Corning 2014).