Researcher: Christopher Watts
How can simulation develop theories in which the meanings and capabilities of objects are not "given" but change and develop in use?
Although innovation is a fundamental engine of social change, there seem to be no models of genuine novelty (and the related concept of an open-ended system) in active use by the social sciences.
To take an example from economics, how are the uses and "preferences for" a qualitatively new innovation (like the aeroplane) created, transmitted and evolved? These cannot usefully be viewed as mere quantitative extensions of existing technologies (like the bullet or kite) or as the realisation of "nascent" preferences (Ironmonger 1972, Witt 2001).
Before aeroplanes were invented, it is simply not plausible that anyone beyond a few visionaries gave sustained (or meaningful) thought to what would happen if they could fly, let alone what would happen when several hundred people could be taken across the Atlantic in a few hours or used to bomb skyscrapers.
This research strand will implement a simulation capable of representing qualitative innovation and the distinctive social regularities it is hypothesised to produce (fads, long latencies before significant adoption, varying kinds of diffusion curves, "fight back" by apparently dominated technologies) using the wide range of case studies available. (For a range of examples across disciplines see Rogers 2003.)
The methodological challenge addressed here, of relevance to economics, sociology, psychology, anthropology, management and linguistics, is how to build a simulation in which the "meanings" and "capabilities" of objects and actors are not specified a priori but arise through interaction in ways that are constrained but not strictly determined.
Rather than defining the set of actions which can be carried out using each object by each agent, as most existing simulations (and other formal models of innovation) do, we must develop simulations in which experiment, imagination and ignorance, the building blocks of genuine novelty, can be made meaningful.
There is a handful of modelling techniques outside social science, such as algorithmic chemistry (Fontana 1991) and genetic programming (Koza 1994), which can be used to provide insights into open-ended processes and how they might be modelled. But in contrast with the strand on repeated interaction, this strand deals with a substantive research absence in the social sciences rather than a substantive presence. It will thus serve as a forceful demonstration of what simulation can do that existing methods apparently cannot.