7: Following Fashion

Much research follows the current fashion. This leads me to develop an argument around two questions. How does ‘fashion’ come about? How should we respond to it? I begin by reflecting on my personal experience.

My research career began in transport modelling (cf. ‘Serendipity’ entry) in the 1960s and was driven forward from an academic perspective by my work on entropy maximising and from a policy perspective by working in a group of economists on cost-benefit analysis. Both modelling and cost-benefit analysis were the height of fashion at the time. (I didn’t choose transport modelling: it was an available job after many failed attempts to find social science employment as a mathematician.) The fashionability of both fields were almost certainly rooted in the highway building programme in the United States in the 1950s: there was a need for good investment appraisal of large transport projects. Planners such as T. R. (‘Laksh’) Lakshmanan and Walter Hansen developed concepts like accessibility and retail models. This leads me to a first conclusion: fashion can be led from the academic side or the ‘real’ policy side – in my case, perhaps unusually, both. It was realised pretty quickly – probably from both sides – that transport modelling and land-use were intertwined and so led by people such as Britton Harris and Jack Lowry, the comprehensive urban modelling field was launched. I joined this enthusiastically.

These narrower elements of fashion were matched by a broader social science drive to quantitative research though probably the bulk of this was statistical rather than mathematical. It is interesting to review the contributions of different disciplines – and this would make a good research topic in itself. The quantitative urban geographers were important: Peter Haggett, Dick Chorley, Brian Berry, Mike Dacey and more – a distinguished and important community. They approached the beginnings of modelling, but were not modellers. The models themselves grew out of engineering. The economists were surprisingly unquantitative. Walter Isard initiated and led the interdisciplinary movement of ‘regional science’ which thrives today. From a personal point of view, I moved into Geography which proved a good ‘broad church’ base. I was well supported by research council grants and built a substantial modelling research team.

By the late 70s and early 80s, I had become unfashionable – which may be an indicator of the half life of fashions! There were two drivers: the academic on the one hand and planning and policy on the other. There was Douglas Lee’s ‘Requiem for large-scale models’ (which seemed to me to be simply anti-science but was influential) and a broader Marxist attack on ‘positivist’ modellers – notwithstanding the existence of distinguished Marxist modellers such as Sraffa. And model-based quantitative methods in planning – indeed to an extent planning itself – became unfashionable around the time of the Callaghan government in the late 70s. Perhaps, and probably, the modellers had failed to deliver convincingly.

By the mid 80s, research council funding having dried up, with a colleague, Martin Clarke, we decided to explore the prospect of ‘going commercial’ as a way of replacing the lost research council funding. That is another story but it was successful – after a long ‘start-up’ struggle. As in the early days of modelling, we had ‘first mover’ advantage and we were valued by our clients. It would be difficult to reproduce this precisely because so much of the expertise has been internalised by the big retailers. But that was one response to becoming unfashionable.

By the 2000s, complexity science had become the new fashion. I knew I was a complexity scientist as an enthusiastic follower of Warren Weaver and I happily rebadged myself and this led to new and substantial research council funding. In effect, modelling became fashionable again, but under a new label (also supported by the needs of environmental impact assessment in the United States, which needed modelling). I feel now, in the 2010s, that the complexity fashion is already fading and new responses are needed. So we should now turn to the new fashions and see what they mean for our research priorities.

Examples of current fashions follow.

  • agent-based modelling (abm)
  • network analysis
  • study of social media
  • big data
  • smart cities

The first three are academic led, the fourth is shared, and the fifth is policy and technology led (unusually by large companies rather than academia or government).

The first two have some substantial interesting ideas but on the whole are carried out by researchers who have no connection to other styles of modelling. They have not made much impact outside academia. In the abm case, it is possible to show that with appropriate assumptions about ‘rules of behaviour’, the models are equivalent to more traditional (but under-developed) dynamic models. It may also be the case, that as a modelling technique, abm is more suited to the finer scale – for example pedestrian modelling in shopping precincts. abm is sometimes confused with microsimulation – a field that deserves to be a new fashion, which is developing, but where there is scope for major investment.

A curiosity of the network analysis is a focus on topology in such a way that valuable and available information is not used. For example, in many instances, flows between nodes are known (or can be modelled) and can be loaded onto networks to generate link loads but this rich information is not usually used. This is probably a failure of the network community to connect to – even to be aware of – earlier and relevant work. In this case, as in others, there are easy research pickings to be had simply by joining up!

The large-scale study of social media is an interesting phenomenon. I suspect it is done because there are large sources of data that can then be plugged into the set of network analysis techniques mentioned earlier. If this could be seen as modelling telecommunications as an important addition to the comprehensive urban model, then it would be valuable both as a piece of analysis and for telecoms companies and regulators but these connections are not typically made.

The ‘big data’ field is clearly important – but the importance should be measured against the utility of the data in analysis and policy – not as a field in itself. This applies to the growing ‘discipline’ of ‘data science’: if this develops as a silo, the full benefits of new data sources will not be collected. However, there is a real research issue to be talked here: the design and structure of information systems that connect big data to modelling, planning and policy analysis.

The ‘smart cities’ field is important in that all efficiency gains are to be welcomed. But it is a fragmented field, mostly focused on very specific applications and there is much thinking to be done in terms of integration with other forms of analysis and planning, and being smart for the long run.

There is one important general conclusion to be drawn that I will emphasise very strongly: fashion is important because usually (though not always) it is a recognition of something important and new; but the degrees of swing to fashion are too great. There are many earlier threads which form the elements of core social science which become neglected. Fortunately, there is usually a small but enthusiastic group who keep the important things moving forward and the foundation is there for when those threads become important and fashionable again (albeit sometimes under another name). So in choosing research topics, it is important to be aware of the whole background and not just what is new; sometimes integration is possible; sometimes the old has to be a continuing and developing thread.

Alan Wilson, April 2015

Leave a Reply

Your email address will not be published. Required fields are marked *