Déjà vu all over again: Spatial science,quantitative revolutions and the culture of numbers

Spatial science (again)

Johnston et al. are clearly unhappy with my representation of spatial science and quantitative analysis. Writing a book on geographic thought is a dangerous enterprise. Each chapter in the book is about 10,000 words and seeks to account for a large body of work written over a long period of time. The book is not a ‘how-to’ book but an account of a few key questions and criticisms. I imagine that somewhere out there may be people equally discontented with my accounts of theoretical trajectories I am generally more sympathetic towards – post-structuralism or feminism perhaps. In the introduction to the book, I warn the readers that there will be much that is left out. And so there was.

Having said that, there is much that I share with Johnston et al. Indeed, my starting point may have been remarkably similar in intent. When writing the chapter on spatial science, I was convinced that spatial science had been given short shrift in recent accounts of geographic thought. I wanted to write more sympathetically and in an open-minded way. I argued that spatial science was not a thing of the past but a vital part of the present. I wanted to make clear that spatial science had brought us, among other things, a novel interest in ‘theory’, a focus on the nature of space and a sophisticated engagement with mobility and process. I also wanted to argue that spatial science did things that other approaches could not – and some of these were politically and socially important. There are, indeed, instances of work in the spatial science tradition that I have found inspirational and compelling (e.g., Forer, 1978; Openshaw et al., 1988; Schwanen et al., 2008). So, just to be clear, I see spatial science as an ongoing tradition in geographical research which has made, and continues to make, valuable contributions to the development of geography as a field of enquiry and which has the capacity to influence the world for the better in ways that few other approaches can.

I was therefore taken aback by Johnston et al.’s response to that chapter. Some of the difference between us appears to be based on some surprising assertions. The authors state, for instance, that almost my whole chapter was based on early work on central place theory, this is despite the fact that it only consists of three pages.

They also assert that the chapter does not do justice to new spatial science which is more focused on place than space – more sensitive to context and ‘place-based heterogeneity’. This kind of spatial science does not seek to generate laws of spatial behaviour but focuses instead on local variation. It is particularly interested in instances of outliers or unusual spatial patterns such as clustering. In a previous era, these would have been considered problematic outliers (Fotheringham and Brunsdon, 1999). This kind of work, in the words of its proponents, focuses on ‘identifying spatial variations in relationships rather than on the establishment of global statements of spatial behavior’ (Fotheringham and Brunsdon, 1999: 341). It is true that these forms of spatial analysis are different from the kind of spatial science that is featured in my book, I still, however, wonder what Johnston et al. mean by place exactly. They seem to mean something resembling ‘location’ – or the fact of certain things, people, characteristics being close to each other and different from those things, people and characteristics somewhere else. This is, at best, a shallow conception of place that really has to be understood experientially and from inside it to be fully present in analysis. I am not sure I can tell from the examples given how place is present in spatial science – or even that it should be. Place is not ‘spatial variation’. Someone fully tuned into the difference that place makes is unlikely to be happy with the findings and suppositions of ‘spatial autocorrelation’ and its arguments that ‘people who live near each other are more likely to be similar in certain characteristics than those who live furthest apart’ (Johnston et al., 2014). A focus on spatial variation through spatial auto-correlation, the expansion method or any other sophisticated form of counting does not, to me, seem like a sensitivity to what any writer I know on place, means by place. Place is not the same as ‘local variation’.

Quantification and the curriculum

The biggest beef that Johnston et al. have with my chapter seems to be the sense that I am reproducing a much more general antipathy towards work in the quantitative tradition. They clearly see spatial science as put upon. My perspective is quite the opposite. One of the central arguments in Johnston et al.’s paper is that quantitative methods need to be a significant part of a geographical education at all levels. To be frank, I think quantitative methods are consistently valorized in geography and beyond. I see the continued emphasis on quantification in human geography training at the postgraduate level as part of a wider ‘culture of numbers’. There is a very notable mismatch between the needs and expectations of PhD students engaged in an in-depth human geography research project and continued and institutionalised insistence of teaching them quantitative methods. I have been through this. As I state in Geographic Thought I had a ‘methods’ requirement as part of my doctoral program at the University of Wisconsin–Madison. At the time a method could either be statistics or a foreign language. I certainly was not in need of either for my research and neither were many of my peers. We could have done with, say, training in discourse analysis, visual methodologies or semiotics, but they would not even begin to count – despite their obvious utility. The situation has generally been better in the UK universities I have been part of where qualitative methods have been a key part of the postgraduate syllabus. But it is still the case that quantitative methods have been underlined as indispensible (despite the fact that they would remain useless for the doctoral researchers who had to learn them).

The Economic and Social Research Council and the various doctoral training programmes around the United Kingdom continue to offer extra support for quantitative research. Part of the reason for this is that relatively few people actually want to do any. Here lies a problem. Why don’t more students come out of their undergraduate geography programs wanting to do more quantitative work? Perhaps they are not being inspired. No one ever inspired me in the many classes I have had to take on quantitative geography and statistics. No one made me want to use them or convinced me that it was an exciting field of enquiry. My guess is that this experience has been quite common. There are many situations in my geography education when I was filled with senses of wonder, enraged with the injustice of the world or intellectually provoked by ways of seeing that opened up new horizons. None of these things happened in a ‘quant’ class. Not even close. Johnston et al. appear to believe there is some kind of structural plot against spatial science where students are being implored not to use numbers. I would suggest that it is more likely that those who teach it are not making a good enough case for it.

‘On exactitude in science’: The culture of numbers and the calculable world

There are striking similarities between the new quantitative revolution and the old one. The first time around, starting in the late 1950s, geography (alongside most of the humanities and social sciences) felt pressed to prove itself in the face of the advances of science. Important departments, such as the one at Harvard, had even closed. One aspect of the turn to numbers, therefore, was a need to attract the admiration of others in the academy as well as the all-important funding that tends to follow numbers wherever they go. In the current age of austerity, similar things are happening in the world of the humanities and the social sciences. Funding and credos are flowing into the science, technology, engineering and mathematics subjects on both sides of the Atlantic while funding to the humanities and social sciences is being frozen or cut. Once again we are being asked to prove our worth. It is in this context that we see the rise of initiatives around both ‘big data’ and its associated ‘computational social science,’ and the ‘digital humanities’. One thing that is clearly happening (again) is so-called ‘soft’ subjects attempting to become more ‘hard’. That, after all, is where the money goes. Numbers and the digital are seen as saviours for humanistic (in the broad sense) endeavours. It is an open question, however, whether the digital will save the humanities or the humanities save the digital.

Despite the undoubted advances in the complexity of quantitative techniques and the increased willingness to frame such techniques within interesting theoretical agendas, I still suspect that many of the critiques of the quantitative revolution from the first time around hold today. The contemporary focus in spatial science on local variability sounds like a numerical version of regional geography (what things gather where) rather than a revolution in explanation and understanding of place and the local. Spatial science, for better or for worse, is often a very elaborate form of description. For this reason, I think particularly of Harvey’s comment that spatial scientists spent a great deal of time and energy in proving that von Thünen’s model was true (substitute any number of more up-to-date models today) rather than thinking about the conditions under which it might no longer be true (Harvey, 1973).

One of the key reasons that is often given for the importance of training in numbers is the fact that there are so many of them in the world we inhabit. Good scholars and good citizens (not just good geographers), it is argued, need to have a grasp of how numbers work in order to not be duped by all those numbers out there in the world. One way in which this argument is formulated goes something like this. Big businesses, governments and an assortment of ill-intentioned people know about numbers and use them, therefore we (well-intentioned people) must do so too.

Most recently this situation has been framed within the notion of ‘big data’. Big data arises from the likes of Google with their insistence that once data sets become big enough (well beyond the abilities of any single computing device to manage or store), then there will be no more need for sampling strategies and algorithms. Data will closely match the world it is about. This situation is remarkably close to the absurdist map of the world in Borges’ prose poem ‘On Exactitude in Science’ in which he concisely accounts for a civilization in which cartography was briefly king and a map was produced of the ‘empire’ ‘whose size was that of the empire’. The map was quickly seen as useless and left to rot in the ‘Deserts of the West’. Many of those who use the term ‘big data’ are quite starry-eyed about the research possibilities that all these numbers entail. Elvin Wyly’s essay – ‘The New Quantitative Revolution’ is, in contrast, a tour de force of insight into all the ways in which are lives are being enumerated – right down to the bibliometrics that inform the assessments of academics throughout the world (Wyly, 2014). This is a world it is certainly important to understand. It is another thing entirely to suggest it is a world we want to encourage.

Two things that are making data suddenly big are the datafication of the individual and the geocoding of everything. Throughout Johnston et al.’s paper, there are references to the use of quantitative methods in government and business. In some instances, this is described as detailed knowledge at the level of the individual.

I am far from sure that I want to celebrate the entry of the individual into the domain of calculability. There are clearly issues for those who are unwittingly becoming sources of data – these are issues of privacy and surveillance as well as a more existential set of issues about the reduction of human subjectivity. There are also issues for those who are not being ‘counted’ – those who exist on the margins and are not therefore part of the algorithms that construct new forms of reality: ‘But Big Data has the potential to solidify existing inequalities and stratifications and create new ones. It could restructure societies so that only people who matter – quite literally the only ones who count – are those who regularly contribute to the right data flows’ (Lerman, 2013: npn). While Johnston et al.’s highlighting of the quantification of the individual is supportive of quantitative methodologies I find it worrying. Do we want individuals to be quantified in this way? Do we want to encourage it by using the data? There are, at the very least, serious ethical and political issues at play here. It is one thing to understand what big data means and it is surely another to produce more of it.

Even if we leave ethical and political issues aside, there is a more profound deficiency in the datafication of the world. It does not matter how much fuzzy logic is invoked, how much data is collected or how sophisticated algorithms become, it simply fails to even begin to approach our subjectivity. This is as true now as it was in the 1950s.

I do not believe that Johnston et al. support the more extreme pronouncements from the advocates of big data. Nor do I suppose that they believe that quantification can get at the deepest levels of human subjectivity. Nevertheless, I would argue that there is a connection between the rise of big data and its advocates and the more mundane ways in which numbers and quantification are valorized in the academy and in the ways we train our students. They are both part of an increasingly powerful culture of numbers. Yes, it is important that geographers know what correlation means (even just to know that it is still not causation) and that they are able to engage with probability. It is important that they know enough to question the stream of numbers that gets pumped out by corporations, governments and the media. I do not, however, want them to be part of the making of a comprehensively calculable world. Now, as in the 1950s, numbers and their aura of science and certainty are given priority over forms of understanding based on the visual or textual, for instance. In this sense, the call for continuous training in the quantitative inhabits some of the same ground that the generation of big data for the production of profit or the manipulation of populations does. This worries me.