On the Future of Spatial Analysis: The Role of GIS

A prognosis on the future of spatial analysis should be treated in much the same way as a weather forecast: both may act as the basis for spirited discussion, but both should be treated with extreme caution when applied to anything other than the near future. For instance, had I been asked to undertake this task seventeen years ago (highly unlikely as I was then a lowly first-year graduate student), I would have anticipated that spatial analysis would be dominated by general equilibrium theorists and systems analysts, both of which are now endangered species, if not already extinct, within spatial analysis. I still remember attending my first North American Regional Science Meeting in Toronto in 1976, fresh off the plane with a fairly traditional (albeit science-based) British geography background at a time when North American spatial analysis was in the grip of a mathematical fervour.(!) It was the zenith of the era in which the more mathematical spatial analysis was, the better it was perceived to be. It was a period when relevance was irrelevant. The meeting was dominated by papers that were not meant to be understandable and one could be forgiven for thinking that many were intentionally obtuse partly as a badge of intellectual machismo and partly to disguise the possibility that little of any consequence was actually being investigated. This was an era when spatial analysis was less mature and spatial analysts perhaps were not as convinced of the uniqueness of their subject as they now are; it was a time when urban and regional economists were our role models. Interestingly, our perceived heroes were probably disparaged by their economics brethren, because urban economics is viewed as the intellectual equivalent of the mad uncle to be kept hidden in the west wing of the castle. Lloyd Rodwin's comment, quoted by Wilson ( 1984, page 1422 ), is particularly apt here and bears repetition: "Putting the word 'urban' before 'economics' has the same effect as putting the word 'domestic' before 'science'". To return to the topic in hand, I do not think that the era I describe above will return in the near future for two reasons. One is that we have matured to the point where we recognise that spatial analysis is not simply aspatial analysis with a distance component thrown into the equation. We have a unique subject matter and increasingly realise that we cannot take techniques developed in essentially aspatial disciplines and expect them to work on our problems. As evidence of this, I would point to the development of spatial choice models based on psychological aspects of spatial information processing (Fotheringham, 1987; Fotheringham and O'Kelly, 1989) as distinct from the traditional aspatial models found to contain undesirable properties such as independence from irrelevant alternatives and regularity when