An assumption underlying policy-driven computer models is that improving model sophistication–more robust statistical techniques or more accurate data–will lead to better policy. Empirically, that assumption is repudiated over and over again but it doesn’t seem to impact most developers of these models.
This is never so evident than in site selection models. Site selection models are traditionally geographic information system (GIS)-based analyses that compare overlapping geographic features to determine the most appropriate location for a facility. Example geographic features are slope, location vis-a-viz a flood zone, adjaceny to like facilities, and existing land use zoning. The facility could be a plaza, a subdivision or, in the most contentious instances, a hazardous waste disposal site. Of all the hazardous waste, the most contentious and most dangerous tends to be nuclear waste. Consequently, you’d want to get that one right and optimize placement of the waste facility. Therefore, siting should be guided by data such as stability of geological formations and should limit the amount of political interference.
Comes a New Scientist article (subscription required, although here’s a report on the sites chosen by Nirex and an article on the political calculations from Scotland’s Sunday Herald) on the sites chosen in the UK to dispose of nuclear waste generated in the UK. New Scientist was one of the organizations that submitted a request for the information about the siting process under UK’s new Freedom of Information Act.
The articles report that, despite the sophistication of the computer siting models, they were systematically ignored when sites were actually chosen. Nirex chose the sites. Nirex is a public private partnership set up by the nuclear industry and UK government to monitor radioactive waste. Unsurprisingly Nirex succumbed to political and personal calcuations, such as political instability in Northern Ireland, which ruled out the entire region and personal threats to Nirex staff, which ruled out all of Wales. Sites ultimately chosen for test bores? Places that already had nuclear power plants because they exhibited a “measure of local support for nuclear activities.” Correlation with stability of geological formations? Zero.
What is surprising is the naivete about politics. Even the New Scientist magazine falls for the assumption, in its article last week entitled, Politics left UK nuclear waste plans in disarray. Of course, politics will significantly impact nuclear waste siting, to the extent of derailing a ‘rational’ process. No one wants them and do what they can to avoid them. Nicholas Chrisman wrote a journal article on the use of GIS to site nuclear facilities in the US. Number of facilities actually built? Zero. What you need is to build political calculations into the process with enhancements like spatial decision support systems. It’s not necessarily pretty. Negotiating with the public or local officials is lengthy and contradictory but then that’s the nature of democracy. It’s not easily quantifiable. How do you equitably measure one community’s opposition compared to another’s? However, if you want one of these sites to ever be built then politics has to be acknowledged as an essential part of the science.
(Whether or not we should be generating the nuclear waste is another matter entirely.)