GIS and Academic Portfolio
The articles we’ve read this week, Establishing a framework for Open Geographic Information science by Singleton et al and Show me the code: spatial analysis and open source by Sergio J. Rey, provide convincing arguments for greater adoption of free and open GIS in academic spaces. Both papers approach the subject by outlining a dichotomy between “academic” GIS, which relies on proprietary software, and open-source models of GIS, which rely more heavily on volunteer contribution networks. I found the analogy in Rey of a cathedral and a bazaar helpful to build the distinction. With the former being built through top-down design, systematically planned and organized, the bazaar is fluid and evolving as people’s needs for it shift. While there is a complication here in that some open-source projects are still centrally controlled, as long as the source code is made available, and thus modifiable, it is still an improvement upon black-box proprietary programs. These contribution networks remind me of the way queer youth are often forced to learn about queerness through informal channels on the internet due to a current reluctance of schools to teach about queer identity. There is something lost by not incorporating less formal information networks into academic spaces.
One critique of academic GIS that the authors raise is a lack of reproducibility in their findings. Reproducibility is a topic that I most associate with disciplines like psychology, where type I statistical errors run rampant. And while there is likely some of that in academic GIScience, Singleton et al note that the primary reproducibility crisis is in the methods. By using proprietary software, supplying only narrative analyses of code, and obscuring the primary data sources (even with the well-intentioned goal of participant privacy), peer review standards for much academic GIScience fall apart. Then not only can the results not be reproduced, but the research as a whole cannot be reproduced. I found the example of the Harvard professors who omitted countries from their calculations particularly telling, and the fact that it took three years after publication for anyone to notice is cause for concern. I’ve always wondered if it would be possible to infiltrate academia and publish falsified findings, and as long as software, methods, and data remain closed from the public the safeguards to prevent this cannot be working.
The dichotomy presented between open and proprietary GIS certainly has its drawbacks. As Rey notes, there is a broad range of definitions and licenses with regard to how ‘open’ and ‘free’ a software can and has to be to meet certain standards. There is also an issue of casting paid softwares as ‘academic’ when schools like Middlebury have begun teaching the open-sourced alternatives, and there would probably be consequences in contributor communities as a result of the decreasing barriers of entry into those spaces. However, these articles have given me a good idea of how these softwares are created, constructed, and used by different communities. It always seemed strange to me that Middlebury teaches open-source GIScience as opposed to the ‘industry standard’ programs, and these readings have given me a better understanding of the thought processes behind this decision.