What is Open Source GIScience?
===
In a prescient 2009 meta-analysis of trends in open source GIS and academic institutions, Sergio Rey describes the social nature of open source, addresses the potential benefits and challenges of merging open source and academic interests, and discusses some of the pros and cons of open source software relative to proprietary tools. To begin with a general definition of open source GIS, it might be described as the collaborative, social, and uninhibited development of ‘free software’ for use in the context of spatial analysis. In this context, free software is distinct from freeware, which is free of cost (like free beer) but not freely modifiable or open (like free speech). Freeware is also retractable; for instance, Google allows free use of Earth Engine but could withdraw access at any time, or begin imposing costs. Additionally, there is a shade of difference between free software and open source – Rey describes open source as a rebrand meant to sidestep possible connotations of low quality in the word ‘free.’ Rey notes that the ‘free software’ movement today retains a greater philosophical commitment to its core freedoms and moral principles, while the open source philosophy is focused more on the practical applications. Sometimes the difference comes down to matters of copyright and licensing. In the freeware movement, stricter copyright licenses prevent code from being used in any future ‘closed’ applications.
In Rey’s vision, there is something profoundly idealistic about open source work, and the way it abolishes hierarchies in favor of an anonymous meritocracy of participation. However, that idealistic strain runs up against tension. Firstly, there is the steepness of the learning curve for beginners, which can make the community somewhat exclusive. Then there is a sense of competition with private companies that might fork code into independent projects. Private companies also tend to employ open source developers, and while this is certainly not categorically positive or negative, we might imagine that this allows companies to maintain some representation of their interests in open source endeavors. This could also begin to separate the ‘haves’ from the ‘have-nots’ within the open source community, as some will have more financial freedom to pursue their projects, while others might not. It is also unclear whether the idealized anonymity of the open source community obscures the same patterns of inequality as are found in academia and the private sector. For instance, Singleton et al. (2016) note that studies have found the open source community to be overwhelmingly male (97.5%) and young (median age of 30). In the final assessment, open source exerts a somewhat democratizing influence on the world of coding, but it has a lot of ground to cover in diversity, representation, and equality.
With regard to the integration of open source in academia, the benefits and opportunities are manifold; notably, researchers can implement novel methodologies through the customization of code and software, and students can download software to personal devices, allowing for greater personal ownership of their education. Rey writes that academic structures that value peer-reviewed articles over code contributions pose a major obstacle to wider adoption of open source principles in academia at large.
Institutional difficulties aside, Singleton et al. make a strong case for the urgency of adopting a reproducible paradigm in geographic research, with openly available data and workflows wherever possible. In particular, lack of reproducibility can lead to major errors going unnoticed and having negative repercussions in the real world. When research done with commercial data comes under question and the data cannot be made available, it can undermine the legitimacy of legitimate research or protect illegitimate research from closer scrutiny. With open data, these issues would be largely resolved, but a couple of other problems might arise, according to Singleton et al. For one, open data will have to be carefully screened to protect the privacy of research subjects and respondents. Open source developers are also at risk of “essentially forfeiting the ability to capitalize on their innovation” (p. 1512), though this could be resolved with the proper balance of openness and copyright protections. Whatever risks open science might entail, they ought to be greatly outweighed by the benefits. Researchers will be able to build off of one another’s findings more efficiently, reproducible research will help guarantee the verifiability of findings, and the community aspect of open science should unify and collectivize geographic research.
How might Open GISc apply to the liberal arts classroom? Learning in the open science world is collective; the sources of knowledge will be inherently eclectic. Random, often anonymous, voices on internet forums may provide the answers when established research does not. It will be critical to learn from the ways others code, but care must be taken not to draw heavily on the code itself. Whenever using more than the most basic line or two of code from an outside source, attribution must be given to the source. Considering the honor code, it would be advisable to work things out independently when possible, and when impossible, to work parallel to the code of others, rather than copy-pasting or drawing directly. In terms of evaluation, code may sometimes simply not work; in such cases, the emphasis should be given to effort – what intermediate outputs were achieved, and how many attempts were made? Collaboration between classmates should propel everyone forward, and assist those who struggle. There is space for individual initiative and potentially radical collective effort.
===
Rey, S. J. 2009. Show me the code: spatial analysis and open source. Journal of Geographical Systems 11 (2):191–207. http://dx.doi.org/10.1007/s10109-009-0086-8
Singleton, A. D., S. Spielman, and C. Brunsdon. 2016. Establishing a framework for Open Geographic Information science. International Journal of Geographical Information Science 30 (8):1507–1521. http://dx.doi.org/10.1080/13658816.2015.1137579
===