About

I am working on a dissertation about geospatial software history and in an effort to make myself a little more accountable to working on the project, thought I would post through it.

What the dissertation is about, briefly (or, things I’m looking at)

• How maps make money (and whether it’s sustainable)
• The labor of doing geospatial (ground truthing/data collection, standards-setting and its discontents, actually super-hard and annoying math, being union-busted)
• What the business and labor constraints and conditions of geospatial software do to the work of doing geo stuff

What the dissertation is about, broadly (cribbed from the proposal)

In my first year as a graduate student, a faculty GIS instructor came to the required all-cohort intro course to share her work. During the question and answer period, a student tentatively raised his hand. “Why is ArcGIS…” he paused. “Why is it like that?” The instructor smiled and nodded, understanding everything left unspoken by like that: why is ArcGIS—the default desktop software for doing work in geographic information sciences, used in geography departments and government agencies around the world—so clunky, why does it have so many dropdown menus and buttons, why is it only available on the Windows operating system, why is it so slow? Her answer was compassionate but somewhat speculative—as an ArcGIS power user, she is well-versed in the quirks of the software but not the processes that made it like that. Another student compared it to Adobe Photoshop: a similarly buttons-and-menus laden software, incredibly powerful for its specialized use cases but with its own learning curve.

The discussion moved on, but this brief moment has stuck with me years into my PhD for a few reasons. For one thing, hearing my peers articulate their own frustrations with ArcGIS was personally validating—it is annoying software! But I also believed there had to be a better answer for the “why is it like that” question than merely “doing geo stuff is complicated.” Not that I lack sympathy for the software engineers at Esri, the company behind ArcGIS. Working with geospatial data involves working on several hard technical problems simultaneously, as varied as planar geometry, boolean algebra, graphics rendering, information design, and computer vision. To do any one thing well is very difficult; to do all of them at once generally means making some compromise. That compromise manifests in user experience. As a result, geospatial software tools tend to fall broadly into two categories: discrete software libraries (many open-source) that do a limited number of tasks which get chained together in code, or overwhelming kitchen-sink desktop software like ArcGIS that for various reasons everybody sort of has to use. Like Tolstoy’s unhappy families, all geospatial software is difficult in its own way, and users of geospatial software—who come from a wide variety of fields including academia, government, journalism, real estate, urban design, activism, extractive industry, and logistics—wrestle with and sometimes begrudgingly come to love these tools despite their limitations because well, it’s what they have to work with. For everyone who doesn’t work with geospatial data and software and mainly encounters digital maps on their phones, this “why is it like that” underbelly largely goes unseen.

Works in GIS history offers few insights for the “why is it like that” question—in part because software has advanced significantly since GIS’ origins, and GIS history tends to emphasize origins more than, say, the last thirty to forty years of software development. As a result, the private sector’s role in establishing key standards and tools of the geospatial trade (whether as contractors to not-for-profit and state efforts, as industry association members collaborating on tools, or as proprietors of key products) has not been subject to as much robust research in geography, history of computing, or STS.

All that being said, I must confess: this dissertation will not get to the bottom of why ArcGIS is like that—in part because of in-progress scholarship on that question from another academic who miraculously has made inroads with Esri corporate and gained access to their archives. Instead, this dissertation takes a broader view at the core tools and methods underlying some of the most seemingly banal, taken for granted qualities—both by expert practitioners and by casual users of consumer-facing digital maps—of geospatial software. I’m looking at things like software libraries and data products and standards. This of course includes looking at companies (including Esri), but I wanted to get away from the tendency within academic geography to frame geospatial software in terms of its impacts and uses in geography itself and that meant looking at products for totally different sectors and markets and use cases.

Acknowledgments

This research received financial support from a Sloan Foundation Dissertation Improvement Grant.