Bill Dollins: ““GIS” as a distinct technological entity is disappearing”

Bill Dollins is the Chief Information Officer at Spatial Networks, Inc., where he is responsible for leading information management and security strategy. He works remotely from his home office in Southern Maryland, leading a team that is focused on optimizing the acquisition, management, analysis, and delivery of geospatial data.

Bill was interviewed for GeoHipster by Atanas Entchev.

Outside of work, he can usually be found spending time with his family, wearing out a pair of Brooks running shoes, or figuring out how to lift heavier things. He blogs less frequently than he used to and is planning to remedy that situation. He can be found on Twitter, LinkedIn, and GitHub. He is a fan of Washington, DC area sports teams, as well as the Alabama Crimson Tide, due to multi-generational family loyalties.

Q: You are the first person to appear in GeoHipster twice. How do you feel about that?

A: I didn’t know that and I feel slightly intimidated. I guess I need to make this good. I’m setting precedent, so should I go with dry sarcasm, self-deprecation, or over-the-top hyperbole? I think I’ll just wing it and see where it goes.

Q: A lot has changed since we last conversed on these pages. Tell us about your new gig.

A: I’ve been working at Spatial Networks, Inc. since February 2017. Many may know us as the company that makes Fulcrum, the leading mobile data collection application for iOS and Android. I joined at a fortunate time, at the outset of a significant period of growth for the company. As a result, we’ve done a lot of hiring and reorganized a couple of times to position the company for continued growth.

In my current role, I wear two hats as CIO and GIO. In the former role, I oversee the implementation and use of corporate systems and also address our corporate technical compliance with regulations such as the EU’s GDPR. In the latter role, I lead the management of our corporate geospatial data assets.

Those assets primarily take the form of data collected to support our Foresight data-as-a-service offering. With Foresight, we offer on-demand geospatial context on any topic in any geography for any duration. Combined with a global footprint, that can make for some unique data challenges and that’s where our data management team picks up. The data goes through QA/QC, normalization and restructuring to make it more consumer-friendly and ready for delivery. We’re using a mix of in-house, commercial, and open-source tools to build and automate processes to ensure consistency and shorten time-to-market. As a result, the last 18 months have seen SQL become my primary development language. I was always comfortable with it, but now it’s where I do most of my hands-on work.

That said, my role in the company is primarily strategy and leadership. That has given me the opportunity to work with an outstanding leadership team to steer the direction of the company and its product line. It’s also given me a chance to appreciate the roles played by design, product management, customer support, sales, and marketing in building successful products. I always understood that conceptually, but seeing people talented in those disciplines performing at a high level has really driven it home for me in a practical way.

I could go on, but I’ll sum it by saying I’m even happier in this role than I expected and I’m looking forward to the growth ahead. Oh….and we’re hiring!

Q: Any other important changes since 2014?

A: In addition to leaving the company in which I was a partner for 15 years, I also sold the house in which I grew up and built a new one. That happened shortly after the first interview, so it’s been quite a while now. It was a freeing experience that I could probably write about at length.

I also dipped a toe back into academia for the first time in a couple of decades by teaching an online course in the Salisbury University Geography program. It’s been a rewarding experience working with the students. It’s a masters-level course, so most are already into their professional careers, which brings a variety of perspectives.

Additionally, my alma mater, UMBC, knocked off 1-seed Virginia in the first round of the 2018 NCAA Men’s Basketball tournament (Go Retrievers!), Alabama won their 16th and 17th college football national championships (Roll Tide!), and the Washington Capitals won their first Stanley Cup (C-A-P-S, Caps, Caps, Caps!).

Q: In your 2014 interview you talked a lot about layers in GIS. What precipitated that, and has your outlook on layers changed since then?

A: I think I was working on some sort of network modelling behavior, which is something I’ve circled back to many times during my career. I re-read that response and I think I was inarticulately trying to say that I find traditional GIS inadequate for modeling our world. I still think that’s true, but maybe that’s also okay. Maybe traditional GIS isn’t meant to do that kind of modeling.

It’s been observed over the last few years that spatial technology is becoming more componentized and spatial analysis is getting embedded within other software tools. This is probably most obvious in things like R and Pandas, which present as more traditional statistical and data analysis tools, rather than primarily as a GIS. It is possible to do sophisticated spatial analysis in those environments, but they don’t drag along all of the overhead of an ArcGIS or a QGIS. I think that trend is accelerating.

There remains, and there probably always will be, a core constituency for traditional GIS. These are things like local government planning, natural resources management, parcel mapping, as well as a fairly exhaustive list of other use cases we can intuitively think of as the core audience for GIS as we’ve come to know it. These aren’t going away anytime soon and I don’t necessarily think they need the kind of modeling that I was discussing previously.

So, I think what I was really talking through was the fact that I was trying to use the wrong tool for the job at hand. The exciting part about that is that there are increasingly modular, component-based spatial analysis tools maturing alongside the traditional, monolithic GIS stacks. Evolution in both approaches means that it’s becoming increasingly easy to find the right fit in terms of use cases for spatial analysis and GIS.

Q: You coined the term “shapefiled”, meaning geodata whose quality has been degraded by converting it (them?) to shapefile format. Yet the shapefile popularity continues to grow. How do you explain an (allegedly) inferior data format’s undisputed reign?

A: Whoa, good pull. I had totally forgotten about that.

Giving the shapefile grief is like shooting fish in a barrel, but there’s an old saying: “Don’t let ‘perfect’ be the enemy of ‘good enough.’” The shapefile is an ideal example of something that is good enough at what it does to meet the needs of a broad audience.

Context matters. The shapefile wasn’t even the best format in the Esri stable at the time of its inception. That was the ARC/INFO coverage and I don’t recall anyone being in love with the shapefile back then. It isn’t conceptually much different than one of its 90s contemporaries, the MapInfo TAB, which was (is?) also a collection of sidecars. So why did the shapefile take off?

In 1998, Esri was under a lot of pressure to publish the binary specification of the ARC/INFO coverage. It was also feeling some heat from the nascent Open GIS Consortium for openness in general, so they published the shapefile. Anecdotally, I had friends who worked for Esri at the time who said the shapefile, since it was non-topological, wasn’t considered a serious format, so it was published to take the heat off the demand for the coverage.

I’m pretty sure that was never an official stance and I could never verify it beyond the anecdotes, but the end result is that the industry finally had the published, royalty-free binary specification of a geospatial format that was already in wide use. It took off. Within a couple of years, all of Esri’s commercial competition supported read/write of the shapefile, but it went beyond that. You (Atanas) may remember that, in its pre-Microsoft days, Visio had a “maps” plug-in where you could make Visio-style cartoon maps. It also supported the shapefile…an office productivity app supported reading a real geospatial format prior to 2000. It was a time when geospatial data was still a mystery outside of GIS, so a useful, open format was pounced upon.

Which brings us to today. The shapefile was so widely adopted so quickly that it litters file systems everywhere. It won’t ever really go away. And, because it is good enough, it presents a challenge to any potential successor that the shapefile simply never had to meet: the compelling reason to change. Thus far, no one has really come up with that reason for people who use shapefiles.

So, while the GIS world continues to search for/debate the perfect format, the one that’s good enough keeps going.

Having said all of that, I will gleefully roast marshmallows over the shapefile’s funeral pyre.

Q: Where is GIS headed? Today “spatial analysis” and “data visualization” are considered parts of “GIS”. But is the term GIS even appropriate anymore? Is spatial still special? When I went to grad school, we called it “Computer Applications in Planning”. These days many universities offer graduate programs in GIS. Is GIS a profession? Or it is a splintering set of tools that many different professions increasingly incorporate into their arsenal?

A: I view technology, especially software, as a concrete manifestation of the knowledge base of its developers and of the discipline in which they operate. So, “GIS,” as a set of software tools is a manifestation of the geographic body of knowledge. In terms of the body of knowledge, I think spatial is still special. A good example of this is a recent Twitter discussion I saw in which Morten Nielsen described the issues involved with unprojecting spatial data (https://twitter.com/JimBarry/status/1014702749102034944). It’s a great encapsulation of what I mean.

Projections are a core concept in geography, and using them incorrectly can result in bad data, erroneous results, and faulty decisions. Morten correctly describes how this works. That’s the body of knowledge. It is manifested in great software tools that have everything you need to correctly address such issues, but many people today see coordinate transformation as plugging a “from” EPSG code and a “to” EPSG code into a dialog box or a function call. That’s a great way to get bad data.

“GIS” as a distinct technological entity is disappearing, as it should. Spatial and cartographic techniques are gradually getting modularized and incorporated into other environments. Most vertical domains already understand how to use location in their activities. They want “just enough” GIS to do what they already know they need to do. For example, is R a GIS? I don’t think of it as one, but it has spatial analysis and visualization capabilities.

But that’s the technology, which doesn’t represent 100% of the knowledge base. Back to the projection example above. Any organization can plug proj4 or something into a piece of software, but they probably still need someone like Morten, who understands the appropriate use of the tools.

So, I see GIS splitting apart and diffusing across application domains. But, as the technology becomes more commoditized, the need for spatial understanding will increase and the value of the larger geographic knowledge base will grow. For the foreseeable future, I see the value of the technology in something of an inverse relationship with the value of the knowledge base.

Q: What would you say to a high school graduate who wants to go into GIS?

A: Don’t. Become proficient at something else and learn how to apply geography and spatial analysis to it. That’s not as contrary to the previous answer as it may seem. If you understand geography at the conceptual and practical level, and aren’t afraid to get your hands dirty with code or technical integration, I think there’s probably still a lot of mileage in being the geographer in an organization that does something else for a living.

Q: The war on cubicle body is raging. Update us on its origins, and the current theatre of operations.

A: I covered the origins in some detail here, but the short version is that 24 years as a defense contracting cube dweller had left me in the worst physical shape of my life. I weighed more than I ever had, I was diagnosed with asthma, and I my cardiac health was not perfect — though not terrible. I have a family history of cardiac issues, so I sat up and paid attention.

I joined a gym and started working with a trainer. The “war on cubicle body” was something I dreamed up to keep myself motivated, as that’s been an issue for me with regard to fitness. I started tweeting and my social media circle, many of whom read GeoHipster, has been incredible in its support. I can’t thank everyone enough.

I chose running as my main line of attack. I find that I need to organize my efforts around a central activity, so I chose running because it’s got a low barrier to entry and it’s easy for me to put on shoes and get a few miles in at lunch time. All of my other strength and core training is centered around getting better at running.

I am currently training for the Army 10-miler in DC in October. It’ll be my longest run yet and I’m looking forward to it. It will be the last race I run in my 40s and is an early birthday gift to myself. I’m certainly not fast, I’m simply looking to enjoy the training process and finish the race.

Q: Levi’s or Carhartt?

A: Mostly Under Armour and Nike Dri-Fit these days. When I have to actually wear long pants, it’s Levi’s 550 relaxed fit, never skinny (see the aforementioned cubicle body).

Carhartt is for people who do real work for a living. I have soft programmer hands and donning Carhartt would be a disservice to those who really need to wear it.

Q: Starbucks, Dunkin, or gas station coffee? Why???

A: Truck stop coffee. I realize there’s debate on this, but coffee is primarily a caffeine delivery mechanism. The best coffee starches your shirt from the inside out and no place does that better than a place that caters to long haul truckers. My order of preference is Flying J, Love’s, and then TA.

Since truck stops are not ubiquitous, I’ve been known to darken the door of a Starbucks or two. Dunkin coffee is generally weak to the point of being worthless.

At home, I brew my own. <shameless plug>I have gotten hooked on the French roast by Maryland’s own Rise Up coffee roasters.</shameless plug>.

Q: Is hipsterism dead?

A: Don’t get my hopes up.

Q: On closing, any updates to the thoughts you left us with in 2014?

A: You are not defined by the tools you use. Do not settle for the limits they impose.