Are web maps recession exhaust? Notes on political economy of geospatial tech in the early 21st century
I’ve been thinking a lot about the conditions that set the stage for the period in which geospatial technologies for web and native apps blossomed in the 2010s. It’s hard to imagine an open source software based/open source tech producing geospatial startup receiving significant VC funding in the current economy, let alone multiple startups competing with each other receiving that kind of funding.
One hypothesis for the influx of money into open source geospatial in the 2010s is the zero interest rate phenomenon (ZIRP)—basically, low interest rates drove a lot of funds seeking higher yields into venture capital, which meant a lot more money sloshing around, which meant even something as unlikely to have a 10X return on investment as open source basemaps-as-a-service could get a piece of the action. Lots of app company shenanigans have been attributed to ZIRP, most notably the subsidizing of the true cost of companies’ services to entice user adoption and lock consumers into particular services: Uber being a cheap alternative to taxis and car services, running taxis and car services that couldn’t compete into the ground, then jacking up prices is a canonical example.
I think there’s some merit to the idea that ZIRP contributed to the growth of new geospatial technologies, especially when looking at open source development and funding for more capital-intensive endeavors like satellite imagery firms. Good geospatial data and tech is often tedious to make and difficult to do well—which is to say, it’s expensive, and it’s not entirely clear to me that it’s profitable (or, at least not hand-over-fist 10x returns profitable in the ways venture capital requires) absent the subsidy of state procurement. Low interest rates make investors willing to throw money at R&D and tedious problems.
Part of why I’m interested in this is because from the perch of academic geography, there have been calls to understand the “political economy” of geospatial technologies. In practice, this means people writing about state-market relations and business models of geospatial data and services. But the aspect of where does the money actually come from is sort of absent—as is a related question, which is whether or not these companies are actually profitable or just heavily subsidized (whether by government, VCs, or by being a Google subsidiary).
I’m trying to establish slightly more empirical arguments for this and I think the timeline for understanding these conditions has to go back a little further than what’s so far categorized as the ZIRP era (late 2008-2015ish and 2020-2022ish). Some of this was covered in (I am serious) one of my qualifying exam essays and I will probably have to turn it into like a for-real journal article and piece of a dissertation chapter; this is more of the brain dump and processing post for thinking through the idea.
Caveat for the state (you gotta hand it to them)
One disclaimer that I want to add here, even though it’s sort of obvious, is that governments do a lot of subsidizing of geospatial software and data projects. In addition to being a reliable client of private geospatial businesses, geospatial tech developed by the federal government (usually released to the public domain) subsidized a lot of private innovation. A super legible example of this comes from navigation tech, where the Census DIME dataset served as a foundation for basically every routing app ever made. Another, probably more obvious example is the ending of Selective Availability for GPS in 2001 opening up precision navigation and precision geospatial data collection to companies.
Pre-ZIRP: Geospatial and the dot-com bubble
Today, early geospatial web businesses are largely regarded as reminders of how far we’ve come and how cumbersome the early web was: imagine reloading a whole webpage to look at a different section of the map,, quelle horreur. But here are some things that I have been thinking about with the 2000-01 crash that might have had an impact on geospatial investment.
First, people who made a lot of money in the first bubble and managed to get out and/or stay rich after the market crash became the venture capitalists and backers of the next stage of tech. This affected lots of tech sectors—Peter Thiel and Marc Andreessen going in on social media, Elon Musk buying the right to be Tesla’s “founder”, etc. (Side note: there’s a great paper on how Musk’s first startup, Zip2, was an early example of the “spatial data arbitrage” business model, highly recommend!) For geospatial, the most legible example of the first-wave super-rich shaping the field is Mark Gorton (the creator of LimeWire) who used wealth accumulated in the first dot-com bubble to create and fund the nonprofit organization OpenPlans in the late 1990s. Originally focused on supporting innovation in urban planning, OpenPlans’ remit expanded to include development of and contribution to a number of open source geospatial software projects. That work, under the umbrella OpenGeo, was eventually rebranded and spun off as Boundless in 2013 (bought by Planet in 2018).
Also, Interest rates after the first dot-com bubble were within or about 1%, so formally speaking this was not a ZIRP moment, but it was a pretty big drop. It’s possible lower interest rates enabled Google’s initial geospatial plays in 2004. Bill Kilday’s book Never Lost Again claims the company offered $30 million for Keyhole, pre-IPO. It’s not clear if that’s what Google ultimately paid for it, and all the coverage of the Where 2 acquisition only offers an “undisclosed sum.” $30 million honestly sounds low for what becomes Google Earth, and Where 2 wasn’t even incorporated when Google acquired it. I would absolutely believe that Google underpaid for both companies (and/or, if they did pay a lot, a 1%-ish interest rate on whatever loans were taken to buy it was a negligible risk). In those first years of Maps and Earth’s development, they were given extraordinary resources mainly because post-IPO, Google was basically made of money. (There’s a longer thread here analyzing Never Lost Again that considers the profit model of Maps and whether organizationally dividing it from Search influenced its profitiability, but I should probably write that up separately.)
A less obvious but still geo-consequential example of a bigger company acquiring another in the post-dotcom pre-recession years is Yahoo acquiring Flickr in 2005 (for somewhere around $22-25 million, which honestly feels like a steal for what Flickr was) and GeoPlanet (for ????). The open sourcing of geospatial datasets from these two acquisitions happened post-2008 crash, which feels very of the Web 2.0 moment: the data largess of the web being released to the general public towards…who knows? But probably something good would come of it.
Finally, there’s the Tim O’Reilly of it all. It’s kind of incredible to realize what a massive impact O’Reilly had on tech writ large through both their publishing and conferences, considering how much smaller they seem today as they’ve shifted to a focus more on selling subscription services to tech companies (not very “open” of you, Tim, lol). While he didn’t coin the term, O’Reilly really pushed web 2.0 as a concept to re-invigorate the tech sector post-dot com crash, including running the Web 2.0 summit starting roughly around 2004 (though it didn’t get called that until like 2006). Notably, a parallel Where 2.0 conference also emerged in this timeframe. O’Reilly made a space for people doing weirdo stuff with geospatial to converge and collaborate, which planted a lot of seeds for future projects. (Also, because I’d be remiss not to mention it: the book Mapping Hacks basically happens because one of the co-authors worked at O’Reilly. That seems like an important data point.)
Pre-ZIRP: 9/11 impact on funding and vibes
As stated before, government is the reliable banana stand of geospatial tech. Consumer navigation and logistics are perfectly fine niche sectors, but they are more prone to the vacillations of the economy. The state, broadly speaking, is always going to need maps and data. And after 9/11, there was more money sloshing around for government tech needs, including geospatial. Some of that money was in research grants and contracts, and some of that money came from In-Q-Tel, the CIA’s venture capital arm.
On the research end, I haven’t found any examples of federal defense funding for academic research that produced specific geospatial tech that was later commercialized, but I also haven’t looked super-hard for examples. There is one example of DoD funding of research with a geospatial component that eventually was spun into a company that leveraged existing geospatial tech: PredPol (rebranded later as Geolitica, which I always think sounds like an STD), the predictive policing software company, emerged from an Army Research Lab grant project using historical data to predict and map potential sites of Iraqi insurgent conflict.
Probably the most famous In-Q-Tel geospatial investment is Keyhole, the company that eventually became Google Earth. That investment, in 2003, is interesting in part because if we assume Bill Kilday’s memory is correct, that investment was teetering on the edge until CNN started using Keyhole’s software during broadcasts about the wars in Afghanistan and Iraq.
Prior to Keyhole, In-Q-Tel invested in MetaCarta, a company that did geographic text search on unstructured documents. As a former MetaCarta employee explained it to me, there’s really only two industries with super-deep pockets who need geographic text search: the feds (e.g., “search for every mention of Al Qaeda and Pakistan in these cables”) and extractive industry (“search for every mention of shale in the Gulf of Mexico”). The company was acquired by Nokia in 2010 and then licensed the product and brand to another company, but MetaCarta’s biggest impact on the mapping space was the slippy map tech it created in-house and released as open source software. (Apparently, the impetus for in-house slippy maps was having intelligence agency clients: MetaCarta had an interface where you could look at a map of the locations mentioned in your queried documents, and they had to make that interface work in airgapped SCIFs so just using the Google Maps API was off the table.)
In-Q-Tel also invested in Boundless in 2013, though I know less about the legacy of that investment—it seems like Boundless’ whole play was figuring out how to get federal agencies off of Esri and onto open source software, operating kind of as a concierge/support service for those open source products? And for various reasons that probably have more to do with contracts than with software itself, Esri still won the contracting game.
There’s another way to look at the Global War on Terror’s impact on geospatial, which is that it galvanized a lot of activists and artists to engage with counter-cartography methods and use new digital mapping tools in unconventional ways. Think of Paul Chan’s 2004 Republican National Convention map or the touring exhibition Experimental Geography in 2008 or An Atlas of Radical Cartography. A lot of the core Mapbox crew came out of (I am serious) Howard Dean’s 2004 presidential run, known for its allegiance to open source and liberal-to-left politics. (This legacy remained pretty legible in the company for a decent time—I mean, back in 2013 a Mapbox employee made a demo showing off the company’s tech using data on US drone strikes in Pakistan (since deleted, but lives on in screenshots).
During ZIRP: the recession and the smart city
This is maybe specious, and largely just a riff on Anthony Townsend’s astute observation that “smart cities” basically were invented in the wake of the Great Recession because big IT firms like IBM were panicking about losing finance clients and decided to push this concept of data-driven urbanism to increase their government contracts. This didn’t necessarily create new investment in mapping startups, but it did increase demand for geospatial tech in one niche market which meant that companies doing geo stuff got some new business out of it.
During ZIRP: geo as scaffolding, moonshots
Similar to the smart city connection, ZIRP produced a lot of companies that needed geospatial tools do make their products work but weren’t exactly geospatial companies. Ridesharing and delivery apps are the most obvious but not the only examples here. Foursquare used Mapbox tiles. When Snap added a maps-based feature, it was powered by Mapbox. These B2B contracts were probably a great thing to show investors: look, we’re an infrastructure for this thriving ecosystem of other startups, and surely the money and enthusiasm for those other startups will never run out, right? (This of course didn’t account for companies building their own in-house mapping stack, which is what Uber and Snap eventually did, but more often the issue is the other companies died.)
ZIRP also encouraged what became Big Tech to swing for the fences by funding ambitious projects, some of which relied on geospatial data. Remember how self-driving cars on highways across the country was just around the corner for multiple years in a row? One of the things that makes that idea seem remotely possible is having a massive, super-detailed geospatial dataset that could be used to create software for a self-driving car—something that Google had been building for a while at that point. And the whole self-driving car buzz got VCs excited to invest in geospatial companies that could in theory help solve self-driving car problems.
Another way of talking about moonshots is more uh, literally: ZIRP was the first time since the early 1990s that saw significant investment into new satellite imaging companies. Satellites are way more capital intensive than software, you’re making a precision instrument that has to do stuff in fucking space. In theory, yes, this is a government-banana stand field that can rely on the ongoing existence of government contracts, but the industry has tended toward monopoly basically ever since Congress passed the law in 1992 that made it possible to have a private satellite imagery industry. The fact that investors were willing to take the risk on Planet and Skybox and Blacksky in the ZIRP years seems like an important data point.
But what does the data say? TBD
Does the actual amount of investment dollars geospatial startups received in the ZIRP era give us any insight into the role of ZIRP on geospatial as a whole? This is a great question that I’ve only started to poke around at. What I’ve seen so far isn’t obviously telling: Mapbox, for example, did funding rounds between 2013 and 2017 and then doesn’t do another round until 2020 (which maps onto interest rates slowly rising, then plummeting because of COVID). However: that 2017 funding round raised more than double the money the company had received in its first three years. This happened because one of the funders was SoftBank, not because Mapbox had some major breakthrough, and I don’t know if it’s fair to generalize the SoftBank effect onto other startups.
Another example that maybe holds more water is Carto. The company went from raising a total of $31 million between 2014-15 to no new funding save a 2020 European Innovation Council grant for €1.8 million, then in 2021—while interest rates are still rock bottom—raises $61 million. Obviously there’s probably a lot of other things going on behind the scenes but as far as “funding number go up when interest rates go down”, that’s at least pretty visible.
But just looking at how much money a given mapping startup received before and after ZIRP misses that ZIRP didn’t just mean “VCs threw money at stuff that wasn’t going to make money.” It meant that VCs threw money at a lot of different stuff, including maps. This was the era of Color Labs and Yo as much as it was of Mapbox and Planet. Maybe a better way of measuring this would be to look at the expansion and contraction of geospatial startups before and after ZIRP, seeing who exited and who didn’t and who folded and who didn’t.
Anyway
When I’ve posed this question to people who work in or have worked in mapping tech the responses have been mixed. One person argued that open source mapping tech succeeded in spite of venture capital, not because of its support (a sentiment I am still trying to unpack). Others have been skeptical because to them ZIRP suggests the most egregious excesses of Silicon Valley and they didn’t specifically experience that in their corner of geospatial startups, or it was relatively small compared to the excesses of WeWork or Google at its peak silliness or something. I haven’t really had anyone disagree that ZIRP was a factor, maybe just that it wasn’t something they were consciously aware of during their time in these companies. Maybe these breadcrumbs I’ve gathered don’t actually lead somewhere, or maybe it’s just the sort of thing that’s more visible in retrospect. I’d love to hear from people on both the geo and the VC side for their takes on this.