Annotated bibliography: Never Lost Again, Bill Kilday
Bill Kilday’s memoir Never Lost Again: The Google Mapping Revolution That Sparked New Inudstries and Augmented Our Reality is an inside account of Kilday’s time first at Keyhole and then at Google as the company dramatically transformed public experience and consumption of geospatial data and maps. Google isn’t a primary focus of my dissertation research mainly because the people I’d need to interview are 1) too rich/busy with whatever people with money do to talk to someone for a dissertation and oral history collection project and 2) probably NDA’ed to hell so there’s only so much they can tell me anyway. (If anyone has an in with any of the core Where2 guys tell them I just want to talk about Web Mercator, LOL.) But I wanted to give Kilday’s book a closer read on the off-chance anyone name-dropped in there would give me the time of day (my analysis: probably no) and just to get a sense of what the Google-approved version of the Google Maps/Earth story looks like. Here are some of my takeaways from the book.
History is contingent, part infinity
The first couple of chapters of Never Lost Again depict Keyhole in its pre-Google era as an extremely cool demo in search of a market. The company’s lineage is in graphics engineering rather than GIS or earth sciences, with a team comprised of mostly ex-Silicon Graphics people. They didn’t really know the mapping space when they entered it, which explains why for the first few years they tried a bunch of angles to keep the company afloat. My personal favorite of their weird deals was a 2001 partnership to make an Nvidia-exclusive version of their product, then called EarthViewer—another reminder of all the contingencies of tech history. Early 2000s Nvidia was viewed by Wall Street as basically too niche since only gamers really needed GPUs and the Keyhole partnership was an effort to diversify and prove its consumer applications.
Keyhole eventually found a decent niche selling its software (the not-Nvidia version) to people in commercial real estate but by 2002 it was still losing money. In January 2003 things were dire enough that employees were given the choice to receive equity in exchange for a pay cut to keep the company afloat. (A lot of employees took the pay cut, with some even forgoing any salary, which I have questions about how they were able to afford that.)
By Kilday’s telling, it’s the War on Terror that helped save Keyhole. A deal with CNN inked in March 2003 led to the broadcaster using Keyhole software segments about the wars in Iraq and Afghanistan. While the plan had been for the Keyhole team to work with CNN’s graphics team on pre-packaged videos, journalist Miles O’Brien instead basically just used EarthViewer live on camera, displaying imagery of a recently-bombed Baghdad and analyzing its impact. (As far as I can tell this wasn’t the beginning of satellite imagery as a tool in journalism, but it might have been the first time journalists could use and analyze satellite imagery that hadn’t been directly supplied to them by the government.) The CNN exposure—which importantly included the company’s URL onscreen every time it appeared—accelerated initially tepid In-Q-Tel conversations and led to a bunch of defense purchases (both for higher-ups who arranged deals for and lower-rank soldiers buying copies for themselves)
To Kilday’s credit, he recognizes that this business breakthrough via a false-pretenses war that killed thousands of Iraqi civilians sounds and kind of is (in his words) “perverse”; his personal opposition to the war added to the ambivalence. (He also notes that commercial real estate remained the foundation of the company’s 2003 revenue.) I’ve seen variations of this part of the Keyhole origin story that frame its use in Iraq war coverage as a mechanism for advocating for peace and empathy, a sort of Overview Effect rendering the place and people of Iraq as more concrete and relatable to the general public. There isn’t really any evidence for this as an intention or an outcome, though.
So on the one hand, there’s a contingency of whether Keyhole was going to make it as a company. On the other, there’s the contingency of the Google acquisition: Keyhole was weighing an investment deal with VC firm Menlo Ventures when Google first approached the company about an acquisition. I don’t think there’s an alternate timeline where Google never gets into geospatial but it is sort of funny imagining one where Keyhole remains a niche satellite imagery viewing and analysis software, making its bread and butter with sales to commercial real estate.
People have opinions about Marissa Mayer
While Kilday doesn’t say anything overtly negative about her, Marissa Mayer is the closest Never Lost Again gets to having something like a villain or enemy character. Google’s geo acquisitions were driven in part by the fact that geospatial queries (i.e., “restaurants in Austin”, “how to get to X place”) made up a significant percentage of Google searches. At the time of the acquisition, Marissa Mayer was in charge of a significant chunk of Google products—most importantly, search, but basically all other consumer products. She apparently wanted to have Google’s new mapping teams report to Search rather than be their own, separate thing. Some of this had to do with an earlier initiative Mayer had spearheaded called Google Local.
Keyhole managed to build out a separate little fiefdom for a while, seemingly mostly because someone in senior leadership saw that the Keyhole team worked really well together under John Hanke and Brian McClendon’s leadership. Maps (formerly Where2) was under Mayer’s purview, and for some time Kilday was in charge of marketing for both Google Earth and Maps meaning he reported to both Hanke and Mayer which was understandably annoying.
The announcement of a competing Microsoft product (Virtual Earth, which eventually became Bing Maps) before the launch of Google Earth (but after the launch of Google Maps) gave Hanke an opening to propose consolidating the geospatial teams of Maps and Earth into Google Geo, wresting control away from Mayer.
It’s not entirely clear to me why Hanke and McClendon run a team reporting to Mayer would have been such a bad thing. Kilday’s hesitation seems to have been “she was smarter than me and I was intimidated by her.” The conflicts between Hanke and Mayer aren’t presented as issues of engineering so much as ego and credit: Hanke bristled at Mayer doing demos of Google Earth, “his” product; Mayer wanted to rename Google Maps “Google Local” after her original project. I didn’t come away from the book feeling sympathetic to Mayer, but I do default to wanting to get a little more context when a narrative about a work conflict more or less plays out as “this woman has too much power and I don’t like it.”
Ironically, when Mayer was moved to running the Geo team years later and usurped Hanke’s position (this would lead to him creating Niantic), media coverage framed it as essentially a demotion for Mayer—which like, it kind of was, and it contributed to her decision to take the job running Yahoo.
But seriously guys (how) do maps make money
Another fun facet to the Mayer-Hanke turf war that isn’t explicitly mentioned in Never Lost Again: Mayer’s importance and influence at Google came out of UX work on Search and the development of AdWords (now Google Ads), which is to say the foundation of Google’s revenue. Having Google Maps under Search initially probably helped with integrating advertising, but beyond that the other path of Google Maps monetization is basically API use.
Keyhole’s primary business model prior to acquisition was selling desktop software. Upon its relaunch as Google Earth, the product was free with a Pro version going for a few hundred dollars (I need to double check the pricing, I think it changed over time—as of 2015 it’s free) and an enterprise product (which also sunset around 2015, the wind-down involved a sort-of confusing handoff from Google to Esri that I don’t know if I really want to spend time learning more about). It’s not clear from the book how lucrative Google Earth Pro or Enterprise sales were. Really, after the acquisition the funding that Keyhole had been desperately chasing for the first couple of chapters becomes an implementation detail as Google leadership gamely poured money into geospatial.
There were some cost-saving efforts within Geo, of course. Ground Truth, Google’s project to transition away from using Tele Atlas data and generate their own base map data primarily using Street View, is a good example (and, in the book, a pretty funny one because it really tries to paint Tele Atlas as baddies for jacking up their prices as Google grew—like I don’t have any loyalty to Tele Atlas but guys, you’re literally Google). But at least in Kilday’s telling, there really was a dedication to the founding mission of “making the world’s information accessible and useful” with little regard for the revenue model. It’s very 2013-coded.
I do wonder how the post-ZIRP, recession-pilled Google makes decisions about maps spending relative to the revenue geo brings in. It would explain why Google Earth Engine’s Javascript API was so fucking bad when it was an experimental free product (I haven’t used it since the full migration to a paid Google Cloud service, someone please let me know if they’ve migrated to ES2015). My impression from knowing ex-Googlers is that the company has a very feudal vibe and the influence or staying power of specific fiefdoms is sort of about revenue but more about relationships and ego, so maybe there’s some insulation on the Geo team.
Just a petty little side note
Sergey Brin’s brief cameos in the book all make him seem like the most tedious person in the world and in at least one of his appearances in the book he’s literally wearing rollerblades.
Someone I interviewed for this project made a comment about the web/not-Esri geo space is basically “Google, and then everyone else fighting for crumbs” and it is admittedly a little weird to be doing a dissertation on geospatial software that’s mostly focused on the crumbs economy. But the very fact that the dominant geo tech does seem to be largely subsidized by other revenue streams does say a lot, and the glimpses available from Google are still useful. For the dissertation work I’ll likely come back to the Ground Truth material because a recurring topic in interviews with geo people from various paths has been the inescapable labor of building out and maintaining basemaps and how hard it is to fully automate that task.