Communication infrastructures and the contest over location positioning

Introduction

This article charts the history of lesser known U.S.-based tech company, Skyhook Wireless, and its efforts to develop a location-determination system that draws positioning information from existing Wi-Fi networks. Skyhook Wireless (now known as Skyhook) was formed by Ted Morgan and Michael Shean in 2003. Their inspiration for the company came from frequent work-related travel. Both men were struck by the proliferation of Wi-Fi signals they encountered when they opened their laptops during business trips. In a 2009 interview, Morgan recalls: “We were amazed by the sheer growth of Wi-Fi. We knew there had to be a new model for mapping location using those signals” (Morgan cited in Wortham, 2009). This epiphany led them to establish Skyhook Wireless and pursue innovative ways of determining a cell phone user’s location in urban space by mapping existing Wi-Fi signals and using this network of signals as a positioning infrastructure for determining a cell phone user’s location. This process was far more effective at calculating one’s position in urban areas than GPS (Torrens, 2008), and had not previously been exploited commercially.

This history warrants recounting for the reason that it forms an important, if little understood, episode in the longer historical development of mobile media and communications technologies. Skyhook Wireless developed systems and software that were later integrated into first-generation smartphones. What is more, the broader methods for location determination they employed are now integrated into and part and parcel of wider smartphone design, manufacture, and use. Skyhook Wireless made, in other words, a less well known yet quite significant contribution to the success of the smartphone as a locative media device. In this sense, this account of Skyhook Wireless constitutes a form of “microhistory,” which, in historiographical terms, “is a genre that aims to make the smallest of . . . episodes luminous with wider significance” (Griffiths, 2016, p. 216).

Through this examination of Skyhook Wireless, this article seeks to contribute valuable new knowledge to our understanding of the location-related operations of mobile devices; the infrastructures associated with these operations; and the businesses that have emerged around, draw on, contribute to, and have come to dominate these infrastructural systems. Understanding specific instances of how these infrastructures were made gives us valuable insight into the often unseen workings of the larger mobile communication ecosystem, the corporate machinations that have shaped and continue to shape this ecosystem, and how it is that certain “keynote species” (Nardi & O’Day, 1999, p. 53), particularly Google, have come to dominate this ecosystem.

Wi-Fi and infrastructures

Writing in 2005, Adrian Mackenzie has suggested that concern for “communications infrastructure falls outside of most accounts of new media and communications” (2005, p. 271). However, a great deal has changed within media and communication scholarship since 2005, and this article builds on a renewed awareness—evident within mobile communication studies, “critical infrastructure studies” (Parks & Starosielski, 2015a, p. 7), and “materialist, non-media-centric media studies” (Krajina, Moores, & Morley, 2014; Morley, 2009)—of the significance of, and need to understand, the material infrastructures that underpin and sustain mobile communication. It is now well understood that Wi-Fi connectivity and the increasing “taken for grantedness” (Ling, 2012) of location-enabled mobile communication belie a complicated and largely hidden set of overlapping infrastructures that make finding and being found (Elmer, 2010) achievable. As Mackenzie (2010, p. 122) notes elsewhere, “the very possibility of wireless and hence locative media relies on antennae and radio signals in some shape or form.” And, as Lisa Parks and Nicole Starosielski (2015a, p. 1) observe, “today, broadcasting, cable, satellite, Internet, and mobile telephone systems are used simultaneously, and sometimes in coordinated ways, to route signal traffic to and from sites around the world.” There are also “beefed-up broadband pipelines, cloud computing systems, digital compression techniques, and protocols” that all play an integral infrastructural role in the movement and storage of this “signal traffic” (2015a, p. 2).

In broad terms, infrastructures are commonly understood as a “system of substrates” (“railroad lines, pipes and plumbing, electrical power plants, and wire”) that enable other things to happen (Star, 1999, p. 380). While communication infrastructures are generally designed so as to disappear, often appearing transparent to users and only visible once they fail (Horst, 2013, p. 148), there is a growing conviction that “renewed attention to the social, technical, political, economic and regulatory forms of infrastructure should be one of the key foci of mobile communication research” (Horst, 2013, p. 151), locative and social mobile media research (Farman, 2015a, 2015b; Frith, 2015, pp. 27–44), and broader media research (Packer & Crofts Wiley, 2012; Parks & Starosielski, 2015a, 2015b; Starosielski, 2015).

Recognition of the importance of communication infrastructures is also evident in established work on other related technical developments that have been crucial to the successful functioning of mobile communication and location positioning. This includes, to name just a few examples, scholarship on mobile phone and Wi-Fi standards and infrastructures, and the politics of their establishment, negotiation, and settlement (see e.g., Doyle, 2015; Goggin, 2006, 2011; Mackenzie, 2010; Sawhney, 2005), and work that documents the manifold ways that existing mobile phone infrastructures—mobile phone network cells, base station transmitters, and GPS—provide important means of determining a phone user’s location (see e.g., Goggin, 2006, pp. 195–197; Ling & Donner, 2009, pp. 31–33).

Much of the existing media and communications scholarship on Wi-Fi and infrastructures focuses on efforts to put infrastructure in place to build wireless networks, whether by community groups (Jungnickel, 2014; Middleton & Crow, 2008; Powell, 2008), cities (Lambert, McQuire, & Papastergiadis, 2014; McShane, Wilson, & Meredith, 2014), or nation states (Anayo & Horst, 2016), as well as evaluations of their successes and shortcomings (Middleton & Crow, 2008).

The emphasis of this article is different. Rather than examine cases involving the building of wireless infrastructures, my focus here is on one firm, Skyhook Wireless, that treated existing Wi-Fi networks as an infrastructure for the determination of mobile phone positioning. Skyhook Wireless is striking for the way that it built its business by piggy-backing on existing communication infrastructures (Wi-Fi, and, later, cellular towers and GPS) in order to create a system for calculating mobile location positioning, which it then transformed into software that was integrated into mobile devices for calculating the position of that device. In this sense, Skyhook Wireless’s systems operate within, and work to form, a complicated “assemblage of space, code, geocoded databases, and sensing devices” (Barreneche, 2012, p. 332). That is to say that Skyhook Wireless’s approach to building its location-finding systems represents a complicated articulation of what Rob Kitchin and Martin Dodge (2011) refer to as “code/space”—the mutual constitution of software and the spatiality of everyday life (p. 16)—particularly given the way that the firm’s operations bring about the intermingling of coded objects (that are reliant on software to perform as designed), coded infrastructures (networks that link coded objects together), and coded processes, to form coded assemblages (pp. 5–7).

Theoretical framework and research materials

Framing this historical account of Skyhook Wireless are two interconnected concepts—“infrastructural inversion” and “infrastructural work”—developed by science and technology studies scholars Geoffrey C. Bowker and Susan Leigh Star. Infrastructural inversion is an approach or critical orientation (if not a formal method) that involves “learning to look closely at technologies and arrangements that, by design and by habit, tend to fade into the woodwork” (Bowker & Star, 2000, p. 34). In the present context, this forms a productive approach given the tendency for Wi-Fi and location to recede from view. In addition to Wi-Fi’s “mundane existence” (Mackenzie, 2010, p. 13), location has also since become thoroughly integrated into a wide array of mobile and social network applications and is now normalized in use (Evans & Saker, 2017, pp. 95–96).

Infrastructural inversion also very much involves the drawing of attention to “infrastructural relations” (Bowker, Baker, Millerand, & Ribes, 2010, p. 99). As Bowker et al. (2010, p. 99) put it, infrastructure “emerges for people in practice, connected to activities and structures” and “consists of both static and dynamic elements, each equally important to ensuring a functioning system.” Furthermore, infrastructural inversion involves “recognizing the depths of interdependence of technical networks and standards, on the one hand, and the real work of politics and knowledge production on the other” (Bowker & Star, 2000, p. 34). Skyhook Wireless, for instance, was able to map existing Wi-Fi networks by exploiting a characteristic of 802.11 Wi-Fi standards that mean Wi-Fi devices constantly communicate with one another (Sandvig, 2004, p. 583).

Looking at Skyhook Wireless through the lens of infrastructural inversion sheds light on the infrastructural work the firm undertook to develop Wi-Fi-based location determination capabilities. What is valuable about the notion of infrastructural work is that it highlights how infrastructure emerges, how it is sustained, and how it functions as a dynamic “sociomaterial assemblage” that produces “intended and unintended outcomes” (Orlikowski, 2007, p. 1445). This focus on infrastructural work here is also important as it treats infrastructure as something that needs to be uncovered because of political, economic, and social struggles occurring there (Mackenzie, 2005, p. 273).

In developing this account of Skyhook Wireless’s historical development and its “infrastructural work,” I draw principally on two sets of key research resources: trade press reportage and patent documents. As reliable corporate data are notoriously difficult to obtain, especially for start-ups and privately owned firms, trade papers in particular remain a vital resource for scholarly research into the recent historical and political economic developments of companies operating within the fast-moving field of communication infrastructures and networked media (Corrigan, 2018; Wilkinson & Merle, 2013, p. 417). Thus, critical engagement with trade sources can assist in building a more complete picture of these arrangements. Consulting a range of trade press and related materials can prove valuable in reducing “information asymmetry” and other factual inaccuracies, thereby enabling greater depth of analysis through the cross-checking of multiple sources, especially around earnings reports, firm-initiated disclosures, and market reactions to these (Bushee, Core, Guay, & Hamm, 2010, p. 1). In the present context, a critical analysis of the trade press provides valuable insight into the rapidly shifting corporate landscape relating to communication infrastructure and location-based services, the business structures of and business decisions made by these firms (Mintz & Schwartz, 1985, p. xvii), as well as everyday representations of these firms.

I also draw on and refer to a select number of patent documents lodged by Skyhook Wireless that relate specifically to Wi-Fi network location determination and associated problem-solving. These documents provide valuable insight into how the company set about addressing the problem of Wi-Fi signal-generated positioning, and building the technological capabilities for capturing these location data from and about Wi-Fi access points, especially as they changed over time.

There are, of course, certain limitations to drawing on patents as empirical data that should be noted here. To begin with, the textual descriptions that are included in patents can be quite generic, or intentionally abstruse, in order to obscure just how the technologies are implemented so as to protect what is most valuable about these corporate properties. What is more, often the technical mechanisms that are represented in patents, both textually and diagrammatically, do not necessarily accord with their actual implementation; these mechanisms can be decidedly mutable and subject to subsequent modification. And yet, while the actual functioning of a technology is often simplified to an abstract model in patents, patents remain valuable instruments for seeking to understand how specific technical problems have been approached as sets of abstract relations and formal operations, and for identifying which aspects of this problem-solving process legal protection is being sought for. Within the discussion that follows, then, Skyhook Wireless’s patents are viewed and discussed not as technical blueprints or patterns, but, following Alain Pottage and Brad Sherman (2010), as “recipes” (p. 71) whose ingredients are “synergistic articulations” (p. 73) of old and new elements that seek, in combination, to “precipitate a particular effect” (p. 71) or achieve a certain outcome.

In the sections that follow, I detail Skyhook Wireless’s efforts to calculate location from Wi-Fi signals, explore the industry impacts of these efforts, and the contest over location positioning that ensued between Skyhook and its competitors (notably, Google). I then conclude by reflecting on some of the larger implications of Skyhook’s efforts.

Calculating location from Wi-Fi signals

In the presmartphone era of mobile communication, there were “broadly three ways of locating a handset or other user equipment with cellular networks” (Goggin, 2006, p. 195). The first took advantage of cellular radio design, and permitted the user’s handset to be identified as being located within a particular cell (2006, p. 196; see also, Ling & Donner, 2009, pp. 31–33). The second involved measurement of the time that signals took to travel from a handset to two or more network base station transmitters—a process known as triangulation (Goggin, 2006, p. 196). And the third was to draw on GPS signals, which permitted the “calculation of position based on propagation delays of different transmissions”—a process known as trilateration (2006, p. 196).

Ted Morgan and Michael Shean took an altogether different approach. In establishing Skyhook Wireless, their ambition was to calculate location using the existing “chaotic patchwork of the world’s Wi-Fi networks” (Wortham, 2009). (Later iterations of their software, as we shall see, supplemented the Wi-Fi data they were collecting with cellular phone tower and GPS data as well.)

Buoyed by several rounds of venture capital investment, totaling US$17 million (“Skyhook,” 2018), which included contributions from handset maker Nokia and chip maker Intel, Skyhook Wireless set about “mapping the contours of Wi-Fi reception” (Coyne, 2010, p. 148) in urban areas in the US that had high population density. The basic premise behind building such a map is set out clearly in the following passage:

Mapping radio sources provides a dynamic constellation of reference points, and such data can help in mapping other points of reference. Internal and external environments are permeated with Wi-Fi and cell phone carrier signals of known frequency. Any point in the city will have a characteristic spatial signature detectable by a suitable radio receiver. GPS signals are weak. . . . The stronger Wi-Fi signals generated deep within buildings can often be detected outside. (Coyne, 2010, p. 172)

Through mapping and recording them, these “privately owned indoor Wi-Fi sources located arbitrarily by their owners can be deployed vicariously to provide positioning to others” (Coyne, 2010, p. 172). This is precisely what Skyhook Wireless set out to do.

Constructing this map of existing Wi-Fi networks was Skyhook’s first major challenge. While Skyhook has been described as a location-positioning “pioneer” (Huang, 2011), the mapping techniques that Skyhook Wireless used had in fact been developed and tested by others. Between 1999 and 2000, data security consultant Peter Shipley pioneered “wardriving,” the practice of searching for and recording information about Wi-Fi wireless networks from a moving vehicle (Bulkeley, 2002). By 2003, the year of Skyhook’s founding, results from the third WorldWide WarDrive were presented at the DefCon 11 hacker convention in Las Vegas (Jardin, 2003). The technical means of mapping Wi-Fi networks via the process known as “beaconing” were also well documented. For instance, Tony Grubesic and Alan Murray (2004) provide a detailed description of the tools required: an 802.11b-ready mobile device; an omnidirectional antenna (pp. 11–12); software, such as NetStumbler (which Skyhook also used), that “listens for 802.11b signals and records information about access points for analysis, including their unique SSID numbers” (p. 12); and, a GPS receiver for recording the longitude–latitude (long–lat) coordinates of each Wi-Fi hotspot detected (p. 12; see also Torrens, 2008).

Using this same suite of technologies, Skyhook set about building a database of available Wi-Fi access points. An initial attempt at doing this involved paying taxi drivers to carry Skyhook’s recording equipment for them in their cabs (Wortham, 2009). This proved ineffective, however, and in its place Skyhook Wireless introduced their own fleet of Wi-Fi scanning (“wardriving”) cars. As per previous wardriving efforts, each car contained a laptop, window-mounted directional antennas, and technology that sent out “short blasts of radio waves, called probe requests, to detect nearby cell towers and Wi-Fi networks” (Wortham, 2009; see Figure 1). As noted earlier, the sending of probe requests works because 802.1 Wi-Fi standards are designed with interoperability in mind, such that Wi-Fi access points are constantly communicating with each other and with other Wi-Fi-enabled devices (Sandvig, 2004, p. 583).

OPEN IN VIEWER

Figure 1.

Skyhook Wireless’s patent illustration showing a scanning system for detecting Wi-Fi networks in order to determine the position of the scanning vehicle.

Based on the information returned via these “probe requests”—which were sent and received from many different angles as the Skyhook car traversed city streets—the source of the Wi-Fi signals was calculated, based on their direction and signal strength. From this information, the position of the Skyhook car was determined. This information was then logged to the on-board computer, and then later transferred to a central network database.

By 2009, Skyhook Wireless’s database was said to include 100 million wireless networks and 700,000 cell towers; the company also reported that it was handling 250 million location requests per day (Wortham, 2009). The following year, Skyhook claimed that their database had expanded exponentially to contain 50 billion scanned records of location signals (Buley, 2010). Skyhook also asserted that these cars had mapped areas containing 70% of the U.S. population, with the company later expanding their operations into Europe and Asia (Wortham, 2009). As a result of these achievements, Skyhook’s Ted Morgan is gushingly referred to as “a modern-day Magellan” (Buley, 2010)—even though, as we have seen, he and Skyhook Wireless were certainly not the first to have employed these methods.

A second, major challenge for Skyhook Wireless was “managing the chaos” of all these data (Morgan, cited in Buley, 2010). This included updating their Wi-Fi database to reflect changes, such as when people move and when new addresses are detected. To achieve this, Skyhook Wireless developed a system that uses “WiFi-enabled devices to monitor WiFi access points in a target area to indicate whether a WiFi access point has moved relative to its previously recorded location” in their database (Skyhook Wireless, 2013). This system works by communicating with Wi-Fi-enabled devices so that “observed WiFi access points identify themselves” through revealing their MAC (media access control) address (Skyhook Wireless, 2013). This information is then checked against Skyhook Wireless’s proprietary database of mapped Wi-Fi access points, in order to determine whether the observed Wi-Fi access point is a new one (and, if so, what its location is, which is calculated using “reverse triangulation” [Skyhook Wireless, 2008b]), or whether it is an existing one, and, if so, whether it is in the same known location or whether this location has changed and, if the latter, what its revised position is (again using “reverse triangulation” methods). Skyhook Wireless’s databases are then updated accordingly (see Figure 2). Access point data filtering and database renewal is a continuous and ongoing process, with newly found and “suspect” access points sent to the Access Point Reference Database (APRD) in real time so that the database can be “up to date at all times” (Skyhook Wireless, 2013, p. 11). Having received feedback data, the APRD then “determines whether to place suspect access points ‘on hold’ so as to prevent them from corrupting another user device’s location request” (Skyhook Wireless, 2013, p. 11). While the precise process and timeframe for resolution of “suspect” access point data is vague, the suggestion is that a “voting scheme” could be brought into play, whereby an access point is regarded as a low-quality reading if only one user has marked it as moved, and higher as more users corroborate its new position (Skyhook Wireless, 2013, pp. 11–12). Such information could then be used to determine whether the recorded access point is a new one or an existing one that has shifted position, and its status resolved within Skyhook Wireless’s APRD.

OPEN IN VIEWER

Figure 2.

A patent diagram providing an example of how observed Wi-Fi access points (APs) are compared against Skyhook Wireless’s proprietary database, how “bad data” are filtered out, and how new access points are added to the database.

A third challenge was to synthesize this data processing into positioning software that could be preinstalled or downloaded onto mobile handsets—Skyhook had already produced a location-based Internet toolbar called Loki (“Skyhook Wireless Announces New Version of Loki,” 2006). The first iteration of this mobile positioning software was called Skyhook WPS (Wi-Fi Positioning System), and it worked as follows. When a phone-based application required a location reading, it would make a request of Skyhook WPS positioning software to calculate the location of the device at that particular moment. That request initiated a complicated sequence of computational procedures—scanning, MAC address data capture, processing, filtering, and smoothing (see Figure 3)—that would then return location positioning information to the application the phone user was accessing.

OPEN IN VIEWER

Figure 3.

A patent diagram showing the procedures that are initiated when a location request is made of Skyhook Wireless’s mobile location positioning software.

Skyhook’s software was said to be able to “get a fix” on a cell phone user’s location within seconds to a distance of within 18 m (60 ft). Ted Morgan claimed Skyhook had solved a “40-year challenge to GPS.” It is widely understood that GPS signals are often weak in high-density urban areas due to the “spread-spectrum” techniques that GPS employs (Ceruzzi, 2018, pp. 62–65). GPS signals are “readily distorted or occluded” (Coyne, 2010, p. 172) by the presence of a proliferation of tall buildings, the deep valleys between them, thick tree canopies, walls, and other factors that create signal shadows or hinder the ability of GPS systems to effectively register location (see e.g., Forlano, 2009; Wilmott, 2016).

While the initial version of Skyhook Wireless’s location-finding system (WPS) “calculated a device’s position by comparing incoming Wi-Fi signals to a huge database of the known positions of Wi-Fi networks in urban areas,” in 2008 the company subsequently brought out a “hybrid” system, called XPS, that combined data from GPS and cellular networks with Wi-Fi signals “to get the most accurate location fix possible in any situation” (Roush, 2008; see also, “Skyhook Wireless Announces XPS 2.0,” 2008).

The previous synopsis highlights the significant “infrastructural work” undertaken by Skyhook Wireless to develop location-capturing capabilities. It is work that drew heavily on the infrastructural relations between technical networks and standards and knowledge production (by them and by others) to develop software that draws positioning information from Wi-Fi networks (and other sources). With their software in place, Skyhook Wireless then began pursuing revenue-generation opportunities.

Markets and the contest over location positioning

Early partnerships were formed with AOL and car navigation firm Navteq (as it was then known). Their technology was also integrated into Eye-Fi’s memory cards for digital cameras (Wortham, 2009). However, Skyhook’s principal source of revenue for its technology was through licensing deals with mobile handset manufacturers, earning the company up to US$2.00 per device each time its software was installed—a similar arrangement to how GPS firms worked with device makers (Graham, 2008). One 2010 estimate suggested that Skyhook was on track to earn US$25 million in revenue from these deals by the end of that year (Buley, 2010).

Skyhook’s big break came when it struck a deal in 2008 with Apple to have Skyhook XPS software added to Apple’s first-generation iPhones, iPads, and iPods. At the time of this deal, Apple also used Google technology for determining location. When an iPhone user launched an application that used location, the Apple device would determine whether it would get the best and fastest information from its own GPS chip, or from other sources (Wortham, 2009). In the latter case, it would first look for Wi-Fi signals, using Skyhook’s system, and, if there were none, then for cell tower signals, using Google’s technology (Graham, 2008). Skyhook had also made concerted, but ultimately unfruitful, efforts to get its location technology onto Android devices (Bray, 2014b).

The contest over location calculation

As Skyhook’s operations matured, and as its location positioning database grew in sophistication and market value, infrastructural relations became decidedly more complicated, with the company facing increasing competition from others interested in location positioning. This competition came from two main sources: smaller, specialist players such as Skyhook’s direct and long-standing competitor, the Florida-based Navizon (formerly, Mexens Technology; https://www.navizon.com/), and bigger mobile and tech players, including Nokia, Apple, and Google.

Despite significant early investments in Skyhook Wireless, Nokia forged its own path and “developed and acquired similar services that use[d] multiple inputs, like Wi-Fi hotspots” (Buley, 2010) to determine location.

Apple withdrew its support in 2010 (Ante, 2010); it was a major blow for Skyhook Wireless. This change was revealed in a letter that was sent by Bruce Sewell, Apple’s General Counsel and Senior Vice President of Legal and Government Affairs, to two U.S. members of Congress, Edward Markey and Joe Barton, who asked Steve Jobs for information about Apple’s privacy policy and location-based services. In this letter, which also revealed that Apple was also no longer relying on Google’s location services for its newer operating system and products, Sewell writes:

For devices running the iPhone OS versions 1.1.3 to 3.1, Apple relied on (and still relies on) databases maintained by Google and Skyhook Wireless (“Skyhook”) to provide location-based services. Beginning with the iPhone OS version 3.2 released in April 2010, Apple relies on its own databases to provide location-based services and for diagnostic purposes. (Sewell, 2010)

Sewell goes on to explain that, in order to operate successfully, Apple’s databases “must be updated continuously” by drawing on cell tower information, Wi-Fi access point (MAC address) information collected from users of its desktop and mobile devices, GPS information, and “diagnostic” information (“Apple may determine the device’s approximate location at the beginning and end of [a phone] call to analyze” dropped call issues; Sewell, 2010). Subsequent to the Apple decision, Skyhook struck deals with Samsung and Motorola (Ante, 2010)—arrangements, as will be discussed next, that would later prove controversial.

As for Google, Skyhook Wireless’s relationship with the tech giant was already under strain in 2007 when Google asked Skyhook to provide it with its Wi-Fi map database, Skyhook’s most valuable asset (Bray, 2014b). Concerned by Google’s motives, Skyhook Wireless declined this request (Bray, 2014b). Only a few months later, Google’s intentions were made clearer when it introduced its own free “My Location” technology, which detected information from cell towers, with this technology used in Apple devices (as noted before), as well as within Motorola, Sony Ericsson, Blackberry, and Microsoft Windows-enabled mobile devices (Graham, 2008).

The release of “My Location,” however, doesn’t reveal the full extent of Google’s consternation about the possible impacts of Skyhook Wireless’s location finding systems on its own operations. In order to fully appreciate the level of their concern, it is necessary to give consideration to Google’s Street View operations, and their controversial collection of Wi-Fi data.

The suggestion I want to draw out in what follows is that Skyhook Wireless’s grip on location finding, and the rich database of Wi-Fi access points they had built, became a key factor in driving Google’s highly contentious Street View program for collecting not only Wi-Fi-related MAC addresses, but also “payload” data as well.

The facts of the Google Street View data collection case are by now well known. Google initially admitted that, like Skyhook Wireless, yes, it was involved in wardriving and that it had been collecting SSID (service set identifier) data, which are 32-character Wi-Fi network name identifiers, and MAC addresses, which are unique numbers given to devices like Wi-Fi routers, and then associating this information with longitude–latitude coordinates obtained from a GPS unit located in the Google Street View vehicle, but they denied they were collecting “payload” data (data content transmitted via Wi-Fi). It was later revealed, however, that Google’s Street View cars had been systematically collecting payload data in addition to the aforementioned identifying information (SSID data and MAC addresses). Skyhook Wireless’s Michael Shean stringently denied collecting anything other than Wi-Fi access point identifying data, stating “we’re detecting, not connecting” (Shean cited in Graham, 2008).

The fallout from the Google Street View Wi-Fi—or “Wi-Spy”—controversy garnered a great deal of attention, both within the tech and popular press, and within academic scholarship. Discussed far less often throughout the whole Google Wi-Fi saga, however, were Google’s motives for wanting to collect Wi-Fi network information and Wi-Fi payload data. Wi-Fi network information improves Google’s location-based services, including those offered through its geolocation application programming interface (API). Google’s Peter Fleischer (2010) described a possible scenario where a user of Google Maps for Mobile enabled “My Location” in order to “identify their approximate location based on cell towers and WiFi access points which are visible to their device.” Thus, Google’s efforts to gather MAC addresses and SSID data further enrich its database of existing Wi-Fi access points (Sterling, 2010). And, “by recording the location and ID of wireless routers, Google can triangulate a smartphone user’s location faster and with less power than using satellites” (King & Gröndahl, 2012). In this way, the Google Street View “wardriving” collected Wi-Fi network data that contribute in vital ways to Google’s master or base map, known as “Ground Truth” (McQuire, 2016, p. 80). Meanwhile, Wi-Fi payload data add contextual richness for Google by revealing user “habits, preferences, routes and routines”—information that is “critical to the data-hungry economy” (McQuire, 2016, p. 79) and vital to Google’s geodemographic profiling efforts (Barreneche, 2012). In broader business terms, these combined streams of geolocation data are core to Google’s commercial success. As Alex Madrigal (2012) notes, “Google’s geographic data may become its most valuable asset. Not solely because of this data alone, but because location data makes everything else Google does and knows more valuable.”

However, a more specific motivation for Google’s Wi-Fi data collection program was to gain clear competitive advantage over Skyhook Wireless (Cleland, 2010). Prior to the launch of Street View, Skyhook Wireless already had one of the most comprehensive databases of Wi-Fi access points, proprietary location determination software (Skyhook WPS and XPS) for mobile handsets, and, as we shall see next, was in the process of developing a suite of mobile location analytics tools (Skyhook Wireless, 2017). In light of these developments, Google’s decision to engage in wardriving to capture location and payload data from unencrypted Wi-Fi channels was seen as part of their “competitive imperative to create a more accurate mobile-location-service than Skyhook” (Cleland, 2010).

The threat that Google perceived Skyhook Wireless posed to its location-related operations became more apparent when, at the same time as the Street View Wi-Fi data collection controversy was unfolding, Google discovered that mobile handset makers Motorola and Samsung had signed agreements with Skyhook Wireless (on top of those already signed with Google) for the provision of mobile location finding services. Google was quick to realize that OEMs (original equipment manufacturers) switching to Skyhook would be “awful for Google, because it will cut off our ability to continue collecting data to maintain and improve our location database” (cited in Patel, 2011). Understanding that “control over core data is critical” (McQuire, 2016, p. 79), especially given Google’s investment in the Android operating system (Lunden, 2011) and the rapid rate of uptake of Android-powered devices, Google acted swiftly and aggressively in response, instructing Motorola and Samsung to withdraw from their deals with Skyhook, which both OEMs subsequently did. Google’s somewhat spurious argument for calling for this withdrawal was that Skyhook XPS software failed Android OS compatibility tests and that Skyhook XPS data would “contaminate” Google’s location database, and confuse users, by registering “Skyhook WiFi / cell tower locations as GPS locations” (Patel, 2011).

Given the threats to its business model posed by Google and other large competitors, and despite its faith in its own software and Wi-Fi map data, Skyhook was, at the same time, also actively pursuing a longer term insurance policy: intellectual property protection through the lodging of multiple patents, with the firm claiming to hold over 450 patents in Europe and the US. As Ted Morgan put it, “We invented this space and we have a lot of intellectual property in this area” (quoted in Ante, 2010). Patent law, Pottage and Sherman (2010, p. 4) argue, works by “fictionalizing scarcity.” This is to say that patent documents have come to form a key instrument for Skyhook Wireless both in establishing geolocation data as a precious and far-from-bountiful resource that is highly sought after and monetizable, and in protecting this scarcity (through patent law) and thus the longer term financial viability of the firm’s location intelligence related intellectual property. Not only would these investments in intellectual property make Skyhook Wireless a more attractive acquisition prospect, should circumstances come to that, but these patents also held the potential to serve as “litigation machines” (Pottage & Sherman, 2010, p. 97) should competitors attempt to replicate their technical systems. This is precisely what came to pass in its dealings with Google.

Skyhook Wireless brought legal proceedings against Google, claiming two things: (a) anticompetitive behaviour on Google’s part, with Skyhook arguing that Google bullied Motorola and Samsung to pull out of signed agreements; and (b) infringement by Google of four location-related Skyhook patents (Hachman, 2010). In 2014, the first case was dismissed as Google was deemed to be “within its rights to demand that Samsung and Motorola break their deals with Skyhook, because of the two companies’ pre-existing contracts with Google” (Bray, 2014a). In the second case, in which Skyhook was “alleging that Google had illegally used its cellphone location technology, which relies on Wi-Fi stations instead of GPS satellites” (Newsham, 2015a), the two firms reached settlement just prior to trial, with Skyhook receiving US$90 million in damages (Newsham, 2015b).

Driving Google’s voracious appetite for personal spatial data was the belief that, when it comes to large datasets, “the sum is more valuable than its parts, and when [one] recombine[s] the sums of multiple datasets together, that sum too is worth more than its individual ingredients” (Mayer-Schönberger & Cukier, 2013, p. 108). While it would be a stretch to suggest that Google’s determination to push Skyhook Wireless from Android location service provision was the sole driver behind its hoovering up of Wi-Fi data, it certainly formed a major motivation for it (Cleland, 2010).

The outcome of this whole messy episode can be viewed in two ways. On the one hand, the Street View data collection program, as well as the Skyhook lawsuits, appear to have been quite high-stakes gamble for Google, given the fallout from the former and the US$90 million fine they were handed in relation to the latter. On the other hand, perhaps the stakes have not been so high after all. Google’s financial punishment for its global Wi-Fi data collection program has been minimal to modest, at best, and, more significantly, it has since emerged with secure longer term control of vital location resources on Android devices, and a position as the clear global corporate leader of consumer-oriented location services. What is striking is that, amidst all the legal and regulatory mess and noise of the Wi-Fi data scandal, this longer term strategic play has received remarkably little coverage, and appears to have succeeded largely unchallenged.

By 2014, Skyhook seems to have seen the writing on the wall. With dwindling opportunities for it to play a key role in the provision of core location positioning services for mobile handsets, the firm was acquired by TruePosition, a provider of wireless solutions and a subsidiary of Liberty Media Corporation (which also owns Sirius XM satellite radio; Skyhook Wireless, 2014). Part of the appeal for TruePosition of the Skyhook Wireless acquisition was also their extensive patent portfolio (Gannes, 2014). In the US, TruePosition provided precise location data for “enhanced 911” or “E911” services for emergency call centres (Ackerman, 2011), as well as “location intelligence” (LOCINT) to law enforcement and intelligence agencies in the US and internationally (Ackerman, 2011)—including, it has been suggested, in response to secret U.S. Foreign Intelligence Surveillance Court warrants issued to agencies such as the NSA in the wake of 9/11 and the passing of the U.S. Patriot Act (Sanchez, 2013). Following its acquisition, Skyhook Wireless continued to operate as an independent subsidiary. In 2016, TruePosition and Skyhook Wireless merged and were rebranded as Skyhook (without the Wireless), which now sits under Liberty Broadband, one arm of Liberty Media.

Under its new ownership structure, Skyhook set about repositioning itself as a broader location-based services platform. For instance, Skyhook developed a new Precision Location SDK (software design kit), with “an extremely small code footprint,” aimed at bringing Skyhook’s location technology to wearables (Miller, 2016). In addition, Skyhook launched its merchant-focused Context platform, for app developers and advertisers. The aim of Context was to convert “location from a simple latitude and longitude to an essential actionable data layer of context” for marketers and other clients to “deliver personalized, relevant content and dynamic experiences to mobile users” (Crowley, 2014). Context consisted of a number of components: geofencing technology, an Internet-based client (My.Skyhook), and a service called Personas for collecting and interpreting demographic and behavioural data.

With Personas, Skyhook sought to establish their credentials within the field of mobile location analytics and mobile geodemographics. As Carlos Barreneche (2012, p. 339) explains, while traditional geodemographics was built around residency (postcodes, households), in location platforms, geodemographic data are built around and from places (“venues”), and the tracking, recording, and interpretation of movement between places. Mobile location analytics, Harrison Smith (2017, p. 13) points out, “does not simply target particular audiences in specific locations in order to fulfil the imperative of relevance.” Rather, it proceeds from the premise that “it is possible to identify new kinds of [consumer] segments by clustering their location frequencies or propensities” (2017, p. 13). This is very much the form of geodemographic profiling that Skyhook is interested in.

Core to Skyhook’s U.S.-focused mobile geodemographics offerings is Personas, a service that is divided into three subcomponents, consisting of Retailer Personas, Power Personas, and On-Demand Personas. Retailer Personas aims to “measure attribution for campaigns associated with the 200 top retailers and brands in North America” (Rogers, 2016). Power Personas assists businesses and marketers to “identify consumer with strong brand affinity based on a high frequency of venue visits and other behaviors” (Rogers, 2016). And, On-Demand Personas works to build profiles “created from data that includes venue visits, venue type, visit frequency, and demographic information” (Rogers, 2016). In developing these offerings, Skyhook’s conception of “personas” very much accords with academic understanding of this term, which has been defined as capturing self-identity construction and agency, and the various forces that are brought to bear on these processes, including our movements through space and interactions with place, our engagement with forms of social connection and social networking technologies, associated affective clusters and micropublics, and how all these things in combination feed the constitution of public identity and identity performance (Marshall, 2014).

Both of the aforementioned strategic changes of company direction—the move into wearables, and embrace of mobile location analytics—have been perceived by industry commentators as attempts by Skyhook to “keep the company relevant in a changing market controlled by giants” (Miller, 2016) of the mobile operating system business (Google and Apple), and established and already strong incumbents in the case of mobile location analytics (Google, PlaceIQ, Foursquare). Of this list, Google in particular now holds an ever-tightening grip on provision of core location positioning and mobile location analytics services. Drawing a comparison with the board game Monopoly, Barreneche (2012, p. 343) notes that, “by owning the database of places, that is, by expanding a monopoly over the world’s metadata, Google gets to charge rents out of the accumulated value produced by users, local business owners, and developers through its advertising system.” Barreneche (2012) is correct to note that, “location-enabled socio-technical systems, and the power relations they embody, are not fixed” (p. 346), and that they are only “temporary stabilizations of ongoing negotiations between social and technical agencies” (Röhle cited in Barreneche, 2012, p. 346). However, in the present settlement of the location-based services ecosystem, location-related “metric power” (Beer, 2016) is held in ever fewer hands, making it increasingly difficult for smaller entrants, like Skyhook Wireless, to maintain a foothold.

Conclusion

In this article, I have traced the historical development of U.S. location-based services firm Skyhook Wireless. In constructing this “microhistory” (Griffiths, 2016) of Skyhook Wireless, I have drawn on Bowker and Star’s (2000) notion of “infrastructural inversion” for the reason that it productively sheds light on the detailed “infrastructural work” (Bowker, 1994) that Skyhook Wireless had to perform to create a software system for calculating mobile location positioning by drawing on existing communication infrastructural resources and data.

Infrastructural inversion also proved valuable for drawing out the complications and intricate infrastructural relations accompanying all forms of infrastructural work. In Skyhook’s case, the effectiveness of their location calculation techniques positioned them as a direct threat to much larger tech firms, such as Apple and Google. In response, these firms sought control over the mobile operating system market as well as control over the provision of mobile location-based services. Indeed, what they discovered was that it wasn’t possible to achieve the former without first achieving the latter. In Google’s case, achieving both these outcomes meant squeezing out Skyhook, including by developing (rather questionably, it turns out) their own database of location positioning information via their fleet of Street View cars.

In his book Wirelessness, Adrian Mackenzie (2010, p. 13) observes that, “Wi-Fi displays constant contractions and dilations . . . sometimes awash in broader shifts in mobility, sometimes stretched or frayed by contact with other infrastructures, media, and events” (Mackenzie, 2010, p. 13). As we have seen, one of the specific challenges Skyhook Wireless faced was that of how to make sense of the mutability and mobility of Wi-Fi stations. This account of Skyhook Wireless’s operations has also revealed how the mutability of Wi-Fi that Mackenzie identifies also extends to the political economy of Wi-Fi and related mobile location services: for Skyhook, the commercial opportunities that Wi-Fi positioning promised rapidly evaporated once firms like Google asserted their industry dominance.

In this way, this “microhistorical” account of Skyhook Wireless’s achievements and challenges makes a valuable contribution to a larger project of tracing the historical evolution of information infrastructures, and the slow but very significant shifts that are occurring regarding the governance of these infrastructures. Agnieszka Leszczynski (2012, p. 73) refers to these governance shifts—following Peck and Tickell (2002)—as being characterized in terms of “‘roll-back’/‘roll-out’ neoliberalism, whereby the welfare state is either being eroded or actively rescinding itself from securing certain services in the public sphere (‘roll-back’),” such as mapping, and where “their provision is opened up to market regimes, which subsume the delivery of these goods and services to their own logics (‘roll-out’).” As part of this larger historical narrative, the case of Skyhook Wireless also reveals certain regulatory failures insofar as the only real avenues open to Skyhook were legal responses (litigation) and legal structures (patent law). These legal responses and structures are not well equipped to deal with complicated public interest issues and implications attending the provision of location-related services, and they do nothing to diminish (increasingly oligopolistic) corporate control and market dominance.

Despite these technical challenges and missed market opportunities (and subsequent regulatory failures), it should be recognized that Skyhook Wireless managed to develop a highly effective system for determining a cell phone user’s location based on Wi-Fi signals (and, later, supplementary data drawn from cell towers and GPS), one that has since been integrated in, and a taken-for-granted part of, most contemporary smartphones. This “microhistory” of Skyhook Wireless thus forms an important episode in the larger story of the development of contemporary mobile communications. My aim in recounting this episode has been to add further definition and detail to present understanding of contemporary mobile, location-related infrastructures, the political economic forces that have defined and shaped them, and the corporate and legal machinations that continue to reshape them and that have led to the present strong concentration in corporate control of these services.