How Google Made Android The World’s Most Ubiquitous Surveillance Platform
Big data has become big news. The New York Times recently reported that Facebook let smartphone manufacturers have access to its users’ locations and other personal information, Sen. Ron Wyden (D-OR) criticized mobile service providers for selling users’ device locations to data brokers, and Google admitted that several of its apps continued to collect users’ locations even after “Location History” was turned “off.”
While big data practices merit scrutiny, the current focus of public ire on Facebook is a mistake. Google engineered surveillance capitalism — the systematic collection and exploitation of consumer data for profit — and Google remains its biggest champion and beneficiary. Google’s profitability depends on its ability to collect personal information and track its users’ locations in real time, an ability Google has mastered like no other. Its Android monopoly is the most ubiquitous and capable surveillance platform the world has ever known.
This fact has received little public scrutiny despite its profound implications for public policy and societal norms. When a company’s profitability depends on tracking people, how can it ever be trusted to stop tracking them? Are cheaper phones worth a total reversal of societal norms surrounding privacy and individual autonomy? What happens when a single company knows virtually everything about everyone? How will it exercise that power?
Any effort to explore these implications should start with an understanding of why, and how, Google turned Android into a Windows-style monopoly for mobile that’s tuned for tracking consumers’ every move. The “why” is straightforward: there’s massive amounts of money to be made through surveillance capitalism.
The “how” is more complex and more hair-raising. Google’s plan for turning Android into a monopoly harnessed the appeal of the “open” internet narrative to seek regulatory advantages for its business model while masking its true intentions from policymakers and the public. In public, Google positioned Android as “the first truly open” platform for mobile devices, one that would unleash the potential of mobile technology (with the government’s help). In private, Google’s proprietary licensing agreements made sure there would be only “one Android platform” — one that Google alone controlled.
Google succeeded in using its openness narrative to build support for Android with policymakers and the public while leveraging its private control over Android to exclude potential competitors from the mobile marketplace and cement its dominance of the digital advertising market. The result is a mobile monopoly that has avoided antitrust enforcement (at least in the United States) and minimized scrutiny of its reliance on surveillance capitalism.
The Android Story
Android’s story begins in October 2003, when four technologists formed Android Inc. According to Android Inc. cofounder Andy Rubin, the company’s purpose was to create a mobile operating system that would know both a user’s location and preferences. Shortly before launching Android, Mr. Rubin told Business Week “there was tremendous potential” in developing mobile devices that could accurately track users’ locations and combine that data with their other personal information. Combining this information would enable advertisers to send targeted marketing information to a mobile device in real-time about nearby products and services in which the user is likely to be interested.
Google also saw the potential for mobile devices to enable this form of surveillance capitalism, and quietly acquired Android for an undisclosed sum in 2005. Google “understood early on that the shift from desktop PCs to mobile internet, which started in the mid-2000s, would be a fundamental change for Google Search” and the company’s digital advertising empire writ large. While the mobile internet experience was still limited at the time, it was already clear that mobile devices would dominate the internet’s future. Cellphones were outselling personal computers 3-to-1 and the gap was widening. Consumers who had never owned a personal computer or had a wired internet connection were going directly to mobile devices with wireless internet access, particularly in emerging markets. To survive the internet’s shift from the desktop to mobile devices, Google would need to ensure that its apps and services were widely available on mobile platforms.
Getting its apps on mobile devices wouldn’t be enough, however, for Google to thrive in a post-PC world. To take full advantage of the shift to a mobile internet, Google’s servers would need access to mobile consumers’ locations, the most critical component of surveillance capitalism on mobile devices. Based on behavioral economics, Google would also need to ensure its apps and services were the default choice for most mobile consumers.
Conquering the mobile market would be much harder than winning the battle for search on desktop PCs, where regulation and antitrust enforcement had resulted in connectivity and operating system standards that enabled developers to reach nearly all PC users with applications coded for a single operating system (Microsoft’s Windows).
Competition among device manufacturers and mobile network operators had resulted in a very different development environment for the mobile internet. In the mid-2000s, there was stiff competition among mobile operating systems (including Symbian, Windows Mobile, and Blackberry), mobile device manufacturers, and mobile network operators (with four nationwide operators in the United States alone), none of whom dominated their respective markets. To ensure widespread availability of its apps and default status in this environment, Google would have to code multiple versions of each app and negotiate separate placement deals with multiple device manufacturers and network operators whose priorities may not be aligned with Google’s.
Getting its apps on cellphones as default options was only part of the problem. To fully implement its plans for surveillance capitalism, Google would also need to negotiate for access to mobile devices’ location tracking capabilities, which raised regulatory and competitive issues.
The regulatory concern related to the privacy obligations of network operators. Mobile network operators are required to collect location information and transmit it to emergency personnel to help them locate 911 callers. When Congress designated “9–1–1” as the universal emergency telephone number in the United States, it recognized a societal norm — that “the privacy of information concerning consumers’ location while using mobile wireless services is critically important.” Congress amended the statute governing network operators’ use of consumer information to include location. It also required that network operators obtain the express prior authorization of a customer (so-called “opt-in” consent) before using or disclosing mobile location information, a standard that is stricter than the “opt-out” consent that applies to other personal information collected by network operators. To the extent Google needed to negotiate with network operators to obtain access to mobile devices’ location information, this law may have subjected Google to the same opt-in consent obligation as network operators. It wasn’t until 2013 that the FCC clarified that mobile operating system and application developers are largely exempt from this regulation.
From a competitive standpoint, there was no guarantee that Google would succeed in convincing device manufacturers to give it access to mobile location information. Google was not alone in recognizing the potential for profit in tracking consumers’ mobile locations. Device manufacturers, network operators, and other companies were all potential competitors, and in the competitive mobile market, device manufacturers and network operators had the ability to tell Google “no.” Google would later use this possibility to claim that regulation of mobile network operators was necessary to facilitate new entry in the mobile marketplace.
When it introduced the iPhone, Apple faced the same market environment as Google, but didn’t believe FCC regulation was necessary to succeed. Apple’s strategy was to offer a device so spectacular that it couldn’t be refused and then sell a lot of them — and it did. The iPhone’s global share of the smartphone market peaked at 23% in 2011 to 2012, and it currently accounts for about 12% of the market. This strategy worked so well that Apple became the most valuable company in the world.
The iPhone business model fit Apple, which had always made its money selling devices that “just work.” But it wasn’t a good fit for Google’s advertising business, where big profits depend on breadth — the sheer number of consumers advertisements can reach — not quality. Even if Google could develop its own phone and achieve a market share similar to the iPhone’s 12% to 23%, it wouldn’t be enough to ensure dominance of Google’s advertising business in the mobile era. To replicate the success of its desktop-based advertising business for mobile, Google needed to make its apps the default option on, and collect location data from, the vast majority of mobile devices no matter who makes them. According to its former Chairman and CEO Eric Schmidt, Google’s vision was that Android would “power thousands of different phone models,” a vision that was “more ambitious than any single ‘Google Phone.’” To achieve that vision, Google needed to make the mobile market work more like the developer-friendly market for Windows PCs, except that Google would control the mobile internet’s monopoly operating system rather than Microsoft.
In hindsight, it seems obvious that turning Android into a Windows-style monopoly was the keystone of Google’s mobile internet strategy. Control over a dominant mobile operating system would give Google the power to ensure its apps and services were default options on devices using its platform and the ability to control potential rivals’ access to that platform, which is what the European Commission recently found that Google has done.
In his book Dogfight, tech journalist Fred Vogelstein wrote that Google always understood that markets for operating systems have historically been winner-takes-all: “if you get enough people using your technology platform, eventually it creates a vortex [through the economic principle of network effects] that forces almost everyone to use it.” Google planned to create such a vortex for Android by licensing it for “free” (i.e., with no copyright royalty), which would maximize its distribution across as many different manufacturers and devices as possible. In a post on Google’s blog, Mr. Rubin euphemistically described the “vortex” economics of network effects as “a virtuous cycle” rather than the source of a potential market failure. According to Daniel Roth at Wired, Mr. Schmidt was more candid in 2007, when he told a group of business leaders at the World Economic Forum in Davos that it would be “the recreation of the PC story,” a story that ended in a durable Windows monopoly.
Bob Lee, then a top Android engineer, predicted Google’s licensing model would “turn [Android] into Windows [for mobile] with a ninety-eight percent market share, and … the iPhone would ultimately end up with just two percent market share.” Android has largely met that expectation, with a worldwide marketshare (excluding China) for licensable smart mobile operating systems of more than 95% and about 80% overall.
While Google’s strategy succeeded, it wasn’t obvious in the mid- to late-2000s that Android would become a monopoly or that Google would get away with abusing it for so long. Dogfight details how Google hedged its bets by partnering with Apple on the iPhone while it continued to develop Android. Apple worked closely with Google to ensure its mapping and other apps worked well on the iPhone, and in return, Apple CEO Steve Jobs sought assurances that Android wouldn’t compete with the iPhone. Mr. Schmidt “made it clear to Jobs that in terms of Google priorities, iPhone came first,” while at the same time Android’s developers were hoping it would be as dominant in mobile as Windows was on the desktop.
Mr. Vogelstein attributes Google’s “two-faced” approach to its uncertainty about who would win the the smartphone race. Though it’s plausible that Google would have focused its efforts on iPhone apps and let Android languish had circumstances turned out differently, it’s unlikely that Google would ever have voluntarily conceded control over the mobile market to Apple. By 2008, it was already clear that Google had no intention of letting Apple get in the way of its ambitions for surveillance capitalism.
In exchange for the use of Google’s mapping technology in the first iPhone, which launched in the summer of 2007, Apple agreed to send the iPhone’s latitude and longitude information to Google when a user opened its mapping app. When the companies attempted to renegotiate their deal in 2008, the talks ended in a shouting match. Google wanted more than basic latitude and longitude data — it wanted the raw data that was used to calculate an iPhone’s location, including the connection technology used (e.g., Wi-Fi or GPS), and Apple refused. Apple worried that Google would use the data to reverse engineer the iPhone’s functionality and that sending raw data to Google would violate users’ expectations of privacy.
Apple also had unmet demands of its own. It wanted to include Google’s turn-by-turn navigation functionality in the iPhone, and Google refused because it felt Apple’s terms were unfair.
In the aftermath of this conflict, Apple decided it couldn’t rely on a potential competitor for something as significant as location services. So Apple developed its own mapping software and removed Google’s mapping software from the iPhone.
It appears Google had come to the same conclusion when it bought Android in 2005: that it couldn’t rely on others’ mobile platforms to obtain detailed and ubiquitous consumer location data. To be a dominate force in surveillance capitalism, Google needed to be in control of the dominant mobile platform.
The Skyhook Saga
The central importance of collecting consumer location data to Google’s motivations for developing Android is evidenced by its relationship with Skyhook Wireless (as detailed in the proceedings in this lawsuit).
Skyhook — which was formed in 2003, the same year as Android Inc. — pioneered the use of Wi-Fi access points for enhancing the accuracy of location services. Earlier services relied on the satellite-based Global Positioning System (GPS) and cellular towers for location data. GPS can be very accurate, but is often slow and doesn’t work well indoors or in dense urban areas; triangulation of cellular towers generally works well both indoors and out, but isn’t very precise. As a result, location-based services that relied solely on these earlier technologies were a niche product.
Using Wi-Fi signals as an additional data point substantially increases the speed and accuracy of location services. Wi-Fi positioning systems rely on a database of known Wi-Fi access points to determine a device’s location. To populate its initial database, Skyhook had drivers survey the streets in tens of thousands of cities and other populated areas worldwide to plot the locations of Wi-Fi access points. Once an initial Wi-Fi database is created, it is continually improved and updated with new data generated by user devices’ location queries (i.e., “crowdsourcing”).
In 2005, the same year it acquired Android, Google obtained an evaluation license for Skyhook’s “XPS” location technology. When the evaluation license expired in 2007, Google asked Skyhook to provide information from its Wi-Fi database to Google, and Skyhook declined. Rather than renew its license with Skyhook, Google decided to compete in the market for location services by creating its own Wi-Fi database using data gleaned from its Street View project, which launched that same year.
In addition to capturing photo data, the Street View project used cars and, in areas that were inaccessible to cars, smaller vehicles known as “Google Trikes,” to collect location and other data from Wi-Fi access points. It appears Google had been preparing to collect this Wi-Fi data since at least 2006 — a year before the Skyhook license expired — which is when Google’s engineers created a Wi-Fi packet sniffer for inclusion in its Street View vehicles. Google appears to have anticipated that Street View’s Wi-Fi sniffer could raise privacy objections. Though Google issued a press release disclosing the project’s use of vehicles to collect photo data, it did not disclose that those vehicles would also be collecting Wi-Fi data. When it was revealed that project also collected Wi-Fi payload data, including usernames and passwords, multiple legal actions were taken against Google for invasion of privacy, violation of the federal wiretap statute, and other laws.
Google’s decision to create a proprietary Wi-Fi location database indicates that it considered direct control of the collection and use of precision location data to be a critical component of its business model. Google’s plan to build its own location database also explains why Google was so aggressive in trying to convince Apple to provide raw location data from the iPhone: Google could have used the iPhone’s raw location data to help populate Google’s Wi-Fi database more quickly through crowdsourcing. Speed was of the essence. It would take Google years to build its Wi-Fi database, and in the meantime, Google would be playing catch-up with Skyhook’s XPS service.
The competitive rivalry between Google and Skyhook blew up in early 2010, after Motorola issued a press release stating that it was replacing Google’s location service with Skyhook and would be “the first Android device maker to abandon Google for its location services.” Google soon learned that Samsung had also licensed Skyhook’s XPS software for use in its Android devices.
Skyhook’s software generated its own raw location data (i.e., latitude and longitude), but used Google’s Network Location Provider (NLP) app to translate that data into street addresses. Google’s NLP app was designed to collect the raw location data obtained from cell towers and Wi-Fi access points separately from the raw location data obtained from satellite GPS, a practice that enabled Google to crowdsource improvements to its Wi-Fi location database. Skyhook’s XPS application, however, combined all its location data together and sent it to the NLP as if it were GPS data only. Otherwise, Google would have been able to use data from its competitor’s (Skyhook’s) service to enhance the accuracy of Google’s own Wi-Fi location database.
The news set off alarms at Google, “particularly with respect to the prospect of losing the opportunity for data collection through Google’s location system.” A Google employee described the news as “devastating, especially since data collection is key to improvements.” Another wrote, “It’s sad to see first Apple, and now Motorola moving away from us, which means less collection for us.” Yet another lamented that Google hadn’t “made an improvement to accuracy in a year and a half for wifi or cell tower. So, the only way we would’ve closed the gap on Skyhook is if they got worse, which is unlikely.” The same employee worried that other device manufacturers would switch to Skyhook, which would “cut off [Google’s] ability to continue to collect data to maintain and improve [its] location database. If that happens, we can easily wind up in the situation we were in before creating our own location database and that is (a) having no access at all or (b) paying exorbitant prices for access.”
Google quickly swung into action to make sure the Skyhook deals wouldn’t disrupt Google’s ability to collect Wi-Fi location data from Android devices. Google declared that reporting Wi-Fi location data through Google’s GPS interface was a “data contamination” issue that was incompatible with Android and demanded that Motorola (and later, Samsung) stop shipment of Android phones with XPS installed. Google’s Andy Rubin described XPS as “a bad bug that breaks google’s [sic] internal systems and cannot be shipped until fixed.”
Google would allow Motorola to include XPS only if Skyhook’s software was revised to report the sources of its location data separately — i.e., in a way that would enable Google to enhance the accuracy of its competitive location service. Skyhook refused unless Google would agree to not collect the data for its own purposes, an offer that Google refused. As a result of this impasse, both Motorola and Samsung ended their arrangements with Skyhook.
There is no evidence that the so-called “data contamination” issue caused any compatibility problems with Android’s actual operation. From a consumer perspective, Android would have functioned fine with XPS installed. In the words of an internal Motorola memorandum, however, “[G]oogle is the final arbiter [of Android compatibility] and [Motorola] could not risk Google stopping shipment” on Motorola’s Android devices.
Skyhook sued Google on contractual grounds and lost. The court found that Google “considered the collection of network location data to be essential for operational and business reasons” and that Google’s efforts to advance its own economic interests through the collection of location data weren’t legally improper interference with the licensing deals between Skyhook and the device manufacturers.
In addition to unmasking the importance of location data to the Android business model, the Skyhook litigation publicly revealed the role that Google’s compatibility testing plays in making that business model work.
The Android Policy Campaign
The Skyhook saga made it clear that the opportunity to collect location data from the vast majority of consumers without the potential for interference from device manufacturers or network operators was essential to Google’s plans for Android. Realizing that opportunity, however, wouldn’t be easy. Google needed to convince device manufacturers and mobile network operators to support Android while downplaying its potential for monopolization and Google’s reliance on the exploitation of consumers’ private information to turn a profit.
Offering Android licenses for free was enough to garner support from device manufacturers, but had limited appeal for network operators. None of them were willing to partner with Google in 2007, when Android was still in development. Verizon and Sprint weren’t interested, AT&T hedged, and even T-Mobile, which ultimately launched the first Android phone, initially declined. Android could succeed, but wouldn’t achieve Windows-style dominance, if the two largest network operators (Verizon and AT&T) refused to support Android devices.
Once Google succeeded in making Android a monopoly, it would face the additional challenge of avoiding the same fate as Microsoft, which was subjected to antitrust restrictions for abusing its power over Windows, the dominant operating system for PCs.
Google’s strategy for overcoming these challenges was to galvanize the support of the tech industry, policymakers and the public for the Android business model before a single device was sold. It launched policy and public relations campaigns in 2007, long before its first phone shipped, that portrayed Android as a white knight of internet openness that would rescue the mobile internet from its allegedly “closed” approach to mobile devices. Google’s narrative painted network operators as villains and Android as an almost absurd philanthropic endeavor — in 2011, Mr. Schmidt testified that he was “not sure Google is a rational business trying to maximize its own profits” — while downplaying its role in expanding Google’s digital ad monopoly and its pervasive tracking of consumers’ locations.
The buzz about mobile “openness” started in February 2007, when Tim Wu, a professor at Columbia Law School who had previously coined the term “net neutrality,” presented his latest paper at a Federal Trade Commission event. Mr. Wu argued that mobile network operators were engaging in anticompetitive practices that were inhibiting the development of mobile applications and devices. He criticized the major third-party mobile operating systems (at the time) for limiting app developers’ access to the underlying capabilities of cellphones — e.g., their location tracking capabilities — and he criticized network operators for requiring certification of apps and mobile devices for use on their networks (similar to Android’s compatibility testing). He proposed that the Federal Communications Commission (FCC) adopt “open access” regulations requiring network operators to allow the use of any compatible mobile device with their networks while prohibiting the “locking” of cellphones to a single network and the “blocking” of mobile applications and content. He also recommended that the industry “work together to create clear and unified standards for [app] developers.”
Mr. Wu’s criticisms and proposed responses went straight to the heart of Google’s challenges in establishing an Android monopoly. His proposed regulations would ensure that Google apps and Android devices could be used with all mobile networks and that network operators couldn’t restrict Google’s access to devices’ location tracking capabilities; and Android would be a perfect fit for Mr. Wu’s recommendation for an industrywide app developer standard.
The tight fit between Google’s plans for Android and Mr. Wu’s paper may not have been pure coincidence. Mr. Wu’s paper was sponsored by the “Wireless Future Program” at the New America Foundation, a think tank that has received more than $21 million in funding from Google and affiliated entities and could boast that Mr. Schmidt was a founding member and eventual chairman of its board. (The New America Foundation’s version of the paper is no longer available on the think tank’s website in .pdf format.) Google’s funding was so important to the think tank that it ousted an entire team of scholars shortly after the team’s director congratulated the European Commission for fining Google for antitrust violations with respect to its search service. Though New America Foundation denied the firing was motivated by Google, publicly-released emails indicate the think tank was concerned that the team’s criticism of Google was “imperiling funding for others” in the organization.
Less than a week after Mr. Wu presented his paper, Skype Communications petitioned the FCC to adopt Mr. Wu’s open access proposals, and Google later urged the FCC to adopt them as part of a “national broadband strategy.” The FCC didn’t grant Skype’s petition, but in the summer of 2007, Google convinced the FCC to impose open access conditions on whoever won a set of wireless frequencies at 700 MHz that were slated to be auctioned for the deployment of 4G networks. This regulation was important enough to Google that it pledged to bid a minimum of $4.6 billion on the frequencies in order to alleviate FCC concerns that open access regulations would dissuade network operators from bidding. Google stopped bidding the moment it met its minimum commitment and the frequencies were won primarily by Verizon, which used them to build the nation’s first 4G LTE network.
Though Mr. Schmidt claimed that Android and its participation in the 700 MHz proceeding were “two separate initiatives,” the fact remains that its regulatory gambit ensured (by law) that Android devices would be able to reach the customers of the country’s largest network operator (Verizon) over its fastest mobile broadband network. This gave Google’s manufacturing partners certainty that Android devices would be widely available on at least one major mobile network in the United States. A Google blog post called this a victory for consumers, who “soon should begin enjoying new, Internet-like freedom to get the most out of their mobile phones and other wireless devices.” As Bryan Gardner, a business reporter for Wired suspected, however, “that internet-like freedom [would] be happening on Android mobile phones.”
Google also championed an FCC proposal to make television frequencies available for Wi-Fi devices that, according to a report prepared by New America Foundation’s Wireless Future Program, would penetrate walls more easily, have greater range, and lower the costs of unlicensed deployments. When asked about Google’s plans for these devices, Daniel Conrad, who was in Google’s strategic partnership division, denied that Google had “some grand master plan” for it. He didn’t mention that the deployment of more Wi-Fi devices with greater reach and ability to penetrate walls could improve the accuracy of Google’s location database, which was still secret at that time.
The Android Public Relations Campaign
Google’s internet openness campaign wasn’t limited to regulatory policy at the FCC. It included a significant public relations component that has served to shield the Android monopoly and Google’s data collection practices from more intense scrutiny.
The public narrative for Android was launched in November 2007, a few months after the FCC adopted regulation assuring that Android would have a home on critical 4G frequencies. That’s when a group of device manufacturers and other companies led by Google announced the Open Handset Alliance, a group committed to the deployment of handsets and services using the Android platform. The alliance called Android “the first truly open and comprehensive platform for mobile devices,” because it would be made available through an open source software license (Apache 2.0). According to the alliance, device manufacturers would “be free to customize Android in order to bring to market innovative new products faster and at a much lower cost.”
Many analyst’s initial reactions to the Android announcement were tepid. It would be nearly a year until the first Android phone was launched commercially, and as Om Malik, founder of GigaOm noted, Google hadn’t explained how Android would benefit its bottom line. Google’s decision to emphasize Android’s openness, rather than its ability to collect accurate location and other personal information from consumers for Google’s advertising business, did not impress the business-minded.
From a public relations perspective, however, the announcement was a hit. Advocacy group Public Knowledge, which supported the Skype petition and the FCC’s 700 MHz regulations, lauded Android as “a step in the right direction to benefit consumers.” In BusinessWeek, Spencer Ante praised Tim Wu as a “freedom fighter,” whose “wireless manifesto was the inspiration for Google’s new mobile software strategy.” (He was apparently unaware that Google had developed Android’s business model before Mr. Wu’s paper was published.) By casting Android as a virtual (and virtuous) David who would rescue the mobile industry from network operators’ Goliath, Google’s openness narrative served to distract policymakers and the public from scrutinizing Android’s true purpose — to become the world’s most ubiquitous surveillance device.
The Android Openness Illusion
As the Skyhook saga would later demonstrate, Android’s openness was illusory. Though Android is largely open for software developers, it’s completely closed for device manufacturers. In the Android context, “openness” means that Android’s source code is distributed under a royalty-free copyright license that permits the creation and distribution of derivative works (so-called “forks”). The Android license does not include the right to use the Android trade name or trademark or the right to distribute any of the marquee applications that most consumers associate with Android devices, including Google Play, Chrome, Google Maps, Gmail, YouTube, and Network Location Provider (which formed the basis of the Skyhook dispute). If a device manufacturer wants to sell a mobile device branded as “Android” or that has Google Play installed — an app that consumers cannot lawfully download themselves — the manufacturer must obtain Google’s consent through a separate, closed source “application license.”
In its antitrust investigation of Android, the European Commission found that Google has used these application licenses as a means of effectively “closing” Android by: (1) bundling all Google apps and significant Android functionality with Google Play, (2) paying large manufacturers and network operators on the condition that they exclusively pre-install Google Search on their Android devices, and (3) prohibiting manufacturers from installing Google apps on any device they make if they make even one competitive device that uses an Android fork (e.g., an Amazon Fire device).
The European Commission found that device manufacturers consider Google Play to be a “must-have” app that consumers expect will be preinstalled on Android devices, a finding that is consistent with the decisions of Motorola and Samsung to abandon Skyhook when pressured by Google. To preinstall Google Play, a manufacturer must also preinstall all of Google’s other apps and configure Network Location Provider as the default network-based location provider on Android devices. This bundling requirement has the effect of ensuring that all Android-branded devices come with all Google apps preinstalled and will transmit detailed location data to Google’s servers. Until 2014, Google also paid the largest device manufacturers and network operators to make Google Search the only search engine that was pre-installed on their Android devices. Google ensures compliance with these conditions by requiring manufacturers of Android devices to do compatibility testing with Google acting as the sole judge and jury, which is how it stopped shipment of devices with Skyhook’s XPS software installed.
The preinstallation of Google apps on Android devices has a significant impact on consumer behavior and the ability of rival apps to compete. For example, the European Commission found that in 2016, more than 95% of all search queries used Google Search when it was preinstalled on Android devices, but only 25% of search queries used Google Search on Windows Mobile devices when Microsoft’s Bing search engine was preinstalled. This evidence confirms what behavioral economics predicts: most mobile consumers tend to stick with a default application if it works reasonably well.
The biggest gut punch to Android’s openness narrative, however, is Google’s insistence that a device manufacturer can’t make Android-branded devices and devices that use “incompatible” Android forks at the same time, a so-called “anti-fragmentation agreement.” When Android was launched, the Open Handset Alliance promised that network operators and device manufacturers would have “significant freedom and flexibility” to customize Android without mentioning that anti-fragmentation agreements were already in place.
In reality, the only companies that can freely modify Android are Google or a company with sufficient resources and incentives to manufacture devices on its own, like Amazon. If a manufacturer wants to sell a version of Android that’s been forked without Google’s permission, that manufacturer won’t be able to sell Android-branded devices or preinstall Google apps on its forked devices. When Acer tried to sell a device running a forked version of Android without Google’s approval, Google threatened to cut off Acer’s access to Google’s apps and the Android brand, and Acer canceled the product. When Amazon tried to partner with device manufacturers on devices to run its custom Android fork, Fire OS, manufacturers were interested but unwilling to risk losing the right to sell Android-branded devices with Google’s apps. Amazon was forced to manufacture Fire devices itself.
Once Android achieved critical mass — a monopoly “vortex” (Mr. Vogelstein) or “virtuous cycle” (Mr. Rubin), depending on one’s perspective — Google no longer needed the openness narrative or the Open Handset Alliance. Though the Open Handset Alliance still has a website, it’s latest “What’s New” post dates back to July 2011, not long before Mr. Schmidt told Bloomberg that Android was “winning the war pretty clearly now” with the iPhone. According to SlashData research director Andreas Constantinou, the Open Handset Alliance has been a myth since at least 2008, when it became nothing more than “a set of signatures with membership serving only as a VIP Club badge.”
The Faustian Bargain
The story of Android illustrates the Faustian bargain of “free” and “open”: you can have a cheaper phone (that still costs real money) only if you’re willing to let Google record your every move, every day, all year long. That’s the price of “free.” The unanswered question is how much it will ultimately cost us.
Disclosure: I was Chief of the FCC’s Wireless Telecommunications Bureau in 2007, when the agency imposed open access conditions on certain frequencies in the 700 MHz band.