Analysis: Android is a Platform, not a Commons
November 21, 2016
Paul MacDonnell, Executive Director, Global Digital Foundation
The European Commission is taking aim at Google which it says may be abusing a dominant position within the Android mobile smartphone market. The Commission believes that Google’s agreements with manufacturers who wish to preinstall Google Mobile Services (GMS), Google phone apps, on their devices come with conditions that harm competition and consumers. Moreover the Commission believes that Google’s agreements with manufacturers not to fork Android—that is not change or neglect the software in a way that would make it incompatible with Google Android—prevents them from selling devices that use competing Android operating systems, thus further preventing them from introducing apps and services that could compete with Google. However, there is no reason to think that Google’s policies are anti-competitive. Google’s efforts to ensure that its installed apps work properly and that phone manufacturers are discouraged from forking Android, the most widely used phone operating system, are not only mild from a competition perspective but, if successful, are likely to benefit competition in the market for smartphones. Furthermore by defining the relevant markets as those for search, phone operating systems and apps the Commission would seem to be ignoring Google’s role in the overall market for smartphones, a market in which it has little power and from which it earns very small revenues.
The Commission’s complaints
The Commission’s complaints boil down to three arguments:
- Agreements that require manufacturers to preinstall GMS, limit their freedom to install apps that compete with Google;
- Agreements that require manufacturers not to fork Android prevent competition from devices running alternative versions of Android and, hence, foreclose competition from apps and services that could compete with GMS;
- Paying manufacturers to exclusively preinstall Google Search reinforces Google’s dominance in search.
The first two of these allegations are wide of the mark. Google’s agreements with manufacturers do not limit them from preinstalling other apps on their smartphones and, in most cases, that is exactly what they and the mobile carriers do. And, far from preventing apps and services on forked versions of Android from competing with Google, Google’s efforts to standardise Android will give its competitors the confidence that their apps and services will work on the same phones as GMS, thus enabling direct competition with Google. The third complaint appears to be an extension of the allegations expressed in the Commission’s separate case that Google is abusing dominance in search and would appear to partly rely on these concerns being validated. However, search is currently in a state of flux with a number of new competitors, including voice-activated search engines, now entering the market and it is not obvious that these competitors are being kept from the market by any actions taken by Google. More important, however, is the question that this allegation begs. In identifying markets for apps and mobile operating systems is the Commission ignoring the market Android and Android’s apps are in, namely, the market for smartphones?
The positive and negative externalities of platforms
On its own Android is a multi-sided platform, a framework that brings together more than two sides, or parties—in this case app developers, phone manufacturers, mobile carriers, users, and providers of services via apps—who benefit from mutual interaction. Multi-sided platforms are not new. The Galeries Royales Saint-Hubert shopping arcade which opened in Brussels in 1847 is a 170-year old multi-sided platform that brings together its owners, shoppers and retailers. Multi-sided platforms can grow rapidly if one side is offered participation free of charge. An early example of this model, where consumers were attracted by a free service, are free newspapers, first seen in Lübeck, Germany in 1885. In this model the parties that pay for the service, the advertisers, are not the parties that use the service, the readers. Free television is the best example of the advertiser-driven approach from the 20th century.
The Internet has allowed many more multi-sided markets to come into existence and, in particular, has allowed the development of many more such markets where a service or product is offered for free. Inherent in all successful platforms are network effects, that is when one side of the platform increases in size or when its members use it more the platform’s value to all other sides increases. Online matchmaking services are the obvious example. As the number of one gender on the site increases the more valuable it is to users of the opposite sex who are looking for spouses.
Economists call the benefits that participants get from a multi-sided platform positive externalities. In offering Android under an open source license Google is seeking to establish a multi-sided platform where the positive externalities generated by offering the product for free will expand the platform thus generating further positive externalities from its wider use. Beneficiaries of these positive externalities include: Google, which receives advertising revenue; manufacturers, whose smartphone development and go-to-market costs are lowered; app developers and companies providing services through smartphones, whose potential customer bases expand; mobile carriers, who upsell consumers to more profitable smartphone contracts; and consumers, who pay less for smartphones.
Sometimes the behaviour of one or more participants in a multi-sided platform can produce bad effects, or negative externalities. For example Atari fell into a trap when, adopting an open software standard in the 1980s, it allowed anyone to write games for its console. Its subsequent inability to impose quality control on the numerous developers who wrote games combined with the fact that users were unable to verify their quality before buying them led Atari to suffer from a lemons problem—that is consumers were unwilling to risk paying for a low quality game. As a result the company went into decline. Another example is the social networking site, MySpace. MySpace’s owners used it as a vehicle for monetisation opportunities, bolting sponsored initiatives, like MySpace Books, and games that linked to surveys and promotions for credit cards, onto the site, thus interfering with and degrading user experience. Further, as MySpace founder, Tom Anderson, has acknowledged, the company's failure to remove sexually explicit material posted by users turned the social network into a "cesspool". Both Atari and MySpace are examples of multi sided-platforms that suffered because their owners did not take steps to prevent negative externalities.
Some of Android’s negative externalities are similar to MySpace’s and Atari’s, particularly Atari’s lemons problem. Because Google relies on manufacturers and carriers to install updates many Android phones are infrequently, or never, updated. Failure to install updates can prevent the phone from working as it should or make it vulnerable to hackers. In addition carriers and manufacturers have frequently freighted their Android phones with unwanted bloatware, software that is difficult to remove, drains battery life and which they are, in some cases, paid to install. The cumulative impact of these negative externalities, caused by the behaviour of actors on two sides of the platform, are a real threat to Android’s future viability and may explain why more users who can afford to, switch from Android to Apple than the other way around. Without governance or rules to protect the integrity of the operating system manufacturers and mobile carriers will inevitably be incentivised to behave in ways that degrade user experience or will be insufficiently incentivised to do what is necessary to protect or enhance it.
Finally all app developers, including Google itself, are vulnerable to Android’s negative externalities. These negative externalities include the missed opportunities to sell to a much wider user base that is the inevitable consequence of fragmentation and forking. The marginal impact of these externalities is, if anything, greater for Google’s less well-capitalised competitors than for Google itself. Google’s response to the problem of fragmentation and forking helps both it and its competitors. First, Google understandably views users of GMS as its customers, and in many cases they will be Google’s paying customers. The company wishes to defend its own brand by setting conditions for manufacturers who wish to install its apps on Android smartphones. The second response is to address a root cause of Android’s sometimes poor user experience for all apps by encouraging greater compatibility across the platform. In trying to establish consistency for Android Google’s strategy is not only aligned with its own interests, it is aligned with its competitors’ interests to the point that it is a precondition to any real competition on the Android platform.
The need for rules and standards
Multi-sided platforms occur in most areas of organised human activity. Many government services and, even, government itself could be described as a multi-sided platform. These include local government services financed by property taxes, and even legal systems—all of which create positive externalities by providing transactional or resolution systems that bring different groups together. Besides shopping centres, commercial multi-sided platforms include fitness centres, that host self-employed personal trainers looking for clients, parcel delivery services, and taxi companies that use self-employed drivers. Given its inherent characteristic of connectivity the Internet has not only been very hospitable to these traditional platforms but, combined with digital innovation, has given rise to a host of new commercial multi-sided platforms that exist entirely within the digital sphere. They include: Microsoft Windows, Apple’s iPhone, eBay, Spotify, iTunes, Pandora and Netflix. All of these platforms have multiple sides. Some, like Microsoft Windows, are software platforms and others, like eBay, rely on software to enable users from different sides of the platform to transact. One characteristic that all of these platforms employ are rules and governance measures to maximise positive and minimise negative externalities. In the case of software platforms these rules will depend upon whether the software is proprietary or open source. Proprietary software platforms such as Microsoft Windows or Apple’s iOS rely on rules made by the owners, and are backed by the owner's’ property rights. Open source software platforms, on the other hand, rely on voluntary rules that require users’ compliance in order to receive, amongst other things, approved updated versions of the software.
The popular view that the open source software movement embodies a democratic alternative to reliance on privately-owned proprietary software is misleading. While it is certainly the case that open source practitioners’ use crowdsourcing to identify and fix problems with software code—exemplified in “Linus’ Law”: With many eyeballs all bugs are shallow—all successful open source software platforms, including Linux, Python, Perl, and Apache have developed and rigorously enforce rules and decision-making systems that preserve their integrity, thereby, ensuring that they remain useful.
The anti-hierarchical / anti-capitalist view may be behind some of the reaction to Google Android in Europe. Former European Commissioner for the Digital Agenda Neelie Kroes has advocated open source software—arguing, in effect, that it is a public good—and associated it with other benefits such as open standards and, even, open data. The only connection between open source software, open standards and open data is that they all share the same word. This simple Rylean category mistake may also be behind a view in Brussels that Android forks are evidence of commercial, creative and competitive diversity brought about by “open standards” when all the evidence is that without rules software fragmentation and forking can be fatal to an operating system and are seldom in their users’ interests.
Google’s grip on Android is not all that certain
Another factor to consider in the case against Google is the extent to which the company has market power within the Android platform. The most recent global figures for smartphones shipped show an increasing bifurcation between Android OS and Apple iOS—over 87 percent of smartphones shipped world-wide during Q2 2016 use Android OS or forked versions of Android while Apple iOS had just under 12 percent. Google bought Android in 2005 and released it under an open source license in 2007. This means that, to the manufacturers of Android smartphones, Android is, to all intents and purposes, their own property which they can tweak or fork from the parent code as they wish.
In 2014 20 percent of Android OS mobile installations were forked, a bigger market share than Apple iOS. To say that this leaves Google with about 70 percent of the mobile OS market would be misleading however. Because the manufacturers controlling this 70 percent also control their own installations of Android, Google has no say, even with these unforked versions, over when or whether updates are installed, what competing apps come pre installed or, in most cases, their position on the screen, a situation that bears no comparison to Apple for example. Google provides periodic updates to Android but manufacturers feel under no obligation to install them. This leads to multiple versions of even Google Android, a problem so serious that the Dutch Consumers Association has sued Samsung for failing to update 82 percent of its Android phones in the Netherlands.
GMS’s position in unforked versions of Android is not always optimal for Google. Many manufacturers wish to create their own apps or to promote apps developed by third-parties other than Google. If Google apps are present on these phones they may be pushed into the background and not immediately visible to the user.
The result is that few people who have used a basic Android phone, even one that is compatible with Google Android, will have had a Google experience comparable to what is possible on Apple’s iPhone, on expensive Samsung devices, or on the new Google Pixel phone.
Cyanogen, which has developed a forked version of Android, provides further evidence that Google’s grip on Android OS is not all that tight. Cyanogen was reported last year to have raised $70m to take on Google Android and in February of this year was also reported to be developing a new version of its Android OS that could serve as a gateway for Microsoft to re-enter the smartphone market, offering a full Microsoft experience. Other companies that have developed forked versions of Android include Amazon, OnePlus, Xiaomi, and Alibaba.
Another factor for Google to worry about is the possibility that particular apps, like Facebook, are already a very powerful presence on Android phones. The presence of Facebook or the expansion of powerful Asian apps, like Wechat, that combine multiple services, represent a threat to Google’s future revenue on the Android platform.
Google’s attempt to address the negative externalities of the Android platform
To tackle a root cause of Android’s performance problems, problems that can affect any app on Android, Google offers manufacturers an Anti-fragmentation Agreement (AFA). This requires manufacturers to agree to build only devices that fulfil a basic standard of compatibility and not to do anything that will fragment Android. Signing this agreement entitles the manufacturer to use the Android compatibility trademark which provides users with reassurance that their apps will work on any device that carries this trademark.
Google also asks manufacturers wishing to preinstall GMS to ensure that their device’s operating system will enable Android apps to work properly by agreeing to sign the AFA and agreeing to abide by an Android Compatibility Definition Document (ADD) a set of standards that ensure that all apps that come preinstalled on the phone will work properly. The agreement is cost free and is aimed not at restricting rivals to GMS but to ensuring that all apps that are preinstalled on the device work properly. Google’s terms for manufacturers bear comparison with Apple’s terms for accepting apps from developers which Apple summarises as “We will reject apps for any content or behavior that we believe is over the line. What line, you ask? Well, as a Supreme Court Justice once said, ‘I'll know it when I see it’”.
As a further step Google offers a Mobile Application Distribution Agreement (MADA) which determines where individual apps in the GMS appear on the screen and standardises other aspects of the user experience. The purpose of the MADA is to make sure that the phone works well out-of-the-box without the need for users to spend a lot of time configuring it. This approach is similar to that taken by Apple, and previously by Blackberry and Nokia, and to what Microsoft plans through its strategic partnership with Cyanogen.
Google’s actions should be seen in the light of the need for governance to be applied to multi-sided platforms in order to minimise negative externalities. Google’s agreements are well within the bounds of those offered to users of other platforms, namely that they will use the platform in ways that do not create negative externalities. Myspace and Atari failed to do this and as a result lost to competitors who avoided their mistakes. It should be noted that none of Google’s measures prevents manufacturers from preinstalling apps that compete with GMS and none of them prevents users from disabling GMS or changing the device’s default settings.
Has DG Competition identified the right market?
In order to establish that a company is abusing dominance in a market. It is necessary for competition regulators to:
Define the market in question.
Establish whether the company dominates this market.
Establish whether the company is abusing this dominance.
The European Commission has said that Google is dominant in “general Internet search services, licensable smart mobile operating systems and app stores for the Android mobile operating system”. But is this the relevant market? In its case against Microsoft—that the company had unfairly prevented rival Internet browsers from competing with the firm’s own browser, Explorer—DG Competition defined the market as the market for PC operating systems. However Microsoft’s market power wasn’t explained by its dominance in PC operating systems per se but by its proprietary PC operating system’s near total long-term dominance of the overall PC market. Similarly, a view of the wider smartphone market is necessary to make sense of Google’s strategy for Android. This reveals that Google’s: a) lack of comparable control over Android, and, in particular, inability to prevent Android’s fragmentation, b) relatively weak position vis-a-vis smartphone manufacturers who can and regularly do ignore requests to update the operating system, c) weak position relative to the smartphone carriers in Europe whose profits are vast in comparison to Google’s earnings from Android, and d) its position relative to Apple which is, by far, the most profitable smartphone supplier in Europe, places it in a very different situation to that of Microsoft in 2009.
Microsoft’s circumstances in 2009—when the European Commission found that the company had abused its dominance by tying its Internet browser, Internet Explorer, with Windows thereby placing rival browsers at an unfair disadvantage—are very different from those of Google today. In that year Microsoft’s dominance was entrenched and long-standing. By 2000 its market share in PC OS had already grown to over 90 percent. By 2009, the year of the final EC decision, Microsoft’s division responsible for Windows and PC software earned $14.7 billion in revenue. In contrast Google’s grip on Android is not secure and its share of the profits from the global smartphone industry are negligible compared to Apple, Samsung and the mobile carriers. In July 2015 the Wall Street Journal reported that Apple earned 92 percent of all operating income from smartphone sales based on a 20 percent market share. Samsung was reported as making 15 percent. The two companies accounted for more than 100 percent of industry profits because other manufacturers—almost exclusively of Android smartphones—broke even or made losses. In comparison Goldman Sachs has reported that, at most, Google earned $3 billion in revenue from Android search in 2014. Apple made almost eleven times this amount from sales of iPhones in the last quarter of 2015 alone. In 2016 the GSMA, the mobile carriers trade association, reported that the global revenue of mobile carriers exceeded $1 trillion—more than 300 times Google’s revenue from mobile search in 2015. In Europe, smartphone subscribers account for a sizeable slice of the mobile market with an estimated 78 percent of all Europeans having a mobile subscription and an estimated smartphone penetration amongst these of 76 percent.
The market that the Commission should be looking at is the smartphone market within which Google has very little market power compared with Apple, Samsung, the Android smartphone manufacturers, and the mobile carriers. By the end of 2015 65 percent of the world’s 3.2 billion smartphones were in the developed world. This year the smartphone market is set to reach 2.1 billion users. It is clear, therefore, that Google does not enjoy the kind of power within the smartphone market that Microsoft enjoyed within the PC market back in 2009. In fact the lack of real competition within Europe’s highly fragmented mobile phone market may be a far bigger threat to smartphone consumers in the EU.
Is the European smartphone market in a Nash Equilibrium?
The competitive dynamics of the European smartphone market suggest that it is an example of imperfect competition with some aspects of a Nash equilibrium. Developed by the mathematician John Nash the Nash equilibrium is part of Game Theory and describes how in a non-cooperative game with two or more players each player, having understood the strategies of all the other players, has nothing to gain by changing their strategy. The outcome can, however, be suboptimal if some of the players could gain more by cooperating. This, however, would require these players to agree to change strategy. A player who changed strategy would lose if other players did not keep their word to change their strategies.
The European smartphone market comprises Apple, Samsung, a host of other Android manufacturers, app developers, mobile carriers and Google, which both curates the open source operating system Google Android and earns revenue from search on smartphones running its applications. Profits from the market are taken mainly by Apple and the mobile carriers. Google’s position can be seen as trying to get some participants—manufacturers and carriers of Android devices—to change their strategies so that outcomes can be improved. The trouble is that the manufacturers’ and carriers’ incentive to do this is not great.
Sellers of Android phones at the mid-to-low price level are, despite what Google says about Android itself, in competition with each other to sell, as price takers, to their all-powerful distributors, the mobile carriers. It is not likely to occur to them to cooperate on a standardised operating system. Apple, with a direct relationship with millions of users, is in a better position to negotiate its terms with carriers and consequently captures a high degree of profits from its phones. What about the carriers? The EU has over 100 carriers in 28 separate markets who, for the most part, function within discrete national oligopolies, hence it may be more accurate to describe European smartphone markets rather than a single market. One result of this fragmentation is that Europe’s carriers of Android smartphones have no interest in aligning their strategies on a global or even European level, as Google would like, if they can earn sufficient profits in their home markets. It should not be forgotten that the European Commission took action against them to reduce, and eventually, eliminate the very high mobile phone roaming costs they have, for years, imposed on Europe’s mobile phone users. Further, carriers have in many cases paid large sums of money for operating licences and remain cash-cows for the governments of their domestic markets in terms of sales taxes and employment. The carriers, therefore, have no interest in Android that compares to Google’s interest in the operating system or, for that matter, to Apple’s interest in iOS. Nor have they any interest in competing to the point where it significantly impacts on their profits.
Google's hopes—that Android can be developed to offer a user experience comparable to the iPhone on many types of cheaper smartphone—are up against the strong incentives of the carriers to ignore such a possibility. Such a commoditisation of high-quality phones could reduce their profits as fewer customers would upgrade to more expensive Samsung and iPhones. Carriers may, in fact, have incentives to regard poor Android smartphone user experience as simply an opportunity to upsell dissatisfied users to more profitable contracts with Samsung or iPhone. It follows that they may also have little incentive to cooperate on OS standards to the point where cheap phones become competitive with Apple.
The fact that the carriers, upon which Android phones rely, have 100 percent of the smartphone market in Europe, from which they earn most of their $150 billion annual revenue—many multiples of Google’s global advertising revenue from Android—speaks volumes about where their incentives lie and about Google’s place in this market.
Is Google keeping rival apps off the Android platform?
Apart from the fact that there is no evidence that Google’s policies deter competition in apps, a heavily fragmented or forked operating system is, if anything, more likely to discourage competitors to GMS. Standardising an operating system does not, absent other factors, discourage competition. Added to this is the lack of evidence that Google dominates Android in any way that is comparable to Microsoft’s dominance of PC operating systems in 2009. Unlike PC users in 2009, or even now, smartphone customers typically change their devices more regularly than their PCs, allowing for more frequent choice about what operating system they use. There is no need for concern about switching costs between different operating systems. Apps seldom cost more than $5. The average cost of iOS apps in the U.S. this year is $1 and Android apps are known to be cheaper on average than iOS apps. Further, building a mobile app is cheaper than producing a PC programme. 73 percent of apps are developed for more than one OS and 62 percent support even more. HTML5 apps can run on all devices and don’t need to be created. Google’s strategy on Android could not be more different from that of Microsoft’s approach to Windows in 1990. In 1990 Microsoft was found to have tried to harm Sun Microsystems Java which ran on all platforms by introducing Java Virtual Machine because Java threatened Microsoft’s dominant position on the market for client PC operating systems. In contrast Google has never tried to hamper another app.
The question for competition regulators may well come down to whether they succumb to the notion that Google Android is a commons of affiliated users that Google is somehow trying to enclose to the detriment of other users, or whether it is a multi-sided platform running on open source software of which Google is the originator, de facto curator and, relative to the overall smartphone market, a minor beneficiary.
Android is a voluntary and, as yet, partly developed multi-sided platform that suffers significant negative externalities from the behaviour of some of its users. A significant part of these externalities is the opportunity cost for all participants from Android’s failure, so far, to deliver smart phones of the highest standard that are easily affordable for most of the world’s population. This failure likely stems from Android’s fragmentation and from the fragmentation of the mobile telephony market.
Google's role as the curator of Android is distinct from its role as a supplier of Apps for Android. If nobody curates Android then it will collapse into an increasing number of fragments and forks. The Commission is correct to examine Google’s dual role as both a curator of the platform and as one of its beneficiaries. Such a dual role gives rise to potential conflicts of interest. However Google's anti-forking and anti-fragmentation policies are, if anything, a weaker version of similar policies that have preserved the integrity and usefulness of other open source platforms. Google’s efforts to prevent forking on Android and to keep the operating system up-to-date will encourage the development of innovative apps and ensure that the platform becomes as reliable, predictable and stable as Apple iOS. If Google is not allowed to curate Android, however, the vast majority of the world’s smartphone users will be saddled with increasingly inferior products.
Views expressed in this article are those of the author and not those of the Global Digital Foundation which does not hold corporate views.