Comparing the Android and Apple Operating Systems as Structuring Tactics in Cloud Surveillance For presentation at the International workshop on Cybersurveillance in everyday life Toronto, May 2011
David J. Phillips, Michael Murphy, and Karen Pollock
INTRODUCTION
This paper asks whether there is a possibility of an open, accessible infrastructure of
surveillance, and more particularly, how closely Google approximates that ideal.
We should be clear at the beginning, especially in this group, what we mean by
surveillance. Surveillance, for the purposes of this paper, is considered as a particular
technique of knowledge production. Surveillant knowledge production requires several
stages of activity. Members of a population are individuated and uniquely identified.
Certain activities or actions are singled out for attention, each individual is monitored and
tracked, and each performance of those activities is recorded. The accumulated data is
subjected to statistical knowledge at the level of the population – groups, types, norms,
and patterns are identified, created, or discovered. That knowledge is the applied back to
individuals and the population as a whole. That is, individuals are treated in accordance
with their relation to those discovered patterns. So surveillance makes populations as well
Surveillant knowledge production is an exercise of meaning making - of semiotic
power. It is in itself neither good nor bad. How surveillance is useful, and to whom,
depends on the context of use. For example, we can see CureTogether as an example of a
surveillance practice in which individuals can make sense of themselves in relation to a
population in a way which preserves and enhances autonomy. Users of Cure Together,
describe their personal experience of the effectiveness of various treatments of specific
diseases. These are displayed on a graph with two axes: number of individuals taking that
treatment and average effectiveness of treatment. It allows users to make sense of
themselves in relation to the population. It is an exercise in semiotic democracy.
“Android, iOS, and Cloud Surveillance”
Cybersurveillance in Everyday Life; Toronto 2011
http://curetogether.com/acid-reflux/ig/treatment-effectiveness-vs-popularity
(On the live site, mousing over each dot reveals the name of the treatment: endoscopic
dilation, reduced lactose diet, AcipHex, etc)
CureTogether, though, is a very simple app, more or less single purpose. We wonder
whether at a larger scale there such a thing as an open, available, common surveillance
infrastructure. What would that look like? In our earlier work, we have suggested that “Android, iOS, and Cloud Surveillance”
Cybersurveillance in Everyday Life; Toronto 2011
such an infrastructure would permit common access to grids, maps and measures, and to
analytic and interpretive tools. It would also permit choices regarding the domain from
which populations are constructed, and uses to which the knowledge is put.
We look at Google to compare their surveillance practice against these ideals. We
choose Google for several reasons. First, they are an all but ubiquitous surveillance
operation. Surveillance is their business. Second, their motto of “don’t be evil” invites
ethical analysis. Third, in many ways aspects of their platform seem to instantiate the
ideals of democratically distributed semiotic power. Google Earth and Google maps,
especially, seem to invite democratic, distributed, multi-purpose meaning making.
We will use the Android operating system as a point of entry into an analysis of
Google’s long term strategies of surveillance. Smartphones are an integral part of cloud
computing, a burgeoning infrastructure of personal interaction and data exchange. The
operating systems that control the handsets are an extremely important structuring
element in that infrastructure. In seeing how Android mediates a certain kind of
surveillance, we can gain some insight into Google’s interest in particular structures of
surveillance. As a comparative foil, we will also refer to Android’s major competitor in
the smartphone OS market – Apple’s iOS.
In our analysis, we will draw on insights from surveillance studies, telecomm and
information policy (especially with respect to open access, common carriage, and public
utility), media studies (particularly the role of audiences in the political economy of mass
media) and technology studies (particularly studies of infrastructure).
“Android, iOS, and Cloud Surveillance”
Cybersurveillance in Everyday Life; Toronto 2011
• How is surveillance implicated in Google’s strategy for profit? How does
• How does Android OS mediate surveillance practice?
• What tactics, strategies, etc. support and maintain that configuration of
• In what ways does/can it support common access to surveillance power?
HOW IS SURVEILLANCE IMPLICATED IN GOOGLE’S STRATEGY FOR PROFIT? HOW DOES SURVEILLANCE PRODUCE VALUE FOR GOOGLE?
This section introduces the economics of surveillance as a technique of profit making
and commodity production. It asks what google produces, how and from what materials
that product is assembled, and how the product is distributed and marketed.
As is clear from their annual and quarterly reports, virtually all of Google’s revenue
(of 6.6 billion in the second quarter of 2010) is from advertising. Google produces
The raw material for the production of the audience commodity are the monitored
and recorded traces of individuals interactions on the web. Google interprets those
interactions as intent, and applies complex and proprietary algorithms to make inferences
from them regarding taste, desire and habitus. The individual producers of those
interactions are grouped and typified, and those groups are ranked, valued and
commoditized as audiences. Access to those audiences is sold to advertisers.
This is basically the same economic model of the audience commodity that has
underlain mass media economics since the penny press. Of course, while the general
“Android, iOS, and Cloud Surveillance”
Cybersurveillance in Everyday Life; Toronto 2011
economic model is old hat, the particulars of its instantiation are in some cases novel, and
The first detail which is a significant departure from the mass media model is that
audience attraction and recruitment occurs through apps and services, rather than through
entertainment. Far and away the most important attractor of the interactions from which
audience is produced is Google Search, though all of Google’s services – Maps, Earth,
gmail, docs, YouTube – have the primary economic utility of amassing records of
interactions, the raw material of audience production. This focus on interaction, rather
than attention, as the primary raw material for audience production is not unique to
Google. At least since the direct marketing industry, everyday transactions, traced
through credit card records, loyalty programs, etc, have been the fodder for the
production of economically useful demographic groups. And more traditional mass media
audience producers, like the Neilson company, have been seeking ways to individuate
mass media viewers, trace their interactions with media content (through eye-motion
meters, for example), and link that with other sources of data regarding the individual’s
market behavior. However, Google has been able to gain a better quality raw material in
several ways. First, because many apps require a gmail account, Google has been able to
individuate the population more effectively, thus permitting finer analysis. Second,
through their Google Analytics service, they have been able to extend the scope of their
tracking capabilities beyond their own apps. Third, their Search service produces very
precise data about individual desires and the cognitive links individuals make between
one linguistic element and another. This facilitates inferences both about those desires
and the ways in which individual thought and behavior can be influenced. These
“Android, iOS, and Cloud Surveillance”
Cybersurveillance in Everyday Life; Toronto 2011
inferences are extremely valuable to marketers, whose stock in trade is the creation and
manipulation of desire. The strategic importance of this continuous monitoring and sense-
making is evident as Google informs its employees that they “stake their bonus” on
“integrat[ing] relationships, sharing and identity across our products.”
Google has developed and patented several techniques for creating relevance ranks.
These combine data from many sources. Google uses datacrawlers to garner textual
hyperlinks among websites and combines these with data of individuals’ interactions in
order to create distinct measures of relevance between users and content. This permits the
production of an audience more closely tailored to the advertisers’ needs. These relevance
algorithms determine which audiences are delivered to which advertiser, (or, conversely,
which ads are delivered to which audiences).
Google employs two mechanisms for audience delivery. The AdWords service
delivers audiences through Google’s Search results page. Advertisers contract Google to
have their ads appear next to Search results whenever certain keywords. Which ad
appears to whom, and which audience is delivered to whom, is a function of the amount
the advertiser has paid, the “keywords” the advertiser has specified, and the complex
relevance ranking assigned to the individual searching and those keywords.
With the AdSense service, advertisers contract with Google to display ads, not on
Google’s network, but on “relevant” web sites with which Google is affiliated. Google
contracts with web site owners to display ads. If the ad works (that is, if it prompts an
interaction with the targeted viewer), the advertiser pays Google, and Google pays the
hosting web site owner. Which ads appear on which sites, (or, again, which audience is
delivered where) is again determined by the relevance ranking algorithms.
“Android, iOS, and Cloud Surveillance”
Cybersurveillance in Everyday Life; Toronto 2011
This relation between advertisers, audiences, Google’s web sites, and Google’s
affiliate web sites is reminiscent of network broadcast economics. There, networks derive
two income streams. They charge affiliates for programming, which the affiliates need to
attract audiences. They also charge advertisers for audience delivery (that is, to show ads
to assembled audiences). Networks pay producers for programming, and they pay
affiliates to carry sponsored ads. The difference with Google is that they extract payment
from affiliates and programmers, not in cash, but in the raw data of audience production.
They monitor individual interactions with affiliates through the Google Analytics service.
To recap: Google’s model of value and profit production is in some ways typical of
all mass media. They produce audiences to sell to advertisers. However, the audience
they produce is more fine-grained, and the social meanings they are able to create around
those audiences is more closely linked with intent and desire. They deliver those
audiences to advertisers using a network/affiliate model reminiscent of commercial
HOW DOES ANDROID OS MEDIATE SURVEILLANCE PRACTICE?
The Android operating system can be understood as a tactic in Google’s efforts to
protect and extend its advertising income. Android is designed to push interactions onto
the web, especially to web services operated by Google. This can be seen both in the
underlying operational paradigm of the OS, as well as in the apps that come pre-installed
Android’s operational paradigm is intended to position Google within the emerging
infrastructure of cloud computing. At its most basic level, cloud computing refers to any
“Android, iOS, and Cloud Surveillance”
Cybersurveillance in Everyday Life; Toronto 2011
network system in which individual computing terminals are able to remotely access
shared data which is stored in large datacentres. As cloud computing becomes the
dominant paradigm for personal computer use, terminals become dumber (often doubling
as a mobile phone handset), data lines become faster and more capacious, and data
storage and processing occurs at ever more vast datacenters. Institutionally, the terminals
are produced by “smart phones” manufacturers, the data lines are operated by mobile
phone networks, and the processing facilities are operated by behemoths like Google,
In general, Android is designed to support Google’s position as a “smart” mediator
among highly configurable apps and highly configurable terminals. In illustration,
consider these capabilities of current and future iterations of Android, as touted at the
public celebration of the launch of Android 2.2. Android can or will be able to cause an
app to be opened on a remote handset. That is, one user could cause another’s handset to
open a navigation program to display directions. It will enable one user to access and
download media from another’s handset. It will include a personalization service
mediating apps and devices and allowing those apps to configured on the fly to run on the
devices. Such a mediating position again strengthens Google’s ability both to monitor
interactions and to mediate between advertisers and audiences.
As another strategy of reinforcing Google’s core advertising revenue, Android comes
with preinstalled apps that push users to Google’s data operations. For example, the
Verizon Android Incredible comes preinstalled with Google Maps, Latitude, Google Talk,
Gmail, YouTube, Google Calendar, and Google Search. All are owned by Google and are
part of its interaction harvesting operations.
“Android, iOS, and Cloud Surveillance”
Cybersurveillance in Everyday Life; Toronto 2011
Android is also closely linked to Google through users’ gmail accounts, which are
essential in order to use an Android handset to its full capacity. When Google offered it’s
own Android handset for sale (the Nexus One) it was impossible to order it without a
gmail account. Google operates the Android Market, the largest clearing house for
Android Apps. One cannot even browse the Market without a gmail account. Android’s
default contacts and calendar apps are referenced by gmail accounts. Presumably, inter-
device operations and personalized app configurations are intended to be coordinated by
gmail accounts. Again, this individuation of members of the population permits Google
to produce more nuanced, refined and valuable audiences. WHAT TACTICS, STRATEGIES, ETC. SUPPORT AND MAINTAIN THAT CONFIGURATION OF SURVEILLANCE POWER?
In order to maintain its position as the web’s dominant purveyor of audiences,
Google has had to protect the conditions that allow it to produce and distribute those
audiences. That is, it must protect its abilities to monitor online interactions, to make
sense of those interactions, and deliver access to the generators of those interactions. As
cloud computing, mediated through smartphones, has emerged, those conditions have
Apple’s model for profit production are very different from Google’s. Apple is
promarily a retailer of terminals, such as the iPhone, the iPod, and the iPad. According to
Apple’s revenue report for the quarter ending 30 June 2010, almost half of its 15.6 billion
in revenue came from sales of iPhones and iPads. However, it also has a significant
interest as a distributer of cultural content, through the App Store and the iTunes Store.
While the 1 billion it received in music sales may seem paltry by comparison to it’s
“Android, iOS, and Cloud Surveillance”
Cybersurveillance in Everyday Life; Toronto 2011
harware sales, the profit margin on music sales far exceeds that on its hardware sales. In
fact, since 2006, Apple has been the largest music retailer in the U.S. It currently
surpasses the combined sales of both of its nearest competitors, Walmart and Amazon.
Retail distribution of popular culture products is a significant source of income for Apple.
Apple’s strategy has been to create a symbiotic relationship between these two
interests by reinforcing its brand, and creating a “walled garden” such that the coolest
content can only be had on the coolest terminals. iOS, the operating system for their
terminals, facilitates this strategy. Whereas Android facilitates synching and sharing
through web services, iOS synchs and shares through iTunes – not a web service, but a
program running on a PC. Thus iOS, by design, avoids web interactions. Nor does iOS
support much peer-to-peer networking. Until recently, it did not support multi-tasking.
Only one program at a time could run on the iPhone. This meant, for example, that it was
impossible to design an iPhone app that would, in the background, exchange data with
other iPhones while still doing something useful in the foreground.
iOS will only install apps that are downloaded through Apple’s AppStore, and Apple
stringently vets the apps available there to ensure that they conform both to Apple’s
family-friendly brand and to technical restrictions designed to foreclose third-party
platforms that extend or modify iOS in ways that Apple feels are detrimental to its
longterm goals. iOS will only run on Apple appliances, like the iPhone, and those
appliances can be “locked” to run only on specific telecomm networks.
This strategy, in which consumer cloud computing is basically an Apple-branded
experience, and in which all interactions are mediated by Apple, is in direct opposition to
“Android, iOS, and Cloud Surveillance”
Cybersurveillance in Everyday Life; Toronto 2011
Google’s interests. It has also been wildly successful. It is in response to Apple’s success,
and as an alternative to iOS, that Android was developed.
However, simply developing a compatible alternative to iOS was not sufficient to
ensure that alternative’s success. Infrastructural re-alignments were also necessary. To
support app development, Google established the Android Market, a centralized point of
distribution that facilitated the kind of sales volume necessary for developers to make a
profit. It also alleviated the difficulties of app downloading and installation.
To weaken the bond between Apple and network carriers, Google entered the U.S.
spectrum auction to ensure that the license stipulated that the licensee keep the spectrum
open to any handheld communications device and enable any software application,
service or content to be downloaded and utilized. To attract network carriers, many of
whom were also threatened by Apple’s ability to dictate the terms of engagement with the
cloud, Google makes Android available in open-source and free of licensing fees. This
makes it very attractive to handset manufacturers and to network operators who can
reconfigure Android, adding features to make it unique to their brand. Carriers faced
barriers to smartphone adoption because smartphones can be bandwidth intensive to an
extent that taxes the carrier’s network capacity. Google addressed this by diverting their
commission from app sales through the Android Market to the carriers.
Apple is not the only source of threat to Google’s profit engine. Facebook looms as
the portal of choice – the attractor of the interactions from which audience is produced. In
response, Google has entered into alliances with industrial producers and distributers of
cultural content. For example, Google Earth integrates content from the BBC and
“Android, iOS, and Cloud Surveillance”
Cybersurveillance in Everyday Life; Toronto 2011
National Geographic (Lee 2010 922). YouTube partners with Universal, CBS, BBC, and
Sony to distribute mass market music and videos.
To recap: While Google pursues its strategy of offering services to individuals in
implicit exchange for traces oftheir interactions, they have also had to enter into retail
markets for apps and handsets and to forge corporate alliances with manufacturers and
network carriers in order to support the viability of Android as an effective alternative to
Apple’s iOS. In addition, they have allied with huge cultural product distributers to
bolster their ability to attract interactions. IN WHAT WAYS DOES/CAN IT SUPPORT COMMON ACCESS TO SURVEILLANCE POWER?
Now that we have an idea of how Google makes money, and the technical and
corporate strategies it has pursued to protect that profit engine, we return to questions of
how these structure the possibilities of semiotic democracy.
It might be argued that, at its best, the Google paradigm offers at least the possibility
of a tremendously sophisticated knowledge production engine, noting patterns, pressures,
and trends in a highly interactive, disperse and diverse technical structure, supporting
interplay among lots of populations, devises, data, and applications. Ideally, and as
reflected in Google’s rhetoric, such a knowledge production engine might be generally
accessible and adaptable to the needs of a wide variety of individuals and organizations.
But such a view neglects to account for the political and cultural economy in which
Google operates. The basic tension is this: Google has an interest in supporting all kinds
of interactions, to monitor and generate new kinds of knowledge, but only insofar as
those interactions can used to produce a commoditized audience.
“Android, iOS, and Cloud Surveillance”
Cybersurveillance in Everyday Life; Toronto 2011
In Manufacturing Consent, Herman and Chomsky present the propaganda model,
which predicts a narrowing of the realm of political and cultural discourse when that
discourse is mediated by profit-seeking, advertiser-driven mass media. This final section
will review some of the structural elements which the propaganda model identifies as
hallmarks of global mass media, and the effects this structure has on political and cultural
discourse. We then suggest how this model might be applied to Google.
Reliance on advertising
The model suggests that corporations that rely for their profit on producing and
selling audiences will produce the kind of audience that advertisers want. This kind of
audience has several hallmarks. They are produced as cheaply while still satisfying the
needs of advertisers, and they are composed of constituents who are willing to buy, who
This suggests that customization – of apps, of handsets, of knowledge – will occur
only insofar as it is profitable. Google’s clients – advertisers – are interested in genres,
brands, and predictable populations, not in bespoke applications or entertainments. They
wish to capture not the attention of lots of individuals, but the zeitgeist. Advertisers and
the culture industry will do their best to ensure that the cloud is instrumental in producing
distributed, branded pop culture. Individuals will be addressed as interactive consumers,
communities as fans and audiences. Google will facilitate this to serve their clients. Ownership by large capitalist organizations
Google itself is one of the world’s largest corporations. Network economics –
economies of scale, network externalities – suggest that Google will continue its trend
toward monopoly. In itself, this positions it as an attractive site for censorship and for
“Android, iOS, and Cloud Surveillance”
Cybersurveillance in Everyday Life; Toronto 2011
antagonistic targeting of individuals or groups by states or corporations (for example by
monitoring dissidents or by tracking copyright infringements).
Google will also protect the core processes and resources that enable them to
produce audiences for profit. They have been very savvy about branding of their look.
They also patent aggressively their techniques of attracting and monitoring interactions,
and making sense of those interactions, and delivering audiences. When patents or
techniques are held elsewhere, they aggressively buy them.
Strategic alignments with other large industrial capitalist orgs
As mentioned in an earlier section, Google especially in the last few years has made
extensive alliances with cultural producers, network carriers, and equipment
manufacturers. Indeed, Google has shown a certain genius for integrating the interests of
various industrial segments in such a way to be beneficial to each, as well as to Google.
The propaganda model suggests that the interlinked interests of many facets of global
capital will promote a kind of détente in culture wars. The interests of each industrial
segment will be tempered in favor of a general support for the interests of the global
This is evident at a structural level in the ways that Google has in fact forced Apple
to retreat in its strategy of controlling and branding all access to the cloud. Apple’s clients
– especially cultural producers who use smartphones as distribution channels – won’t
stand for losing Google (through whom they access audiences) or general accessibility to
cloud. Apple has been forced to loosen its grip and recognize the interests of other
“Android, iOS, and Cloud Surveillance”
Cybersurveillance in Everyday Life; Toronto 2011
Control of editorial policy by owners
The propaganda model suggests that when push comes to shove, owners retain the
right to control editorial content. While that control is usually delegated to professionals
to maintain an aura of objectivity, owners will step in in a crisis. Google explicitly
retains the right to editorial control as far as possible, stating in its user agreement that
“…your use of the Google Brand Features [that is, Google Earth, Maps, Analytics, …]
will inure to the benefit of Google. You agree not to challenge or assist others to
challenge the Google Brand Features (except to the extent such restriction is prohibited
CONCLUSION
In conclusion, we restate our argument. Google is in some ways a typical mass media
corporation. It produces audience to sell to advertisers. Thus we may effectively use
analyses of the political economy of mass media to understand and predict the cultural
and political ramifications of this form of profit-making. That analysis suggests that
Google is also increasingly typical of mass media in the alliances, dependencies and
market relations it enjoys with other industrial segments, particularly content owners and
network distributers. Given this, we can expect surveillance in the cloud to operate much
as market surveillance does now, in a cyclic process of individual actions informing the
creation of structures which channel those actions to the benefit of global capital.
REFERENCES My apologies for posting, at this very late date, a version immaculate of reference. A more complete version will be supplied soon. This entirely my fault, not that of my co- authors. djp
“Android, iOS, and Cloud Surveillance”
Cybersurveillance in Everyday Life; Toronto 2011
Pain is defined by The International Association for the Study of Pain as an unpleasant sensory and emotional experience associated with actual or potential tissue damage, or described in terms of such damage. In the elderly many complex issues may contribute to the emotional experience of pain including: a long history of various pain experiences, multiple chronic comorbid conditions, proximity t
Patent expiries: Which drugs have price reductions? Adapted from Drug patents: which will expire in 2012 and 2013? UK Medicines Informationharmacists for NHS healthcare professionals The patents of 25 drugs currently marketed in the UK are due to expire in 2012, with a further 21 in 2013. Some of these will be marketed as generics shortly after the patent expiry, although at present i