IEEE Technology and Society Magazine - June 2022 - 81

audience itself, for example, when viewers share
content on the individual, through their own outlets,
creating the so-called identity coperformance [20].
Such social privacy control is currently promoted
by some platform operators-they study what makes
people feel more secure to share and the results
inform privacy options and policies [21]. However,
those options that limit institutional privacy
breaches (e.g., required by laws such as the GDPR
in Europe) are usually far more difficult to access.
In effect, social privacy options do not increase the
agency of the individual in data sharing, as the control
or security she feels is spurious. In fact, the mere
existence of privacy policies sometimes creates a
false feeling of security, which serves to further the
interests of platform owners [7]. The links between
content for different audiences, and between the different
identities, are not removed-the companies
still have them.
This is also true when the individual makes the
effort to divide her online activities across various
profiles and accounts (sometimes against platform
policies). The companies operating
those
platforms still have the links. Interestingly, there is
a trend to promote such advanced social privacy
control. Currently, Instagram is promoting multiple
accounts [22], acknowledging the new trend of the
so-called " fake " accounts (e.g., fInsta) that are used
to self-present to different audiences [23]. Presumably,
the companies noticed that individuals tend
to restrict the content they post when it is viewed
by a public that might be perceived as too broad.
Self-restriction on the part of users limits the data that
platform operators get and it might have prompted
them to promote sharing content through different
accounts. In effect, these data collectors are taking
advantage of the psychological need to perform differently
to various audiences and of the individuals'
lack of awareness of the differences between social
and institutional privacy.
Finally, even when accounts are created on
different platforms (run by different companies),
the third-party cookie arms race is in action. Even
though regulators and browser creators try to limit
the placement of cookies, tech giants try to be one
step ahead [24]. Thus, activities can be gathered
across many digital spaces and linked to a single
user profile. Moreover, this can be done even for
users who do not have accounts on the platform at
all or who log out of their accounts when browsing
June 2022
other online spaces-the so-called " shadow profiles "
are created for these. Such profiles, while not
as rich in data, are still a good source for the prediction
of user behavior [25].
Once such an overreaching profile-or a massive
set of profiles-is created, alongside other uses, this
data can be used to ensure that individuals share
more or to limit their true privacy options. First,
the best narratives can be selected to push public
debate away from the crucial points. For example,
discussion of the need for, or lack of, censorship on
social media, and of specific instances thereof (e.g.,
censoring Holocaust denial) draws attention away
from the general problem of profiling through data
collection and linkage and toward more specific,
often minor issues [26]. Google's plan to switch
off the ability of other companies to place cookies
on users' computers through the Chrome browser,
which effectively centralizes all data linkages for the
interest of Google, is sold to the public as an increase
of privacy protection [27]. Availability of privacy
options, which are promoted as extending care for
privacy on behalf of users, at the same time pushes
the responsibility toward individuals (they need to
protect) and away from companies (who should
not collect in the first place). Such narrative construction
and control are relatively easy for platform
operators who have knowledge of users' problems
and concerns as well as vast resources for lobbying
policy-makers.
Another tactic, also made possible through access
to a huge population of users, may be employed.
Dark patterns-interface and service design that
prompts suboptimal choices [28]-are possible only
because they can be tested on and adjusted toward
many different profiles of users. The richer the data,
the more targeted such practices can be. Alongside
other abuses, dark patterns can be used to nudge
for more data sharing. An example of this, brought
up earlier, is the interfaces and privacy options
that make users feel more secure and share more.
Another way is to construct those privacy options
that are truly important for the companies (i.e., institutional
privacy protection) in such a way that users
never access them. For example, with almost a quarter
of humanity on board, it is easy for Facebook to
test what types of options individuals-in all their
variety of cultural and psychological specifics-will
never reach. Not surprisingly, arguably the most difficult
to access privacy option on Facebook is the
81
IEEE Technology and Society Magazine - June 2022

Table of Contents for the Digital Edition of IEEE Technology and Society Magazine - June 2022
