First presented at Future of Digital Identity, British Library, 7 Jan 2010. Updated for presentation at Privacy and the Law Conference, London, 1 Dec 2010, and for an OxPILS seminar at Balliol College, 23 May 2012.
2. Data: Pew Research Center, Internet and American Life Surveys
Graphic: Jenise Uehara Henrikson, Search Engine Journal, 30/8/11
3. Which data are social networks
storing?
• In 2011, Austrian law student
Max Schrems received 1,222
pages of data after subject
access request to Facebook
• Searches, pages viewed,
relationships with friends,
events, like button tracking,
cookies, conversation tracking…
• Facebook now has data
download tool, supplies 39 of 84
data categories held TouchGraph Facebook Browser
4. Data: CareerBuilder, online, fieldwork 17/11/09-2/12/09, probability
sample of 407 UK private sector employers, sample error ±4.41%
Graphic: Jenise Uehara Henrikson, Search Engine Journal, 30/8/11
5.
6. Reasonable expectations?
• Oxford students fined on basis of Facebook photos of exam
celebrations. Whose “fault”?
• Students who didn’t take appropriate security measures
using available tools?
• Oxford proctors for snooping on a “private place”?
• Facebook because it did not provide the right defaults for
a “reasonable expectation of privacy”?
• A29WP: “SNS should ensure privacy-friendly and free of
charge default settings are in place restricting access to self-
selected contacts”
• Canadian Privacy Commissioner: “Facebook’s default
settings in respect of photo albums and search engines do
not meet users’ reasonable expectations”
7. Data shared about third parties
• How far can users control what
is “tagged” with their identifier?
• Facial recognition
• A29WP: “Users should be
advised by SNS that pictures or
information about other
individuals, should only be
uploaded with the individual’s
TouchGraph Facebook Browser
consent.”
8. Facebook Platform
• Over 9m apps and integrated websites as of March 2012
• “Social Plugins” are revealing browsing behaviour across the Web to
Facebook and Twitter – used by latter for profiling (10 days of
records). Banned in Schleswig-Holstein
• X’s app consent may reveal personal data about Y, and transmit
user IDs to ad tracking companies
• Canadian Privacy Commissioner: “Facebook should be doing much
more to ensure that meaningful consent is duly obtained from users
when developers access their personal information [and]
technological safeguards that will not simply forbid, but effectively
prevent, developers’ unauthorized access to personal information
that they do not need.”
9. Young people and privacy
• Most young people see Internet as private space for talking to
(already-known) friends, and target information to peer group
• Lenhart et al. (2007) found stricter access controls on photos/videos
by teens than adults (76% v 58% most of time/sometimes)
• Teens showed higher privacy concerns with parental monitoring;
parental discussions increased privacy concerns and reduced
disclosure
• Human impulse to connect and share information with friends, but
when mediated can easily be replicated and spread to places never
intended. Teens less good at managing collapsed contexts (boyd,
Marwick)
• Adult users of social media are developing similar behaviours –
consequence of mediation, not age (Marwick et al. 2010)
10. Young adults and privacy
• Hoofnagle et al. (2010) found very limited understanding
of privacy laws among young adults – 42% answered all 5
questions incorrectly
• Jones et al. surveyed 7,421 students at 40 US colleges.
75% concerned about passwords, SSNs, credit card
numbers but few about SNSes due to insignificant
consequences (2009)
11. Student information disclosure
What kind of personal information do you post online? (first year N=177, final year N=133) Oostveen
(2010)
12. Sampling Facebook Experiences
Facebook Application
Location disclosed to friends
Server
Data are collected
in our server
Questions are
sent to participants
Aims
To understand:
through SMS
Why do students share
their location
How (text, picture)
When, to whom they
Mobile phones carried by students share this location
Location retrieved with embedded GPS At what locations are
Subjects answer questions (e.g., sharing choices) they more willing to share
13. Location sharing
• 40 participants responded to over 2000 questions over 2 weeks
• Participants are more willing to share their location when they are in ‘Leisure’ or
‘Academic’ locations than in the ‘Library’ or in ‘Residential’ areas.
Abdesslem, Parris & Henderson (2010)
14. Privacy is contextual
• “Contrary to the assumption … that people have
stable, coherent, preferences with respect to
privacy, we find that concern about privacy … is
highly sensitive to contextual factors”
• Privacy salience primes concerns
• “People, it seems, feel more comfortable providing
personal information on unprofessional sites that are
arguably particularly likely to misuse it.”
• “Covert inquiries … do not trigger concerns about
privacy, and hence promote disclosure.”
John, Acquisti and Loewenstein (2011)
15. Homo economicus vs. sapiens
• Bounded rationality
• Privacy risks are highly probabilistic, cumulative, and
difficult to calculate
• Most individuals bad at deferred gratification, and have
time-inconsistent preferences
Acquisti (2009)
16. How to further privacy in social
networks?
• Is the consent given when signing up for Facebook (and apps) good
enough? Informed? “Explicit” for sensitive data? EC: “general
principle of transparent processing”, “improving the modalities for the
actual exercise of the rights of access, rectification”, “clarifying and
strengthening the rules on consent”
• Should current consent expose users to future risks? “The eternal
memory of Google” vs. the “right to be forgotten”; “data portability”
• Can T & C which exclude liability for privacy and security breaches be
potentially void as unfair consumer terms? EC: “general personal
data breach notification”, “extending the power to bring an action
before the national courts”, “strengthening the existing provisions on
sanctions”
• EC: “further promoting the use of PETs and the possibilities for the
concrete implementation of the concept of ‘Privacy by Design’”,
“continue to promote the development of high legal and technical
standards of data protection in third countries and at international
level”
17. Individuals ≠ data controllers
• How sustainable is Lindqvist?
• A29WP: “when access to a profile is
provided to all members within the SNS
or the data is indexable by search
engines, access goes beyond the
personal or household sphere.”
• Better privacy protection by
infomediaries?
• Defaults/Nudges?
• Expedited temporary restrictions on sharing?
20. Is competition regulation required?
• OpenSocial and related efforts may reduce switching
costs,
• but network effects will still act as a barrier to entry
• vertical integration will limit consumer choice
Competition authorities could:
• impose ex ante interoperability requirements
• upon dominant social utilities
• between vertically integrated value chains
• to minimise network barriers
21. Three models
Model 1: Must-carry obligations
• on broadcasters and Electronic Programme Guides
Model 2: API disclosure requirements
• on Microsoft from DoJ and EC rulings
Model 3: Interconnect requirements
• on telcos, especially with SMP
• for IM clients, from Time Warner/AOL merger case
22. Model comparison
• API disclosure requirements are necessary but not
sufficient - ability to program platform apps is of little use if
they cannot run
• Must-carry obligations enable one platform to “break in” to
another (eg Flickr app on Facebook)
• Interconnect requirements most likely to lead to seamless
user experience that will create real competition