Big data analytics: from
threatening privacy to
challenging democracy
Paola Mavriki, PhD Candidate, pmavriki@aegean.gr
Maria Karyda, Associate Professor, mka@aegean.gr
Department of Information and Communication Systems Engineering,
University of the Aegean
Focus of presentation
 What are the privacy threats stemming from
the use of big data for political purposes?
 What is their impact on democracy?
The research work
 Identifies and investigates privacy threats stemming from the use of
big data for political communication,
 argues that mechanisms are required to protect the privacy interests
of groups as in the case of the release of demographically
identifiable information (DDI) for example.
 Focusing on the European landscape and the democratic value of
privacy the research work shows that main features of democracy
such as election fairness and pluralism of views may be undermined
through big data techniques.
Background: The professionalization of
political communication
 The online dimension of election campaigning is the latest
phase in the professionalization of political communication.
 The Internet is changing the way campaigns strategize and
communicate with their constituents.
 Persuadable voters are identified by profiling and targeted
through various personalisation methods.
 Political branding is gaining increased attention.
The European digital political
landscape
 Despite the differences between the European countries and
US, there are common trends such as the intensive use of
new ICTs and a marketing-oriented strategy of campaigns.
 There are several examples proving the integration of the
Internet and marketing tools in the political communication in
European democracies:
 in the 2007 Finnish general election for instance, political marketing
procedures have been used for targeting young voters,
 in the 2003, 2006, 2010 and 2012 Catalan elections technological
advances were used by the political parties,
 in the 2016 Brexit referendum behavioral science, big data analysis and
ad targeting have been used by the “ Vote Leave” campaign.
The democratic value of privacy
(1/2)
 In political democracies, ‘privacy provides opportunities for
political expression and criticism, political choice, and
freedom from unreasonable police interference; it provides
opportunities for people and organizations to prepare and
discuss matters in private…’ (Westin, 1967).
 Privacy is a fundament of our democracy, because without
privacy people might exercise power to influence others
(Janssen and van den Hoven, 2015).
 Privacy allows us to form our political, democratic selves
(Voice, 2016).
The democratic value of privacy
(2/2)
 Privacy is essential to the maintenance of democracy
because it ensures that citizens are able to hold elected
governments to account and place limits on the expansion
of the state. Privacy’s public value also stems from its
importance to the exercise of political rights (Regan, 1995).
 Individuals need to be free to share those thoughts with
others without being subject to the watchful, possibly
critical, eye of the state (Goold, 2010).
Privacy threats related to
surveillance (1/3)
 Contemporary ubiquitous systems gather vast amounts of
information about people accumulating a large amount of
personal information in new ways with new technologies.
 Public transport smart cards, smartphones, digital payment
systems, smart wearable devices for tracking biometrics,
Wi-Fi providers having access to geolocation and internet
browsing practices, Facebook tracking users’ likes and
preferences, Google tracking browsing and searches, credit
card use, car GPS systems, are some examples of the new
ways of monitoring people’s activities.
Privacy threats related to
surveillance (2/3)
 For the observed, surveillance can lead to self-
censorship and inhibition.
 Surveillance ‘threatens not only to chill the expression of
eccentric individuality, but also, gradually, to dampen the
force of our aspirations to it’ (Solove, 2003).
 Surveillance inhibits freedom of choice, impinging upon
self-determination. It also creates a trail of information
about a person.
Privacy threats related to surveillance
(3/3)
 Surveillance impedes one’s anonymity, inhibits one’s
freedom to associate with others, and alters one’s
freedom of movement.
 Surveillance’s inhibitory effects are especially potent
when people are engaging in political protest or dissent.
 People can face persecution, public sanction, and blacklisting for
their unpopular political beliefs.
 Surveillance can make associating with disfavored groups and
causes all the more difficult and precarious.
Privacy threats related to predictions
(1/3)
 Machine learning technologies become increasingly
pervasive in behavioral prediction.
 Campaigns need accurate predictions about the
preferences of voters, their expected behaviors, and their
responses to campaign outreach.
 Big data technologies allows practitioners, researchers,
policy analysts, and others to predict the spread of
opinions, political trends, etc.
 Campaigns may use data to construct predictive models to
make targeting communication more efficient .
Privacy threats related to predictions
(2/3)
 Based on existing publicly available information, big data
analysis tools can generate a predictive model of what has
a high probability of being PII, essentially imagining an
individual’s data.
 Kosinski’s (2013) method for example, is able to predict
political party affiliation about 85 percent using only
Facebook “likes”.
Privacy threats related to
predictions (3/3)
 Deducing extensive information from digital footprints and the kinds
of predictions that can be made from this are far too vast and
complex, and are evolving too quickly, for people to fully assess the
involved risks.
 A person may leave behind in several years thousands of pieces of
data which does not affect her/him negatively. As time passes, data
analytic techniques improve, and a newly created algorithm may
process the previously harmless digital footprints and deduce
harmful information.
Privacy threats related to automated
profiling (1/2)
 stem from the accumulation of personal data: fragments
of data regarding an individual user may be linked peace
by peace until an individual profile is entirely expose,
 aggregation:
 may cause dignitary harms because of how it unsettles
expectations.
 can lead to power asymmetries, as it can increase the power that
others have over individuals.
 moreover, data compilations may be both telling and
incomplete.
Privacy threats related to automated
profiling (2/2)
 In the case decisions are taken based on data mining
and profiling, undesirable discrimination may occur,
 In political context, a person may be profiled inaccurately
as an extremist movement or party adherent or
supporter.
Group Privacy threats related to
automated profiling (3/3)
 New technologies and powerful analytics collect and analyse large
amounts of data in order to identify patterns in the behaviour of
groups, communities and even entire countries.
 Some are concerned about the privacy of specific groups such as
non-governmental organisations (NGOs) or journalists. For example,
they argue that activists are not just individual targets, but, because
of their work, they have been targeted as a group.
 In the case of the release of demographically identifiable information
(DII) group privacy risks may occur.
Group Privacy
 Group privacy may be defined broadly as the interest a
group has in controlling the information privacy related to
the group.
 P. Mavriki, M. Karyda, Big Data analysis in political
communication: Implications for Group Privacy ,
International Journal of Electronic Governance, 2018.
Long-term implications for democracy:
obstructing the free flow of communication
 The “filter bubbles” which personalization tools induce,
obstruct the free flow of information, ideas, debates and
the communication between citizens and the power
holders of society.
 Consequently, the own opinion is reinforced, while the
ability to handle different points of view is reduced
facilitating in this way the polarization of society.
Long-term implications for democracy:
manipulation
 Content is selected for citizens on the basis of criteria unknown to
them and is calibrated not to their proximate selection decisions, but
to big data–generated assumptions about where those citizens
would want to focus their attention or where marketers need those
citizens’ attention to be focused.
 The “filter bubbles” and financial means of potent influencers
facilitates the spread of inaccurate, misleading or even wrong
information. They become increasingly personalized, manipulative,
and deceptive, spreading oversimplified messages or
misinformation.
Long-term implications for democracy:
fragmented political messages
 A party may highlight a different issue for each voter, so
each voter sees a different one-issue party. In this way,
microtargeting could lead to a biased perception
regarding the priorities of that party.
 Online political microtargeting could lead to a lack of
transparency about the party’s promises. Voters may not
even know a party’s views on many topics.
Long-term implications for democracy:
surveillance (1/2)
 Political rights are embodied in the unlimited validity of
the right to freedom of speech and opinion and freedom
of speech, expression, of association, etc.
 Privacy’s public value stems also from its importance to
the exercise of these political rights.
 Surveillance, may disrupt this structural dimension of a
democratic public sphere.
Long-term implications for
democracy: surveillance (2/2)
 One of the greatest harms occurring from mass
surveillance is the potential chilling effect on political
discourse, on the ability of both individuals and groups to
express their views and on the possibilities for
whistle‐blowing and democratic activism.
 New surveillance technologies can lead to “social
sorting,” where discrimination and privilege are
entrenched through the unplanned consequences of
data gathering and analysis.
Long-term implications for
democracy: power asymmetries
 The current ecosystem around big data creates a new kind of
digital divide: the big data rich and the big data poor.
 This environment favours incumbents who already are in the
possession of valuable data, also entrenched and moneyed
candidates within parties, as well as the data–rich among
existing parties.
 Data-driven campaigning might even be a form of
“cartelization” where large parties erect barriers to protect
their dominance from new entrants. The high expense of new
campaigning techniques is a significant disadvantage for
smaller and newer parties.
Overall conclusions
 Overall, parties and politicians try to discover, reach and
target influenceable potential supporters and voters.
 Consequently, the big data technologies they use, pose
new types of threats to voters’ privacy with multilevel
consequences such as the potential impact on
democracy.
Further research
 It seems that predictive analytics and automated decision-making
entail unique privacy risks and implications which need further
research.
 In addition, since big data analytics is blurring the boundaries
between individual and group data, there is a need for further
research related to the nature and the types of groups as entities.
 As personal data is the fuel of processes such as manipulation and
discrimination] there is a need for a broader and somewhat different
notion of data protection which might counter asymmetries brought
about by technology.
Thank you for your attention!

Big data analytics: from threatening privacy to challenging democracy

  • 1.
    Big data analytics:from threatening privacy to challenging democracy Paola Mavriki, PhD Candidate, pmavriki@aegean.gr Maria Karyda, Associate Professor, mka@aegean.gr Department of Information and Communication Systems Engineering, University of the Aegean
  • 2.
    Focus of presentation What are the privacy threats stemming from the use of big data for political purposes?  What is their impact on democracy?
  • 3.
    The research work Identifies and investigates privacy threats stemming from the use of big data for political communication,  argues that mechanisms are required to protect the privacy interests of groups as in the case of the release of demographically identifiable information (DDI) for example.  Focusing on the European landscape and the democratic value of privacy the research work shows that main features of democracy such as election fairness and pluralism of views may be undermined through big data techniques.
  • 4.
    Background: The professionalizationof political communication  The online dimension of election campaigning is the latest phase in the professionalization of political communication.  The Internet is changing the way campaigns strategize and communicate with their constituents.  Persuadable voters are identified by profiling and targeted through various personalisation methods.  Political branding is gaining increased attention.
  • 5.
    The European digitalpolitical landscape  Despite the differences between the European countries and US, there are common trends such as the intensive use of new ICTs and a marketing-oriented strategy of campaigns.  There are several examples proving the integration of the Internet and marketing tools in the political communication in European democracies:  in the 2007 Finnish general election for instance, political marketing procedures have been used for targeting young voters,  in the 2003, 2006, 2010 and 2012 Catalan elections technological advances were used by the political parties,  in the 2016 Brexit referendum behavioral science, big data analysis and ad targeting have been used by the “ Vote Leave” campaign.
  • 6.
    The democratic valueof privacy (1/2)  In political democracies, ‘privacy provides opportunities for political expression and criticism, political choice, and freedom from unreasonable police interference; it provides opportunities for people and organizations to prepare and discuss matters in private…’ (Westin, 1967).  Privacy is a fundament of our democracy, because without privacy people might exercise power to influence others (Janssen and van den Hoven, 2015).  Privacy allows us to form our political, democratic selves (Voice, 2016).
  • 7.
    The democratic valueof privacy (2/2)  Privacy is essential to the maintenance of democracy because it ensures that citizens are able to hold elected governments to account and place limits on the expansion of the state. Privacy’s public value also stems from its importance to the exercise of political rights (Regan, 1995).  Individuals need to be free to share those thoughts with others without being subject to the watchful, possibly critical, eye of the state (Goold, 2010).
  • 8.
    Privacy threats relatedto surveillance (1/3)  Contemporary ubiquitous systems gather vast amounts of information about people accumulating a large amount of personal information in new ways with new technologies.  Public transport smart cards, smartphones, digital payment systems, smart wearable devices for tracking biometrics, Wi-Fi providers having access to geolocation and internet browsing practices, Facebook tracking users’ likes and preferences, Google tracking browsing and searches, credit card use, car GPS systems, are some examples of the new ways of monitoring people’s activities.
  • 9.
    Privacy threats relatedto surveillance (2/3)  For the observed, surveillance can lead to self- censorship and inhibition.  Surveillance ‘threatens not only to chill the expression of eccentric individuality, but also, gradually, to dampen the force of our aspirations to it’ (Solove, 2003).  Surveillance inhibits freedom of choice, impinging upon self-determination. It also creates a trail of information about a person.
  • 10.
    Privacy threats relatedto surveillance (3/3)  Surveillance impedes one’s anonymity, inhibits one’s freedom to associate with others, and alters one’s freedom of movement.  Surveillance’s inhibitory effects are especially potent when people are engaging in political protest or dissent.  People can face persecution, public sanction, and blacklisting for their unpopular political beliefs.  Surveillance can make associating with disfavored groups and causes all the more difficult and precarious.
  • 11.
    Privacy threats relatedto predictions (1/3)  Machine learning technologies become increasingly pervasive in behavioral prediction.  Campaigns need accurate predictions about the preferences of voters, their expected behaviors, and their responses to campaign outreach.  Big data technologies allows practitioners, researchers, policy analysts, and others to predict the spread of opinions, political trends, etc.  Campaigns may use data to construct predictive models to make targeting communication more efficient .
  • 12.
    Privacy threats relatedto predictions (2/3)  Based on existing publicly available information, big data analysis tools can generate a predictive model of what has a high probability of being PII, essentially imagining an individual’s data.  Kosinski’s (2013) method for example, is able to predict political party affiliation about 85 percent using only Facebook “likes”.
  • 13.
    Privacy threats relatedto predictions (3/3)  Deducing extensive information from digital footprints and the kinds of predictions that can be made from this are far too vast and complex, and are evolving too quickly, for people to fully assess the involved risks.  A person may leave behind in several years thousands of pieces of data which does not affect her/him negatively. As time passes, data analytic techniques improve, and a newly created algorithm may process the previously harmless digital footprints and deduce harmful information.
  • 14.
    Privacy threats relatedto automated profiling (1/2)  stem from the accumulation of personal data: fragments of data regarding an individual user may be linked peace by peace until an individual profile is entirely expose,  aggregation:  may cause dignitary harms because of how it unsettles expectations.  can lead to power asymmetries, as it can increase the power that others have over individuals.  moreover, data compilations may be both telling and incomplete.
  • 15.
    Privacy threats relatedto automated profiling (2/2)  In the case decisions are taken based on data mining and profiling, undesirable discrimination may occur,  In political context, a person may be profiled inaccurately as an extremist movement or party adherent or supporter.
  • 16.
    Group Privacy threatsrelated to automated profiling (3/3)  New technologies and powerful analytics collect and analyse large amounts of data in order to identify patterns in the behaviour of groups, communities and even entire countries.  Some are concerned about the privacy of specific groups such as non-governmental organisations (NGOs) or journalists. For example, they argue that activists are not just individual targets, but, because of their work, they have been targeted as a group.  In the case of the release of demographically identifiable information (DII) group privacy risks may occur.
  • 17.
    Group Privacy  Groupprivacy may be defined broadly as the interest a group has in controlling the information privacy related to the group.  P. Mavriki, M. Karyda, Big Data analysis in political communication: Implications for Group Privacy , International Journal of Electronic Governance, 2018.
  • 18.
    Long-term implications fordemocracy: obstructing the free flow of communication  The “filter bubbles” which personalization tools induce, obstruct the free flow of information, ideas, debates and the communication between citizens and the power holders of society.  Consequently, the own opinion is reinforced, while the ability to handle different points of view is reduced facilitating in this way the polarization of society.
  • 19.
    Long-term implications fordemocracy: manipulation  Content is selected for citizens on the basis of criteria unknown to them and is calibrated not to their proximate selection decisions, but to big data–generated assumptions about where those citizens would want to focus their attention or where marketers need those citizens’ attention to be focused.  The “filter bubbles” and financial means of potent influencers facilitates the spread of inaccurate, misleading or even wrong information. They become increasingly personalized, manipulative, and deceptive, spreading oversimplified messages or misinformation.
  • 20.
    Long-term implications fordemocracy: fragmented political messages  A party may highlight a different issue for each voter, so each voter sees a different one-issue party. In this way, microtargeting could lead to a biased perception regarding the priorities of that party.  Online political microtargeting could lead to a lack of transparency about the party’s promises. Voters may not even know a party’s views on many topics.
  • 21.
    Long-term implications fordemocracy: surveillance (1/2)  Political rights are embodied in the unlimited validity of the right to freedom of speech and opinion and freedom of speech, expression, of association, etc.  Privacy’s public value stems also from its importance to the exercise of these political rights.  Surveillance, may disrupt this structural dimension of a democratic public sphere.
  • 22.
    Long-term implications for democracy:surveillance (2/2)  One of the greatest harms occurring from mass surveillance is the potential chilling effect on political discourse, on the ability of both individuals and groups to express their views and on the possibilities for whistle‐blowing and democratic activism.  New surveillance technologies can lead to “social sorting,” where discrimination and privilege are entrenched through the unplanned consequences of data gathering and analysis.
  • 23.
    Long-term implications for democracy:power asymmetries  The current ecosystem around big data creates a new kind of digital divide: the big data rich and the big data poor.  This environment favours incumbents who already are in the possession of valuable data, also entrenched and moneyed candidates within parties, as well as the data–rich among existing parties.  Data-driven campaigning might even be a form of “cartelization” where large parties erect barriers to protect their dominance from new entrants. The high expense of new campaigning techniques is a significant disadvantage for smaller and newer parties.
  • 24.
    Overall conclusions  Overall,parties and politicians try to discover, reach and target influenceable potential supporters and voters.  Consequently, the big data technologies they use, pose new types of threats to voters’ privacy with multilevel consequences such as the potential impact on democracy.
  • 25.
    Further research  Itseems that predictive analytics and automated decision-making entail unique privacy risks and implications which need further research.  In addition, since big data analytics is blurring the boundaries between individual and group data, there is a need for further research related to the nature and the types of groups as entities.  As personal data is the fuel of processes such as manipulation and discrimination] there is a need for a broader and somewhat different notion of data protection which might counter asymmetries brought about by technology.
  • 26.
    Thank you foryour attention!