Social Networks and Data Privacy
Privacy is the ability to restrict and control access,
transfers, and exchanges of one’s information (Tavani 2011). Internet users
constantly share information on social networks that third parties gain access
to. This data exchange occurs through weak or non-existent user privacy
settings, through the application programming within these social networks, and
tracking technology such as cookies (Al-Saggaf & Islam, 2015). Data miners collect
this large amount of personal information and analyze it for meaning and
problem solving. Many fields use data mining to gain a deeper understanding of
human behavior (Barbier & Lui, 2011). While the information is often
directly provided by the users and public, data mining practices raise questions
regarding the ethical nature of these practices and possibly harmful
implications. Until stronger user protections exist or users have complete
control over their data, internet users must take necessary precautions and
only share what they deem acceptable for public use.
Web 2.0
Web 2.0 began a new internet age. No longer is the
internet purely a one-sided information source, but rather a communication hub
where users can create their own material and respond to other users (Bowles,
2013). This significant addition along with the development of improved access
to high-speed internet and affordable portable computer devices allows users to
stay constantly connected. Some popular internet tools for sharing content and
engaging with others are blogs LinkedIn, Facebook, Google+, and Wikipedia.
The first blog was published in 2000 and by 2011, 181
million bloggers were sharing ideas through websites and social networks
(Bowles, 2013). People also use the internet to connect to colleagues, family,
and friends. LinkedIn is the largest professional network with 740 million
global users from over 200 countries (LinkedIn, 2021). Facebook is a popular social
network that began in 2004 (Bowles, 2013). On the Facebook platform users can
connect with friends and family through posts, reactions, comments, messages,
and pictures. Google+ is another social network, and Facebook competitor, that
allows a larger circle of acquaintances since the website makes friend
suggestions according to other connections (Bowles, 2013). These social
networks allow users to strengthen connections and conduct professional
business from their homes.
With all the benefits Web 2.0 offers, there are also
potential disadvantages. Users are vulnerable to cyberbullying privately or
publicly across various digital channels (Bowles, 2013). Users have the
capability to share everything about their lives, which depending on the
information, can have harmful consequences for the users. Hackers may obtain
personal information and gain access to private accounts. Everyone has access
to public data, such as future employers, which could prevent users from being
hired (Bowles, 2013). Companies and websites can easily obtain personal data,
track online behavior, and use it for profit or research. Currently, internet
users have only a limited control over who obtains their data and how others
use this information.
Implications of the Data Mining
As of 2015, Facebook had one billion users, while the
internet had two billion users globally (Al-Saggaf & Islam, 2015). Upon
joining Facebook and other social networks, users build profiles that contain
personal data such as names, education, age, employment, gender, relationship
statuses, location data, and photos. In the article Data Mining and Privacy of Social Network Sites’ Users: Implications of
the Data Mining Problem, researchers discuss potential privacy issues due
to data mining. In the study, researchers
collected data and then applied a SysFor algorithm to test the accuracy of the
data mining algorithm to gauge the threat level to privacy. SysFor is a data mining
algorithm comprised of multiple decision trees, “allowing the exploration of
more logic rules from a dataset compared to the number of logic rules that can
be explored by a single tree” (Al-Saggaf & Islam, 2015, p. 942).
Many professions use data mining as a means to
effectively reach consumers, which at first glance is not threatening or an
invasion of privacy for several reasons. First, the data collected is readily available
online and often disclosed publicly by the owners themselves. Secondly,
hypothetical trends are inferred from a larger data-set (Al-Saggaf & Islam,
2015). The accuracy of the conclusions
drawn are questionable and not a clear violation of users’ privacy. However,
privacy issues do exist with secondary usage of the data including profiting
and potential for profiling
Researchers Yelsam Al-Saggaf and Md Zahidul Islam
conducted the study to provide empirical evidence to the argument that data
mining is harmful to users. In 2012, two coders searched for public Facebook
statuses through a website youropenbook.org with
the search terms “lonely” and “connected” and analyzed the first 308 accounts
of females over the age of eighteen within the results (Al-Saggaf & Islam,
2015). Researchers collected 45 real attributes from 616 Facebook user’s
profiles. Next researchers applied the SysFor algorithm to the real data, which
detected 800 logic rules (Al-Saggaf & Islam, 2015). One logic rule SysFor
gathered is that women who are interested in both genders, and do not disclose
information about their language are lonely. Other logic shows that users with
a profile picture of their face, with a significant other, or with family feel ‘connected’,
whereas users with the “in a relationship” status who speak multiple languages
are ‘lonely’ (Al-Saggaf & Islam, 2015). The accuracy of SysFor’s predictions is less
important than the patterns that surfaced. From the resulting logic analysts
can take minimal information and make potentially inaccurate assumptions, which
applied to a population may group individuals erroneously.
Ethical Implications of the 2016
Election & Cambridge Analytic
In November of
2016, polls predicted that Hillary Clinton would win the presidential election,
so when Donald Trump was declared the 45th President of the United
States many sought to understand how. Cambridge Analytica became the sole focus
for some analysts for the company’s part in Trumps success. Cambridge Analytica
claimed to have classified 220 million Americans from data collected from
Facebook and purchased data from brokers. With this data the company used
behavioral microtargeting to send individualized messages to users according to
behavioral profiles playing on “hopes, fears, prejudices, and fancies that
message recipients may not, themselves, have even been aware of” (Ward, 2018,
p. 133).
Ward looks at Cambridge Analytica’s actions through the
lens of Kant’s principles or categorical imperative asking three questions to
determine the ethicalness of the company’s practices. Ward answers the first
question, “Does Kant’s categorical imperative provide ethical bases for opposing
the collection, aggregation, and instrumentalization of digital data for the
purpose of behavioral microtargeting?” as yes, explaining that behavioral
microtargeting practices infringe on the targeted users autonomy preventing
them from acting rationally (2018, p. 145). The next question asks if there are
limitations for the application of Kant’s categorical imperative. Ward explains
that since the actions of an individual is insignificant to massive data
already accumulated by Cambridge Analytica, autonomous actions are meaningless
in this instance (2018). In other words, the only action a user can take is to
avoid social networks, which is still not a viable solution. The final question
recommends applying other ethical principles to the incident as Kant’s
principles are grounded in the idea of acting according to one’s duty and does
not fully address social implications (Ward, 2018).
Ward concludes that behavioral microtargeting is a threat
to privacy and autonomy, but additional research is required to understand the
full ethical implications (2018). Ward also concludes that the incident
occurred because social media networks and data companies like Facebook
supplied Cambridge Analytica the data, which has ethical implications also
(Ward, 2018). The 2016 election is one example where secondary data mining
practices may harm those targeted and society.
Comparison and Analysis of the
Articles
Data mining on its own is not necessarily a harmful
practice. Users share the information publicly and willingly. Marketers can use
that compiled data to match consumers with beneficial products and services.
Consumer data allows companies to gain an understanding of their consumers’
preferences and create enjoyable products and satisfying shopping experiences
for each one. However, issues of privacy and autonomy arise regarding secondary
data in which data-miners profit financially or otherwise (Tavani, 2011). Both articles
research the use of secondary data in different circumstances. Yelsam Al-Saggaf
and Md Zahidul Islam research what type of inferences analysts can make from
this data, while Ward covers the ethical practices of Cambridge Analytic and
data suppliers during the 2016 election.
Both articles support the idea that data-mining can have
harmful implications to users. Al-Saggaf’s and Islam’s article concludes that data
mining software and algorithms can profile users inaccurately assigning
incorrect assumptions to subgroups of users. Ward asserts that it is ethically
questionable to use those assumptions as a means to manipulate users without
their knowledge. Saggaf and Islam do not address ethical issues with how the
data is obtained, but rather the secondary use of the data, while Ward notes
the ethical responsibility of social networks and data providers.
Conclusion
The fact that social network users have limited control
over their data is a privacy threat.
The unethical practices
of Cambridge Analytica during the 2016 election the ethical research compiled
for Al-Saggaf’s and Ward’s study reflect that. Both third parties were able to
obtain personal information of many social network users. While one data mining
example is for election campaigning and the other for research, the users were
still unaware of the secondary use of their public data. Even if social
networks list these uses within the terms of service or privacy agreements,
many users may not be aware of the full scope of these contracts.
Social networks share information with search engines and
websites, and while there are some privacy standards and guidelines in place,
public data is available to a variety of third-parties. Until social network users
have primary control over how their data is used or user data is completely
protected from external sources, the best defense for users is to refrain from
sharing any information with social networks. If that is not practical or possible
within this digitally-dependent society then users should limit and monitor the
information they share publicly. Whether the data is used in a socially
positive way for research or as a manipulation tactic, a user’s best course of
action is to filter the information that is public domain.
References
Al-Saggaf,
Y. & Islam, M. (2015). Data mining and privacy of social network sites’
users: Implications of the data mining problem. Science & Engineering Ethics, 21(4), 941-966. https://doi.org/10.1007/s11948-014-9564-6
Barbier
G., Liu H. (2011) Social network data
analytics: Data mining in social media. Springer, Boston, MA. https://doi.org/10.1007/978-1-4419-8462-3_12
Bowles,
M. D. (2013). Introduction to digital
literacy [Electronic version]. Retrieved from https://content.ashford.edu/
LinkedIn.
(2021). About LinkedIn. LinkedIn. https://about.linkedin.com/
Tavani,
H. T. (2011). Ethics and technology:
Controversies, questions, and strategies for ethical computing (4th ed.).
John Wiley. https://futuresinitiative.org/futureofwork/wp-content/uploads/sites/216/2018/07/Ethics_and_-Technology-Controversies-Questions-and-Strategies.pdf
Ward,
K. (2018). Social networks, the 2016 US presidential election, and Kantian
ethics: applying the categorical imperative to Cambridge Analytica’s behavioral
microtargeting. Journal of Media Ethics,
33(3), 133-148. https://doi.org/10.1080/23736992.2018.1477047
No comments:
Post a Comment