Darker side of social media: breaking down Cambridge Analytica’s use of Facebook data

Graphic News
Graphic News |

Latest Articles

Privacy concerns arise with government Covid-19 tests

Coronavirus: What’s the difference between quarantine and isolation?

#MoreViralThanTheVirus warns that students are not immune to Covid-19

Delving into Facebook’s big data scandal.

Today we will look at the technology, psychology and values behind Facebook’s big data scandal.

Facebook has become a part of our lives. Without it, we wouldn’t be able to keep in touch with friends and family, or do a host of dumb quizzes to find out what kind of dog we are.

It was such a quiz that snared Facebook in its biggest scandal yet.

The scandal involving Cambridge Analytica’s use of Facebook data to help Donald Trump’s campaign win the 2016 US presidential election began three years earlier with a paper in the Proceedings of the National Academy of Sciences.

The study by Michal Kosinski of Stanford University, David Stillwell of the University of Cambridge, and colleagues, involved 58,466 adult American users of Facebook. They provided information about themselves and took a test to classify their personalities in five broad categories: degree of openness, conscientiousness, extroversion, agreeableness and neuroticism – or “Ocean”, as the test is sometimes called.

The researchers then correlated the “big five” personality traits with each person’s Facebook likes, an average of 227 per subject. This data led to a computer model which could predict a person’s race with 95 per cent accuracy, gender (93 per cent) and political orientation (85 per cent).

In 2014, Aleksandr Kogan, who runs the London-based Global Science Research (GSR), approached Kosinski and Stillwell. Kogan wanted access to the Ocean data set on behalf of Strategic Communication Laboratories (SCL). The deal did not go ahead.

GSR developed its own version of the data set using Amazon’s crowdsourcing marketplace Mechanical Turk (MTurk). Kogan’s subjects were paid by SCL to

take the Facebook survey. But, for every person recruited on MTurk, Kogan also captured the same data for each subjects’ unwitting Facebook friends – an average of around 340 friends per individual. GSL suddenly owned a massive data pool of profiles of 50 million individuals across the United States.

GSR’s data set was used by Cambridge Analytica in the Republican Party’s Iowa caucuses in support of Ted Cruz. Cruz won. The impressive bit, says Cambridge Analytica’s chief executive Alexander Nix, is to expand the findings from those who took the personality tests to the entire American electorate of 230 million. Nix says Cambridge Analytica also has “4,000-5,000 data points” – pieces

of information – on every single adult

US citizen.

In June 2016, the Trump campaign hired Cambridge Analytica. The campaign developed “dark posts” on Facebook to target uncertain left-wingers, black Americans, and young women to “suppress” their vote. Dark posts target users with specific profiles in specific locations; for example, news aimed at black Americans in which Hillary Clinton refers to black men as predators.

That’s how Facebook became weaponised, and Trump became President of the United States of America.

Points to consider:

  • What do you know about Ocean, or the big five character traits test? Do you think character tests are valid?
  • What advances in technology allowed Cambridge Analytica to go after unsuspecting people? Is there technology around now that could have prevented it?
  • Facebook’s creator Mark Zuckerberg has earned a lot of money off Facebook being able to sell data. Is this ethical? How is it different from the way Cambridge Analytica used the data? If people use Facebook, should they expect to have its vast range of services for free?