THE TANGLED TALE OF CAMBRIDGE ANALYTICA

Did a little-known company use profiling and big data to manipulate the US presidential elections?

“His scienti c work on the Ocean models cautioned that these profling and prediction techniques could be used in ways to manipulate people.”

In December of 2016, an article in the German Das Magazin magazine caused a stir. Reporters Hannes Grassegger and Mikael Krogerus wrote that a little-known company called Cambridge Analytica had used pyschometrics, big data and social media to help swing the US presidential elections in favor of one Donald J. Trump.

In a presentation at the Concordia Summit, Alexander Nix, CEO of CA, revealed how the company had crunched big data to influence the elections. It had bought personal data from various sources, aggregated this data and calculated a personality profile based on five personality traits, also known as the Ocean Method.

Nix told the crowd that by categorizing voters based on their psychology, the company could design specifc ads to target their values. “For a highly neurotic and conscientious audience,” Nix said, he might be shown the threat of a burglary, to better sell him a gun. Nix told Motherboard that on the day of the third presidential debate, Trump’s team tested 175,000 different Facebook ad variations, to find the ones with the best influence.

Michal Kosinski, a leading expert on psychometrics – a data-driven branch of psychology – has demonstrated how much you can do with the Ocean Method, when combined with enough data. In 2008, Kosinski, then a student at the University of Cambridge in England, wrote a small quiz app for Facebook called MyPersonality, based on the Ocean Method. Millions of people took the quiz, which, together with the subjects’ Facebook activity, gave Kosinski and his team a wealth of big data to work with.

In 2012, Kosinski demonstrated something startling: From a mere 68 Facebook likes, his team could reliably predict a person’s skin color, sexual orientation, and whether they leaned Democrat or Republican. With 150 likes, they could profile a person better than their parents, with 300 likes, Kosinski’s profiling techniques could predict a person’s behavior better than their partners.

Neither Kosinski nor Cambridge University are associated with Cambridge Analytica, but the process appears to be identical to the models that Kosinski developed and that he had warned people about. His scientific work on the Ocean models cautioned that these profling and prediction techniques could be used in ways to manipulate people — and it looked like a company had found a way to weaponize it for political gains. But had it really? Following the buzz that arose with Das

Magazin’s report, the New York Times spoke with Republican consultants and former Trump campaign aides, along with current and former CA employees, who say the company’s ability to exploit personality profiles are “exaggerated.” According to the

NYT, CA executives “now concede that the company never used psychographics in the Trump campaign.” And for all the apparent success CA had with the Trump campaign, there was one inconvenient fact that they couldn’t escape. Before Trump, CA had worked with Republican presidential hopeful, Ted Cruz. Cruz eventually lost the Republican primary to Trump and suspended his campaign for president.

Whither the truth about how powerful Cambridge Analytica’s techniques are, perhaps the fact that we are so able to believe their claims and not just dismiss them as science ction is the more telling story. We live in a surveillance economy, where powerful companies like Google and Facebook already know more about us than we do about them, and we take online tracking for granted more often than not.

We believe a company like CA is possible because the technology already appears to be here, or if not now, then soon. And that caps this tangled tale with a more worrisome ending than anything.