Cambridge University psychologist Aleksandr Kogan, the scientist whose app “thisisyourdigitallife” harvested tens of millions of Facebook profiles for disgraced and now-shuttered election data firm Cambridge Analytica, wants everyone to know that his data set is not responsible for the way the 2016 federal elections went down.
In fact, Kogan would like it to be known that he thinks if Cambridge Analytica used the psychometric profiles he sold them for campaign purposes, that was “entirely ineffective” and “stupid,” according to his testimony before the US Senate Commerce, Science and Transportation subcommittee on Tuesday. Per the Guardian:
“People may feel angry and violated if they think their data was used in some kind of mind-control project,” Aleksandr Kogan, the now notorious Cambridge University psychologist whose app collected data on up to 87 million Facebook users, said during a US senate hearing. “This is science fiction. The data is entirely ineffective.”
“If the goal of Cambridge Analtyica was to show personalized advertisements on Facebook, then what they did was stupid,” Kogan said, arguing that it is much more effective for any advertiser to use Facebook’s own advertising targeting tools.
Cambridge Analytica denies using the harvested Facebook profiles on Trump’s campaign, something Kogan agreed in April was “unlikely.” According to Kogan’s testimony on Tuesday, Facebook’s own advertising tools—which even Facebook admits Donald Trump’s campaign made better use of than his opponent Hillary Clinton—were much better suited for campaign purposes.
While Cambridge Analytica goons were caught on camera by Channel 4 bragging that their research was key to helping Trump clinch the election, former clients including the Trump campaign have alleged that the company’s psychometric profiles were overpriced and worthless. A staffer for Ted Cruz’s campaign told Gizmodo in March that one of the company’s products, a software suite called Ripon designed to help classify voters by personality types, never worked and was essentially “vaporware.”
Other former clients told the LA Times that Cambridge Analytica’s powerful Republican backers, the billionaire Mercer clan and Steve Bannon, made the firm able to land deals. But some also alleged that the company’s tools were unimpressive, convoluted, and never proven to have worked.
Whether Cambridge Analytica’s tools were particularly useful is a separate issue from whether the decision to harvest the profiles and the permissions granted by Facebook to apps like Kogan’s were ethical.
Prior versions of Facebook’s advertising API allowed apps to access extensive information on the friends of people who used them without their consent; Cambridge Analytica’s decision to partner with Kogan to gather data for their own purposes may have technically been a violation of Facebook policy, but at the end of the day both firms took a flippant approach to user privacy that left millions angry. The wide-ranging permissions Facebook gave to apps may have also violated a Federal Trade Commission consent decree, which could still result in hefty fines.
In the Senate hot seat, Kogan argued that tech companies like Facebook took such a cavalier approach to privacy to continue improving their internal ad tools.
“They are under enormous financial pressure to gobble up more and more of our data so they can deliver better and better personalized ads,” Kogan said, according to USA Today. “And the dirty secret in the industry is that these ads right now are just not that effective. Not useless, but not as effective as we’d want. So companies want more, not less, data, so they can do better.”
Yet research has shown “not useless” is still pretty powerful, the Guardian noted, with one 2013 study conducted by his former colleagues at Cambridge University’s Psychometrics Centre finding that Facebook likes can predict a wide variety of “private traits and attributes.” According to the New Yorker, research conducted by the UK-based Online Privacy Foundation’s Chris Sumner and Matthew Shearing found that while psychological messaging tools might not work on the individual level, they worked on the group level on Facebook audiences:
By rewording … ads to appeal to the respondents’ underlying psychological disposition, the researchers were able to influence and change their opinions. According to Sumner, “Using psychographic targeting, we reached Facebook audiences with significantly different views on surveillance and demonstrated how targeting . . . affected return on marketing investment.” Psychological messaging, they said, worked.
As the New Yorker noted, this was accomplished using Facebook’s audience tools, and the company’s own research has concluded it can manipulate people’s moods by changing their feeds or raise voter turnout in elections.
Whether Facebook was decisive in the outcome of the elections is a question that may never be solved and includes other players like alleged Russian troll farms, but one thing that is clear is that the Trump campaign did make heavy use of those built-in tools. An internal Facebook campaign report found team Trump spent $44 million on the site, testing 5.9 million different versions of his ads, while Clinton’s campaign spent $28 million testing 66,000 ad variants.