The Brexit vote and the US presidential election were swung by shadowy firms using sophisticated but underhand new techniques, say critics. Simon Wilson reports.
What’s the worry?
A series of newspaper articles over the past few months claim that shadowy consulting firms, dominated by wealthy right-wing backers, helped to swing both the Brexit vote in the UK and the Donald Trump victory in the US. They did so, it is alleged, by marrying cutting-edge expertise in the capture and analysis of big data with psychological profiling techniques – enabling them to analyse people’s personalities using data mined from social media and then tailor advertising (in this case political messages) to their psychological profiles.
How does it work, exactly?
Cambridge Analytica (CA), a US-based spin-off from an established British company, Strategic Communication Laboratories (SCL), is at the centre of much of the fuss. One of its biggest shareholders is Robert Mercer, a billionaire backer of Trump. The firm claims to have played a key role in Trump’s victory (although this is disputed by some involved in the campaign).
Its psychometric (or “psychographic”) model has its origins in research by scientists at Cambridge University’s Psychometric Centre, which involved a personality quiz on Facebook that went viral. More than six million people took the test, based on a well-established model (dating back to the 1980s) which assesses people based on the “big five” or “Ocean” personality traits: openness, conscientiousness, extraversion, agreeableness and neuroticism.
What did the quiz reveal?
It showed that Facebook profiles, especially people’s “likes”, could be correlated across millions of others to produce a very accurate picture of their personality. With knowledge of 150 “likes”, the model could predict someone’s personality better than their spouse. With 300, it understood you better than yourself. Cambridge Analytica (which has no link with the university, despite exploiting some of the research that originated there) developed this model by comparing its psychometric information with data it harvests about other aspects of consumers’ lives.
The idea, says Gillian Tett in the Financial Times, is that “if someone knows how you shop, live, communicate, travel and so on, you can extrapolate backwards to create messages that resonate in a psychological sense”. For the Trump campaign, says CA boss Alexander Nix, the firm built a “massive database of 4,000-5,000 data points of every adult in America”. It then used its data to identify voters who could be “flipped” to support Trump; Hillary Clinton supporters who could be nudged into staying home; and which cities and messages Trump should be targeting.
Why is this dangerous?
Because there are no checks against dishonest or misleading claims. These are adverts targeted specifically and exclusively at receptive audiences. So, as Dominic Cummings of the Vote Leave campaign puts it, “it’s as if someone is doing TV ads but nobody in the media sees them and writes about it”. As a result, bogus claims are much less likely to be spotted and censured. The Times journalist Hugo Rifkind wrote that he had contacted Ofcom, the Advertising Standards Authority and the Electoral Commission, all of whom “sent me to each other”.
None regards the “verification of political claims as its remit”, even when made widely and in public (such as the dubious £50m-a-day figure on the Brexit bus). So when it comes to microtargeted claims, all bets are off. Cambridge Analytica, initially “chatty, stopped replying when I asked who checked its communications were actually true”, writes Rifkind. “The real answer, almost unbelievably, is nobody.”
Is that the only worry?
It’s not just about honesty, but about transparency, influence and money. Cambridge Analytica, for example, told The Observer that it had not been “employed” on the Leave campaign. It “is a US company based in the US. It hasn’t worked in British politics.” In fact, Andy Wigmore of the Leave.EU campaign confirmed to the paper’s Carole Cadwalladr that the firm had indeed worked for them unpaid, at the behest of Robert Mercer. Under electoral rules, services-in-kind are supposed to be declared. This gratis work was not. This raises the question of whether a foreign billionaire influenced the referendum without it being apparent.
Did Vote Leave use these techniques?
It’s hard to say with confidence, since firms’ techniques are proprietary. But Vote Leave did spend £2.7m with one small Canadian digital marketing firm called AggregateIQ. That’s well over a third of Vote Leave’s budget. And two other campaign groups that received donations from Vote Leave spent a further £765,000 with the firm, according to Jasper Jackson in The Guardian. What services did they provide?
The invoices provided to the Electoral Commission under referendum rules is filled with jargon: AggregateIQ was running a “targeted video app installed and display media campaign”. But there’s no clue as to where the ads appeared or who saw them. “Did they spend more money on reaching under-45s in Hull or pensioners in Canterbury? There’s no way of knowing, not least because the Electoral Commission doesn’t ask for the information.”
How to fake it
Psychological profiling is not the only campaign technique that potentially threatens democracy, says George Monbiot in The Guardian. Other examples are lobby groups funded by billionaires that disguise themselves as think tanks; “astroturf campaigning” (employing people to masquerade as grassroots movements) and “botswarming” (creating fake online accounts to create the impression that large numbers of people support a political position).
All these are “current threats to political freedom”, says Monbiot. “Election authorities such as the Electoral Commission in the UK have signally failed to control these abuses, or even, in most cases, to acknowledge them.” So the chances of them getting to grips with personalised digital advertising are slim indeed.