Already a subscriber? Log in
Listen to Paul Wood on Trump and the data-crunchers
To the detractors, this is pure fantasy. One Republican political consultant said: ‘Their thesis is people don’t know what they think about politics but we can anticipate what they will think based on their personality types. That’s nonsense.’ Political consulting — the business he was in — was full of characters like the Duke and the King in Twain’s Huckleberry Finn, schnorring the yokels and relieving them of their wallets. ‘It makes perfect sense that Donald Trump used these people. You’re talking about a guy who lost money on red meat and professional football, booze, sex, gambling, discount airfares and higher education in the United States. He went broke doing things it’s impossible to lose money on in this society, so he’s very good at being played for a rube.’
A Republican data scientist for a rival firm said he did not use psychographics. ‘If you get a voter on the phone, why are you asking them what their favourite ice cream is or what their favourite colour is — why don’t you just ask them who they’re going to vote for?’ He added: ‘They’ve got a smooth-talking Brit wearing Savile Row suits who gives you a great pitch and wows you a little bit; they’ve got a great PR operation, but with psychographic profiling, there’s nothing there. They’re really, really smart people. It’s like they’re a bunch of board-certified doctors who decided to make a lot more money selling snake oil.’
That charge is wearily familiar to Matt Oczkowski. At 29 he is a veteran of eight years of political campaigns and he ran Cambridge Analytica’s Trump data operation. ‘The snake-oil salesman stuff is probably just someone who never believed in the methodology to begin with. This is much more a case of jealousy than it is legitimate and scientific argument. Most Republican political consultants have no conception of what a data programme is.’ He was tired of having to defend the company. ‘Given the results of the election, I feel like other people will have to start defending themselves to us.’
Actually, there was very little psychographics in what he did for Trump. There wasn’t time: they’d had to set up the entire data operation ‘from scratch’. Instead, they did a huge amount of more conventional polling and psychographics was used — he maintains — to get better predictions out of it, updating ‘methodologies that have been used the same way since the 1970s’. They simply had a better picture of what the electorate looked like, Oczkowski said.
The problem, says Jill Lepore, a historian of polling at Harvard, is that while traditional polling was bad for democracy and could be unreliable, data science was still more damaging. Politicians’ views were dictated by consultants, not principle, while voters were told only what they wanted to hear. ‘The effects of data science on the political process are probably considerably worse. Data science is the solution to one problem but the amplification of a much bigger one — the political problem.’
She was talking about the key to it all: ‘microtargeting’. Cambridge Analytica sliced and diced the electorate so Trump could talk to small groups directly, or even one at a time: one email or letter to the timid introvert at No. 22 who cares about jobs and limited government, another to the loud extrovert next door who cares about gun rights and Isis. Psychographics was used in what Oczkowski calls ‘sentiment analysis’ and ‘tonality’, that is in deciding what to say to people and how to say it. Almost 50 years after The Selling of the President was written about the advertising industry’s intrusion into politics, this is the ultimate attempt to manipulate the electorate.
Nix, Cambridge Analytica’s chief executive, is evangelical about this. ‘What we’re seeing is not about politics. Communication is fundamentally changing. Creative-led blanket advertising is being replaced by data-driven individualised advertising. This can only be good. The advertisers get a much better return on investment; the consumers are not being bombarded with adverts for things they don’t care about.’ Don Draper is dead — replaced by a twentysomething chugging Diet Coke at a laptop.
The clients seem to like it. Trump reportedly spent only a third as much as Clinton on TV advertising. ‘Cambridge Analytica bullshit a lot,’ said a senior figure in the campaign here for Brexit, echoing the American consultants, but ‘if you are a smart client you can definitely get value out of them’. Trump had already announced his rust-belt strategy before Cambridge came on board. But Oczkowski says he ‘greatly influenced’ where the candidate travelled, based on the ‘density of persuadable voters’. ‘We certainly don’t take credit for any strategy,’ he was careful to say, but ‘we reinforced decisions that Mr Trump had already decided to take’.
You might think from a casual reading of the Cambridge Analytica press release that they predicted the outcome of the election. They did not. A company spokesman called reporters before election day to say that Trump had only a 20 per cent chance of winning. That was increased — in an internal assessment — to 30 per cent as people went to vote. Having put the entire nation on the couch, Cambridge Analytica got it just slightly less wrong than all the conventional pollsters. ‘We saw it coming in terms of the trends,’ Oczkowski said, ‘but in terms of actually predicting a win, we knew it was a possibility but we couldn’t say definitively that it was going to happen.’
This is comforting. The human heart remains unknowable. Elections remain fickle things. And despite data science, the warning to journalists from this astonishing presidential election still stands. Never again should we begin a report: ‘According to opinion polls…’
Because you read about advertising
Also by Paul Wood
Latest
Comments
Join the debate for just $5 for 3 months
Be part of the conversation with other Spectator readers by getting your first three months for $5.
UNLOCK ACCESS Just $5 for 3 monthsAlready a subscriber? Log in