Skip to Content

The Wiki Man

Had 533 people voted differently, the Tories would have a full majority

31 March 2018

9:00 AM

31 March 2018

9:00 AM

Nine years ago, before Cambridge Analytica existed, I caught wind of a research project at Cambridge involving the online measurement of the ‘big five’ personality dimensions. These are usually listed by the acronym OCEAN or CANOE: Conscientiousness, Agreeableness, Neuroticism, Openness and Extraversion. I made a note to go to Cambridge to learn more but, being low on conscientiousness, I never got around to it.

Perhaps I dodged a bullet by not getting involved. When people say ‘It’s the early bird that catches the worm’, I always reply, ‘Yes, but what about the early worm?’

Besides, does personality alone predict behaviour? Perhaps not as much as Cambridge Analytica promised. True, had Donald Trump done less or worse digital advertising, he might have lost. But that’s not to prove psychographic data was decisive. Trump had a smaller budget, after all. But unlike Hillary, he was sufficiently interesting that people wanted to see him. He also savvily used his plane to return home to New York every day after campaigning, rather than spending exhausting consecutive nights on the road as she did. (In this, Trump was again inspired by British thinking — by Peter Grant and Led Zeppelin, who pioneered this approach when touring in the 1970s.)


Political advertising is always more dishonest than commercial advertising. The reason is simple: commercial marketing is a repeated play; political marketing is a one-shot game. Dishonest commercial advertising — for products bought repeatedly — might pay in the short term but costs you double in the long term as people cotton on to your deceit. In politics, only the short term counts. Digital targeting makes deception easier, as it’s easier to lie to people one at a time.

(This long-term/short-term distinction is even found in nature: some orchids produce petals advertising the presence of nectar to bees, even though they can’t deliver it — the Nick Cleggs of the botanical world. This only pays off early in the season, after which bees spot the deception.)

Could dishonest, micro-targeted advertising sway a US election? Given the peculiarities of the American electoral system, yes. Just 20,000 people duped into voting (or staying home) in the right part of one swing state could be decisive. What is less convincing is the media’s recent contrived attempt to blame similar data shenanigans for the Brexit result. In a referendum every vote counts equally.

Forget your stereotype of a Brexit voter, trapped in some dismal post-industrial town. Geographically, the Brexit vote was widespread. Nine out of 12 UK regions voted to leave, including the most populous and prosperous — the south-east. Overall 408 constituencies voted Leave against 242 which voted Remain (55 of which were in Scotland). The reason the result is not seen as decisive is largely because those places most in favour of Remain (Scotland, central and west London, Cambridge, Oxford, Brighton, the mind of A.C. Grayling) all possess a sense of their own importance out of all proportion to their size. Scotland actually has a smaller population than Yorkshire.

If anyone should be investigating election fakery, it’s Theresa May. Had 533 people voted differently across nine constituencies, she’d have a full majority. This media frenzy isn’t about CANOE — it’s about confirmation bias: post-selecting facts to suit pre-existing prejudice.

Never forget that intellectuals and educated people are far more prone to feats of self-delusion than ordinary people. Amexiteer Benjamin Franklin spotted this when he wrote: ‘So convenient a thing to be a reasonable creature, since it enables one to find or make a reason for everything one has a mind to do.’


Show comments
Close