The Wiki Man

Why targeted political advertising is not to be trusted

Has an algorithm simply identified you as someone who might care about a particular issue?

20 May 2017

9:00 AM

20 May 2017

9:00 AM

When you get into a taxi, there’s usually a framed sheet of paper describing what you pay for your trip: the cost of every mile travelled at different times of day, and the price of waiting time.

As digital screens become ever cheaper, it won’t be long before someone suggests that there is no need to have these things any more. Instead a button will appear on the taxi’s new seatback touchscreen which will reveal the tariff when pressed.

All very sensible, you may think. Except for this. The nature of a promise displayed on paper is subtly different to a promise displayed on a screen. Anything writ in liquid crystal should always be viewed with a little added suspicion.


There is a reason why we use phrases such as ‘tablets of stone’ to refer to promises. It isn’t only that stonework endures; once carved, it is also very difficult to edit. And when a declaration is carved in a public place, the reader can see that the promise you are making to him is the same one you have made to everyone else. Public promises carry more weight: hence why the words ‘as seen on TV’ are more convincing than ‘as seen on Facebook’.

But when your taxi displays the prices on a digital screen, you have no way of knowing whether it is displaying ‘the true price’ or ‘the drunk gringo tourist price’. For all you know the driver may have pushed a button under his seat, hiking the tariff according to how much he can gouge from you. Similarly when you visit a travel website using an expensive laptop computer, you may not be quoted higher prices — but by default the website will automatically display a swankier (and pricier) range of hotels.

Today, when your political party makes a promise to you via some highly targeted social media campaign, can you be really sure it is seriously committed to the issue at hand? Perhaps some algorithm simply identified you as one of a number of people in marginal constituencies who care disproportionately about the subject. This distinction matters. I need to know not only what politicians are promising to me, but what they are promising to everyone else. If there is one thing to be said for the ‘Ed Stone’, at least there was only one of them.

The interesting thing about automatically targeted advertising is that it does not require some evil mind for this two-facedness to emerge. It will happen automatically as a product of what the software is taught to do. People on lower earnings will be told all about taxing the rich: no one in richer areas will hear a thing. People in Richmond, Yorkshire, might learn about the re-legalisation of hunting; in Richmond–upon-Thames they won’t.

This is something which bothers me about ‘programmatic’ advertising: that at times it may be too effective. No one in a human meeting would seriously suggest placing booze advertising near the locations of AA meetings. But algorithms may learn through ethics-free trial and error to blitz reformed gambling addicts with ads for online casinos. In the old analogue days, recovering addicts could avoid temptation by skipping the racing pages, or staying out of the bookmakers’. These digital temptations could follow you wherever you go.

In political advertising, the potential for digital hypocrisy matters a great deal. The value of democracy lies not only in choosing a government, but in giving it the legitimacy to govern. If everyone has access to the same information on which to decide, you can reasonably describe an election result as a consensus and expect people to defer to the popular will. This is far from certain when everyone is voting for a digital manifesto customised to their own personal tastes.

Rory Sutherland is vice-chairman
of Ogilvy Group UK

Subscribe to The Spectator today for a quality of argument not found in any other publication. Get more Spectator for less – just £12 for 12 issues.


Show comments
Close