Google says it will limit the targeting of political ads to make it harder to sneak misinformation to impressionable voters. That puts the company ahead of the pack when it comes to making the political business of big internet platforms look less threatening. But the efficiency of political micro-targeting is questionable, and Google is responding to a moral panic rather than any real danger to democracy.
Since the 2016 US presidential election, the public has become aware of techniques that allow advertisers to aim their messages at narrow groups of people, sliced not just by place of residence, age and sex, but also by consumer and political preferences, browsing histories, voting records and other kinds of personal data. This culminated in the Cambridge Analytica scandal in 2018, when news reports showed that the U.K.-based micro-targeting firm had improperly harvested lots of private user data from Facebook. The platforms were on the spot to do something.
Twitter has banned political ads entirely, but then it didn’t sell many, anyway, serving instead as a free platform for political messages. In an op-ed in the Washington Post following Twitter’s announcement, Ellen Weintraub, chairwoman of the U.S. Federal Election Commission, called for an end to political micro-targeting instead of an ad ban. “It is easy to single out susceptible groups and direct political misinformation to them with little accountability, because the public at large never sees the ad,” she argued.
That was a controversial proposal. Writing in the same newspaper, Chris Wilson, who had been responsible for digital strategy in Senator Ted Cruz’s 2016 presidential campaign (which was the first in that election to hire Cambridge Analytica), countered that micro-targeting has helped increase voter turnout and drive down advertising costs for campaigns. His suggestion was to make the targeting more transparent.
Google, however, found it more expedient to go along with Weintraub’s proposal than to fight an uphill battle using Wilson’s arguments. In a blog post on Wednesday, the company said it would no longer let advertisers target messages “based on public voter records and general political affiliations (left-leaning, right-leaning, and independent).” Only basic targeting by age, gender and postal code would be allowed.
This, is course, is no more than Russian trolls would have required in 2016 — as Wilson pointed out in his Washington Post op-ed. Their propaganda campaign was largely geographically targeted.
There’s still no proof that micro-targeting is more effective than other forms of advertising. Academic work on the subject has tended to be rather theoretical, while experimental evidence is scarce. In a paper published this year, German researcher Lennart Krotzek concluded after an experiment matching ads to personality profiles that “candidate messages are more effective in improving a voter’s feeling toward a candidate when the messages are congruent with the voter’s personality profile, but they do not result in a higher propensity to vote for the advertised candidate.”
Internet platforms have done little to further the study of political targeting.
Google offers a transparency report on political ads placed on its various properties — search pages, YouTube, the sites of media partners. It says that the biggest U.S. advertiser in the last 12 months is the Trump Make America Great Again Committee, which has spent $8.5 million. The report discloses that the pro-Trump group has targeted its most recent ads at all people older than 18 throughout the U.S., but offers no clues as to whether any more precise targeting was used. That’s the case with the rest of the advertisers, too.
Facebook’s transparency report is just as opaque when it comes to the precise targeting of ads by voters’ interests and political leanings.
It’s easier for Google than for Facebook to abandon precise targeting, because one of its key strengths is being able to link ads to search words.
That’s a form of rather precise targeting not affected by Google’s policy change. Slicing and dicing the audience is at the heart of Facebook’s offering to advertisers, so it’s understandably hesitant to disable the feature, thought it, too, has been mulling some targeting curbs.
But Facebook doesn’t have to make the sacrifice. It would make more sense to reveal exactly how each political ad is targeted — and to cooperate with researchers interested in evaluating the ads’ efficiency. Facebook has the means to deliver messages from such researchers to the target audiences, which would help them recruit subjects for experiments. Google should have done the same instead of introducing drastic curbs that probably won’t do much to raise the level of political discourse, anyway. Policymakers need data, not hype, to make informed decisions on how to regulate modern advertising.