The role of technology in the presidential election

EARLY in America’s presidential campaign, pundits compared the contest between Hillary Clinton and Donald Trump to a fight between a large tanker and Somali pirates. This turned out to be particularly true of the digital campaigns: a massive data battleship lost to a chaotic flotilla of social-media speedboats. The big question now is what this means for future elections, both in America and abroad.

Mrs Clinton’s campaign was arguably the most data-driven in American history. Her organisation employed dozens of data scientists who designed statistical models that determined, for instance, which Democrat-leaning voters should get a knock on the door from a friendly volunteer to remind them to do their civic duty. The campaign’s master programme ran six times as many simulations a day as the one that helped Barack Obama get re-elected in 2012.

That may have bred complacency. Her campaign may have had “too much faith in what the data told them,” says Daniel Kreiss, a campaign and data expert at the University of North Carolina. In Michigan, which had voted for the Democratic presidential candidate since 1992, Mrs Clinton was caught flat-footed. Mr Trump had visited the state frequently, despite polls that showed him far behind. Only at the last minute was the vulnerability discovered, and Mr Obama hurriedly dispatched to stanch the bleeding. But it was too late.

All statistical models are premised on a set of assumptions. When these are not met, the resulting estimates can be “biased”, statistician-speak for wrong. Blunter members of the profession put it another way: “Garbage in, garbage out.” Two senior advisers to Bernie Sanders, whom Mrs Clinton defeated in the primaries, have argued that her campaign may have inadvertently turned out voters who were likely to cast their vote for her rival.

By contrast, Mr Trump dismissed data as “overrated”. Only in June did his campaign hire Cambridge Analytica, a data firm based in London. After Mr Trump had secured the Republican nomination in July, he was also able to use the party’s digital operations, but these are considered less sophisticated than the Democratic ones.

Mr Trump’s data team seems to have helped to identify a path to victory. Early voting and absentee ballots showed a large increase in rural voters, explains Matt Oszkowski of Cambridge Analytica. The Trump campaign redoubled its efforts to turn out rural voters, betting—correctly, it turned out—that this would erase polling deficits in traditionally Democrat-leaning states.

But if any technology helped Mr Trump win, it was social media rather than data science. He and Steve Bannon, his controversial campaign boss and now his chief strategist, understood how the Facebook and Twitter-driven media landscape worked. Whether a piece of news spreads online does not depend on whether it is true and coherent, but whether it is surprising, shocking and confirms prejudices. It can bounce endlessly in virtual echo-chambers—even if it is patently false.

This phenomenon is particularly pronounced on Facebook, whose algorithms are tuned to maximise “engagement”, meaning they present users with the type of content they have already been shown to like in order to keep them on the site as long as possible and comment on it or share it. “Trump and his camp essentially hacked Facebook’s algorithm,” says Matthew Hindman of George Washington University. Hundreds of online publications spewed out pro-Trump propaganda, as well as a steady stream of fake news. Left-wing sites published reams of the stuff too, but their output is estimated to have been lower in volume.

Mr Trump’s most valuable media tool was his Twitter feed, currently followed by 15.4m people. It brought him attention and billions of dollars in free advertisement when mainstream media picked up his often outrageous tweets (see chart).

How much all this really mattered in this campaign is already the subject of debate. Some argue that Mrs Clinton’s loss proves the danger of relying too much on data and statistical methods. “We need to throw out some of the science and focus more on the art,” says Scott Goodstein, who runs Revolution Messaging, a left-wing campaigning outfit, and was in charge of the digital side of Mr Sanders’s campaign. It also made much use of technology, but often with a human touch. One important tool was Hustle, a messaging app, which allows volunteers to quickly organise rallies by sending them personal texts.

The role of social media is the focus of most attention. Pressure is mounting on Facebook to change its algorithm to filter out fake news. Initially, the firm was in denial. Its boss, Mark Zuckerberg, at a conference after the elections, called the idea that fake news had an effect on the elections “pretty crazy”. But this position seems otherworldly: nearly half of American adults now get their political news on Facebook.

The firm has since shifted its stance somewhat. When Google, another online giant, said on November 14th that it would bar fake-news sites from using its advertising services, Facebook quickly followed suit. (A day later Twitter suspended the accounts of personalities of the “alt-right”, as America’s new extreme right-wing movement is called.) On November 19th, in a detailed blog post, Mr Zuckerberg explained how the social network could deal with the problem. Options include making it easier for people to report stories as fake and relying on respected fact-checking organisations. But he insisted that Facebook itself does not want to decide whether something is fake or not. “We do not want to be arbiters of truth ourselves, but instead rely on our community and trusted third parties,” he wrote.

Mr Trump’s success on social media may be the result of a set of unique circumstances, such as his personality and a rival who was particularly vulnerable to attack. But in Britain’s Brexit referendum, too, social media played an important role in spreading half-truths and outright lies. The French presidential elections in April, which will probably see the candidate of the extreme-right National Front, Marine Le Pen, reach the second round, will be the next test case to see how influential the big online platforms are. They have long hidden behind the argument that they are mere technology companies. After Mr Trump’s victory this position, never persuasive, rings even more hollow.