Was it Big Data wot won it? Political campaigning today

Sunday 29 October, 10:0011:30, Frobisher Auditorium 1Tech futures

How could so many people be convinced to vote for Donald Trump? Why did so many Brits vote to leave the EU, despite almost unanimous advice from experts, political leaders and celebrities that we should remain?

One suggestion is the power of Big Data, and specifically the high-tech psychological marketing techniques of a company called Cambridge Analytica. The company has claimed a key role in both the winning campaigns, citing its use of ‘big data and advanced psychographics to grow audiences, identify key influencers, and move people to action’. Cambridge Analytica combines information pulled from social media and various other available records to target messages to individuals. Such targeted campaigning opens up many questions. For example, there is the potential for a candidate or campaign to put out one message to one group of voters and a contradictory message to others. If we don’t know what a party or campaign really stands for, where does that leave democracy? There are also concerns about how our data is being used. In May 2017, the UK information commissioner, Elizabeth Denham, launched an inquiry into the potential misuse of personal data, particularly from social media.

But is some perspective required? Using data to try and swing undecided and persuadable voters to vote one way isn’t unique to these campaigns. Similar techniques (and software) were used by both major UK political parties in the 2015 General Election, and by Barack Obama in his two successful election campaigns. In 2017, Labour’s election campaign used a tool called Promote to allow specific groups of voters to be targeted on Facebook. Others argue that this is just a political version of snake oil. After all, Ted Cruz lost to Donald Trump in the Republican primaries, despite having Cambridge Analytica working on his campaign. There are serious practical problems with targeting messages as accurately as the proponents of such methods claim, too. One insider describes the aim as not so much to sway millions of voters as ‘to find tiny slivers of influence that can tip an election’. Perhaps the people who are really getting conned are not voters, but the campaigns themselves.

Indeed, how important were data techniques to Trump’s victory? The outcome was close – in fact, Clinton won the popular vote – so the final result came down to relatively small numbers of votes in ‘swing’ states. It seems plausible that such techniques were powerful enough to influence the decisions of enough voters to hand the White House to Trump. Yet others argue that more important factors were at play. Despite claims of enormous blue-collar support for Trump, the big story of the campaign was Clinton’s inability to hang on to the voters won over by Barack Obama. Many argue Trump’s victory was down to a lacklustre Democrat campaign that took for granted voters in rural areas and ‘flyover’ states. The big story of the Brexit vote appears similar: voters who had been taken for granted for years took the opportunity to kick back at an out-of-touch establishment, in Westminster as much as Brussels.

So to what extent is it true that Big Data methods can swing important votes? What are the implications of this approach for privacy and democracy? Does the assumption that a few targeted messages can swing an election actually reveal a dim view of voters?