One of the biggest news stories about the Internet in Britain and America right now concerns Facebook, a company called Cambridge Analytica, and the impact that they might have had on democratic decisions. Two decisions in particular: the narrow vote in Britain to leave the European Union, and the election of Donald Trump.
This story’s about many things: the power that data gives to data companies like Facebook; the potential that it gives those companies and others to influence opinion (not just which film to watch but who to vote for); the impact of new media on old, on politics and populism.
Changing views about the Internet and politics
The story about Cambridge Analytica’s important too because it reflects a shift in thinking about the Internet’s impact on politics. Most Internet insiders used to think it would empower citizens and give them greater influence over decision-making at the expense of those with power (and money). Now, there’s more concern that it’s giving those with power (and money) more opportunities to manipulate opinion and decisions in their favour.
What’s the story?
The story’s quite contentious, so we must be careful. Two points are crucial. A small company called Cambridge Analytica got hold of a lot of data on Facebook users: at least 87 million of them. It’s used these and other data to target political messages on those its evidence suggested were most likely to support the party, candidate or cause that paid it for promotion – to reinforce their views and reinforce their likelihood to act (or vote) upon them.
One part of this story is concerned with data management and privacy; another with the integrity of politics. This week I’ll focus on the latter; next week on wider implications.
The core is ‘psychographics’
The process here’s called ‘psychographics’, and it depends on data. The insight’s that individuals’ behaviour, and the personality traits that it reveals, are strong predictors of their wider worldview. Politicians can use this, like advertisers, to ‘micro-target’ messages to individuals or small groups of people in ways that will influence their behaviour, including how they vote as well as how they shop.
‘Psychographics’ isn’t harmful in itself. Health professionals might use it, for example, to promote safe sex or discourage smoking, reinforcing what is called ‘nudge theory’. But there’s a fine line here between influencing people’s behaviour and manipulating them.
I’m not going to comment directly on Cambridge Analytica or anything that it might or might not have done (though It’s worth noting that it’s been at work not just in Britain and America but also in developing countries including Kenya and Nigeria). I want to raise instead some general questions about what micro-targeting might do to the way in which politics occurs in democratic systems.
My thanks, along with all who comment on this story, by the way, to the exceptional reporting of Carole Cadwalladr and colleagues on The Guardian and The Observer, who broke the story and have followed it for more than a year. Good old fashioned investigative journalism, of a kind few newspapers can now afford.
Five impacts on politics
So how do social media in general, and psychographics in particular, change the ways of politics. After all, political activists have always sought to influence the ways that people think and vote. Social media’s just another vehicle for this, you might say; digital data just another step along the road from the notepads that canvassers have always carried.
What’s new here, and why might it be troublingly transformative. I’ll raise six issues that have emerged as this story’s gone on.
Today’s targeting is more effective …
First, targeting’s more personal, more precise and more effective. People curate their lives online these days, and most voters in most democratic countries use just one platform, Facebook, to do that. The amount of data available to those who’re using Facebook – 98 data points about each individual – far exceeds anything that advertisers or politicians previously had available; and they now use computer power to do their targeting. Messages get maximised for individuals rather than for whole communities.
… and more manipulative …
Second, it’s more manipulative. As a leading former politician from my country put it in a recent lecture, ‘People vote with their hearts, not with their heads.’ Few voters spend much time thinking about politics or reading up on different policy options. What drives them in the polling booth is gut reaction, emotion, who they think’s most likely to do what’s in their interests.
Voters’ decisions have always been subject to a lot of influences – peer groups, employers, unions, religious leaders, others. Old media too (‘It’s The Sun Wot Won It’ said one British newspaper of the election in which I was a candidate in 1992).
It’s easier to influence (or manipulate) people’s emotions than it is to convince with facts and figures. And so: four reasons why that’s also easier on new media than old.
… but less transparent
Point one. Online messages are subject to much less scrutiny than those offline. In the past, political advertising and campaigning were transparent: advertisements and leaflets were published for all to see, which did a lot to keep what they had to say in bounds.
There’s no such transparency in targeted messages, because they’re seen only by the individuals they’re sent to, not – crucially – by political opponents too. That makes it much easier to lie and smear, to abuse and sneer, to exploit racial and other community divisions. Who’ll get to know?
Point two. Attitudes to truth have changed. Democratic societies, Christopher Wylie, Cambridge Analytica whistleblower, puts it, ‘need a common understanding of reality.’ Populist politicians today are inclined to shout ‘fake news’ at anything, whether true or not. The ‘truth’ for their followers lies not in evidence but in who is saying something. You’re either with Fox News or you’re against it.
And the revenue-earning model of social media platforms, sadly, favours fear and anger over hope and compassion, outrage over accuracy. It’s the outrageous post that generates most clicks and most clicks that generate most revenue. Populist politicians have fed on this. The frontier between fact and fiction’s likely to be distorted further by fake videos which could become widespread within the next few years.
Point three. Much of this may be down to algorithms, but algorithms are being and will be manipulated by powerful actors, especially if they’ve access to the data that is used in psychographics. Those powerful actors may include traditional political parties and interest groups, but may also include foreign governments and criminals.
Point four. Expenditure controls have lost their meaning in elections. Most democracies have limited campaign expenditure in order to create a level playing field – to stop elections being bought by those who could pay for blanket advertising, for example. That could be monitored when campaigns were conducted offline and subject to transparency. It’s proved almost impossible to follow what money’s being spent and where online.
As often in these posts, I’m saying here that this is complex and that binary approaches to it are not going to work. I’ll end with three big problems that it raises.
First, big changes are taking place within the public sphere. We’d expected the Internet and social media would make politics more participative, more like a public meeting. Instead, we’ve seen more polarisation and aggression arise as social media’s displaced old media and public discourse. It’s opinions, rather than debates, that have been reinforced.
Second, elections are usually won and lost on small changes in voting, not on big ones. Micro-targetting’s especially effective in constituency systems. Donald Trump won three million fewer votes than Hilary Clinton in 2016, but he won the crucial states that made him president by tiny margins. His campaign focused on influencing critical voters in those states.
Third, the ways of democratic politics have changed in this age of social media. In the last few years, results suggest that's favoured populists who’ve exploited ways to trade on nationalist emotions. Liberal democrats need to find a way of counteracting this if the pendulum is to swing back.
Election laws clearly need updating to meet new circumstances in the digital age, or they’ll be bypassed. New ways of engaging citizens need to be found. But platforms also need to recognise that the power and the data that they wield can be and have been abused. This requires new thinking on their part too about the responsibility that comes with power, rather than a weary shaking of the head.
Next week: some wider implications for Facebook, data protection and the Internet in general.
Image credit: A protest following the Cambridge Analytics and Facebook data scandal with Christopher Wylie and Shahmir Sanni. By Jwslubbock, via Wikimedia Commons