How can we explain the new populism, this rise of social media-empowered, elite-bashing, partisanship-promoting politics? Is it due to economic inequality, the revenge of the “left behind”? Or is it down to culture, the “localists” pushing back against the “globalists”? Is it all because of Neo-liberalism? Neo-Nationalism? Old-fashioned racism? Every time someone proposes a single solution (The economically disadvantaged voted Trump!) someone else comes along to show how the model doesn’t fit (The rich voted Trump too! The poorest voted Democrat!).

As Francis Fukuyama expertly laid out in these pages, all of the above are relevant. But there might be another angle from which to view the phenomenon, one that looks at it from the perspective of the campaign manager—a perspective that sees populism not so much as a result of socio-economic forces, but as a strategy for success.

Consider the landscape from the point of view of the spin doctor. Increasingly you need to target voters split up into little causes on social media. They have vastly different interests: animal rights and hospitals, guns and gardening, immigration and parenting. Some of these interests might be overtly political, while others are private. Your aim is reach out to these different groups in completely different ways, tying the voting behavior you want to what they care about most. Animal rights groups need to be persuaded that your campaign has some sort of relation to their cause; anti-immigration activists that it relates to theirs. So, for example, if you were running a campaign for exiting the European Union, you would argue to animal rights groups that the EU is bad for animal welfare due to EU regulations, which demand that animals be cruelly transported across long distances before they are slaughtered, or because the EU supports farmers who raise bulls for bullfighting. Meanwhile, you would target anti-immigrant groups with a message about how the EU enables migration.

The animal rights supporters may actually have a very different stance on immigration—they may well be for it—but that doesn’t matter as you are sending different, targeted ads to various groups which the others won’t see.

You can even avoid using your main candidate in ads to audiences who don’t like him. During Donald Trump’s election campaign, ads targeted at some middle-class voters conspicuously avoided showing the main man, and focused instead on touchy-feely messages quite out of sync with Trump’s vitriol. (Jamie Bartlett’s BBC documentary about the dangers Silicon Valley poses to democracy is a must-see.)

This sort of hyper-targeting, where one set of your voters shouldn’t know about the others, means you need some big, empty identity to unite all these different groups, something so broad these voters can project themselves onto it—a category like “the people,” or “the many.” The “populism” which is thus created is not a sign of the people coming together in a great groundswell of unity, but a consequence of the “people” being more fractured than ever, of their barely existing as one nation. To seal this improvised identity one needs an enemy. Best to keep it, too, as abstract as possible so anyone can invent their own version of what it means: the “establishment” will do, or “elites.”  Of course this means you’ll lose some voters who are the said “enemy,” but if you keep the definition of the enemy elastic enough, people will be able to define their own version of who exactly the enemy is.

One could even imagine a situation where one group of voters you have seduced chant about some awful “elite,” while in truth the “enemy” they have in mind is another group you have persuaded to vote for you. The Labour Party in the UK is pulling this off rather well. By defining themselves as being “For The Many, Not The Few” they have created a semantic space wherein metropolitan middle-class types vote Labour considering themselves as opposing the “Few” super-rich and landed gentry, while Labour voters in the north of England see themselves as “the Many” opposing the privileged middle-class metropolitan “few.”

This approach is the opposite of “centrism,” which tried to find a pot of common policies to bring different camps together under one big tent, and to reduce polarization. That made sense when campaigns played out on a limited number of TV channels that people of different political persuasions watched together—when you had to engage in debate and to win many over simultaneously. Now your job is to fire up little disparate groups, to energize and polarize them as much as possible, not to win a rational argument in a disappearing common public space. This pattern is most advanced in America, where cable TV began the fracturing work that social media is now augmenting, with its inhabitants living in parallel informational worlds of alternative facts. But even in countries less far gone, one can see emerging signs of a similar future. Researchers from Oxford University have recently shown how Japan’s President Abe preached one centrist message on TV about economics, while pursuing a hidden campaign online which reached out to his right wing. The combination of the two helped him to victory in 2014.

If the secret to a successful campaign is to speak in different tongues to different groups, then it doesn’t make sense to try to defeat the data populists by choosing one message oneself. The debate among U.S. anti-Trumpers, about whether a “centrist” or “economically radical left-wing” platform is best suited to defeat the current regime is already a case of asking the wrong question. Instead, they should be asking how you create a campaign liquid enough to reach out to centrist dads, millennial economic radicals, the “left behind,” and whoever else you need to win.

This was the mistake the British Tory party made in the general election immediately following Brexit. Misinterpreting the referendum result as simply being about immigration, they over-focused their subsequent campaign on nationalism. It flopped as it ignored all those other causes, from animal rights to saving the health service, which the original Brexit campaign had skillfully played on. Labour, on the other hand, have kept their policies vague, attracting Europhiles with one message, while keeping their anti-European base in the north. I remember being struck by the use of gerunds in their election campaign literature, which featured phrases like “creating an economy which works for everyone,” “leading richer lives,” and “extending democracy.” Gerunds can be a sign that the author doesn’t quite know, or doesn’t want to reveal, what they really think. Many editors don’t like them. But in this case they were just what was needed. Gerunds, it turns out, don’t always decline.

But before one rushes into whole-heartedly copy-catting populism as a strategy, one should consider its fatal flaw. Having sold itself in such different ways to different groups, it struggles when it comes to actually governing. In Britain the first party to use social media effectively in a campaign was the Scottish Nationalists, who managed to market themselves in different ways to the cosmopolitan bourgeoisie (won over by the idea that Scottish nationalism meant being European and England was holding that back), “little Scotland” nationalists (inspired by anti-foreigner sentiment), and the economically deprived (convinced that England was stealing Scotland’s wealth). When it came to ruling, that unity began to fracture: finding one tax policy to fit all, for example, is hard. Similarly, Donald Trump has struggled to pass policies, and the ones he has passed, like the massive tax cut, barely register with the “left behind.” In Italy the digital-populist movement par excellence, Five Star, sold itself as addressing concerns ranging from immigration to the environment to basic income, all the fault of “the elite,” but is struggling to make those promises a reality since gaining power in regional governments.

That’s the thing about this digital “populism”: it’s great at campaigning, less good at creating any sort of sustainable politics. It’s of no surprise that those who have used it best continue in an ersatz campaign mode once they have won, constantly searching for new enemies (“Remoaners” for Brexiteers in the UK, “the Swamp” for Trump) to maintain coherence. Instead of arguments about ideology, policy, or a vision of the future, the challenge for the digital populist is to create mirages of membranes, imaginary walls that don’t actually have anything on the other, “enemy,” side, but which can, for the length of a few tweets at least, give their disparate followers the sense that their interests are aligned. Thus conspiracy theories become their favored idiom. As the Bulgarian political scientist Ivan Krastev has pointed out, you can’t argue and engage with a conspiracy the way you can with a policy or ideology; you have to be either on the side of people who believe in it, or on the side of who don’t.

And of course, social media itself is about creating soap bubbles of identity rather than engaging in ideas. And like so much on social media, digital populism is a phenomenon full of sudden surges and seeming meaning, that dissolves as soon as you try to reach out and find something real.