Three long years ago, on a glorious summer day in 2016, I went to Brooklyn to meet the team running Hillary Clinton's presidential campaign. They were based in a hipsterish office space, in a room crammed with brightly coloured posters and T-shirt-wearing millennials.
They seemed ebullient. "We've got this!" one told me, explaining that the Democrats were far ahead of their rivals when it came to using digital tools to fight political campaigns.
Naively, I assumed they were right. After all, I reasoned, Silicon Valley skewed progressive, Obama's team had a vast voter database, and his victories in 2008 and 2012 showed how brilliantly his team had harnessed the power of internet campaigning.
How times change. Last month, I headed to another trendy location — a WeWork office in central Washington, with the requisite cold-brew coffee and exposed pipework — filled with more idealistic young Democrats.
But their confidence has been replaced by underdog defiance. No wonder. Soon after I visited Clinton's Brooklyn HQ in 2016, it became clear that Donald Trump's digital team had quietly built an insurgent campaign, using groups such as Cambridge Analytica to target voters, helping to propel Trump to victory. (Full disclosure: the FT briefly used Cambridge Analytica for a market research project in the past.)
Three years later, their tactics remain deeply controversial. Revelations about the Trump team's use of personal data for microtargeting on social media, sending crafted messages to particular demographics, provoked widespread unease (even though these methods are commonplace in consumer advertising, using data that we all constantly give up in exchange for online services).
The yet more potent criticism, informed by the reporting of investigative journalist Carole Cadwalladr, was that Cambridge Analytica also used furtively garnered Facebook data to shape these messages, or "hack the minds" of American voters, as Christopher Wylie, a former employee who has now turned against the company, puts it.
Cambridge Analytica not only did this for Trump, but worked in more than 60 countries around the world, according to a trove of internal documents posted online by Brittany Kaiser, another former Cambridge Analytica employee turned whistleblower. And as the 2019 documentary The Great Hack shows, these campaigns featured a host of dirty tricks: deliberate dissemination of misinformation; incitement to extremism; voter suppression tactics; the attempted blackmail of politicians. "It may never be possible to have fair and free elections again [without digital reform]," says Karim Amer, co-director of The Great Hack.
READ MORE:
• How Cambridge Analytica's whistleblower became Facebook's unlikely foil
These revelations have prompted regulators to intervene, and Cambridge Analytica has now collapsed. But the controversy has not disappeared. Far from it. With the US election looming, Trump's digital team is gearing up for a new campaign, and his Democratic opponents are faced with an urgent question: should they try to outsmart the Republicans at their own digital game? And if they do ape Trump's tactics, can they avoid losing their souls?
The young Democrats in the Washington WeWork office think the answer is a resounding "yes". They belong to a US$50 million ($75.4m) entity called Acronym, created in 2017 by Tara McGowan, a former journalist who oversaw digital advertising for part of the failed 2016 Democrat campaign.
"Unless Democrats realise that we need to get serious about this digital war, we cannot win," she says, explaining that Acronym's mission is to "build the digital infrastructure needed to power the progressives".
Some Democrats applaud this: David Plouffe, head of Obama's 2008 campaign, has joined the board. But McGowan's tactics may spark controversy: not only is Acronym using the consumer and voter data now regularly sourced by campaigns to send personalised messages, but it is also building quasi "news" operations in swing states such as Wisconsin and Arizona, simulating the local newspapers. These will place Democrat-friendly stories in social media feeds, using cutting-edge ad tech tools to target specific users.
"We have to ask, is [ad tech] what politics is about?" says Tristan Harris, former Google design ethicist, who admires McGowan but worries about the bigger implications of ad tech in politics. "Using data to make predictions and whisper completely different things to every ear, depending on whatever resonates with your lizard brain? That's not democracy."
Either way, the stakes are rising. Michael Bloomberg, the billionaire late entrant into the Democratic primary race, has built his own secretive data group called Hawkfish, which has hired former Facebook employees and is also talking to former Cambridge Analytica employees. It reportedly plans to spend US$100m on the 2020 fight, topping even Trump's campaign.
"The battle is just getting more intense," says Nicco Mele, a Harvard professor — and most voters barely know anything about it.
The story of the 2016 election is often viewed as a case study in sordid skulduggery, with a dash of cold war drama. Quite apart from the revelations about the tactics used by Cambridge Analytica and Trump, it also emerged that Russian groups spread misinformation on social media to undermine democracy too.
However, another way to understand this history is to consider the lessons that business schools teach about competition and disruption: this is also a classic example of how an overconfident incumbent can be disrupted by a nimble, desperate upstart with nothing to lose. "Necessity is the mother of invention," observes Mele. "This was central here."
This story starts in 2008, when the Obama campaign was trying to beat the then-incumbent Republicans. They worked closely with key Silicon Valley players, among them Eric Schmidt of Google, to deploy cyber innovations to reach voters and raise money.
Initially they did this with email. Then they expanded their reach to social media: as Kaiser explains in her autobiography Targeted, an Obama Facebook app interacted with supporters and collected their contact details.
Then, in 2012, Obama's team discovered that there were loopholes on the Facebook system that not only let them access data about their supporters, but their supporters' contacts too (the "social graph" of users). This data could be used to send personalised political messages, something that raised privacy implications about which none of the Democrats seemed too concerned.
"Facebook was surprised we were able to suck out the whole social graph, but they didn't stop us once they realised that was what we were doing," Carol Davidsen, the former head of Obama's digital campaign, admitted in a tweet in 2018. "They [Facebook employees] came to [the] office in the days following election recruiting & were very candid that they allowed us to do things they wouldn't have allowed someone else to do because they were on our side."
The 2008 and 2012 victories filled the Democrat digital team with a sunny sense of self-confidence. However, as so often, success sowed the seeds of failure.
After the Obama victories, Republican donors scrambled to fight back. The Koch brothers used their largesse to fund a digital outfit called i360, with a vast (and swelling) database of voter information.
Meanwhile Robert Mercer, a hedge fund manager and conservative donor, took another route: he teamed up with Steve Bannon, co-founder of the alt-right website Breitbart, and poured money into Cambridge Analytica.
This was initially a small London-based consultancy that emerged soon after the turn of the century to offer psy-ops tools for military and political purposes, primarily in emerging markets (The Great Hack shows that in Trinidad and Tobago, for example, it ran manipulative grassroots campaigns designed to "increase apathy" among young Afro-Caribbeans, to enable a majority-Indian party to win).
Its old Etonian founder, Alexander Nix, was initially uninterested in tech. However, after Schmidt's daughter did an internship at his company, Nix spotted the power of digital data.
Although many of his staff backed leftwing causes, Nix wanted to work with the Republicans. That was partly because his own politics leaned right. But like any business entrepreneur, Nix also responded to the rules of supply and demand, and there was little demand for more data science on the ultraconfident Democrat side.
"The Republicans were where all the business opportunity was," recalls Julian Wheatland, former chief financial officer at Cambridge Analytica.
To pitch for work, Nix needed to catch up with the Democrats. And in 2013 Wylie stumbled on an idea: psychologists at Cambridge university were developing online surveys on the Facebook platform to conduct research. This essentially used the same loophole (which has since been closed) that the Obama team had uncovered in 2012: if one Facebook user answered a survey question, they not only handed over their own data, but data from their entire social graph.
How useful this Facebook data really was is unclear. Wylie says it was used to create psychological profiles on almost 90m voters to unleash highly manipulative, targeted propaganda. "It was a total mindf***," Wylie says. Cambridge Analytica salesmen echoed this at the time. "If we can understand the personality of the electorate we can design a message which really resonates with that person," Jack Hansom, a data scientist for the company, explained at a 2015 conference.
Unsurprisingly, other former Cambridge Analytica staff are now keen to downplay this. "The Facebook stuff didn't really work," says Wheatland, the former CFO.
Critics will dismiss this as special pleading or a desire to avoid legal attacks. But others untainted by scandal agree. "I don't think Trump won because of dark arts," Acronym's McGowan says, suggesting the key advantage that Trump enjoyed was using digital innovation to meet "voters where they were" — that is, online. Or as Harvard's Mele echoes: "I am sceptical that Cambridge Analytica's models did anything which really mattered."
The only thing that is crystal clear, however, is that the slick sales pitch about "psychographic" targeting worked: the group worked for John Bolton, then won the mandate for the primary campaign of senator Ted Cruz, and in 2016, after Cruz dropped out, the Trump team hired Cambridge Analytica too. The company duly expanded its operations in Washington. It also created a dedicated digital Trump campaign team in San Antonio, Texas. The code name was Project Alamo.
Last year I travelled down to Texas to visit the site of Project Alamo. It is hard to imagine a more stark contrast with Clinton's Brooklyn HQ: a nondescript rented corporate office, next to a La-Z-Boy furniture store, by a giant Texan highway where traffic constantly thunders. The unprepossessing site was chosen because it was dirt cheap and because Brad Parscale, Trump's head of digital operations, lived in San Antonio. And it had another crucial advantage: it was so distant from Washington that it was out of sight from the world of mainstream politics. "Almost nobody had any idea what was happening [in San Antonio]," Bannon recalls. "Data science was not something most Republicans understood at all."
That created extraordinary freedom. The digital team was headed by Matt Oczkowski, a bearded whizz-kid who worked for Scott Walker, the arch-conservative Wisconsin governor, before joining Cambridge Analytica. Oczkowski knew he faced fearsome odds.
So in the summer of 2016, his group engaged in wild experimentation, unleashing rapid-fire tests of political messages, using every personalisation and persuasion trick they could dream up. If a tactic worked — say, sending the phrase "Build The Wall!" to social media users who embraced alt-right causes — they immediately replicated that. If not, they dropped it.
Oczkowski was able to do this partly because his scrappy team was free from the rigid bureaucratic divisions that normally plague political campaigns, separating content creation teams, say, from ad buyers.
As underdogs, they were willing to test ideas from any source. Facebook, meanwhile, reckoned it had a duty to help both political parties to use its targeting tools to maximum effect (just as it might do with commercial advertisers), so a Facebook employee called James Barnes sat in the San Antonio office as an "embed" — an embedded consultant. "We had . . . spent years mapping out what is the strategy that we think the ideal candidate would use, using all the products we have, using all of the thinking that we've done," Barnes recently told an Acronym podcast. "We kind of came ready with that playbook. We'd written it."
Another Facebook embed, Tatenda Musapatike, was placed as an adviser to independent expenditure groups on the left. But the Democrats were divided into many different teams with rigid protocols. "There were established ways of doing things and I think Democrats were really, really cautious to change," recalls Musapatike. "[Innovation] was not adopted on the left as it was on the right." Disruption had flipped.
Today Musapatike and Barnes no longer work for Facebook. Instead they have both joined Acronym in Washington, hoping to topple Trump in 2020. Other ex-Facebook staff have also joined the cause, including Chris Cox, the powerful former head of product, who is a key financial backer. Armed with this firepower, McGowan is trying to learn every lesson possible from 2016 to enable Democrats to switch the digital game.
Her group has smashed down many of the bureaucratic structures that impeded campaigns in the past (separating ad-buying and content creation, say). It is gathering consumer and voter data to conduct microtargeting. And perhaps most controversially, it has created a media entity called The Courier in crucial battlegrounds such as Arizona and Virginia, which creates local news content that can be disseminated on social media to counter the rightwing offerings from entities such as Breitbart.
Critics might argue this risks fuelling polarisation, particularly since these media sites do not trumpet their political origins. However, McGowan insists this is the only way to counter Republican "misinformation" in the void created by the decline in independent local news. And the content itself is more wholesome-seeming than that found on rightwing platforms. (Top headline in this week's Virginia Dogwood: "These hospitals are giving their employees relief from Virginia's dismal minimal wage".)
Others are joining the cause too. Bloomberg's Hawkfish operation is expanding fast.
Donors are also funding a rich ecosystem of Democrat digital start-ups. Betsy Hoover and Shomik Dutta, two Democrat stalwarts, have created a Y Combinator style incubator called Higher Ground Labs that has put about US$15m of donor money into 36 progressive digital start-ups.
"If you can deeply understand and predict someone's personality that is a very effective way to do marketing and sales and campaigns," says Dutta. "We have to be careful that we are not creepy and manipulative but if you can appeal to someone's core values that is totally fair game."
Can this innovation swing the pendulum again? "What pushes campaigns to innovate is often necessity," argues McGowan. "The Republicans had more incentive to innovate [in 2016] when Facebook and Google were changing their algorithms and models to monetise attention. But I feel very confident that Democrats are going to be able to use every tool and channel available to them to organise and mobilise in 2020."
However, one problem for the anti-Trump camp in 2020 — as in 2016 — is that many senior Democrats are digital dinosaurs. "The Democrats keep putting their money into television even though their voters don't live there," says Harvard's Mele. "The Democrat candidates have come up in a non-digital culture."
The other — more existential — challenge is that the Cambridge Analytica scandal has left a legacy of fear about tech among Democrats. McGowan insists there are ways to use ad-tech without dirty tricks, and that the Democrats must do so in order to win. "I believe fiercely in the need for regulation in social media," she says. "But it is critical to meet voters where they are — online, on their devices — and we can do this without stooping to the immoral level of Republicans in spreading lies and encouraging voter suppression."
The Bloomberg data team clearly agrees. So does Kaiser. "It is not the technology which is the issue, but how it is used."
But the more intense the 2020 battle becomes, the more anguished the debates around the issue will become. "I am desperate to beat Trump! But does the end justify the means?" laments a luminary of the 2016 Democrat campaign who insisted on anonymity since she feels deeply ambivalent about the merits of ad tech. "I don't want a world where politicians use our personal data. But can anyone win without it?"
Written by: Gillian Tett
© Financial Times