The Fake News About Cambridge Analytica

The analytics firm likely didn’t swing any elections. But that doesn’t mean our personal information is safe

Illustration of Social Media Icons Tied to a Puppet-master Hand
The Walrus
The Walrus
The Walrus

In recent weeks, you would have been hard pressed to open your social media, turn on a news channel, or pick up a newspaper without seeing ample mention of Cambridge Analytica, the firm reputed to have turned the tide on the US presidential election, the Brexit campaign, and untold other races, all with the help of a trove of data pilfered from Facebook.

Most infamously, it deployed a Facebook personality quiz in order to gauge voter’s likes, dislikes, and political affiliation. That is, if a survey respondent was a “leader,” liked Hello Kitty, and preferred trade-skeptical Democrats from the Rust Belt, the firm could deliver highly targeted ads to users who bore similarities. They turned and weaponized that data, according to whistle-blower Christopher Wylie, to shift elections in the favour of their right-wing clients.

It’s a fantastic story—a shadowy organization inched us closer to a Black Mirror–like dystopia without anyone ever knowing it was happening. There’s a core problem with this tale: Cambridge Analytica isn’t that special. It is, at best, a serviceable advertising firm that lucked into some raw data. It is not an omniscient political machine. It has not clinched any election through predicting and influencing voters. And it does not have a “secret sauce” that gives it a window into the motivation of the masses. In fact, the hysteria over Cambridge Analytica may be the firm’s most successful bit of psy-ops yet.

But if Cambridge Analytica’s special powers turn out to be fiction, that doesn’t mean we shouldn’t be furious. On the contrary, we ought to widen the scope of our outrage.

I first started looking into Cambridge Analytica last year, asking around on both sides of the border about the firm that, according to some media outlets, flipped the results of both the US presidential race and the vote to pull the UK from the European Union. While almost nobody, save Wylie, is clamouring to go public, I’ve gotten a good sense of the ocean that these fish swim in.

Not that the company ever really managed to stay under wraps in the first place. In 2017, the Observer billed Cambridge Analytica’s aid to the Brexit campaign as “the great British Brexit robbery.” Motherboard billed the firm’s information trove “the data that turned the world upside down.” French investigative website Mediapart called the company “le Big Brother electoral.”

Separating the truth from the bluster is tough. Cambridge Analytica reportedly began in 2013 as the red-headed stepchild of the SCL Group, a UK behavioural-research company that was primarily concerned with top-secret Pentagon work. While its parent company traded on secrecy, Cambridge Analytica built itself up as an election consultancy firm and made a go of creating its own infamy. It promised to be able to get into voters’ heads, identifying persuadable citizens with its powerful database and top-tier advertising acumen. What’s more, it promised to craft ads that could nudge undecided voters in the right direction. It was, for Republican and right-wing causes, supposed to be a breakthrough moment in their digital arms race against the Democrats. For years, the Democrats had eaten the GOP’s lunch on high-tech campaigning. By 2014, companies such as Cambridge Analytica promised to close the gap. Nationalism and conservatism were finally breaking into the second decade of the twenty-first century.

But, even then, Cambridge Analytica’s claims attracted skepticism. There were already many multi-million dollar firms with impressive voter registries, capable of serving ads to social-media users with increasingly impressive levels of accuracy. The firm’s claim to be above the rest, at least dating back a couple of years, was already falling apart.

In a seminal 2017 feature on the company’s less-than-illustrious electoral track record, Mother Jones went through the list of American races that the firm jumped into. There was the 2013 gubernatorial race where the firm’s advanced data analysts were asked to spit out a list of soft Democrats and turned over a list of firm Republicans instead. There was one race where its English staff didn’t know what a precinct was. And there was its work on the failed nomination races of both Ted Cruz and Ben Carson (though no data wizardry in the world could’ve saved the latter candidate).

But, sometimes, Cambridge Analytica got it right, ensuring its ability to sell itself to the next campaign. It threw its weight behind Republican Tom Cotton in his 2014 Senate race, and he won with a commanding 56 percent, unseating the Democratic incumbent.

And yet, even then, the company wasn’t even the biggest player on that multi-million-dollar campaign. Cotton paid the firm just $20,000 for “data consulting,” according to election finance records. Voter contact and data firm FLS Connect was paid ten times as much. The campaign also paid thousands to data firm i360, a firm whose ubiquitousness in the world of Republican politics dwarfs the scrappy UK company. As Time wrote in 2016, i360 has “hundreds, if not thousands, of pieces of information on each and every American, gleaned through documents [the] team picks up from courthouses, auto dealerships and frequent shopper programs. No magazine subscription is too insignificant, no online purchase too small to note.” Its funding comes directly from infamous Republican mega-benefactors the Koch brothers, and the firm has a working relationship with the Republican party.

Cambridge Analytica, it’s true, has its own deep-pocket endorsers, such as the Mercer family, which also bankrolls far-right efforts including Breitbart News. The Mercers installed Steve Bannon onto the Cambridge Analytica board of directors to act as their surrogate. But the point here is that Cambridge Analytica was not the only firm that thought to scrape data from Facebook. And it is not the only firm with a database on American voters. If anything, other firms—such as i360—likely had significantly more advanced and impressive databases of the electorate and its likes, dislikes, habits, and so on.

But mixed reviews from both the Cruz and Carson camps didn’t stop the Trump campaign from hiring Cambridge Analytica anyway, for just shy of $6 million (US). The firm deployed the the targeted ads strategy that has now become synonymous with the Trump campaign. Pro-life Democrats might be served up Youtube video ads about Hillary Clinton’s support for Planned Parenthood. Blue-collar union members’ Instagram pages might be blanketed with details of the Democrats’ open-door immigration policies. And so on.

There’s no question that $6 million is a step up from the paltry fee Cambridge Analytica got from Cotton’s campaign, but it’s also a bit misleading. That money would have included the cost to actually pay for the ads. But, even still, Cambridge Analytica won that gig (or, perhaps more accurately, was set up with that gig thanks to help from Mercer) because the company, and its charismatic CEO, Alexander Nix, bragged of a “secret sauce” of data that would make its ads better, its targeting more laser guided, and its results all the more pronounced.

A big part of that sell was Cambridge Analytica’s semi-secret Facebook-scraping operation, conducted with the help of Aleksandr Kogan, a University of Cambridge psychology lecturer. He described the process he used to obtain the data in a letter obtained by The Outline. “Through the app, we collected public demographic details about each user (name, location, age, gender), and their page likes (e.g., the lady gaga page). We collected the same data about their friends whose security settings allowed for their friends to share their data through apps,” he wrote.

That process provided Cambridge Analytica with some 30 million Americans’ “predicted personality scores.” But, Kogan wrote, “the predictions themselves are actually not useful for micro-targeting.” He continued: “In fact, from our subsequent research on the topic, we found out that the predictions we gave [Cambridge Analytica] were 6 times more likely to get all 5 of a person’s personality traits wrong as it was to get them all correct.”

He concluded that “even if the data was used by a campaign for micro-targeting, it could realistically only hurt their efforts.” Indeed, another source who spoke to me said that Cambridge Analytica came to the same conclusion as Kogan pretty quickly and that the “predictive personality scores” were never used in any real election campaign.

The data scraped through this quiz—your Facebook likes and any public profile data—was, however, very useful. But Cambridge Analytica was hardly the only company collecting that data. Some people have come out publicly in recent weeks to admit that they did similar data collection. Also worth mentioning is that this data was dated to 2014 (Facebook put a stop to such scraping the same year), meaning it became dated pretty quickly. In the big-data world, up-to-date intel is crucial.

Given this, it’s curious to watch ex-employee (and ex–Liberal Party of Canada staffer) Wylie do the media rounds to spin his former employer’s work—indeed, his work—as “cyberwarfare for elections” or a “dirty MI6.” Because, for all of their bragging about weaponizing sex workers as political props and offering bribes, there is little publicly available evidence that Cambridge Analytica was paid to do anything more than come up with likely voter models for undecided voters and deliver ads on social media.

One wonders if Cambridge Analytica’s PR coup, selling themselves as the home of evil geniuses, may be bigger than anything they’ve delivered to their clients.

Much of the public reaction to Cambridge Analytica has taken the tone of shock—shock that political firms are maintaining complex databases on our likes, dislikes, pressure points, and political positions in order to deliver us tailored ads and polemics.

Indeed, the grievances against Cambridge Analytica, and its alleged nefariousness, are laid out in a lawsuit filed in an Illinois court on March 23 by a state attorney—one of several filed in the country. The company, the lawsuit alleges, “exfiltrated the personal data of more than 50 million Facebook users in the United States, including millions of users in Illinois. This data trove included Facebook users’ ages, interests, pages they’ve liked, groups they belong to, physical locations, political affiliation, religious affiliation, relationships, and photos, as well as their full names, phone numbers, email addresses, and physical addresses.” That data, it goes on, “was supposedly private.” (It later notes that “this kind of mass data collection was not only allowed but encouraged by Facebook.”)

The litigation argues that, by convincing 270,000 Americans to download and participate in a personality quiz—scraping their Facebook data and that of their friends and, in the process, collecting information from some 50 million pages—Cambridge Analytica had disregarded and misused “sensitive, personal data” of those millions.

The lawsuit, which also targets Facebook, effectively describes the basic business model for scores of firms which deal and trade in big data. Without passing judgement on whether it’s right or wrong (and, perhaps more relevantly, actionable or not) it is hardly new or unique to Cambridge Analytica.

The shock and betrayal has also mixed with a need to rationalize Trump’s victory. Some have comforted themselves with the conclusion that Russian meddling was so widespread and systematic that it flipped the results in a number of states. Others now want to turn to the Cambridge Analytica scandal as evidence that it was all a high-tech plot. Few have been willing to face reality: Trump’s economic and racial nationalism hit home with millions, delivering him the electoral college and thus the presidency.

The same sort of public derangement also explains the intense attention paid to Cambridge Analytica’s role in the Brexit campaign. Amongst Wylie’s litany of explosive allegations was, as the Guardian reported, that there was a “common plan” to avoid election spending laws in the Brexit referendum. It’s been a familiar refrain from the paper and its sister publication, the Observer, and is now the subject of a probe by UK election authorities. And yet, how would Wylie know? He left Cambridge Analytica in 2014. (To his credit, Wylie did say that he was “absolutely convinced” there was such a plan, not that he knew first-hand—a fact omitted from the Guardian headline.)

It’s unclear whether Cambridge Analytica was formally involved in the Brexit campaign. One of the firms that Leave campaigns allegedly used to sidestep campaign-spending limits was British Columbia’s AggregateIQ. For a time, AggregateIQ was advertised as “SCL Canada,” allegedly functioning as the Canadian wing of Cambridge Analytica. That designation, however, was dropped in 2017. Plans to have the UK company take over the Canadian firm were, apparently, unsuccessful. The two have severed ties.

While Cambridge Analytica promised to be a company who could sway elections through its well-honed magic, AggregateIQ was more involved in building tools. Some of those tools recently leaked online. They are more conventional, albeit seemingly advanced, applications which can help measure and improve the impact of online advertising and help campaigns organize and direct their efforts.

For example, AggregateIQ systems could track, pixel-by-pixel, how your mouse hovers over the ad, using that data to inform how to tailor its next set of ads. If analysis of the firm’s leaked system files are correct, Ripon—the software that it built—gave phone canvassers and volunteers access to databases built with voters’ personal information that may have been scraped without their knowledge by other companies. But that is also standard operating practice in this world.

Cambridge Analytica may be a perfect caricature of our fears regarding big data in politics, but it is important that it not become the only focus of our newly awakened concern with the privacy-indifferent nature of the current political game. The reality is that regulation and legislation have not caught up with technology. Governments have been embarrassed by massive data breaches, the weaponization of social media by foreign governments, and the wholesale indifference to citizens’ privacy on the part of these massive technology companies. And despite that slapstick incompetence, governments have responded largely with inaction.

The European Union has made leaps forward to empower citizens with their privacy rights—regulations set to come online in May could slap fines on tech companies, including Facebook, that collect users’ data without permission or abuse data they do collect—while legislators in Ottawa and DC have simply continued to talk around the problem. Canada and America have both threatened to slap new regulations on Facebook, but the plans have amounted to little more than hot air at this point.

The media has worked itself into a lather speculating which laws have been broken by Cambridge Analytica, but it is likely that everything it has done over the course of its political life has been specifically legal. Or, at least, not illegal.

Yes, Cambridge Analytica exploited its Facebook app to collect data on 50 million Facebook users. But it is just one of tens of thousands of apps with that power. What of the makers of Candy Crush? Or Farmville? Their apps had many more users and thus had access to untold amounts of data.

The thrilling conspiracy that has been affixed to this scandal—one that involves Bannon, the Russian government, and seedy data scientists—masks the banality of the problem. Cambridge Analytica was just a small fish in murky waters.

Justin Ling
Justin Ling is an investigative journalist and host of the podcast Uncover: The Village.