Any effort to combat disinformation on social media through regulation will inevitably revive arguments about free speech, its limits and how digital technology should serve the public interest. What often gets overlooked is that disinformation is as old as human conflict, and the media environment, as a whole, is a willing participant.
Ève Gaumond reminded us of this during a recent podcast interview with CBA National. Gaumond, a researcher on AI and digital technologies at Laval University, cited Harvard’s Yochai Benkler, whose research shows that mainstream news outlets and cable news, not social media, are primarily responsible for the disinformation crisis. Take the fabricated accusations of voter fraud during the 2020 presidential election, driven primarily by Donald Trump, and then mainly circulated through mass media. "Social media only played a secondary role," says Gaumond. Similarly, in the last Canadian federal election, when a video tweeted by Deputy Prime Minister Chrystia Freeland was labelled "manipulated media" by Twitter, most people heard about it through traditional media, according to the Canadian Election Disinformation Project.
"It's what Benkler called the feedback loop," said Gaumond on the podcast. "So, the news arises from mainstream media or a politician and then it reverberates into social media, but it doesn't start from there."
Still, social media is a formidable tool that allows candidates, through micro-targeted ads, to tailor their messages to different audiences without them knowing about it. In the media environment that dominated most of the 20th century, pleasing different audiences was a far more subtle art because everyone could hear a candidate's speeches broadcast on the news or read them in print.
Laws and regulations imposing campaign spending limits on advertising also predate the emergence of social media. As law professor Michael Pal has argued, the low cost of online advertising has undermined the effectiveness of those rules. "The total amount of spending permitted by the limits, particularly for political parties, may be so large as to no longer provide a meaningful restriction in relation to social media," he wrote last year.
Pal proposes some reforms to correct the imbalance, including a lower ceiling for social media spending. He also wants enhanced disclosure rules around online advertising by political parties, candidates, and third parties. These should go beyond existing requirements introduced in 2018 under the Canada Elections Act that impose an obligation on certain large online platforms to publish a digital registry of all regulated ads, including information about who authorizes them. Pal would also see to it that we have better transparency on how viewers are micro-targeted, and which search terms are used to target voters. In a 2020 interview with Michael Geist, he even suggested limiting other objectionable practices, such as targeting racist people with political ads.
Gaumond supports some of these policy recommendations in large part because they would help us enforce laws already on the books. Transparency, she explains, "is a way to understand the problem properly and see how our actions are working or not." She cites as an example the prohibition against foreign actors paying for political ads in Canada. "If we have no means to see a […] microtargeted ad paid for by a foreign actor, then we cannot sanction it," she says.
But more transparency is hardly a cure-all, argues Léonid Sirota, a law professor in public law who is about to take up a position at Reading Law School in the UK. "The harm isn't in transparency itself, but in the costs that its pursuit imposes," he wrote in an email to CBA National. "Canadians tend to talk as if disclosure requirements are a democratic free lunch, but […] in the context of registration and disclosure requirements imposed on 'third parties', compliance can be costly, and these costs are a strong deterrent to engaging in political speech." Case in point: Google banned political advertising on its platforms ahead of the 2019 Canadian federal election because of the ad transparency rules passed the year before. What's to stop other tech behemoths from packing it in?
Transparency may not entirely repair our fragmented politics, but at least it's a feasible policy option compared to other measures being bandied about, like banning surveillance advertising altogether. The latter is an idea gathering steam as calls grow to reform major platforms like Facebook and Google. The European Union's lead data protection supervisor, Wojciech Wiewiorówski, is pushing for EU lawmakers to phase out targeted advertising tied to pervasive tracking. But the European Commission has done little more than conduct public consultations on the issue. A U.S. congresswoman has also reintroduced legislation (supported by Benkler) to prevent large online platforms from allowing advertisers to target messages based on their users' demographic or behavioral data. As the bill tries to make it through Congress, expect plenty of pushback that it amounts to unconstitutional regulation of political speech (proponents will argue that it’s just data).
There are at least two good reasons why we should give such a ban careful consideration in Canada. For starters, eliminating micro-targeted political ads could put well-established parties in a better position to blanket the airwaves and the internet with ads designed for mass consumption. Newcomers and independent groups who rely on targeted fundraising appeals to challenge the status quo are more likely to get crushed. "[As] a public choice devotee," says Sirota, "I have to wonder whether the silencing of third parties ― and even the channelling of parties' advertising into expensive channels that the big ones can afford much more easily than their competitors ― doesn't rather suit the [Liberals and Conservatives], one or the other of whom is always writing the rules."
Second, a ban on political microtargeting is probably unconstitutional in Canada as well, according to Gaumond. Writing for Lawfare, she's taken the position that a ban probably wouldn't "make it past the third step of the Oakes test, the one requiring minimal impairment to the right in question." Measures aimed at preventing discriminatory practices stand a better chance of success, she added.
For his part, Sirota questions whether that is even worth the effort. "For one thing, politicians are regularly accused ― sometimes fairly ― of "dog-whistling" to this or that basket of deplorables," he says. "Some, of course, don't even dog-whistle […] Considering how tolerant we are of such things in other formats, I find it perplexing, and not a sign of clear thinking, that online advertising is seen as somehow special, and so I worry that, due to a failure to think clearly, we will not adequately consider the costs of regulation, as well as its hoped-for benefits."
Surely, political campaigns will always find ways to target, divide and conquer their electorate. It’s also unlikely that increased transparency will ever produce perfect accountability when it comes to telling tall tales. But at least it's a step toward restoring a common marketplace of ideas.