Combating the Spread of Misinformation on Social Media Platforms

The Viral Nature of Misinformation Revealed in Toronto Attack Coverage

In the weeks following the 2016 U.S. presidential election, Facebook CEO Mark Zuckerberg declared that his company takes misinformation seriously. Since then, combating “fake news” has become a critical concern for both technology companies and governments worldwide. Yet despite widespread recognition of the problem, opportunities to observe misinformation dynamics in real-time remain surprisingly rare.

A unique case emerged following the 2018 Toronto van attack when CBC journalist Natasha Fatah inadvertently created a natural experiment in misinformation spread. Fatah posted two competing eyewitness accounts on Twitter: one incorrectly describing the attacker as “angry” and “Middle Eastern,” and another accurately identifying him as “white.”

The results were revealing. The tweet containing the incorrect information about the attacker’s ethnicity generated substantially more engagement than the accurate one. Within approximately five hours after the attack, the misinformation tweet had spread significantly faster and wider than its accurate counterpart. Even more troubling, this pattern persisted over the next 24 hours, with the accurate information never catching up in terms of engagement.

This stark contrast raises important questions about why misinformation travels so quickly through social networks and what can be done to address the problem.

The mechanics behind this phenomenon stem from both human psychology and technical factors. Twitter’s algorithm, like those of Facebook and YouTube, prioritizes content that generates high engagement. The company’s engineering team has revealed that its deep learning algorithm promotes content that has already received significant interaction through retweets and mentions.

Human cognitive biases fuel the initial stage of this process. People are naturally drawn to content that confirms existing beliefs or taps into emotional triggers. Once a provocative or inflammatory tweet receives those initial interactions, the algorithm amplifies it, showing it to more users who then engage further. This creates a self-reinforcing cycle that can transform social media into what some critics call a “confirmation bias machine.”

This pattern played out clearly with Fatah’s tweets. A subset of her followers immediately engaged with the inflammatory, though incorrect, information about the attacker being “Middle Eastern.” This triggered the algorithm to show that tweet to more users, creating exponential growth in engagement. Meanwhile, the accurate tweet received minimal initial engagement and thus remained relatively invisible in the platform’s ecosystem.

The persistence of misinformation on Twitter extends beyond just the initial event. Ten days after the Toronto attack, searches for the attacker’s name still surfaced tweets falsely claiming he was Muslim, potentially leading users to wrongly associate the attack with Islamic terrorism rather than its actual motivation.

Addressing these challenges requires both technical and human solutions. Twitter could implement several measures to curb misinformation spread during crisis events. The platform could prioritize official police or government accounts during emergencies, display warnings about the unreliability of early eyewitness accounts, or restrict visibility of trending misinformation once authorities have provided accurate information.

The company could also improve its search functionality to prevent misinformation from dominating results long after an event. Options include establishing an editorial team to monitor and remove false information from trending topics or creating a user reporting system specifically for misinformation.

However, technological solutions alone cannot solve the problem. Users, especially journalists, must develop greater awareness of how their actions can inadvertently spread misinformation. In Fatah’s case, she shared an unverified account without proper corroboration—even though the eyewitness himself had expressed uncertainty about his own observation.

This incident provides valuable insight into how misinformation operates in our digital ecosystem. While we can reasonably expect platform algorithms not to amplify our worst instincts, the responsibility also falls on users to exercise greater care in what they share and how they consume information online.

As social media continues to serve as a primary news source for many people, understanding the mechanics of misinformation spread becomes increasingly vital for maintaining an informed public discourse during critical events.