.svg)
From our previous work with SAMbot, it is clear that a candidate’s lived experience on the digital campaign trail can’t be captured by numbers alone. Quantitative data, the type of numerical data that SAMbot produces, has helped us measure and quantify the state of online civic discussions at a massive scale. However, qualitative research is also necessary to understand how online abuse manifests in relation to individual candidates, and how this affects participation in municipal politics. This means we need to look deeper to uncover who or what topics attract abuse, why certain candidates or topics are targeted, and the precise nature of the abusive conversation.
Comparing the respective online engagement of Olivia Chow and Chris Saccoccia, also known as Chris Sky, two very different mayoral candidates, illuminates how experiences of online abuse can greatly differ between people running for office — even if the quantitative data is similar.
These two candidates saw the highest engagement and received the highest amount of abuse on Twitter in the 2023 Toronto mayoral by-election. Exploring their experiences on the digital campaign trail demonstrates two important takeaways: the lived experience of online abuse varies widely depending on the context, and LGBTQ+ content has become a concerning vector of abuse in municipal politics.
Olivia Chow won the election with 37% of the vote. Chris Saccoccia finished with 1% of ballots cast.
Despite much different results at the polls, Chow and Saccoccia had similar outcomes in online engagement and abuse.
Chow and Saccoccia are very different Canadian political figures. Chow was elected mayor of Toronto in 2023. She had a long and diverse political career preceding this result, including serving as a school board trustee, Toronto city councillor, and MP for the Trinity-Spadina riding. She previously ran for mayor of Toronto in 2014, and placed third (with 23% of the vote), behind candidates John Tory and Doug Ford.
This election marked the first time that Saccoccia ran for political office. He had built a large digital following during the COVID-19 pandemic as a leading Canadian voice in anti-mask, anti-vaccine, and conspiracy theory spaces online. He has been arrested and received criminal charges for uttering death threats against public figures multiple times, including during the by-election.
These candidates’ approaches to social media use were very different. Chow put out strongly curated tweets, sharing her policy goals, images promoting her campaign, and videos with high production value, taken from media appearances.
Saccoccia tweeted frequently during the election campaign — from May 1 to June 30, he tweeted 984 times. For comparison, Chow tweeted 374 times in the same period. His tweets often focused on identity politics, conspiracy theories, and policy issues outside of municipal jurisdiction.
Saccoccia engaged in online arguments during the campaign, sharing toxic and abusive content in the process. He practiced a form of negative campaigning, a political strategy focused on criticizing other people, groups, or policies. This kind of behaviour is in part reinforced by social media algorithms that tend to incentivize and promote negative content.
While both candidates received abusive content on Twitter, Saccoccia also used abusive content to engage followers and spread his views and campaign. Saccoccia posted videos promoting conspiracies about other candidates and about the city and election, often arguing with Twitter users publicly. Digital platforms’ content recommendation systems tend to favour incendiary and divisive content, which encourages and incentivizes rhetoric like Saccoccia’s, as this can draw a larger audience. Comparing Chow and Saccoccia demonstrates how candidates who use social media in different ways can experience very different online realities.
Closely comparing the abusive language in tweets sent to a candidate, and the tweets sent by candidates themselves, then, shows us that identifying the volume of abusive tweets received by candidates doesn’t tell the whole story. It is critical to explore how this identified abuse manifests to understand the whole picture. This type of inquiry needs to be done on a case-by-case basis when evaluating targets of significant online abuse, as different aspects of a candidate, such as their identity, background, and their willingness to share toxic and divisive content, can all contribute to the amount of abusive content they receive.
Posts focused on LGBTQ+ issues coincided with an increase in both engagement and abusive tweets for Chow and Saccoccia during the election. This follows a concerning trend in abusive anti-LGBTQ+ rhetoric emerging at the forefront of Canadian political conversations, which we previously observed in our Online Abuse in Local Elections: The SAMbot Municipal Report.
Anti-LGBTQ+ hate is on the rise in Canada, and LGBTQ+ rights have become major political topics both internationally and in Canada. The forms of engagement directed at Chow and Saccoccia related to LGBTQ+ rights reinforce that even at the municipal level, these discussions are significant drivers of online engagement and toxicity.
The machine learning models that power our SAMbot data analysis are not designed to discern context or the target of a message in the way people would expect. This is especially evident in cases where attacks on identity or a specific community are involved. This means that, for example, tweets are categorized as abusive in cases where hateful anti-LGBTQ+ tweets are sent to candidates, whether that tweet is speaking in agreement with the user they’re replying to or not. We observed that since Saccoccia shared hateful messages about LGBTQ+ Canadians, he received many tweets that were evaluated as abusive that were in support of him as a candidate and his views on LGBTQ+ people. Conversely, he also received many tweets that were labelled as abusive that were critical of his anti-LGBTQ+ rhetoric. This is an example of why quantitative data is not enough to tell the whole story when exploring the issue of abuse online.
Throughout the election, Chow’s volume of tweets remained relatively steady day to day, aside from two large spikes in engagement — one on election day, and one on May 30, 2023. This spike was largely in response to the following tweet in reference to the York Catholic District School Board’s (YCDSB) decision to not fly the pride flag at its Catholic education centre:
Tweet: May 30, 2023
From: @OliviaChow
Flying the Pride flag says to 2SLGBTQ+ students: you are welcome here.
Everyone should feel safe at public schools.
YCDSB has made a decision to leave room for homophobia and transphobia–it’s shameful and doesn’t represent the majority of people.
Eleven percent of all abusive tweets Chow received were on May 30. She received five times her average amount of daily abuse on that day.
May 30 ended up being the second-most abusive day online during the election, only behind election day. Nearly half of the abusive tweets on this day were sent to Chow.
Saccoccia also received engagement spikes from his commentary about LGBTQ+ issues. On June 5, Saccoccia tweeted that Toronto drag performances were deliberately attempting to “sexualize” children. Saccoccia received his highest total of abusive tweets in one day on June 5, aside from election day.
Disclaimer: Chris Saccocia’s Twitter (now called X) account was suspended days before this report was published. As a result, some links below may be unavailable or not include the content we intended to source. In the future these links may become viewable again, or they may remain permanently unavailable.
Note: Chris Saccocia’s Twitter account was suspended at the time of this report’s publication. As a result some links may be temporarily or permanently unavailable.
He directed anti-LGBTQ+ hate towards fellow candidates via Twitter as well.
Tweet: May 30, 2023
From: @chrissaccoccia1
I TOLD YOU. @oliviachow wants to push SOGI and the LGBT agenda on YOUR CHILDREN.
We need to protect them.
Chris Saccoccia for mayor!
Retweet with the #chrissky
And put Chris Saccoccia for mayor in your profile!!!!
#topli #mayor #toronto #family #children #sogi #lgbt
Linked: A screenshot of Olivia Chow’s aforementioned tweet about the York Catholic District School Board
Saccoccia also spread transphobic sentiments in debates with Twitter users.
Saccoccia received high volumes of abuse on days when he shared posts promoting conspiracy theories about 15-minute cities, and allegations of voter fraud. But in general, the most significant vectors of online abuse that we measured during this election coalesced around anti-LGBTQ+ sentiment.
Although Saccoccia only received 1.1% of votes, it’s vital to recognize that while he may not have found significant success at the polls, he nonetheless has a significant impact on online discussion spaces. As of publication, months after Chow was elected mayor, Saccoccia still has approximately 50,000 more followers on Twitter than Chow.
.png)
.png)
Understanding the topics that attract toxicity, the nature of the abuse received, and the motivations behind such discourse are vital to understanding the conditions of work for political candidates — particularly because understanding these working conditions may reveal why our body of elected officials does not reflect the diversity of Canada’s populations. Although neither of these candidates identify as LGBTQ+, the rhetoric that they both received, and that Saccoccia shared, is a threat to our democratic diversity. LGBTQ+ Canadians who aspire to run for office encounter these abusive, anti-LGBTQ+ messages on the digital campaign trail — it’s a condition of work. Volumes of abuse received (and shared) by municipal candidates are visible to the many Canadians who engage with municipal politics online. This sends a powerful message about who is or is not welcome in online civic discussions, and in turn may dissuade LGBTQ+ candidates from running for local office, out of fear of receiving hate or facing threats or violence.
Although LGBTQ+ rights were not a topic of particular debate in this mayoral election, some of the largest influxes of engagement and abuse on Twitter concerned this topic. This is a prime example of how online hate and abuse directed at marginalized communities threaten the democratic participation of groups who have historically been, and continue to be, underrepresented in our political systems. Understanding the nuances of the receiving and sharing of abusive content by municipal candidates can illuminate significant barriers to representation and inclusion affecting our city councils, as well as other Canadian legislative bodies.
For more information about our research approach see our methodology section and data release
.png)
.png)
.jpg)
%20(1).png)
.png)
.png)
.png)
.png)
.png)
%20(1).png)
.png)
-min.png)
.jpg)
.png)
-min%20(1).jpg)
.jpg)
.jpg)
.jpg)
