Is Canada Racist? The Honest Answer Nobody’s Feed Will Give You

By Milad Ghobadibeygvand, BScN (Western University, 2014) — Co-founder, Zeus eBikes Canada. Published July 25, 2026.

I came to this country as a refugee. I have defended it in every argument I've ever had about it — sometimes against people who were born here. Seven days ago, a video from a parking lot in Moncton, New Brunswick started circulating: a Muslim woman being punched repeatedly in front of her seven-year-old son at a car meet, after — according to the National Council of Canadian Muslims — the group shouted "go back to your country." I watched it, and then I watched what the internet did with it. This article is my attempt to answer the question honestly — is Canada racist? — with the discipline the question deserves: real numbers, real history, real science, and a hard look at the machinery that decides what you believe about your own country.

What Happened in a Moncton Parking Lot

On Saturday, July 18, 2026, at about 7:30 p.m., RCMP were called to an altercation in a parking lot on Mountain Road in Moncton, New Brunswick, during a local car convoy. A video later posted by the National Council of Canadian Muslims shows a group confronting a Muslim woman before one woman punches her repeatedly while a man shouts at her — in front of the victim's seven-year-old son. Police say they are investigating whether hate was a factor (CBC News, July 2026).

That paragraph matters more than it seems, because everything in it is verified and everything past it, for now, is not. According to the NCCM, the family had been enjoying the car parade when a group allegedly shouted racist and anti-immigrant remarks — including "go back to your country" — before the assault began. The woman suffered several serious injuries, including a concussion (CTV Atlantic, July 2026). The RCMP say they believe they have identified the people involved. As of this writing, no arrests or charges have been announced, and no one in that video has been publicly identified. This article will not change that. Canada's courts decide guilt; a comment section does not.

Two things in that video deserve to be named precisely. The first is the child. A seven-year-old watched his mother being beaten and, in the footage that circulated, no adult in frame moves to comfort him. The second is what the video did next: it travelled. Within days it had been viewed millions of times, sliced into reaction clips, stitched into compilations, and pushed — by ranking systems we will spend half this article examining — into the feeds of people who will never set foot in Moncton, as their primary data point about what Canada is.

The aftermath followed a pattern Canadians now know by heart. The NCCM published the footage and called for a full investigation; human-rights advocates connected the assault to a broader rise in anti-Muslim and anti-immigrant rhetoric; political figures expressed disgust; and the RCMP asked for patience while they did the slow work of determining whether the Criminal Code's hate provisions apply. Meanwhile, in the parallel jurisdiction of the feed, the verdict was in within hours — a thousand different verdicts, actually, each one certain, each one selected for the audience most likely to be aroused by it. Some captions offered the clip as proof Canada is irredeemably racist; others framed it to inflame a different audience entirely. The one thing no version served was the sentence investigators keep repeating: we are still determining what happened.

Here is the uncomfortable fact that the rest of this article has to live with: the reason you and I are talking about a parking lot in Moncton at all is that an algorithm decided this footage would hold your attention. It was right. The question is what we do with that attention — and whether we can tell the difference between what the video proves and what the feed wants us to feel it proves.

The second event: within days, one parking lot in Moncton became the primary evidence about Canada in millions of feeds.

I Came Here as a Refugee. Here Is My Ledger.

Before the numbers, you should know who is writing this and what I owe this country. I arrived in Canada as a refugee. I have since co-founded a company here, built a life here, and been welcomed here in ways that still surprise me. If this article were a court case, I would have to declare a bias: mine runs in Canada's favour.

Did I experience racism here? Yes — and for years I barely registered it, for a reason that may sound strange: my baseline was different. I had watched how Afghan refugees were treated in Iran, where I came from. I want to be careful with this comparison because it is testimony, not statistics — but by comparison, what I found in Canada felt like a different moral universe. There are far more Iranian refugees thriving in Canada than there are Afghan refugees permitted to thrive in Iran. When I hit prejudice here, I mostly filed it under something almost forgivable: humans are tribal animals, tribalism is old software, and some individuals run it unexamined.

But here is the part of my ledger that I think matters most, and it is the thing I have repeated in every argument about this country: when I met a racist gatekeeper in Canada, there was always a system behind him that let me route around him. A manager who wouldn't take me seriously was not the end of the road, because there was another employer, a human-rights framework, a licensing body, a court, a complaint process — an architecture built on the assumption that the gatekeeper might be wrong. Racism existed at the level of individuals. The system, imperfect and slow as it is, kept refusing to let those individuals have the last word on my life. That distinction — racist people versus a racist architecture — is the most useful lens I can offer for everything that follows.

The first minute of a new life. For those of us who arrived with one suitcase, this door — and what waited behind it — is the Canada we keep arguing for.

What I saw from the waiting room in Turkey

Before Canada, there was Turkey, and waiting. Around us were other families in the same limbo, and I learned there were two roads into Canada: ours — the UN-referred stream — and another one I could hardly believe existed when it was first explained to me. Ordinary Canadians, mostly through their churches, were personally sponsoring refugee families: pooling their own money, signing their own names, taking on the legal responsibility so completely that if the family fell onto social assistance, the cost fell on the sponsors. Strangers, an ocean away, betting their own savings on people they had never met. Some of the families around us had waited eight, nine years for their turn — and Canadians they'd never seen kept the commitment alive the whole time.

When our flight finally left for Canada, one of those church-sponsored families was on the same plane. Years later we reconnected, and I got to see the whole arc of what those congregation members had done — met at the airport, housed, furnished, walked into their first jobs. Two families on one flight: one carried by the United Nations, one carried by a church basement's worth of ordinary Canadians. When I wrote, later in this article, that private refugee sponsorship is the machinery that won the people of Canada the UN's refugee medal, understand that I am not citing a history book. I watched it work, from the waiting room, on people I know by name.

The country that taught me the opposite

My first classroom in Canada was an ESL class, and it was the United Nations with worse furniture: Russia, Japan, Afghanistan, Congo, Venezuela, Colombia, Cuba, Mexico, Egypt — Palestinians and Israelis in the same room — all of us butchering English together. We cooked for each other. We made fun of each other's accents, constantly, and everyone gave as good as they got. I won't argue about whether that kind of teasing passes today's standards; I can only report what it felt like from inside: not hatred — play. It was the first place I unlearned the habit of sorting people by colour and shape, and it was Canada that ran the classroom.

Years later, in nursing school, my family hit a mental-health crisis and I started falling behind. One day my teacher — a white woman, since this article is about race — walked me to the counsellor's office, another white woman, who asked me one question: "What's going on?" I burst into tears in that office. What followed was compassion, accommodation, and practical help until I was through it. And when the crisis needed more than a school could give, Canada's healthcare system — the nurses, the psychiatric care — gave my family a second chance at survival. I remember sitting with exactly one thought: do you understand how fortunate we are to be here? That is my ledger's light column, written by actual Canadians, most of them white, none of them owing me anything.

What my ledger cannot see

Before you weigh my ledger, weigh its blind spot. I am a tall man who, physically, reads as white. I do not stand out in a Canadian crowd, and much of what this article examines never has to find me. The people who taught me that were Canadians of colour I worked beside. My nursing teacher — a Nigerian woman whose competence ran the room — carried experiences of this country I have never had to carry. A colleague of mine, a Black Canadian RN I worked with in psychiatric nursing, once told me about the number of times he had been pulled over in his Cadillac — stop after stop, because somebody wanted to satisfy themselves about a Black man in that car. His Canada and my Canada share a border, but they are not the same country. So let me be precise about what my testimony is: it is one family's true experience, not a verdict on anyone else's. Nothing in my good fortune minimizes the person who faced what I didn't. But the reverse discipline also holds — if we build our verdict on one incident, or three, we miss the whole picture. Which is exactly why this article keeps returning from stories, mine included, to the full dataset.

The switch I watched in 2023

Then I watched something change, and I want to describe it precisely because it is the closest I have come to seeing this article's thesis operate in the open. Around 2023, seemingly in one season, a surge of anti-Indian sentiment arrived — at work, in conversations, everywhere at once. Too many of them. They're the reason for everything. Meanwhile my actual Indian colleagues were humble, hardworking, generous newcomers — many of whom had sold family farmland to pay tuition here, on promises made by colleges and universities our own government licensed and let recruit. While those institutions made money hand over fist — while restaurants finally had staff and dealerships were selling out of cars — nobody complained. Then, suddenly, uniformly, everybody did. The numbers section of this article shows where that went: hate crimes targeting South Asian Canadians rose another 15% in 2024, one of the sharpest climbs of any group (Statistics Canada).

Here is what I could not stop noticing: organic prejudice does not switch on like that. Real attitudes move slowly, unevenly, person by person. This moved like a broadcast — the same complaints, the same phrasing, the same targets, in the same months. And it did not travel through the CBC, CityTV or any newsroom with a masthead. It travelled through Facebook, Instagram and TikTok — the same engagement machinery this article documents. A policy debate about immigration volumes is legitimate, and Canadians were owed one. What arrived instead was a mood, delivered by ranking systems, and billed to my coworkers' faces. That is the difference between a country having a hard conversation and a feed having a profitable one.

So when I say the Moncton video shook me, understand what it shook. Not my belief that racists exist here; I knew that. It shook my confidence that the story I'd been telling — "the system routes around them" — still describes the direction we are moving. Because the woman in that parking lot could not route around anything. And the crowd around her, in the moments the camera caught, was not behaving like the Canada in my ledger.

How the world treats its strangers: the comparison, with data

My comparison between Canada and my birthplace is no longer just testimony — 2025 turned it into a dataset. In that single year, Iran expelled or returned nearly two million Afghans to Taliban-ruled Afghanistan; roughly 1.2 million of those — about two-thirds — were deportations, with a surge of forced family returns that the UN's migration agency called "a new and concerning trend," and children making up roughly a quarter of the deportees (Amnesty International, December 2025; UNHCR/IOM, 2025). These were people fleeing the same kind of danger I once fled. Many had lived in Iran for decades. Their children had been born there. There was no human-rights tribunal to route around the gatekeeper, because the gatekeeper was the system.

Now the same year in Canada. In 2025, a majority of Canadians told pollsters immigration levels were too high — and Canada's response to that souring mood was to lower planned admission targets through its normal, published, debated policy process, while remaining a country where a refugee can land, claim protection under law, appeal a refusal in an independent court, and — as the person writing this can attest — co-found a company and criticize the government in print without consequence. I do not offer this comparison to grade Canada on a curve of atrocities. I offer it because "is Canada racist?" is usually asked in a vacuum, and moral judgment without a comparison class is how both propagandists — the "Canada is a white-supremacist project" one and the "Canada has no racism" one — do their work. Compared to its own ideals, Canada falls short, measurably, 4,882 times a year by police count alone. Compared to how nation-states actually treat their strangers in 2026, Canada remains, by the best international measure we have, the top of the class. Both gradings are real. A country, like a person, should be judged against both.

Is Canada Racist? The Actual Numbers

The honest answer is that three things are true at once. Canada ranks first in the world on Gallup's Migrant Acceptance Index, at 8.46 out of 9. Police-reported hate crimes have also risen 169% since 2018, reaching 4,882 incidents in 2024 (Statistics Canada, March 2026). And a majority of Canadians — 56% — now tell pollsters immigration levels are too high (Environics Institute, fall 2025). Any answer that uses only one of these numbers is propaganda.

The light column

Start with the number that most "Canada is racist" essays never print. When Gallup last ran its global Migrant Acceptance Index — asking people in 145 countries whether migrants living in their country, becoming their neighbours, and marrying into their families were good things — Canada scored highest in the world (Gallup). Not top ten. First. And it earned that ranking while acceptance was falling across most of the world. Whatever is wrong here, the best comparative instrument humanity has says this remains one of the most accepting societies on Earth — which matches the ledger of my own life, and probably the ledger of most immigrants you know, most days.

There is even good news inside the hate-crime data: incidents targeting East and Southeast Asian Canadians have now declined three years in a row from their pandemic-era spike, down another 5% in 2024 (Statistics Canada). Spikes can recede. They are not destiny.

The dark column

Now the ledger's other page, from the same March 2026 Statistics Canada release — all police-reported, all 2024:

Source: Statistics Canada, "Police-reported hate crime in Canada, 2024," The Daily, March 30, 2026.

And sit with the most lopsided line in that table: seven of every ten religion-motivated hate crimes in Canada target Jewish Canadians — a community that is roughly one percent of the population. I have watched jokes about Jewish people get treated as socially acceptable in rooms where no other group could be spoken of that way. So the principle this article keeps applying to Muslims and immigrants gets applied here, in full: being Jewish does not mean endorsing the policies of Israel's government, any more than being Muslim means answering for any state or any group. When a Jewish community centre in Canada is attacked, or Jewish children are afraid on their way to school because of what another country is doing on another continent, that is not OK — full stop, no "but." Communities do not stand trial for nations.

Read the trend, not just the total. From 2018 to 2024, police-reported hate crimes more than doubled. The 2024 plateau (+1%) is genuinely better news than another 34% jump — but a plateau at the top of a 169% climb is not recovery. And Statistics Canada itself cautions that police-reported figures are a floor: most incidents are never reported to police at all. The 229 recorded anti-Muslim hate crimes in 2024 are the cases that produced a file number — not the sum of what Muslim Canadians experienced.

The mood column

Between the light and dark columns sits public opinion, and it has moved. In 2025, 56% of Canadians told the Environics Institute's Focus Canada survey that Canada accepts too many immigrants — the first sustained majority in roughly a quarter-century of that question, though it levelled off (down two points) after two years of steep increases. Environics' own analysis attributes the shift primarily to housing costs, economic strain and perceptions of a poorly managed system — not to a surge in ethnic hostility (Environics Institute, fall 2025). That nuance is not a comfort blanket; it is a warning with a mechanism attached. Economic grievance is historically the kindling that opportunists — human and algorithmic — convert into ethnic grievance. The kindling is now stacked.

The Ledger Canada Already Wrote: 140 Years of Receipts

Can a country that ranks first in migrant acceptance have a racist past? Canada answers that question itself, in its own official documents: a head tax, an exclusion act, a turned-away refugee ship, mass internment, and a residential school system its own commission called cultural genocide. Canada has formally apologized, in Parliament, for nearly all of it. The receipts and the apologies are both part of the honest answer.

This section exists because "is Canada racist?" cannot be answered with 2024 data alone. The country kept its own records. Here are five of them, told at the length they deserve.

1885–1947: The head tax, and then the honesty of exclusion

When the Canadian Pacific Railway was finished — built in its most dangerous western sections substantially by Chinese labourers — Canada's response to the men who built it was the Chinese Immigration Act of 1885: a $50 head tax charged to Chinese immigrants and no one else, raised to $100 in 1900 and $500 in 1903, roughly two years' wages. When the tax failed to stop immigration entirely, Parliament stopped pretending it was about revenue: the Chinese Immigration Act of 1923 — remembered by Chinese-Canadians as the Exclusion Act — banned Chinese immigration almost outright. It came into force on July 1, Dominion Day, which many Chinese-Canadians thereafter marked as "Humiliation Day." The ban held until 1947. Families were split across the Pacific for a generation. On June 22, 2006, Prime Minister Stephen Harper rose in the House of Commons and apologized, offering symbolic redress to surviving head-tax payers and their spouses.

1914: The Komagata Maru sits in Vancouver harbour

In May 1914, a chartered Japanese steamship carrying 376 passengers from British India — nearly all Sikhs, all British subjects with formal rights of movement within the Empire — arrived in Vancouver to test Canada's "continuous journey" regulation, a law engineered to exclude Indian immigrants without saying so. Canada did not let them land. For two months the ship sat in the harbour, passengers running short of food and water, while officials, courts and mobs on the shore made Canada's answer clear. On July 23, 1914, the military escorted the Komagata Maru out to sea. When the passengers reached India, colonial police met the ship; 19 passengers were killed in the confrontation that followed and others imprisoned (CBC News). Prime Minister Justin Trudeau delivered Canada's formal apology in the House of Commons on May 18, 2016 — 102 years after the ship was turned away.

Two months at anchor: an editorial evocation of the Komagata Maru, kept off Vancouver’s shore in the summer of 1914.

1939: The MS St. Louis, and the sentence that named an era

In June 1939, the German liner MS St. Louis approached Canada carrying 907 Jewish refugees fleeing Nazi Germany, having already been refused by Cuba and the United States. Canada said no. The ship returned to Europe; more than 250 of its passengers were later murdered in the Holocaust. Around the same period, a Canadian immigration official, asked how many Jews Canada should admit after the war, gave the answer that historians Irving Abella and Harold Troper later took as the title of the definitive book on the era: "None is too many." On November 7, 2018, Trudeau apologized in the House of Commons, noting that Canada's silence had been part of the world's permission structure: the government "failed" the test of its humanity (Prime Minister's Office; The Globe and Mail; CBC News).

1942–1949: The internment of Japanese Canadians

Beginning in 1942, Canada forcibly uprooted more than 20,000 Japanese Canadians — the majority of them Canadian-born or naturalized citizens — from the British Columbia coast, interned them in camps and work sites, sold off their homes, farms, boats and businesses without consent, and used the proceeds to pay for their own detention. None were ever charged with disloyalty. The restrictions outlasted the war itself, with the last controls lifted only in 1949. Prime Minister Brian Mulroney apologized on September 22, 1988, and signed a redress agreement including individual compensation for survivors.

1883–1996: Residential schools — the record Canada wrote about itself

For over a century, Canada removed roughly 150,000 First Nations, Métis and Inuit children from their families into church-run, government-funded residential schools whose explicit purpose, in the founding generation's own words, was to "kill the Indian in the child." The Truth and Reconciliation Commission, after hearing nearly 7,000 witnesses over six years, concluded in 2015 that the system "was always more than simply an educational program: it was an integral part of a conscious policy of cultural genocide." The commission documented more than 4,100 children who died in the schools — of disease, neglect, abuse and suicide — and its chair, Justice Murray Sinclair, estimated the true toll at 6,000 or more (TRC, 2015; CBC News). The last federally funded school closed in 1996. That is not a historical date. That is within the lifetime of most people reading this sentence.

The pattern in the receipts

Sources: Prime Minister's Office; CBC News; Truth and Reconciliation Commission of Canada, 2015.

Why print this in an article about a 2026 parking lot? Two reasons. First, because anyone who answers "is Canada racist?" with "it never was" is arguing with Canada's own Hansard. Second — and this is the part the doom-scroll version of history omits — because the same country prosecuted, documented, and apologized for these acts in its own Parliament, changed its laws, and then became the country that tops Gallup's acceptance index. Nations do not get to be innocent. They get to choose a direction. The record shows Canada's direction over 140 years, with reversals, has been away from those receipts — which is exactly why a measured six-year doubling of hate crimes should alarm us: direction is the thing we know can change.

The Other Ledger: When Canadians Were the Opposite of Racist

Has any people ever been formally honoured, as a people, for welcoming strangers? Once, in the entire history of the United Nations' refugee agency — and it was the people of Canada. In 1986, the UNHCR awarded the Nansen Medal, its highest honour for service to refugees, to the Canadian population as a whole. No other nation's people have received it, before or since. That fact belongs in this article with the same weight as the head tax — because it is the same country, and mostly the same families.

The doom-feed version of Canadian history keeps only the receipts. Here is the ledger it deletes — and read it noticing who did these things: overwhelmingly ordinary Canadians, most of them the white Canadians the algorithm now casts as the villains of every clip.

• 1793 — the Empire's first anti-slavery law. Upper Canada's Act to Limit Slavery, passed under Lieutenant-Governor John Graves Simcoe, was the first legislation restricting slavery anywhere in the British Empire — four decades before abolition. In the decades that followed, British North America became the terminus of the Underground Railroad: tens of thousands of African Americans escaped bondage by walking, hiding and sailing toward British territory where slavery had been outlawed and the slave-catcher's writ did not run. Canadians — Black and white, Quaker farmers and small-town churches — hid them, ferried them and settled them.
• 1979–80 — the boat people, and the invention Canadians gave the world. When Vietnamese, Cambodian and Laotian refugees were drowning in the South China Sea, Canada created something no country had ever had: the private sponsorship of refugees program (1978–79), letting any five citizens or a church group personally resettle a refugee family at their own expense. Ordinary Canadians privately sponsored 34,000 of the 60,000 Indochinese refugees Canada took in 1979–80 — church basements, Rotary clubs, farm towns — a per-capita intake that led the Western world and won the 1986 Nansen Medal (Canadian Museum of Immigration at Pier 21; UNHCR; The Conversation). The model has since been copied by more than a dozen countries. Refugee sponsorship is a Canadian export — and, as I described in my own ledger above, a machine I watched work firsthand from a waiting room in Turkey, carrying a family who flew to Canada on my own flight.
• 1971 — the world's first official multiculturalism. Canada was the first country on Earth to adopt multiculturalism as official state policy (1971), then wrote it into section 27 of the Charter of Rights and Freedoms (1982) and the Multiculturalism Act (1988). Imperfectly lived, endlessly debated — and still a world first that dozens of countries later imitated.
• 1980s — Canada against apartheid. When much of the Western alliance hedged on apartheid South Africa, Prime Minister Brian Mulroney — a Conservative — led the Commonwealth push for sanctions over the open resistance of Canada's closest allies. Nelson Mandela chose Canada's Parliament for one of his first foreign addresses after his 1990 release, and in 2001 became the first living person granted honorary Canadian citizenship. When Mandela wanted to thank someone, he flew here.
• 2015–16 — 25,000 Syrians in a hundred days. Canada resettled 25,000 Syrian refugees in roughly four months, with the Prime Minister greeting the first planeload at the airport with winter coats and the sentence "You're home" — and the private-sponsorship queue told its own story: as was widely reported at the time, sponsor groups formed faster than the government could match them with families — Canadians literally waiting in line for refugees.
• 2017 and 2021 — what Canadians did after the attacks. After the Québec City mosque shooting, tens of thousands of Canadians stood in vigils across the country, and Toronto congregations — synagogues prominent among them — formed human "rings of peace" around mosques during Friday prayers. After the London attack, thousands lined the streets outside the London Muslim Mosque in one of the largest multi-faith gatherings in the city's history. And within days of Moncton, community groups and political leaders across New Brunswick were publicly condemning the assault and organizing responses (CBC News, July 2026). The attackers in this article's three cases number a handful. The Canadians who showed up afterward number in the hundreds of thousands.

A thousand church basements, 1979: private sponsorship — the machinery that won the people of Canada the UN’s refugee medal — as it looked. Editorial re-creation.

This ledger is not offered to cancel the other one — that arithmetic doesn't exist. It is offered because the question "is Canada racist?" is really asking "what should I expect from my neighbours?" — and the honest base rate is this: for every Canadian who has ever screamed at a stranger in a parking lot, there are thousands who quietly co-signed a refugee family's first apartment. I know, because I have met both, and only one of them shaped my life.

The Names Before Moncton

Moncton was not the first time a Canadian Muslim family's worst day went national. Twice in the past decade, Islamophobic violence in Canada turned lethal — six men murdered in a Québec City mosque in 2017, four members of one London, Ontario family killed in 2021 — and in both cases, the court record documents the same thing behind the attacker: a screen, and a feed.

Québec City, January 29, 2017

A 27-year-old walked into the Centre Culturel Islamique de Québec during evening prayers and murdered six men — fathers, a grocer, a professor, a pharmacist — and injured many more. He pleaded guilty to six counts of first-degree murder and six of attempted murder. What makes his case essential to this article is what investigators found on his devices, entered into the court record at sentencing: in the month before the attack — January 1 to January 29, 2017 — he searched for Donald Trump 819 times (417 on Twitter, 337 on Google, 63 on YouTube, twice on Facebook). He obsessively checked the feeds of far-right figures — Alex Jones, Richard Spencer, David Duke — alongside mainstream right-wing outlets, and researched previous mass killers (National Observer, April 2018; The Washington Post; CBC News). To be precise about what this proves: it documents his media diet, not a causal verdict — courts convict people, not feeds. But 819 searches in 29 days is not a man reading the news. It is a man being fed.

London, Ontario, June 6, 2021

The Afzaal family — Salman, 46; his wife Madiha Salman, 44; their daughter Yumnah, 15; and grandmother Talat, 74 — went for an evening walk near Hyde Park Road in London, Ontario. A 20-year-old deliberately drove his pickup truck into them because they were Muslim. Four died; the nine-year-old son survived his injuries. Canada came to know them as Our London Family. The attacker was convicted of four counts of first-degree murder and one of attempted murder, and in February 2024, Justice Renee Pomerance ruled his crimes were terrorism — the first time white-nationalist violence was formally labelled terrorist activity under Canadian anti-terror law, a "textbook example" in her words (The Globe and Mail; PPSC, February 2024). The court record on how he got there is stark: he radicalized alone, in his apartment, online — months of reading and writing extremist material, no cell, no recruiter, no in-person movement. The pipeline was a browser.

Radicalization’s real address: no cell, no recruiter — an ordinary door, and a feed on the other side of it.

The pattern, stated carefully

Two convicted attackers; two court records; one shared feature — heavy, documented consumption of online hate and extremist content in the period before the violence. Add Moncton, where the investigation is open and no equivalent record yet exists, and the careful claim is this: in the two lethal Islamophobic attacks in modern Canadian history where we have full court records, the road ran through a feed. That is not proof that feeds cause violence in general — the science section ahead is honest about how contested that is. It is proof that for the specific men who did these specific things, the radicalizing library was delivered to their screens, sorted by relevance.

What the Platforms Knew

Did the platforms know their systems could push ordinary users toward extreme content? Yes — because they tested it, documented it, and in the worst case, watched it help fuel an atrocity. The three exhibits in this section are not critics' claims. They are Facebook's own internal research, Facebook's own ranking formula, and the United Nations' finding on Myanmar — all on the record.

Exhibit A: Carol's journey — two days to QAnon

In the summer of 2019, Facebook's own researchers created a fake user: "Carol Smith," a fictional conservative mother from Wilmington, North Carolina. Her profile followed Fox News, Christian pages, and Melania Trump — mainstream conservative interests, zero conspiracy content. The researchers then simply followed Facebook's own recommendations. Within two days, the platform was recommending Carol join QAnon groups. Within a week, her feed was, in the researchers' description, full of content violating Facebook's own rules. The internal study was titled "Carol's Journey to QAnon," and it became public only through the documents Frances Haugen disclosed to the U.S. Securities and Exchange Commission (NBC News, October 2021). Remember the timeline the next time someone says radicalization takes years: the platform's own experiment measured the on-ramp in days. I ran a version of this experiment on myself, deliberately, and I will tell you what it did to my own mind in a later section.

Exhibit B: five points for anger

From 2017, Facebook's News Feed ranking formula counted emoji reactions — including "angry" — as worth five times a like when deciding what to show people next. To be precise: the 5× weighting applied to all reaction emojis, not anger alone. The problem was what anger correlated with. By 2019, Facebook's own data scientists had established that posts drawing angry reactions were disproportionately likely to contain misinformation, toxicity and low-quality news — and an internal staffer had warned the weighting could fill feeds with a "higher ratio of controversial than agreeable content." Facebook eventually cut the angry reaction's weight to zero (The Washington Post, October 26, 2021, from the Haugen documents). For four years, though, the arithmetic of the world's largest information system paid a five-fold premium for the emotion most useful to people who traffic in hate.

And Facebook knew feed composition changes minds, because it had proven so in its own name: in a study published in PNAS, company researchers manipulated the feeds of 689,003 users — suppressing positive posts for some, negative for others — and measured users' own subsequent posts shift in the direction of what they'd been shown (Kramer, Guillory & Hancock, 2014, PNAS 111(24)). The emotional-contagion experiment is infamous for its consent ethics. It should be equally famous as the platform's own proof that the dial works.

Exhibit C: Myanmar — the precedent with a body count

If you want to know what unaccountable amplification can contribute to at full scale, the United Nations already wrote the report. In March 2018, the chair of the UN Independent Fact-Finding Mission on Myanmar said Facebook had played a "determining role" in stirring up hatred against the Rohingya Muslim minority ahead of and during the 2017 atrocities (TIME, March 2018). Facebook spent months minimizing — then in November 2018 published the human-rights assessment it had commissioned from BSR, which concluded the platform had been "used to foment division and incite offline violence," with Facebook conceding it "can and should do more" (Meta/BSR, November 2018). Amnesty International's 2022 follow-up, The Social Atrocity, went further: Meta's engagement-driven algorithms "substantially contributed" to the atrocities (Amnesty International, 2022). This is a pattern my generation of immigrants knows by heart and Canadians should study: denial, delay, partial admission — after the harm. Canada is not Myanmar; our institutions are incomparably stronger. But the machinery being examined is the same machinery, running on Canadian phones tonight.

The Public-Square Sales Pitch

How did private advertising companies convince the West they were the infrastructure of democracy itself? By saying so, eloquently, for fifteen years — Facebook casting itself as a "Fifth Estate" for free expression, Twitter branding itself "the free speech wing of the free speech party" — while their own documents show they were quietly weighting, whitelisting, blacklisting and self-promoting the whole time. The pitch was neutrality. The product was curation.

The new town square: the gazebo went dark, and the screen answers to no one.

Why the West believed it

Before examining the pitch, it is worth remembering why intelligent people bought it, because the belief was not stupid — it was hopeful, and for a moment it looked true. In January 2010, U.S. Secretary of State Hillary Clinton delivered her "Internet Freedom" address, elevating an unrestricted internet to the level of a foreign-policy doctrine: connection technologies as the samizdat of our age. A year later came the Arab Spring, and Western media christened the uprisings in Tunisia and Egypt "Twitter revolutions" and "Facebook revolutions" — young crowds organizing through the very platforms Silicon Valley had built, toppling dictators who had controlled every newspaper and TV station but not the feed. The platforms did not have to argue that they were democracy's infrastructure. CNN argued it for them, nightly.

What got lost in the celebration was a quieter question nobody wanted to ask while the squares were full: if a feed can organize a crowd against a dictator, what else can it organize a crowd against — and who decides? The Rohingya of Myanmar received the answer within six years, from the same platform, running the same machinery. The tool was never pro-democracy. It was pro-mobilization, and agnostic about the cause. The West spent a decade mistaking a megaphone's first customers for its character.

The pitch

The high-water mark of the sales pitch was October 17, 2019, at Georgetown University, where Mark Zuckerberg wrapped Facebook in the First Amendment, Frederick Douglass and the civil-rights movement, described social media as a "Fifth Estate" alongside the traditional pillars of democracy, and declared that "when it's not absolutely clear what to do, we should err on the side of greater expression" (speech text: The Washington Post; Meta Newsroom, October 2019). Twitter's version came earlier and pithier: in March 2012 its UK general manager, Tony Wang, called the company "the free speech wing of the free speech party" (The Guardian, 2012). The framing did enormous work. If the platform is the public square, then regulating it is censorship, criticizing it is attacking free speech, and whatever the ranking system does is simply "what people want to see."

The hand on the dial

Here is what the documentary record shows was happening behind the neutral-square framing — each item from primary reporting, each about mechanics rather than any one ideology:

• "We do not shadow ban" — while building reach-suppression tools. In a 2018 blog post titled "Setting the record straight on shadow banning," Twitter's legal and product chiefs wrote flatly: "We do not shadow ban… And we certainly don't shadow ban based on political viewpoints or ideology." Internal materials reported in December 2022 showed the company operated what staff called visibility filtering — Trends blacklists, search blacklists and "Do Not Amplify" tags applied to specific accounts, described by one senior employee as "a way for us to suppress what people see to different levels… a very powerful tool" (Twitter blog, 2018; Twitter Files reporting, December 2022). Precision matters here: Twitter's denial used a narrow definition (making content invisible to a user's own followers), and the internal tools limited reach rather than existence. That gap — between the public word and the internal instrument — is the point.
• The VIP whitelist. The Wall Street Journal's Facebook Files reporting (September 2021) revealed "cross-check" or XCheck: a programme that by 2020 exempted at least 5.8 million high-profile users — politicians, celebrities, athletes — from the enforcement systems applied to everyone else, while the company told its own Oversight Board the programme affected only "a small number of decisions" (CNBC; Bloomberg, September 2021). A public square where 5.8 million people stand on a platform the rest cannot see is not a square. It is a stage.
• Project Amplify. In 2021, Facebook began tests — approved by Zuckerberg — using its own News Feed to push positive stories about Facebook, some written by the company itself, into users' feeds in three U.S. cities (The New York Times, September 2021; Forbes). The referee, it turned out, would also occasionally play striker.
• The owner's thumb, weighed in public. In February 2023, after a Super Bowl tweet by U.S. President Joe Biden drew 29 million impressions to Elon Musk's 9.1 million, Musk ordered X's engineers to fix the disparity; internal documents reported by Platformer showed his tweets were then hard-coded with a 1,000× ranking multiplier (Platformer; Rolling Stone; Fortune, February 2023). Musk disputed the reporting, attributing it to a disgruntled employee — both his denial and the documents belong in the record. What is not disputed is X's official doctrine, announced by Musk in November 2022: "freedom of speech, not freedom of reach" — an explicit statement that the platform decides whose speech travels.
• And in Canada, the square simply closed. When Parliament passed the Online News Act (Bill C-18), Meta responded by blocking news links for Canadian users entirely — a ban in force since August 2023. The company that told Georgetown it existed to give people voice removed Canadian journalism from Canadian feeds rather than pay for it. Our companion piece on Canada's information dependence documents the fuller exposure: the platforms Canadians use most are 100% foreign-owned.

Notice what this list is and is not. It is not a claim that one political side was throttled everywhere and always — accounts across the spectrum have been suppressed or boosted at different times, and the loudest partisans of every stripe have real grievances and real blind spots. The narrow, documented claim is this: reach on these platforms has always been curated — by ranking weights, blacklists, whitelists, self-promotion campaigns and, at least once, the owner's personal multiplier — while the companies marketed themselves as neutral ground. A newspaper that did all this would be called what it is: an editor. An editor with three billion readers, no masthead, and no press council.

What the Best Science Actually Shows

Does the algorithm actually radicalize people at scale? The honest answer from the strongest research: the amplification mechanism is proven, individual pathways are documented, and the population-level effect is genuinely contested. Anyone who tells you the science is settled — in either direction — is selling something. Here are both stacks, at full strength.

The evidence that the feed amplifies division

• Twitter's own audit. In the largest platform self-study of its kind, Twitter researchers compared the algorithmic timeline against a chronological control group of millions of users. The algorithm amplified mainstream political-right content more than the left in six of seven countries studied — including Canada (Huszár et al., 2022, PNAS 119(1)). The study did not find the far ends amplified more than moderates; its significance is simpler — the platform confirmed, in its own name, that the ranking system tilts political reach.
• Out-group hostility is rocket fuel. Analyzing 2.7 million posts from U.S. news media and members of Congress, researchers found posts about the political out-group were shared roughly twice as often, and each additional out-group word raised the odds of sharing by 67% — the strongest predictor found, several times stronger than negative or moral-emotional language generally (Rathje, Van Bavel & van der Linden, 2021, PNAS 118(26)). Earlier work found each moral-emotional word in a political post raised retweet rates by about 20% (Brady et al., 2017, PNAS 114(28)). An engagement-ranked feed does not need to hate anyone. It just pays whatever earns engagement — and contempt for the other tribe earns.
• Outrage is learned. Follow-up work showed users whose outrage got engagement expressed more outrage over time — the feed operates as a social-learning machine, teaching people what version of themselves performs best (Brady et al., 2021, Science Advances 7(33)).

The evidence that complicates the story

• The biggest experiment ever run found no attitude shift. In a landmark 2020 U.S. election collaboration between Meta and independent academics, tens of thousands of consenting users had their algorithmic feeds replaced with chronological ones for three months. The result: no significant change in issue polarization, affective polarization, political knowledge, or key attitudes — though users spent less time on-platform, and notably, the chronological feed increased exposure to untrustworthy content while the algorithmic feed had been suppressing roughly half the hateful content they would otherwise have seen (Guess et al., 2023, Science 381, and companion papers). Meta's leadership claimed vindication; the authors and Science publicly objected that three months on one platform in one country cannot acquit "the algorithm" — but the finding stands and belongs in every honest telling.
• Most extreme-content viewers go looking for it. Large-scale browsing studies of YouTube found little evidence the recommender systematically drags unsuspecting users to far-right content; most consumption of extreme material came from subscriptions and external links, concentrated among users who already scored high on gender and racial resentment (Hosseinmardi et al., 2021, PNAS 118(32); Chen et al., 2023, Science Advances 9(35)). The rabbit hole, on YouTube at least since its 2019 changes, is less a trapdoor than a door people walk to.
• Even the "echo chamber" fix backfires. When researchers paid U.S. partisans to follow a bot retweeting the other side's politicians for a month, Republicans became measurably more conservative, not less (Bail et al., 2018, PNAS 115(37)). Popping the bubble is not automatically medicine; hostile exposure can harden identity. Any regulation built on "just show people the other side" is built on a folk theory the data rejects.
• Leaving helps — modestly. In the largest deactivation experiment, people paid to quit Facebook for four weeks before the 2018 U.S. midterms showed small reductions in polarization on policy issues, a measurable bump in well-being — and knew less about the news (Allcott, Braghieri, Eichmeyer & Gentzkow, 2020, American Economic Review 110(3)). The feed is neither harmless nor the sole author of division; it is one dial among several, and turning it down has real but modest effects.

How both can be true

Hold the three registers apart and the picture resolves. Mechanism: proven — engagement-ranked feeds amplify outrage and out-group content, by the platforms' own math and audits. Individual cases: documented — Carol Smith in 2 days, a mosque shooter's 819 searches, a London killer radicalized alone at a screen. Population effect: contested — the best experiments cannot yet show the feed shifting the average citizen's attitudes in months. The most defensible synthesis is that the feed functions less like a mind-control ray and more like a distribution system for accelerant: most people walk past it; the vulnerable few — the resentful, the isolated, the primed — get an inexhaustible, personalized supply, faster than any radical bookstore or basement pamphlet network in history could have managed. Population averages stay flat. Tail risk explodes. And terrorism has always been a tail-risk phenomenon: it does not need the average Canadian to change. It needs one man in an apartment in London, Ontario.

This is also why Canada's own Foreign Interference Commission reached the conclusion it did. After two years examining hostile-state operations, Justice Marie-Josée Hogue's final report identified not spies or tanks but "information manipulation (whether foreign or not)" as "the single biggest risk to our democracy" (Hogue Commission, January 28, 2025). When the instinct says the feed is a national-security problem, the instinct is not paranoia. It is the finding of a royal commission.

Not Your Eyes — Your Milliseconds

Is your phone watching your eyes or reading your face to decide what to show you? No — no evidence supports that as of July 2026, and this article will not launder the claim. The documented truth is less cinematic and more unsettling: the platforms do not need your camera, because your thumb confesses everything. The measured signals — how many milliseconds you hesitate before scrolling past, what you rewatch, what you finish, what you share at 1 a.m. — map your arousal more efficiently than any eye-tracker could.

The confession the platforms actually read: hesitation, measured in milliseconds, on every scroll.

The ranking signals platforms themselves document are behavioural: watch time (the strongest, especially the first seconds), completion and rewatch rates, likes, shares, comments, follows, skips, and session patterns. Nothing in the documented stack involves the camera. The most vivid demonstration remains The Wall Street Journal's 2021 investigation "Inside TikTok's Algorithm": reporters deployed over a hundred bot accounts that expressed interests through nothing but how long they lingered on videos — no likes, no follows, no searches. Dwell time alone was enough to steer each bot into a narrow content tunnel, sometimes in under an hour (WSJ, July 2021). The bots had no eyes to track. Hesitation was the whole confession. I flag this precisely because this article is hard on the platforms: the case against engagement optimization is strong enough that it must not lean on a single unverified claim. The moment we assert things we cannot prove, we become the misinformation we are describing.

Now the personal testimony I promised — labelled as exactly that. Over the past year I investigated this machinery the only way a civilian fully can: from inside. I went in deliberately and with a method: a sustained mindfulness practice first, so I could watch my own reactions as data rather than be them; then, for two weeks, I let go completely — no curating, no resisting, no correcting the record in my own head — and let X and Facebook feed me whatever my hesitations taught them to feed me. I watched what one hesitation over one Islamophobic clip did to the next hour of my feed, and the next day of it. And here is the part I need you to hear, because it is the whole reason this article exists: within two weeks, I could feel Islamophobic sentiment forming in my own thinking. Me. A refugee. A man whose closest friends are Muslim, writing an article like this one. When I later described some of those creeping thoughts to Muslim friends, they were shocked — and so was I. If two weeks of surrender could bend my mind against people I love, what is a year of it doing to a lonely nineteen-year-old who never chose the experiment and doesn't know he's in it? Facebook's own researchers measured Carol Smith's journey in days. My replication agrees. (One person's experience proves nothing at scale, and I ran no control group. But I no longer needed convincing that the mechanism section above describes something real.)

There is a second entry in my self-experiment file, and it points the other direction — which is how I know it is the machinery and not one ideology. Through the war in the Middle East, my feeds served me, morning after morning, starving children and bodies torn apart by bombs. Nobody should be unmoved by real suffering, and I am not minimizing an atom of it. But notice the conditions under which millions of us were processing it: fresh out of COVID, working from home, talking to screens, seeing fewer humans in a week than we used to see in a day — atrocity delivered daily, grief with no one in the room to metabolize it with. And I caught the drift again: I found myself having to deliberately summon the faces of the Jewish nurses and friends who have been good to me for years — people whose kindness I have watched extended to everyone — because the feed was teaching my reflexes something my own life contradicted. That is the tell. When you need to actively remember your own friends to counter what a ranking system is installing, the question is no longer what you feel. It is who is doing the installing, and why.

The commercial logic is not a mystery and requires no conspiracy: attention is the inventory, advertising is the revenue, and outrage holds attention. The 5×-anger weighting, the two-day QAnon on-ramp, the out-group sharing premium — none of it was designed to spread hate. It was designed to sell ads, and hate turned out to be a high-performing genre. That is somehow worse. A conspiracy can be exposed. A business model has to be regulated.

The Feed Is Not the Country

If you learned Canada only from X, Facebook and TikTok, would you recognize the country outside your door? No — and that gap is measurable. The feed is engagement-ranked, which means it is a highlight reel of the worst days of 41 million people. The country is what happens at the base rate — and the base rates keep telling a different story than the timeline does.

Start with the ugliest trope the Moncton video seemed to confirm: "nobody helps." The largest real-world study ever done on this examined CCTV footage of 219 actual public conflicts in three countries and found that in 91% of them, at least one bystander intervened (Philpot et al., 2020, American Psychologist 75(1)). Intervention is the human norm. That finding makes the passivity captured in the Moncton footage more damning, not less — it was a deviation from how people usually behave, not a revelation of how they always do. And it means the lesson "no one will help you in Canada" — the lesson the clip taught millions of feeds — is statistically backwards.

Canada has a structural aggravator here that most countries do not: on the largest social platforms Canadians use, the journalism is missing. Since August 2023, Meta has blocked news links for Canadian users rather than pay under the Online News Act — which means a video like Moncton's circulates on Facebook and Instagram stripped of the reporting that would contextualize it. The CBC's careful account — what is verified, what is alleged, what investigators actually said — cannot legally travel alongside the clip on the platform where the clip lives. Canadians get the arousal; the facts are banned. If you designed a national information system to maximize misreading of events like Moncton, it would look like this.

Run the same correction on the country itself. In the feed's Canada, every parking lot is Moncton. In measured Canada, there were 4,882 police-reported hate crimes in 2024 — each one a real harm, and collectively about 13 a day in a nation of 41 million people making billions of daily interactions. In the feed's Canada, "go back to your country" is the national greeting. In measured Canada, the country leads the world in migrant acceptance and most immigrants' ledgers look more like mine than like the video. Both Canadas contain truth. Only one of them is weighted by what sells ads.

This is what I meant when I said I let myself be the guinea pig: after two weeks inside the feed, the outside world felt like a factual error. Then I would go outside — ride through my own city, talk to actual neighbours — and reality would reassert itself within the hour. The disagreement between your timeline and your street is not a tie. Your street is the primary source. My honest prescription, and yes, I sell a product adjacent to it so weigh my bias accordingly: log off and go be in the country you are worried about. Walk it. Ride it — any bike, including ours. Attend the car show anyway. The feed's Canada shrinks to its true size the moment you stand in the real one.

Measured Canada, at human speed: the country the feed cannot sell ads against.

What the schoolyard knows that the feed forgets

Here is the progress the doom-feed never shows you: watch children in a Canadian schoolyard in 2026. They do not sort each other by the race theories that ran the early 1900s — mostly they sort by running shoes and video games. That instinct was built by exactly the unsupervised mixing this article's light column describes: my ESL classroom, scaled to a country. It is also unguarded. Those same kids will spend their adolescence inside recommendation systems with no supervision at all — and the pipeline this article documents does not care that they started out colour-blind.

Two clips from this year show how the pipeline works on all of us. In April, in Woodstock, Ontario, a youth was arrested and charged with uttering threats and criminal harassment after video circulated of him shoving an elderly Sikh man on a sidewalk and screaming at him to get out of "my country" — police called it a suspected hate-motivated incident, and under the Youth Criminal Justice Act the accused cannot be named (CP24; CTV News, April 2026). Then in July, a video of a Sikh man being asked to leave a Windsor gym over his kirpan — after another member raised a safety concern — passed half a million views within days (Yahoo News Canada, July 2026). On the second one, the legal question was actually settled by the Supreme Court of Canada back in 2006: a blanket kirpan ban violates the Charter, and institutions have a duty to accommodate. But set aside who was right in that gym, because the feed did not serve that clip to millions of strangers to settle a Charter question. Ask the only question this article keeps asking: what is the benefit to society of a ranking system putting that confrontation in front of millions of people who were never in that gym? There isn't one. There is only the benefit to the system — outrage holds attention, and attention sells. Democracy means free speech; it has never meant that an advertising algorithm gets to decide, unsupervised, which two minutes of a country's day 41 million citizens will marinate in. Are we going to let what maximizes engagement determine the kind of society we live in? On current policy, that is not a rhetorical question — it is the default setting.

How to read the next viral video

There will be a next one. Before it tells you what your country is, run it through five questions — they take thirty seconds and they are the entire discipline of this article in miniature:

1. What is verified, and by whom? Separate what a reporter confirmed from what a caption asserted. If the only source is the caption, you have an allegation, not an event.
2. What happened before the clip started? Nearly every viral confrontation video begins mid-story. Not knowing the before does not excuse anything — it just means you do not yet know what you are judging.
3. Who benefits from my reaction? Someone clipped, captioned and posted this for an audience. The ranking system profits from your response either way. Neither fact makes the video false; both should slow your verdict.
4. What is the base rate? Ask what the measured frequency of this event actually is — 13 police-reported hate crimes a day in a country of 41 million is both "too many" and "vanishingly rare per interaction." Hold both.
5. What would I conclude if my street were the evidence? Your actual lived experience of your actual neighbours is a dataset too — usually a larger one than your feed's.

Should Canada Build a Firewall Like China?

After everything above, the tempting conclusion is the one I once reached myself: if the feeds are the vector, wall them off — build a Canadian Great Firewall. The honest answer is no: China's firewall does solve the specific problem of foreign-platform manipulation, but it is inseparable from domestic censorship, and Canada can take the transferable pages — algorithm regulation, minors' defaults, real penalties — without buying the cage.

What a firewall is when you stand in front of it: a wall somebody else owns, across your own road.

What the firewall actually is, and what it actually buys

China's Great Firewall is IP-blocking plus deep packet inspection at national scale: Facebook, Instagram, X, YouTube, Google and most major Western outlets simply do not reach Chinese screens (Britannica; Comparitech). Grant the argument its strongest form: a 2016-style foreign-platform influence operation is structurally impossible there — no foreign company's algorithm decides what Chinese citizens see. Meanwhile Canada's exposure is total: the platforms Canadians use most are 100% foreign-owned, and one of them has banned our news. If your only metric is "can an outside power run our information space," the firewall wins.

But watch what the same machinery does facing inward. The wall that keeps Meta out also erased #MeToo posts, Uyghur-language content, and the White Paper protesters of 2022 — people arrested for holding blank sheets of paper, because even blankness read as criticism. China's protective children's rules and its censorship are not two systems; they are one system with two outputs. Reporters Without Borders ranks China 178th of 180 countries for press freedom, with 121 journalists jailed (RSF, 2026). A firewall is not a filter you aim only at enemies. It is a switch someone owns — and Canadians would have to trust every future government with it, forever. That is not a trade a free country makes, and it is not one I am willing to make even after Moncton.

The pages that do transfer

Strip the censorship out and look at what China's regulators did to the attention machinery itself — because this part, remarkably, the democracies are now copying:

• Algorithm accountability. China's algorithm regulations (in force March 1, 2022) are the world's most extensive: platforms must offer a one-tap opt-out of personalized recommendation, let users view and delete the interest tags built on them, and are barred from addiction-engineering designs, with special protections for minors and the elderly (China Law Translate; China Briefing, 2022). Imagine simply having the legal right to see the tag file X has built on you — and delete it.
• Minors' defaults. Douyin — TikTok's Chinese-market sibling, same parent company — limits under-14s to 40 minutes a day, locks the app overnight, and weights their feed toward education. The company's Chinese children get the protective version; everyone else's children get the engagement-optimized version. That asymmetry alone tells you the risk is understood at the source.
• The democratic template already exists in Europe. The EU's Digital Services Act — fully in force for the largest platforms since August 2023 — requires them to assess and mitigate the systemic risks their recommender systems create, disclose the main parameters those systems use, offer users at least one feed option not based on profiling, and open their data to vetted researchers, with fines up to 6% of global revenue. No wall, no censor's office — just the end of the black box. Canada could adopt this framework largely as-written.
• And the West is already copying the homework. Australia's under-16 social-media ban — covering Facebook, Instagram, TikTok, YouTube, Reddit, Snapchat, Twitch and others, with penalties up to A$49.5 million — took force December 10, 2025, a world first (TIME; CNBC, December 2025). Malaysia has announced a ban for 2026; New Zealand and Indonesia are drafting; France has proposed an under-15 ban with a night curfew for teens; Spain is moving to 16; South Korea banned phones in classrooms from March 2026 (TechCrunch, July 2026). None of these countries built a firewall. All of them stopped pretending the feed is a neutral square.

Where Canada actually stands: late, and not yet law

Canada tabled its answer — Bill C-63, the Online Harms Act — in February 2024. It died on the Order Paper when Parliament was prorogued on January 6, 2025, and only on June 10, 2026 did a successor — Bill C-34, the Safe Social Media Act — reach first reading (Parliament of Canada, 45-1; Canadian Privacy Law Blog, June 2026). Read that against the timeline above: Australia's regime has been in force for seven months; Canada's answer spent eighteen months dead and has yet to pass a single vote. Our own commission of inquiry named information manipulation the single biggest risk to our democracy — and the legislative response to that finding is still not law.

It is not that Canada lacks the muscle. When Ottawa ordered TikTok's Canadian subsidiary wound up on national-security grounds in November 2024, then negotiated, the result — announced March 9, 2026 — was TikTok operating in Canada under binding undertakings, including stronger protections for minors (ISED; CP24, March 2026). Canada can extract enforceable conditions from a platform when it decides to. It has simply only decided to once.

The Scoreboard: Governments That Fought the Platforms

Has any government actually gone toe-to-toe with Facebook, Instagram, TikTok or X — and won? Yes: Brazil, India, Australia and the EU all have wins on the board, and every win shares the same three features — penalties that genuinely hurt, leverage the platform could not ignore, and a demonstrated willingness to lose the platform entirely. Canada has fought three times and split the series.

Follow the money: the Canadian ledger of interests

My reader asked the right question about all of this: who allowed it, and what were the financial interests? Here is what the public record shows, names and dollars attached:

• The government is a paying customer of the platform it failed to regulate. Ottawa was spending roughly $10 million a year advertising on Meta platforms when the C-18 fight began. It suspended those ads in July 2023 in protest of the news ban (CBC; Al Jazeera) — a real, creditable stand — and then, in February 2025, quietly resumed buying Meta ads while the news ban was still in force, a resumption confirmed by the Privy Council Office (Canada's National Observer; The Canadian Press). Read that sequence again: the platform banned Canadian journalism, outlasted the boycott, and won back the government's ad budget without conceding anything.
• The cheapest outcome for a platform is a deal that pre-empts regulation. Google's $100M-a-year payment bought it an exemption from the Online News Act — a fraction of what a genuine revenue-share regulation might have cost, as analysts noted at the time (The Conversation; Nieman Lab). The deal is real money for Canadian journalism, and it is also the price of making the law not apply.
• Everyone who could regulate the feed also campaigns on it. Every federal party buys its reach from the same recommendation systems Parliament has failed to govern — the structural conflict nobody has to be corrupt to be caught in.
• And the sums are asymmetric. Australia's maximum penalty (A$49.5M) and Brazil's collected fines (US$5.2M) are rounding errors against platform revenues — which is why the wins above were never really about the money. They were about credibility: Brazil proved it would flip the switch, India proved it would live without the app, Australia proved it would pass the law anyway. Canada's C-63 proved the opposite.

What You Can Actually Do

National regulation is Parliament's job, but the accelerant supply chain runs through individual phones — yours included — and there are evidence-aligned moves available to you today. None of them require a hashtag. Most of them are quiet.

• Starve the machine of the victim's worst moment. Don't share raw footage of a person being beaten — not even in outrage. Sharing it pays the exact ranking systems this article documented, and re-publishes a family's trauma. Share the reporting instead; the CBC's coverage of Moncton needs no accompanying beating to be believed.
• If you witness an incident: be the 91%. Call 911 if anyone is in danger. If safe, go stand with the person targeted and speak to them — "I'm staying with you until this is over" — rather than debating the aggressor. Record only if it is safe, and give the footage to police and the victim, not to your followers. And if a child is present, someone's whole job is that child.
• Report what you see. Hate crimes to your local police (it is the count in next year's StatCan table); platform content through the in-app reporting tools — which feed the enforcement records regulators rely on.
• Audit your own feed the way I did — briefly, and on purpose. Watch one session with one question: "what is this ranking system learning from my hesitations?" Then use the controls that exist: turn off personalized recommendations where offered, prune your subscriptions, and treat the "angry" reaction as what Facebook's own scientists found it to be — a vote for more of the same.
• Say the boring, load-bearing sentence out loud. When someone bills a community for an individual — in either direction — say the sentence: "individuals did this; communities don't stand trial." It is the single most protective social norm this article knows of, and it costs eleven words.
• Tell your MP one specific thing. Not "do something about online hate" — ask why Australia's minors-protection regime has been in force since December 2025 while Canada's first attempt died in January 2025 and its successor, Bill C-34, is still not law — and when it will be. Specific questions are the ones that get answered.

The 91%: in real public conflicts, someone stays (Philpot et al., 2020). Be the someone.

Sources & Further Reading

Every factual claim in this article traces to the primary sources below, grouped by section. Where a source is behind a paywall, a second open source reporting the same document is listed where available.

The 2026 incidents (Moncton · Woodstock · Windsor)

• CBC News — Police in Moncton investigating "whether hate was a factor" in altercation depicted in video
• CTV Atlantic — N.B. RCMP investigate possible hate crime as assault video circulates
• Global News — New Brunswick RCMP investigates whether Moncton altercation was hate-motivated
• CP24 — Youth charged after shoving, threatening Sikh man in Woodstock, Ont. (April 2026)
• Yahoo News Canada — A Sikh man with a kirpan was asked to leave a GoodLife gym: what to know (July 2026)

Canadian data and history

• Statistics Canada — Police-reported hate crime in Canada, 2024 (The Daily, March 30, 2026)
• Gallup — Canada No. 1 for Migrant Acceptance
• Environics Institute — Canadian public opinion about immigration and refugees, Fall 2025
• Truth and Reconciliation Commission of Canada — Final Report (2015), via the National Centre for Truth and Reconciliation
• Prime Minister's Office — Apology for the fate of the passengers of the MS St. Louis (November 7, 2018)
• Prime Minister's Office — Komagata Maru apology in the House of Commons (May 18, 2016)
• Amnesty International — Millions unlawfully deported to Afghanistan in 2025
• Canadian Museum of Immigration at Pier 21 — The boat people, private sponsorship and the 1986 Nansen Medal
• The Conversation — The story behind the world's first private refugee sponsorship program

The two attacks, from the court record

• National Observer — Bissonnette searched online for Trump more than 800 times before the mosque attack
• CBC News — Court documents on the Québec City shooter's online activity
• The Globe and Mail — The London attack was terrorism, judge rules
• Public Prosecution Service of Canada — Terrorism finding in the Veltman case (February 2024)

The platforms' own documents and studies

• NBC News — "Carol's Journey to QAnon": what Facebook knew about how it radicalized users
• The Week — Facebook gave emoji reactions five times the weight of a like (Facebook Files reporting)
• Kramer, Guillory & Hancock (2014) — Experimental evidence of massive-scale emotional contagion, PNAS 111(24)
• Lawfare — Facebook’s Role in Myanmar: the UN mission’s "determining role" finding and the BSR report
• Meta/BSR — Human Rights Impact Assessment: Facebook in Myanmar (November 2018)
• Amnesty International — The Social Atrocity: Meta and the right to remedy for the Rohingya (2022)
• CNBC — Facebook shields millions of VIP users from standard moderation (XCheck)
• Forbes — Project Amplify: Facebook tested pushing positive stories about itself
• Rolling Stone — Musk had engineers boost his tweets after the Super Bowl
• Meta Newsroom — "Mark Zuckerberg Stands for Voice and Free Expression": the Georgetown speech, full text (October 2019)

The peer-reviewed science, both directions

• Huszár et al. (2022) — Algorithmic amplification of politics on Twitter, PNAS 119(1)
• Rathje, Van Bavel & van der Linden (2021) — Out-group animosity drives engagement, PNAS 118(26)
• Guess et al. (2023) — How do social media feed algorithms affect attitudes? Science 381
• Hosseinmardi et al. (2021) — Examining the consumption of radical content on YouTube, PNAS 118(32)
• Chen et al. (2023) — Subscriptions and external links drive resentful users to extremist channels, Science Advances 9(35)
• Bail et al. (2018) — Exposure to opposing views can increase polarization, PNAS 115(37)
• Allcott et al. (2020) — The Welfare Effects of Social Media, American Economic Review 110(3)
• Philpot et al. (2020) — CCTV footage shows intervention is the norm, American Psychologist 75(1)

Law and regulation

• Hogue Commission — Public Inquiry into Foreign Interference, Final Report (January 28, 2025)
• Norton Rose Fulbright — Bill C-63: The Online Harms Act (status)
• Parliament of Canada — Bill C-34, the Safe Social Media Act (tabled June 10, 2026)
• Canadian Privacy Law Blog — "Online Harms" is back: an overview of Bill C-34 (June 2026)
• CNBC — Australia's under-16 social media ban takes effect (December 10, 2025)
• China Law Translate — Provisions on the Management of Algorithmic Recommendations (in force March 1, 2022)
• CP24 — TikTok to continue operating in Canada subject to conditions (March 2026)
• CNBC — Brazil's X ban puts Musk's Starlink in the crosshairs (September 2024)
• NBC News — X and Starlink face nearly $1M daily fines in Brazil
• Al Jazeera — Brazil unfreezes X and Starlink accounts after fine transfer
• Al Jazeera — India makes its TikTok and Chinese-app ban permanent (January 2021)
• SCOTUSblog — Supreme Court upholds the TikTok divest-or-ban law (January 17, 2025)
• CBC News — Google signs the $100M-a-year deal for Canadian news
• CBC News — Ottawa suspends its advertising on Facebook and Instagram (July 2023)
• Canada's National Observer — Federal government is once again advertising on Meta platforms (February 2025)

The Bottom Line

Is Canada racist? Canada is a country with racists in it, a documented racist past it has formally owned, hate-crime numbers that have doubled in six years, the world's highest measured acceptance of migrants, and the only population on Earth ever awarded the UN's refugee medal as a people — all at once. The woman in the Moncton parking lot deserves justice, and her seven-year-old deserves a country that tells the truth about what happened to his mother: it was real, it is being investigated as a possible hate crime, and it is not the whole story of Canada.

But the deeper finding of this article is about the lens, not the country. Between you and Canada sits a ranking system that pays five points for anger, routes test users to conspiracy groups in two days, doubles the reach of out-group contempt, marketed itself as your public square while running whitelists and multipliers, and — in the one place the UN examined at full scale — played a determining role in real hatred. That system is what showed you Moncton. It showed you the truth, selectively, at the angle of maximum arousal, and it will show you the next one the same way. Our own commission of inquiry named this machinery, not any foreign army, the single biggest risk to Canadian democracy — and Canada currently has no law in force that governs it.

So hold both resolutions at once. Demand the boring, Australian-style guardrails from Parliament — protecting citizens from corporate machinery that profits on division is a government's job description, not censorship, and a country cannot outsource its moral sense to Facebook — and refuse the feed the verdict it keeps offering you about your neighbours. I came here as a refugee. I have met Canada's racists, and I have met Canada. They are not the same size. Do not let a ranking system convince you otherwise.

Visuals created by Playcut.ai