Grief Is the Gateway: The Cognitive Engineering of South Africa's Xenophobic Machine

South Africa's latest wave of anti-foreigner mobilisation began with the emotions linked to a missing person. Understanding why that matters is the difference between fighting symptoms and fighting the machine.

In early 2026, emotional appeals about the disappearance of a young man, Mazwi Kubheka, spread across X under hashtags like #BringMazwiBack and #JusticeForMazwi. Within days, a community-driven search had mutated into a national anti-immigrant mobilisation, complete with deadlines, marches, and deaths. A year earlier, the same thing happened with a Vosloorus spaza-shop owner reported murdered by a "Somali syndicate", a man later found alive.

This piece is about the engineering behind that shift, and how emotion becomes the raw material used to capture our attention. The feeling is rarely the point; it is the vehicle for mobilising a crowd.

The machine is old; only the sophistication is new

The Daily Maverick reported on the Murmur analysis of nearly four million posts between January and May this year, which found highly active online communities repeatedly amplifying immigration-related content: a textbook signature of inauthentic coordination.

It reminded me of the Bell Pottinger scandal of 2017, when a commercial firm built fake accounts to manufacture racial division for a paying client in South Africa. Political demand meets manipulation-for-hire. South Africans have polled negatively on migration for as long as we have measured it. What is new is how that diffuse anxiety gets converted into targeted action.

Why a missing youth, and not a statistic

The playbook has a consistent step: hijack an emotionally charged, unresolved event before the facts are in. It is one of the most cognitively efficient entry points available, for reasons that map cleanly onto decades of research.

The affect heuristic.

When feeling runs high, we substitute "how do I feel about this?" for "what actually happened?" Emotion operates as a fast, pre-rational shortcut for judgment. A missing young man triggers protective fear and grief that arrive long before any verification can.

The identifiable victim effect.

A single named, faced person mobilises far more emotion, and far more action, than abstract numbers about migration ever could. The movement personalises deliberately: one Mazwi outperforms a thousand data points.

Negativity and threat bias.

Threat-related content, like kidnappings, syndicates and crime, is processed faster, weighted more heavily, and shared more readily, feeding our evolved threat-detection machinery.Platforms reward outrage, not accuracy. Brady and colleagues found that each moral-emotional word in a message measurably increases its spread. That emotion is converted into engagement, and engagement into reach. Outrage also does double duty: online "mobbing" lets people attack a designated wrongdoer while affirming their own virtue, so joining the pile-on feels less like cruelty than like justice.

Manufactured consensus and the speed of the lie

Once an event is seized, several mechanisms lock the narrative in place. The medium does much of the work: platforms optimise for velocity over veracity, and with the old editorial gatekeepers gone, a falsehood outruns its correction by default. Much of what spreads is forwarded on the strength of a headline, for social approval, before anyone has weighed it.

On top of that sits illusory truth through repetition. When "buzzer" cohorts repost the same claim in synchronised bursts, they create a cognitive fluency that feels like truth. The most effective disinformation convinces a few people that thousands are already convinced: deliberately fabricated social proof.

We are herd animals; a manufactured majority is enough to move us.

Then there is the asymmetry between a lie and its correction. Brandolini's principle holds that refuting nonsense costs far more effort than producing it. By the time AFP Fact Check has debunked a recycled video, or Durban authorities have confirmed that of 457 foreign nationals checked only two lacked papers, the emotional payload has already detonated, and the continued-influence effect means the false belief lingers even after the correction is accepted. Verification is structurally too slow to be a defence on its own.

Grimes calls the industrial version the firehose of falsehood: the high-volume, multi-channel model refined by Russian influence operations, flooding every platform with a fast, repetitive and often contradictory stream of claims. The aim is not to win an argument but to exhaust the audience's capacity to tell true from false, and to fabricate consensus through sheer volume. South Africa's synchronised buzzer cohorts are a local, lower-budget expression of the same logic, and it is the same instrument that resurfaces around the pro-Rosatom, anti-NGO narratives the Daily Maverick reported on.

From metaphor to permission

There is a darker linguistic layer worth naming. The movement's vocabulary, "Abahambe" (they must go), "invasion," "Clean SA", is the language of cleansing and infestation. Nick Haslam's work on dehumanisation identifies exactly these metaphors as the markers that strip a group of full human status. This matters because dehumanising language lowers the moral inhibition against violence: once people are framed as contamination to be removed, harming them is recoded as hygiene. The metaphor does quiet work an explicit call to violence cannot, which is also why it survives content moderation.

By staying "within bounds" while signalling, the speaker lets the audience supply the extreme inference themselves, and a conclusion you generate yourself is far more persuasive than one you are handed. Plausible deniability is a powerful persuasion technique.

The bait and the switch: foreigners first, NGOs next

Here is where the mechanism becomes genuinely alarming, because the affect is often a delivery vehicle for something the mobilised crowd never signed up for.

Consider one of the most polished campaigns in the current wave, examined in a Daily Maverick investigation by Rebecca Davis. South Africans for Constitutional Reform (SACR) presents itself as a citizens-before-foreigners movement: the most-repeated donor message on its crowdfunding page was simply "South Africa for South Africans," alongside the familiar isiZulu "abahambe", they must leave. Same affect, same grievance, same emotional fuel as the street movements.

But the campaign's actual centrepiece, buried beneath the anti-migrant framing, is a demand to rewrite the Constitution to bring NGOs under state oversight. Its founder and chair, Princy Mthombeni, is a nuclear-industry figure who has spent years arguing that the abandoned Zuma-era Rosatom nuclear deal should have gone ahead, a position that aligns neatly with Russian strategic interests and sits inside a broader narrative recasting civil-society and media opposition to that deal as illegitimate "interference with sovereignty."

Princy Mthombeni, source: X

Look at what has happened structurally. The emotionally resonant issue, foreigners, is the front-end that recruits the crowd and the donations. The payload is entirely different: constraining independent civil society and advancing an energy-and-sovereignty agenda with a distinctly pro-Moscow tilt. Supporters believe they are defending jobs; the constitutional fine print does something else.

This is affect transfer weaponised as bait-and-switch: the strong feeling attached to migration is transferred onto a logically unrelated demand, and because the crowd is already mobilised and already trusting, it carries the second payload without inspecting it.

Two further mechanics deserve naming. The first is legitimacy laundering: SACR translates raw street xenophobia into the sober register of "constitutional reform," sanitising an ugly affect into a demand that can be presented to Parliament with a straight face. The second is the choice of target. A constitutional clamp on NGOs, justified as protection against foreign interference, echoes the "foreign agents" template used to hollow out civil society elsewhere, most notably in Russia. That parallel is not proven coordination, but the pattern is the point. Once you can move a population on feeling alone, you can attach almost any payload to the vehicle.

Why this works on us, and what actually defends against it

A coherent story will always beat an accurate one. "South Africa is under siege from foreigners and an incapable state" explains unemployment, crime, and collapsing services all at once. It offers a villain, and therefore a sense of agency, to people living with real and unaddressed frustrations. Their grievances are genuine; but the attribution is engineered. That is what makes the narrative so resistant to correction: you are not arguing with a fact, you are arguing with a feeling that has found a home.

Two features of the medium harden it further. Echo chambers mean the narrative is mostly met among people who already accept it, so it rarely encounters friction. And the framing is relentlessly binary: you are with South Africans or with the foreigners, patriot or sell-out. This manufactured false dichotomy is itself a manipulation, collapsing a complex policy reality into a loyalty test and foreclosing the nuance any workable solution needs.

The target throughout is fast, affective, System 1 cognition; the whole architecture is designed to stop slow, deliberate System 2 thinking from ever engaging. Which is why the defence cannot be fact-checking alone. By the time you are correcting, you have already lost the speed race.

What can we do about it?

The more promising interventions act before the payload lands:

• Prebunking: exposing people to the manipulation technique itself, and pre-positioning the counter-narrative, so the pattern is recognised next time, regardless of content. In a speed contest, only the message already there can compete.

• Friction: anything that interrupts the reflexive share and gives System 2 a moment to engage. The share button is the weapon; the pause is the defence.

• Digital mindfulness: the capacity to notice when one is being moved at speed, and to treat that feeling as a signal to slow down rather than act.

  
  

The movement's ominous 30 June "deadline" for foreigners to leave came and went without the mass violence many feared. That is a relief, and a credit to the civil-society groups, journalists and communities who pushed back. But it would be a mistake to read the quiet as failure. Spectacular violence was only ever one possible output. The real yield of an operation like this is the mobilised audience, the normalised narrative, and the constitutional payload now sitting in the national conversation.

A pogrom that does not happen does not refund the attention already captured, or retract the demand already laundered into respectability.

Attention is the perimeter now, and it is breached daily through our oldest and most human responses: fear for a missing child, grief for a dead neighbour, the need to belong to a righteous majority. The networks exploiting South Africa's xenophobia have grasped something our institutions have been slow to see: you do not need to change what people know if you can control what they feel, and when.

Sources

Brady, W. J., Wills, J. A., Jost, J. T., Tucker, J. A., & Van Bavel, J. J. (2017). Emotion shapes the diffusion of moralized content in social networks. PNAS, 114(28), 7313–7318.

Brandolini, A. The bullshit asymmetry principle.

Crockett, M. J. (2017). Moral outrage in the digital age. Nature Human Behaviour, 1, 769–771.

Davis, R. (2026). Foreigners first, NGOs next: The buried demand inside a viral constitutional campaign (with accompanying video, "South Africa's anti-migrant campaigns: What's really driving them?"). Daily Maverick.

Findlay, K. (2026). How South Africa's xenophobic online machine was rebooted in 2026. Daily Maverick.

Grimes, D. R. (2019). The Irrational Ape: Why flawed logic puts us all at risk, and how critical thinking can save the world. Simon & Schuster.

Haslam, N. (2006). Dehumanization: An integrative review. Personality and Social Psychology Review, 10(3), 252–264.

Kahneman, D. (2011). Thinking, Fast and Slow. Macmillan.

Lewandowsky, S., Ecker, U. K. H., Seifert, C. M., Schwarz, N., & Cook, J. (2012). Misinformation and its correction: Continued influence and successful debiasing. Psychological Science in the Public Interest, 13(3), 106–131.

Slovic, P., Finucane, M. L., Peters, E., & MacGregor, D. G. (2007). The affect heuristic. European Journal of Operational Research, 177(3), 1333–1352.

Small, D. A., Loewenstein, G., & Slovic, P. (2007). Sympathy and callousness: The impact of deliberative thought on donations to identifiable and statistical victims. Organizational Behavior and Human Decision Processes, 102(2), 143–153.