What a petrol bomb buys in Belfast

Two nights, one script

On the night of 8 June, Hadi Alodid, a 30-year-old Sudanese man, attacked Stephen Ogilvie with a kitchen knife in north Belfast; Ogilvie survived but was hospitalised with major injuries, and a court later heard he lost his left eye. Alodid was charged with attempted murder the same day. He entered the UK in 2023 and was granted refugee status that year, with leave to remain until 2028.

Then the script ran. Masked men claiming to be "getting the foreigners out" kicked in doors and windows on the Lower Newtownards Road; at least three houses, a Middle Eastern supermarket, a Glider bus and numerous vehicles were set ablaze across Belfast. By the second night, at least 16 people had been arrested and 12 officers injured, with police firing water cannon at demonstrators who had lit fires in the street near Newtownabbey.

The targeting was administrative in its precision. A list of Belfast addresses where immigrants were said to be living circulated on social media, and a nursing union official said ethnic minority nurses had been chased by masked men on their way to work, while police patrolled neighbourhoods on what she called a "hit list". Northern Ireland Secretary Hilary Benn was blunt: "if you are targeting people on the basis of the colour of their skin, how else can you describe them?"

One more fact, the one this column turns on: Northern Ireland has now seen anti-immigration unrest for three consecutive summers.

The deep problem: violence that delivers

The surface issue is racism, and the racism is real. The deep issue is an incentive structure.

Go back twelve months. In June 2025, riots broke out after two Romanian Roma teenagers were charged over an alleged assault in Ballymena. Across two weeks, 107 police officers were injured and 56 people were arrested, with 27 remanded into custody. The legal system did its work on individuals. But look at the outcome that mattered to the rioters: of the roughly 1,200 Roma who lived in Ballymena before the riots, two thirds were reported to have left. As the Guardian put it at the time, "The rioters, after all, got what they wanted. They won." A bitter coda: the charges against the two boys were dropped in November 2025 after new evidence meant the case no longer met the threshold for prosecution. The pretext evaporated; the expulsion stood.

This is why I resist calling this week an eruption. It is a transaction with a known exchange rate. Thomas Schelling taught that deterrence is not about punishment after the fact but about credible commitment: an adversary must believe, in advance, that the move cannot pay. The UK state has the opposite record in Northern Ireland. It prosecutes rioters and preserves their winnings. Every conviction says crime does not pay; every emptied street says it does.

There is a second layer, and it deserves a different evaluation. Judith Shklar argued that the first obligation of a liberal state is protection from cruelty and fear, before any loftier goal. By that standard, a Black nurse chased to work has suffered a state failure even if no law was broken in her particular case, because the state's job was to make that fear impossible to manufacture. SDLP leader Claire Hanna called the scenes "a race-based pogrom", and the word is technically apt: a pogrom is defined not only by mob violence but by the authorities' failure to prevent or reverse it.

A third layer: the amplification machine. Tommy Robinson posted planned demonstrations across the UK early after the attack, and Elon Musk was among US right-wing figures calling for demonstrations. This layer is real but secondary. Amplifiers can only sell a product that works locally.

Three tempting responses that miss

"Full force of the law." Keir Starmer vowed to crack down on anyone fuelling the division, and prosecution is necessary. But Ballymena had 56 arrests and the Roma still left. Punishing individuals while ratifying the collective outcome is exactly the bargain rioters have already priced in.

"This reflects legitimate immigration concerns." Concerns about asylum policy are a legitimate democratic topic. But the people best positioned to invoke the attack rejected this move: Ogilvie's own family said the attack should not be used to incite violence and that they "do not want this terrible tragedy to be used to divide people or fuel hostility". A policy debate does not require a hit list of addresses. Whatever needs the riot serves, immigration policy is not among them.

"Blame the platforms." Justice Minister Naomi Long criticised far-right commentators who were clearly trying to stoke racial tensions, and she is right about the accelerant. But moderation fights the spark, not the payoff. If expulsion stops working, the posts lose their customers.

The fix: make the riot fail, visibly

My verdict: the decisive failure here is not rhetorical or even prosecutorial. It is the state's refusal to treat the demographic outcome of intimidation as something it must reverse. The intervention is a governance commitment, jointly owned by the Northern Ireland Office, the Stormont Executive and the PSNI:

1. A funded right of secure return. Every household displaced by intimidation gets a guaranteed, protected, financially supported path back to its home or an equivalent one in the same area, with PSNI protection attached for as long as needed. Return must be the household's free choice, but the offer must be real, budgeted and standing.
2. Prosecute the lists as the weapon they are. Circulating addresses of immigrant households is targeting infrastructure, not speech about immigration. The Public Prosecution Service should charge distribution as intimidation, and the PSNI should say in advance that it will.
3. Publish the scoreboard. Six and twelve months after any episode of communal intimidation, the Executive publishes how many displaced residents returned. This single number is the deterrent. It tells the next mob, before it gathers, that the exchange rate is now zero.

The best case against me

The strongest objection is humane, not cynical: displaced families may not want to return to streets where neighbours watched them burned out, and a state that engineers returns to satisfy a deterrence metric is using frightened people as policy instruments. I concede the core of this. Return must never be pressured, and some families will rationally choose to rebuild elsewhere; that choice deserves the same funding.

But the objection mistakes the mechanism. The commitment is not that every family returns; it is that the option is guaranteed, protected and paid for, so that any departure is a choice rather than a verdict imposed by a mob. Deterrence does not require coerced returns. It requires that rioters can no longer point to an emptied street as proof of victory. There is also a capacity objection, that the PSNI is overstretched, and that is true; it is an argument for budgeting the commitment explicitly, not for continuing to improvise.

Bottom line

Benn's phrase, racist thuggery, is accurate and inert. Thuggery is a moral category; what Belfast faces is a working strategy, refined over three summers, with a proven product: streets with fewer foreigners on them. States do not end strategies by deploring them. They end them by changing what the strategy yields. Condemn less, restore more, and publish the number.