The Returns
Europe tested its deportation machine. It failed. So it became the law of the continent.
Gjadër is not a place many Europeans could find on a map, or would know how to pronounce. In the Albania of my childhood it was a rumour, coming with a real runway: a military airbase near Lezha where the dictatorship had cut tunnels into the side of a mountain and parked its MiGs inside the rock, safe from the invasion that never came. Readers of last week’s essay will recognise the species of place. Sazan was the regime’s island; Gjadër was its cave. The same paranoia that sealed a green island in the Adriatic hollowed out a mountain in the north, and both stood empty for thirty years after the system that built them died.
Then, in October 2024, an Italian navy ship arrived on the coast below, and Gjadër found its second career. Italy had built two camps on Albanian soil — a screening hotspot at the port of Shëngjin, a detention centre at the old airbase — to hold asylum seekers picked up in the Mediterranean while their claims were processed remotely, under Italian law, by judges sitting in Rome. Giorgia Meloni called it a model for the whole continent. It cost somewhere between €650 million and a billion euros, depending on who is doing the counting and what they are willing to count. It was, by every measurable standard, an unqualified failure.
Hold that thought, because on the first of June, in a Brussels negotiating room, that failure became the model. The Council and the European Parliament struck a deal on the new Return Regulation, the strictest migration law the bloc has ever agreed, and at its heart sits the very mechanism that has spent twenty months dying in the Albanian courts: deportation centres outside Europe, in countries the deported may never have set foot in. Officials briefed the press, without apparent embarrassment, that the Albanian centres would serve as the proof of concept. The migration commissioner, Magnus Brunner, supplied the language of the week: control. “A really very important step in making sure that we have control over what is happening in the EU, over who comes but also who has to leave.”
I want to take that sentence apart slowly, because three different stories are folded inside it — a practical one, a political one and an older one that begins on a beach. None of them is about control.
The official story is a percentage
Here is the case for the regulation, as Brussels tells it. Only around 20 per cent of the people ordered to leave the European Union actually go. A legal order that is enforced one time in five is not a law but a suggestion; a system like that corrodes public trust faster than any far-right poster could; therefore the system must be made faster, simpler, more effective. Every interview, every press release, every leaked negotiating note circles back to the number. Europe has a returns problem, and the proof is 20 per cent.
So let us do something nobody quoting the number ever does, and ask who the other 80 per cent are.
One of them is a Bangladeshi fruit picker in Puglia whose appeal is still moving through a court, which is to say a man exercising a right the European Union wrote down on purpose. One is a woman whose country of origin will not issue the documents to take her back, a refusal she can do exactly nothing about. One is a man who cannot be put on a plane to Kabul because European judges have ruled on what awaits him there. One is the same person counted twice, in two member states, a single human being doing the work of two statistics. Stack up the cases, as researchers at CEPS and elsewhere have done, and the scandal shrinks with every file you open: a large share of the famous gap between ordered and executed is not administrative failure at all. It is the law, working — appeals, non-refoulement, the inconvenient machinery that Europe says distinguishes it from the regimes people flee. That machinery has a more familiar name: European values; the same ones the new regulation has been drafted to streamline away.
Notice, too, what the vocabulary is doing before the law does anything at all. Returns, as though the journey were a parcel’s, as though somewhere a sender was waiting. Hubs, a word borrowed from airports and logistics, warm with the promise of onward connections. Place of residence or other relevant premises, a phrase that had to be written by someone who knew exactly which doors it would open and preferred not to list them. This is the real craft of the document: a question about rights — who may be locked up, for how long, what happens to their children, whether a state may fly them to a country they have never seen — has been re-described, clause by clause, as a question about throughput. And once the problem is just a number, every tool that raises that number reads as competence. Detention stretched from six months to two years, a six-month extension on top, no limit at all for anyone labelled a security risk: efficient. Entry bans doubled to ten years, lifetime for the security cases: efficient. The power to search a migrant’s home: efficient. Families with children eligible for removal to an offshore hub, only unaccompanied minors spared the hub, though not the detention: efficient. The old requirement that you could only be deported to a country you had some connection with: deleted, as friction.
Eleonora Celoria, a lawyer with the Italian legal network ASGI, read the home-search clause and gave Euronews the only gloss it needs: “The provision is vague on purpose, to allow a broad interpretation in the different member states. It opens the doors to home raids and also raids in the premises of associations helping migrants and healthcare facilities.” Keep her sentence in mind. We will need it again at the end.
The pilot failed in the right direction
Now back to Gjadër, because the practical story is the strangest of the three. Europe did not adopt this model sight unseen. It ran the experiment, at full scale, with real money and real people and the results are public. I can be more precise than that: someone I love went and looked.

On 22 November 2024, Francesca Romana D’Antuono, co-president of the pan-European party Volt Europa and, I should disclose, my partner, entered the Gjadër centre with four members of European parliaments. Six weeks earlier the Italian government had declared it open and operational. What she found was a building site pretending to be an institution: work machines parked and switched off with no one near them, a first-reception block with 350 of its planned 880 places finished, a repatriation wing with twenty-four. The cooperative contracted to run the centre had just sent its workers home to Italy; they were paid per migrant and there were no migrants. What remained, on a site of seventy-seven square kilometres, was a handful of Albanian cleaners, seven administrators and twelve Italian agents out of the 295 the plans called for, guarding cell after empty cell. Don’t you get bored here? Francesca asked some of them. A shrug came back and an answer I have not been able to forget, because half of twentieth-century history fits inside it: “There’s nothing to do. But we have to stay, because we were asked to.”
The head of logistics, at least, kept faith. If the procedures were ever activated, he told the visitors, the centre could process up to three thousand asylum requests a month. Three thousand a month. Italy, as a whole country, processes roughly that many in a year.
The procedures were activated, after a fashion. The first group of migrants Italy shipped over in October 2024 was ordered back to Italy by a Rome court within days; so was the next, and the one after that. For over a year the scheme ran as a kind of judicial tennis — men transferred at naval expense, the ship alone costing €300,000 a day, lawyers filing, judges returning them — until the Grand Chamber of the Court of Justice ruled in August 2025 that the “safe country of origin” designations underpinning the fast-track procedure could not stand without real judicial scrutiny and Italy’s did not survive it. In its first nine months the operation processed 111 people. Follow the Money costed the construction at €72,461 per bed; in one five-day stretch of actual operation the centres burned €114,000 a day. By the spring of 2025 the government had quietly converted Gjadër into an ordinary Italian pre-removal detention centre that happens to sit in another country, holding a few dozen people transferred not from the sea but from Italy itself. The hotspot at Shëngjin is, in practice, closed. Through all of it, one line in the budget never failed, one service was never interrupted, one performance never missed a night: the empty cells were guarded. Whatever else Gjadër could not do, it could always pay men to watch nobody.
A normal policy process buries a pilot like this and never speaks of it again. Instead, the European Union wrote the model into a regulation for 450 million people and the Italian prime minister who built it greeted the news by promising that now, with the new rules, the centres in Albania will finally work.
For once, I believe her.
Because you have to ask what the experiment was actually testing. If the metric was returns, people lawfully and durably removed, Gjadër failed beyond rescue. But if the metric was the performance of control — headlines about ships leaving, footage of a state visibly doing something hard and far away — then Gjadër succeeded every single week of its existence, including the weeks the courts emptied it. The judges who blocked it did not spoil the show; they joined the cast, as the obstacle that proves the government is fighting against the establishment. And the new regulation has been drafted, with some care, to write that obstacle out of the script. The pilot did not fail and get scaled anyway. It failed at the stated goal, succeeded at the real one, and the law has now been rebuilt around the success. It failed in the right direction.
They saw the horror first
The political story is the one I most want my European friends to sit with, because it destroys the comforting idea that this happened in ignorance.
Europe watched American immigration enforcement throughout 2025 the way you watch a house fire across the street: horrified, grateful for the distance, unable to look away. The masked agents, the courthouse arrests, the people who died in custody; by the Guardian’s count, more than thirty deaths connected to ICE operations in 2025 alone. In January 2026, federal immigration officers in Minneapolis killed Renee Nicole Good and Alex Pretti, and the names crossed the Atlantic within hours, into every feed, under every clip. I watched the revulsion move through my own comment sections in real time. For once it was not performative. It was the sincere shock of people discovering what an unaccountable enforcement agency looks like.
And here is the part that matters: when the horror came close, Europeans moved. In January, news broke that ICE agents would be present at the Milan–Cortina Winter Olympics. The mayor of Milan, Giuseppe Sala, did not reach for diplomatic language: “This is a militia that kills. A militia that enters people’s homes signing its own permission slip. They are not welcome in Milan.” A petition to deny the agents entry passed fifteen thousand signatures; people marched in Milan in February. And the pressure worked, in the way pressure actually works, not as a miracle, as a fence: the interior minister was dragged before parliament for an urgent briefing and made to promise, on the record, that ICE would conduct “no operational policing activity“ on Italian soil. The agents were confined to an operations room in the American consulate, watching the Games like everyone else. Italians had seen the horror, named it and caged it. Remember that. It is the proof, filed in advance, that none of what follows was inevitable.
Because while citizens were building the cage in Milan, the toolkit was being copied in Brussels.
To follow what happened next you need one piece of post-war furniture. Since 1945, the mainstream parties of Europe have lived by an unwritten rule with a French name, the cordon sanitaire: you may lose votes to the far right, you may even chase its themes, but you do not pass laws that depend on its hands going up. The rule is not etiquette. It is a load-bearing wall, the practical form of the promise that the catastrophe would not be allowed a second run. Every European parliament has its version, and for eighty years, with lapses, it held.
Now watch the votes. In March, in the European Parliament’s civil liberties committee, the official rapporteur’s text on the Return Regulation, already harsher than the Commission’s proposal, was voted down. What passed instead was an alternative drafted by François-Xavier Bellamy of the EPP, the centre-right family of Ursula von der Leyen, and it passed because three groups to the EPP’s right voted for it: the national-conservative ECR, Viktor Orbán’s Patriots for Europe and the AfD-led ESN, all of them different shades of hard-right. On the defining rights legislation of the decade, the centre did not merely brush against the far right. It built its majority out of it. Two weeks later the full Parliament confirmed the mandate, 389 votes to 206. The rapporteur, Renew’s Malik Azmani, defended the method in a sentence that deserves preservation in amber: “As rapporteur, I was always inclusive in this parliament that goes from the left to the extreme right, because for me the content is important.“ Read it twice. Inclusive: the gentlest word European liberalism owns, repurposed to describe the dismantling, brick by brick, of the wall built to keep the extreme right away from the law.
The far right, in other words, did not need to win the institutions. It needed the centre to decide that the far right’s toolkit was simply administration. And the centre obliged, in the same season its own analysts were warning, in so many words, that the regulation would “ICE-ify” European migration policy. That CEPS commentary was published on the morning of the plenary vote. It described the detection raids, the expanded detention, the policing turn, the American playbook adopted “precisely as that playbook’s costs are now becoming abundantly clear.” Nobody can claim the warning came too late. It came the same day and it has been read thirty-eight thousand times since, which is to say: they knew and we knew and it passed anyway.
This is the move we must learn to catch, here and everywhere: when the centre absorbs the far right’s premises and re-labels them efficiency, nothing visible happens. There is no torchlight march, no constitutional rupture, nothing for the cameras. There is a trilogue that runs late on a Monday, a commissioner saying control and a Green MEP, Mélissa Camara, telling reporters a sentence from that night that history is likely to keep: “The legal arsenal serving a xenophobic ideology is now complete.”
The boomerang does not check your papers
Now the older story, the one that starts on the beach.
In February 1991 I was a boy of six in Fier, and the world as I understood it was coming apart in the most exhilarating way. My mother had taken me to a protest that year where I saw, with a child’s eyes that did not know what they were seeing, the toppled bust of Enver Hoxha — the man whose name adults had spent my whole short life lowering their voices around, suddenly face-down and harmless as a broken toy. The giddy happiness all around me, I still remember, even without full understanding. I knew nothing of what happened that August on the water. I learned it later, the way you learn the history of your own house.
What was happening on the water was this: a freighter named the Vlora sailed out of Durrës carrying close to twenty thousand Albanians, people so desperate to reach the world that had just been unforbidden to them that they clung to the rigging, and put them ashore at Bari. Italy’s answer was improvised in a weekend: the crowd was locked inside the Stadio della Vittoria, a football stadium ringed by police, water and bread dropped to them from helicopters as though to a besieged garrison and then, within days, nearly all of them were deported, many lured onto planes with the promise they were being transferred elsewhere in Italy. It was Europe’s first rehearsal, in my lifetime, of mass migrant detention and the people in the cage were mine. A few years later I crossed the same sea myself, at eleven, unphotographed, and grew up Albanian in that Italy that the Vlora footage had taught what an Albanian was. A country where Albanian fast became the worst insult.
Now hold the two images side by side, because the symmetry is not a writer’s trick; it is the thing itself. In 1991, Italy locked twenty thousand Albanians in a stadium and deported them with a lie. In 2026, Italy detains its unwanted in a camp in Albania and now the European Union has made the arrangement law. Thirty-five years and the cage itself has crossed the sea. The desperation that filled the stadium has new nationalities now, Bangladeshi and Nigerian and Afghan and the bay it is processed in is one I used to swim in. Nothing about this is poetic. A cage that changes its address is still a cage; the only thing that travelled is the lesson: that there is a category of person to whom you can do this and that the category can be moved.
There is also a name for the lesson, and it is older than the European Union. Aimé Césaire was a poet from Martinique, a Black French colonial subject who became one of the great minds of post-war Paris. In 1950 he wrote a furious short book, the Discourse on Colonialism, to tell Europeans the one thing they could not bear to hear about the catastrophe they had just survived: that it was not an aberration. Europe preferred to call the catastrophe a Zivilisationsbruch, a rupture of civilisation, something that broke in from outside history. Césaire’s answer: nothing broke in. The techniques that had devastated Europe, he argued, were Europe’s own colonial techniques, come home. A choc en retour, a return shock. Hannah Arendt, writing a year later from the other shore of the same catastrophe, called it the boomerang effect. The methods a power rehearses on people it has decided not to fully count as people — in the colonies, at the border, on the periphery — do not stay there. They are perfected where rights are thin and then they come home.
And in the decades since, historians have turned the metaphor into a documented genealogy. Germany’s first concentration camps were not built in the 1930s; they were built in 1905, in its colony in south-west Africa, today’s Namibia, and they were already called Konzentrationslager. The prisoners were Herero and Nama, two peoples Germany had set out to exterminate; tens of thousands died in the camps, on an island called Shark Island most infamously, and as David Olusoga and Casper Erichsen showed in The Kaiser’s Holocaust, the line from that island to the 1940s runs through ideas, methods and, chillingly, individual careers: the race scientist Eugen Fischer, who conducted his “research” on Shark Island’s prisoners and ended up training the doctors of the Third Reich; the first imperial commissioner of the colony, whose son, Hermann Göring, needs no introduction. The boomerang is not a literary figure. It has a paper trail and names.
I rehearse this history for one reason: to make a single sentence land with its full weight. The tools do not check your papers. They never have. Ask the Americans, who are running the experiment in real time: ProPublica has documented more than 170 US citizens — citizens, with the passport and the birth certificate — held by immigration agents in the current enforcement wave, kicked, dragged, detained for days. A man born in Denver was deported to Mexico in April after a traffic stop, his proof of citizenship ignored. A five-year-old American citizen was put on a plane to Honduras with her deported mother. The machine built for them turned out, the moment it reached speed, to have no reliable way of recognising us; because no such machine ever does. This is exactly how the boomerang works.
Europe’s version is now on the launch rail. A two-year detention norm exists in European law, a number available to any future emergency that finds it convenient. A search power covers “other relevant premises” which is to say the NGO office, the clinic, the parish hall, the places citizens work and volunteer and pray. Health organisations warn that the regulation could let medical data travel between authorities and onward to third countries, in the service of removals. Mixed-status families live at single addresses; the door the law opens for one resident is open for all of them. Every one of these powers was written for them. So was every power in this essay’s history, at the start.
Is all lost or can we act?
But the story is not finished and this is the part the despair merchants always skip. The deal struck on the first of June is provisional: the Parliament’s committee votes at the end of this month, the plenary after that and a law this size then has to survive years of implementation, transposition and challenge. The courts that emptied Gjadër, transfer after transfer, have not been abolished; in February, Italian judges affirmed the right of civil society organisations to enter Gjadër and look — and the right to look, as the ASGI lawyers who won it understand, is the one right that makes all the others enforceable. Milan proved in February that ordinary public pressure can still put an enforcement agency in a box: a mayor pushed by the crowds to speak up, fifteen thousand signatures, a minister forced to make promises in parliament.
The scheme, remember, runs on the performance of control. Performance has a weakness as old as theatre: it cannot survive an audience that has stopped suspending disbelief and started taking notes. Watch it like that, as a witness, not a spectator. Write to your MEP before the plenary vote; they count the letters, especially the polite ones. Fund the people who go and look, the ASGIs and the PICUMs, who win in courtrooms what cannot be won in comment sections. None of this is dramatic. Neither was the trilogue.
Before you file all of this under other people’s misfortune, walk the radius of the new powers through one ordinary day. The man who delivered your groceries this morning. The woman who looks after your grandmother. Your colleague’s husband, whose permit lapsed with his job. The clinic where you give blood; the parish hall where you vote; the address where one resident’s missing paper opens the door for everyone behind it. You do not need to be deportable for these powers to enter your life. You only need to know someone who is. And you do, whether you know it or not. And if your passport has already done the quiet arithmetic for you, the one that says these powers are for them, remember that in America the machine needed less than a year to reach the people the arithmetic called safe: the passport-holders, the Denver-born, the five-year-old citizen. The distance between them and you is not a category. It is a speed.
So ask the question that the percentage was designed to bury. Ask your own government, calmly, in writing, the way you’d ask about any procurement: Which of these powers would you give back if every migrant vanished tomorrow? The two-year detention? The home searches? The data-sharing, the lifetime bans, the offshore camps with their sunk costs and their salaried guards watching the empty cells? You already know the answer, because the answer is the entire history of state power in one word. None.
The regulation, at least, is honestly named. Everything in it returns.


