Same Target, No Agreement

Why horseshoe theory reaches for the wrong explanation, and what a better one predicts

In the first week of November 1932, Berlin’s buses and trams stopped. The city’s transport workers had struck, and for four days the depots were blocked, track was torn up, and the few trams driven by strikebreakers had their windows put through.

The dispute was over two pfennigs an hour. The transport company had wanted a much deeper pay cut, and the Social Democrat-aligned union had bargained it down to that and accepted the settlement. The Communist trade union rejected the compromise and called a strike. So did the Nazi one. The strike leadership that ran the walkout was dominated by the Communists, but Nazi representatives sat on it too, and the government’s own records called it the communist-national socialist strike leadership.

They were not allies. Within months of Hitler taking power, his regime would be arresting Communists by the thousands, and over the years that followed tens of thousands of the party’s members were imprisoned or killed. Nobody on that picket line was confused about who the other side was. They were there anyway — not only against a pay cut, but against the settlement the Social Democrats’ union had signed, and the union’s authority to sign it.

That picket line is worth keeping in mind, because it is the cleanest available answer to a question that keeps coming back.

The popular version of the question is horseshoe theory: the idea that the political extremes curve round towards each other until they almost touch, closer to one another than either is to the centre. It is intuitively satisfying. It explains the strange bedfellows. And it asks you to believe something that is usually false — that people who despise each other’s entire view of the world are nonetheless alike underneath, sharing a temperament or a psychology that the centre lacks.

The trouble is that it treats every case of opposites lining up as the same phenomenon, when there are at least three. Sometimes two movements genuinely share a political programme on some question, which is ordinary overlap. Sometimes they share a way of doing politics — the contempt for process, the appetite for enemies — while wanting entirely different things, which is closer to what horseshoe theory is reaching for. And sometimes they share nothing but a target. Treating all three as proof that the extremes are secretly alike is a category error, and it hides the one that matters most here.

The Communists and the Nazis on that picket line were not alike underneath. They did not need to be. Each was running its own doctrine, and each doctrine happened to point at the same target.

The Communist line was set in Moscow. Under what the Comintern called its Third Period, Social Democrats were not a rival on the left but the principal obstacle to revolution — “social fascists,” whose moderate reforms propped up capitalism and delayed its collapse. The German party put it plainly: fighting fascism meant fighting the SPD just as much as it meant fighting Hitler. You will often see the bet summarised in a phrase attributed to the party’s leader, Ernst Thälmann — “Nach Hitler, kommen wir,” after Hitler, our turn. Its provenance is disputed and I am not going to lean on it. The doctrine it summarises needs no help; it is in the party’s own statements.

The Nazi union had its own reasons, and they were not the Communists’ reasons. There was a federal election three days into the strike, and backing a popular walkout was a way to reach workers who would otherwise never have looked at them.

Two doctrines. No shared programme, and no need of one. One shared target — not the pay cut as such, but the arrangement that had produced and defended it: the Social Democrats’ union, and its claim to speak for the workers.

I want to give this a name, because naming it is what lets you test it. Call it the common enemy effect. Two actors, ideologically distant, each acting on a doctrine of its own, arrive at the same institution as the main obstacle to what they want. That is enough to produce convergent rhetoric, or even direct cooperation, on that front — without any convergence of values or temperament behind it.

The precise word for what it produces is conditional convergence. The alignment depends on the target. Take the target away and what remains is incidental at best — nothing strong enough to have put them on the same side of anything.

And that is what makes it more than a relabelling of horseshoe theory. It predicts something horseshoe theory does not. If this is what is going on, the alignment should be substantially stronger wherever the shared target is involved than anywhere it is not. Two actors who turn out to converge just as strongly across a whole range of unrelated questions are not showing this effect; they are showing ordinary overlap, which is a different thing. Horseshoe theory can absorb either result, which is exactly what is wrong with it. This one can fail.

So test it.

George Galloway, who is standing in the Holborn and St Pancras byelection on 8 October, spent nine years presenting a programme on Russian state television, from 2013 until Ofcom revoked RT’s British licence in 2022. Neil Oliver, the former BBC presenter now on GB News, has cited Galloway by name as an authority on the weakness of the British army. On Ukraine and NATO the two arrive in roughly the same place. In The Permission Slip I set out how differently they get there — one says the hardest part out loud, and the other never has to.

That difference in route is not a complication for this argument. It is what the argument predicts. Galloway reaches the target through an explicitly geopolitical position: he has said in terms that he supports Russia’s war as part of an anti-imperialist framework. Oliver reaches it through distrust of the institutions reporting the war, and of the governing class he sees behind them. Those are my characterisations of two records, not labels either man uses, and the reader should weigh them as such. But the pattern they describe is visible in the record: same destination, different machinery.

Now look away from the target. Galloway’s Workers Party runs on redistributive socialism. Oliver’s politics are the populist right’s. On the fundamental question of what an economy should be organised to do, their programmes point in very different directions.

They are not wholly unrelated elsewhere. Both are fluent in distrust of the establishment, and a sceptical reader would be right to point that out. But notice what that shared ground is made of. It is not a shared programme. It is another shared target — the establishment itself — which is exactly what the effect would lead you to expect. Where they overlap, they overlap against something. Where the question is what to build, they part company completely.

A framework that confirms everything you point it at is not a framework. So it is worth showing one it rejects.

Donald Trump spent months calling Zohran Mamdani a communist lunatic. Then, in November last year, the two met in the Oval Office and were warm to the point of unsettling their own supporters. That looks like a candidate — two men at opposite ends, suddenly friendly. There is even a possible shared enemy, since Mamdani came to power by beating the Democratic establishment’s candidate.

It doesn’t fit, and the reason it doesn’t is instructive. When Trump explained the warmth, he did not describe a common opponent. He described common ground: housing, affordability, a safer city, the shared hometown. Some of Mamdani’s ideas, he said, were the same as his own. That is overlap on particular policies, plus a meeting that suited both men politically. The effect predicts convergence against a target; this was convergence on a handful of issues, and a deliberate willingness by both men to emphasise their common ground.

There is a third reading, and it is cleaner than either. Back in September last year, before the election and while he was still calling Mamdani a communist, Trump predicted he would prove to be one of the best things ever to happen to the Republican Party. That is not a shared target and it is not shared policy. It is the calculation that a radical-looking opponent is useful to have in view — and it requires nothing of the other man at all, no common enemy, no common ground, not even a meeting. A sceptical reader might reach for it as the explanation, and they would have a case. The framework puts it in the other pile, which is what it should do.

Which leaves the distinction worth carrying away. Opposites meeting is not this. Opposites agreeing is not this. Opposites cooperating is not this either. Opposites converging specifically against the same target, and only there, may be.

One thing needs saying plainly. The mechanism in 1932 and the mechanism now are the same. The stakes are not remotely the same, and nothing here should be read as suggesting they are. A byelection in central London is not the last winter of the Weimar Republic. I am comparing the shape of a reasoning error, not the size of its consequences.

And the common enemy effect has a second property, which is easier to miss than the first and matters more.

It tells you nothing about what happens next. The Communists’ bet was that a shared enemy’s fall would clear the ground for them. It did clear the ground. It cleared it for whoever was organised to walk onto it, and that was not them. The aftermath of a common enemy’s collapse is not automatically inherited by the people who helped bring it about. What follows will be shaped by whoever is organised and positioned to take advantage of it.

The mirror-image mistake was made in the same months by the conservative circle around Franz von Papen, who helped put Hitler into the chancellorship believing a cabinet of conservatives could keep him in check. They thought they could use his political strength without surrendering control of it. Same error from the other direction. The Communists thought a force would clear the way for them. The conservatives thought they could harness one. Neither had asked who would be holding it afterwards.

That is the question to bring to any strange alignment, and it does not require believing anyone is secretly alike. When two people who agree on nothing start pointing at the same target, the useful thing to ask is not what they have in common. It is which of them has thought about the day after.

Standing on the same picket line tells you what two people are against. It tells you nothing about which of them will be standing when it is over.

Sources and acknowledgements

The Berlin transport strike ran from 3 to 7 November 1932, and was called by the Communist trade union opposition and the Nazi factory cell organisation. Nazi representatives joined the Communist-dominated strike leadership, which the Reich Chancellery’s own files describe as communist-national socialist. The strike leadership also included non-union workers and some Social Democrats, and the later label of a joint “Nazi and Communist” action against democracy has been contested by historians; the account here keeps to what the record shows. The KPD’s formulation that fighting fascism meant fighting the SPD as much as Hitler comes from the party’s own statements of the period. The phrase “Nach Hitler, kommen wir” is widely attributed to Ernst Thälmann; its attribution is disputed and the argument does not rest on it. Galloway’s programme on RT UK ran from November 2013 until Ofcom revoked the channel’s licence in March 2022, and his framing of Russia’s war was reported by Novara Media on 14 September. The Trump and Mamdani meeting took place in the Oval Office on 21 November 2025 and was reported by the Associated Press, CBS and Al Jazeera. His earlier prediction about the Republican Party was posted on 29 September 2025, before the mayoral election.

The piece refines two earlier strands of work here: the horseshoe formulation, and The Permission Slip.

On the evidential line. The historical events and the two men’s public positions are documented. The characterisation of each man’s route to the shared target is my inference from their records. That both cases share one mechanism, and that the mechanism predicts alignment concentrated on the target rather than spread across a programme, is my argument — with one confirmed contemporary instance, which is not yet a pattern.

Same Target, No Agreement

Why horseshoe theory reaches for the wrong explanation, and what a better one predicts

In the first week of November 1932, Berlin’s buses and trams stopped. The city’s transport workers had struck, and for four days the depots were blocked, track was torn up, and the few trams driven by strikebreakers had their windows put through.

The dispute was over two pfennigs an hour. The transport company had wanted a much deeper pay cut, and the Social Democrat-aligned union had bargained it down to that and accepted the settlement. The Communist trade union rejected the compromise and called a strike. So did the Nazi one. The strike leadership that ran the walkout was dominated by the Communists, but Nazi representatives sat on it too, and the government’s own records called it the communist-national socialist strike leadership.

They were not allies. Within months of Hitler taking power, his regime would be arresting Communists by the thousands, and over the years that followed tens of thousands of the party’s members were imprisoned or killed. Nobody on that picket line was confused about who the other side was. They were there anyway — not only against a pay cut, but against the settlement the Social Democrats’ union had signed, and the union’s authority to sign it.

That picket line is worth keeping in mind, because it is the cleanest available answer to a question that keeps coming back.

The popular version of the question is horseshoe theory: the idea that the political extremes curve round towards each other until they almost touch, closer to one another than either is to the centre. It is intuitively satisfying. It explains the strange bedfellows. And it asks you to believe something that is usually false — that people who despise each other’s entire view of the world are nonetheless alike underneath, sharing a temperament or a psychology that the centre lacks.

The trouble is that it treats every case of opposites lining up as the same phenomenon, when there are at least three. Sometimes two movements genuinely share a political programme on some question, which is ordinary overlap. Sometimes they share a way of doing politics — the contempt for process, the appetite for enemies — while wanting entirely different things, which is closer to what horseshoe theory is reaching for. And sometimes they share nothing but a target. Treating all three as proof that the extremes are secretly alike is a category error, and it hides the one that matters most here.

The Communists and the Nazis on that picket line were not alike underneath. They did not need to be. Each was running its own doctrine, and each doctrine happened to point at the same target.

The Communist line was set in Moscow. Under what the Comintern called its Third Period, Social Democrats were not a rival on the left but the principal obstacle to revolution — “social fascists,” whose moderate reforms propped up capitalism and delayed its collapse. The German party put it plainly: fighting fascism meant fighting the SPD just as much as it meant fighting Hitler. You will often see the bet summarised in a phrase attributed to the party’s leader, Ernst Thälmann — “Nach Hitler, kommen wir,” after Hitler, our turn. Its provenance is disputed and I am not going to lean on it. The doctrine it summarises needs no help; it is in the party’s own statements.

The Nazi union had its own reasons, and they were not the Communists’ reasons. There was a federal election three days into the strike, and backing a popular walkout was a way to reach workers who would otherwise never have looked at them.

Two doctrines. No shared programme, and no need of one. One shared target — not the pay cut as such, but the arrangement that had produced and defended it: the Social Democrats’ union, and its claim to speak for the workers.

I want to give this a name, because naming it is what lets you test it. Call it the common enemy effect. Two actors, ideologically distant, each acting on a doctrine of its own, arrive at the same institution as the main obstacle to what they want. That is enough to produce convergent rhetoric, or even direct cooperation, on that front — without any convergence of values or temperament behind it.

The precise word for what it produces is conditional convergence. The alignment depends on the target. Take the target away and what remains is incidental at best — nothing strong enough to have put them on the same side of anything.

And that is what makes it more than a relabelling of horseshoe theory. It predicts something horseshoe theory does not. If this is what is going on, the alignment should be substantially stronger wherever the shared target is involved than anywhere it is not. Two actors who turn out to converge just as strongly across a whole range of unrelated questions are not showing this effect; they are showing ordinary overlap, which is a different thing. Horseshoe theory can absorb either result, which is exactly what is wrong with it. This one can fail.

So test it.

George Galloway, who is standing in the Holborn and St Pancras byelection on 8 October, spent nine years presenting a programme on Russian state television, from 2013 until Ofcom revoked RT’s British licence in 2022. Neil Oliver, the former BBC presenter now on GB News, has cited Galloway by name as an authority on the weakness of the British army. On Ukraine and NATO the two arrive in roughly the same place. In The Permission Slip I set out how differently they get there — one says the hardest part out loud, and the other never has to.

That difference in route is not a complication for this argument. It is what the argument predicts. Galloway reaches the target through an explicitly geopolitical position: he has said in terms that he supports Russia’s war as part of an anti-imperialist framework. Oliver reaches it through distrust of the institutions reporting the war, and of the governing class he sees behind them. Those are my characterisations of two records, not labels either man uses, and the reader should weigh them as such. But the pattern they describe is visible in the record: same destination, different machinery.

Now look away from the target. Galloway’s Workers Party runs on redistributive socialism. Oliver’s politics are the populist right’s. On the fundamental question of what an economy should be organised to do, their programmes point in very different directions.

They are not wholly unrelated elsewhere. Both are fluent in distrust of the establishment, and a sceptical reader would be right to point that out. But notice what that shared ground is made of. It is not a shared programme. It is another shared target — the establishment itself — which is exactly what the effect would lead you to expect. Where they overlap, they overlap against something. Where the question is what to build, they part company completely.

A framework that confirms everything you point it at is not a framework. So it is worth showing one it rejects.

Donald Trump spent months calling Zohran Mamdani a communist lunatic. Then, in November last year, the two met in the Oval Office and were warm to the point of unsettling their own supporters. That looks like a candidate — two men at opposite ends, suddenly friendly. There is even a possible shared enemy, since Mamdani came to power by beating the Democratic establishment’s candidate.

It doesn’t fit, and the reason it doesn’t is instructive. When Trump explained the warmth, he did not describe a common opponent. He described common ground: housing, affordability, a safer city, the shared hometown. Some of Mamdani’s ideas, he said, were the same as his own. That is overlap on particular policies, plus a meeting that suited both men politically. The effect predicts convergence against a target; this was convergence on a handful of issues, and a deliberate willingness by both men to emphasise their common ground.

There is a third reading, and it is cleaner than either. Back in September last year, before the election and while he was still calling Mamdani a communist, Trump predicted he would prove to be one of the best things ever to happen to the Republican Party. That is not a shared target and it is not shared policy. It is the calculation that a radical-looking opponent is useful to have in view — and it requires nothing of the other man at all, no common enemy, no common ground, not even a meeting. A sceptical reader might reach for it as the explanation, and they would have a case. The framework puts it in the other pile, which is what it should do.

Which leaves the distinction worth carrying away. Opposites meeting is not this. Opposites agreeing is not this. Opposites cooperating is not this either. Opposites converging specifically against the same target, and only there, may be.

One thing needs saying plainly. The mechanism in 1932 and the mechanism now are the same. The stakes are not remotely the same, and nothing here should be read as suggesting they are. A byelection in central London is not the last winter of the Weimar Republic. I am comparing the shape of a reasoning error, not the size of its consequences.

And the common enemy effect has a second property, which is easier to miss than the first and matters more.

It tells you nothing about what happens next. The Communists’ bet was that a shared enemy’s fall would clear the ground for them. It did clear the ground. It cleared it for whoever was organised to walk onto it, and that was not them. The aftermath of a common enemy’s collapse is not automatically inherited by the people who helped bring it about. What follows will be shaped by whoever is organised and positioned to take advantage of it.

The mirror-image mistake was made in the same months by the conservative circle around Franz von Papen, who helped put Hitler into the chancellorship believing a cabinet of conservatives could keep him in check. They thought they could use his political strength without surrendering control of it. Same error from the other direction. The Communists thought a force would clear the way for them. The conservatives thought they could harness one. Neither had asked who would be holding it afterwards.

That is the question to bring to any strange alignment, and it does not require believing anyone is secretly alike. When two people who agree on nothing start pointing at the same target, the useful thing to ask is not what they have in common. It is which of them has thought about the day after.

Standing on the same picket line tells you what two people are against. It tells you nothing about which of them will be standing when it is over.

Sources and acknowledgements

The Berlin transport strike ran from 3 to 7 November 1932, and was called by the Communist trade union opposition and the Nazi factory cell organisation. Nazi representatives joined the Communist-dominated strike leadership, which the Reich Chancellery’s own files describe as communist-national socialist. The strike leadership also included non-union workers and some Social Democrats, and the later label of a joint “Nazi and Communist” action against democracy has been contested by historians; the account here keeps to what the record shows. The KPD’s formulation that fighting fascism meant fighting the SPD as much as Hitler comes from the party’s own statements of the period. The phrase “Nach Hitler, kommen wir” is widely attributed to Ernst Thälmann; its attribution is disputed and the argument does not rest on it. Galloway’s programme on RT UK ran from November 2013 until Ofcom revoked the channel’s licence in March 2022, and his framing of Russia’s war was reported by Novara Media on 14 September. The Trump and Mamdani meeting took place in the Oval Office on 21 November 2025 and was reported by the Associated Press, CBS and Al Jazeera. His earlier prediction about the Republican Party was posted on 29 September 2025, before the mayoral election.

The piece refines two earlier strands of work here: the horseshoe formulation, and The Permission Slip.

On the evidential line. The historical events and the two men’s public positions are documented. The characterisation of each man’s route to the shared target is my inference from their records. That both cases share one mechanism, and that the mechanism predicts alignment concentrated on the target rather than spread across a programme, is my argument — with one confirmed contemporary instance, which is not yet a pattern.

The Permission Slip

Why the argument for letting Russia win keeps arriving in the same three parts

On Monday, on his own programme, George Galloway announced he intends to stand against the Green leader Zack Polanski in the Holborn and St Pancras byelection. In the course of the same show he named Polanski’s support for Ukraine as one of his reasons for opposing him, praised Russia’s military prowess after a drone strike near the Ukraine-Poland border, and said he supports Russia’s war as part of a wider anti-imperialist, anti-colonialist framework. He said all of it from Russia, where he has been living since counter-terror police stopped him at Gatwick.

None of that is new. Galloway spent years hosting shows on RT and on Kremlin-owned Radio Sputnik, until Ofcom revoked RT’s British licence after the invasion. He has occupied this ground for a long time.

Separately — different programme, different outlet, no connection between the two occasions — Neil Oliver appeared on The Duran and cited Galloway by name as an authority on how weak the British army has become. Oliver’s politics and Galloway’s have almost nothing in common. One speaks the language of anti-imperialism, the other of anti-globalism. They are not working together and I am not going to suggest they are.

But strip the vocabularies away and the same three parts can be reconstructed underneath both. That reconstruction is mine, not a template either man is consciously following. It is also, I think, the most useful thing to notice about either of them.

The first part is true. This war has cost an enormous number of lives and an enormous amount of money, on both sides, and that fact does not depend on being right about anything else. Anyone who waves it away is not worth listening to.

The second part is contested, and comes from somewhere serious. NATO’s expansion after the Cold War helped create the conditions for the invasion — an argument made by John Mearsheimer and others within a real academic tradition, long before it became a talking point, and defensible on its own terms whether or not you accept it.

The third part follows from neither of the first two on their own. Because the war costs too much, and because the West bears some structural responsibility for the conditions that produced it, Russia’s victory should therefore be accepted, or actively hoped for.

Read those three again and notice what is missing. Nobody ever explains the join. It is simply performed, as though the third sentence were a continuation of the second rather than a different kind of claim altogether. A war can be too costly to continue and Russia can still be the invader whose victory extends the costing rather than ends it. Cost tells you a war should stop. It tells you nothing whatsoever about who should be standing when it does.

So here is what would show I have this wrong. If someone ran this argument and then stopped at part two — made the cost case and the causation case, and said plainly that Russia is nonetheless the invader and its victory is nonetheless a bad outcome — there would be no slippage to name, only a hard argument I happen to disagree with.

I set that test before I went back through Oliver’s record. It half fired, and the half that fired is more interesting than the test passing cleanly would have been.

Because Oliver does say he doesn’t trust Putin. He has said it from the beginning. In his first substantial monologue on the war, in early 2022, he told his audience he didn’t know what was happening in Ukraine and didn’t understand it, that he ignored most of the mainstream media because he didn’t trust it, and that whatever Putin was doing, the West had to accept a share of the blame. He was widely mocked for it at the time, mostly for the sequence: professing ignorance, then supplying conclusions anyway.

What he does not do, then or since, is state the third part. He does not say Russia’s victory should be accepted. He says he doesn’t know, that nobody can really know, that the people telling you they know are not to be trusted — and then he arrives at the destination anyway, without ever having to put his name to the sentence that would be hardest to defend.

That is the more efficient version of the machine, not a counterexample to it. Galloway states the third part and owns it. Oliver never states it and gets there regardless, which means there is nothing to hold him to. Hedge the credential, keep the conclusion.

So how does the third part arrive if nobody says it?

There is a fourth move, and it is never stated out loud either. Russia is going to win anyway. Once that sits quietly underneath the argument, the third part stops looking like a moral claim and starts looking like realism. You are not being asked to prefer an outcome. You are being asked to accept one somebody else has already decided. That is what makes the missing join invisible. If the result is fixed, there is nothing left to argue about, only a cost to stop paying.

It is also why the disclaimer costs nothing. If the outcome is settled, not knowing what is happening is no obstacle to knowing what should be done about it.

Which is why it is worth checking whether the result is fixed.

Over the twelve months to early September, Russia’s net territorial gain was about 779 square miles on the Institute for the Study of War’s data — roughly 0.3 per cent of Ukraine. The Center for Strategic and International Studies put the casualty ratio at between two and three Russian casualties for every Ukrainian for most of the war, rising to nearly eight to one in the first half of this year — though hold that one loosely too. Casualty estimates are contested, both sides have reasons to shade their own, and the officer figures the two armies report are close enough to sit awkwardly with a ratio that wide. Ukraine’s military intelligence chief, Oleh Ivashchenko, told The Times last week that Ukrainian intelligence believes Russia’s generals have asked Putin for another 800,000 troops — there is no Russian confirmation of it, and it is one side’s assessment of the other, so hold it loosely. ISW assessed three days ago that Ukrainian counteroffensives have reduced Russia’s net gains since March to close to nothing, and on Tuesday Ukraine’s Third Army Corps announced another operation in northern Donetsk.

None of that means Ukraine is winning, and I am not claiming it. Wars are not decided by square miles. What it means is narrower and sufficient: the specific proposition the argument quietly rests on — that this is settled, that the outcome is already fixed and resistance only extends the bill — is not what the front line currently shows. The people telling you the war is already decided are not reading the battlefield. They are reading each other.

Take the fourth move away and the third part has to stand on its own. In these arguments, it doesn’t. In Oliver’s, it never even has to appear.

Now the part I want to be careful about.

Galloway and Oliver did not coordinate. There is no evidence they did and I am not going to imply it. The link between them is not a meeting. It is that the same structure works just as well in anti-imperialist language as in anti-globalist language, so both of them reached for it independently and neither had to invent anything. That is the whole point. A form available to anyone does not require a director.

There is a second distinction underneath that one, and I had it wrong until I went back through the record.

Galloway was paid for years to host programmes on Kremlin-controlled media. His route to the conclusion looks less like being persuaded than like starting there. Oliver has nothing equivalent — his link runs through The Duran, an outlet whose director hosts a Kremlin talk show, which is a structural tie rather than a payslip, and everything else about his record reads like a man who believes it. The distinction between someone working a position and someone who has arrived at one honestly is a real distinction and it matters, and I am not able to collapse them just because the output is similar.

What I can say about the output is that it does not survive checking. In the same appearance where he cites Galloway, Oliver claims Putin’s warning about European troops went unreported in Western media. Associated Press wired it and it ran across dozens of outlets in the same news cycle. Sincerity does not make that true. It only means the mistake was made in good faith, which is a statement about him and not about the claim.

And the audience is doing something else again. Some of it is reasoning backward too. But a good deal of it is people who have watched institutions lie to them for a decade and have no reserve of trust left to spend on a harder answer. I am not going to call that cowardice. I have never met them and it would be an unfalsifiable thing to say about a stranger.

What I will say is what the argument does, mechanically, for whoever picks it up — believer, operator or bystander. It does not persuade. It excuses. It takes a true thing about cost, welds it to a contested thing about causation, and uses an unstated assumption about inevitability to carry you somewhere neither of them goes — and it does this while sounding like reasoning, which is the only reason it works.

There is one more feature, and it is the one I find hardest to get past. The people I have watched advocate acceptance of Russian victory are doing so from positions of personal safety, in countries that are not Ukraine, about a population that will bear the direct consequences of the outcome being recommended on its behalf. Nobody advocating acceptance has anything to accept.

The hard position is not complicated. The war is real. The cost is real and unbearable. NATO expansion is a legitimate subject of argument. Nothing about the outcome is settled. And none of that tells you who should win, which remains a separate question requiring a separate answer that nobody wants to give.

That is an uncomfortable place to stand, and standing there costs something. Saying you don’t know is not the same as standing there. It is the cheapest seat in the room, and it has a clear view.

A permission slip is not an argument. It is a note excusing you from one. And the reason it works is not that you believe it — it is that somebody respectable signed it.

Sources and acknowledgements

Galloway’s byelection announcement, his remarks on Polanski, his praise of the Yahodyn strike and his anti-imperialist framing of Russia’s war were all made on the same edition of his own programme on 14 September, and reported by Novara Media the same day. His tenure on RT UK and Radio Sputnik ended when Ofcom revoked RT’s licence in 2022. Oliver’s early-2022 monologue is on GB News’ own site and was widely reported at the time. His appearance with The Duran is a separate occasion entirely; the material quoted from it here comes from a full transcript, and The Duran’s Kremlin-linked personnel are a matter of record. The territorial figures are the Institute for the Study of War’s, as compiled by Russia Matters on 9 September; the casualty ratio is the Center for Strategic and International Studies’, from 1 July. Oleh Ivashchenko’s assessment was given to The Times on 11 September. ISW’s assessment of reversed Russian gains is dated 12 September, the Donetsk operation was announced on 15 September, and Putin’s remark on European troops was carried by the Associated Press and others on 12 September.

The piece extends earlier work here on the Convergent Ukraine Error, and on why convergence between hostile camps does not imply coordination between them. John Mearsheimer’s argument is criticised here only for what others do with it, not for itself.

On the evidential line. What both men said is documented from their own words, on separate occasions, with no contact between them established or claimed. The three parts are my reconstruction, not a template either is consciously following, and only one of the two states the third of them. That the structure functions as permission rather than persuasion is my inference. So is the reading of Galloway’s route to it; I have not made the same inference about Oliver, and the difference between the two is set out above rather than smoothed over. A single case is not a pattern — this needs a second, non-Ukraine instance before it is worth calling a framework.

On Saturday afternoon at Farmleigh House in Dublin, standing beside the Taoiseach, the President of the United States said two things about Britain.

The first was about the Falklands. He wasn’t sure the British would be willing to travel that far again. Eight thousand miles. Weeks by boat. He added that Britain had a bit of a problem with immigration, and a bit of a problem with energy, and some other things to be getting on with. He expected he’d be called in to settle it.

The second was about Ireland. He’d love to see it unified. It was going to happen anyway, so it may as well happen now.

Both remarks landed on the same government, in the same hour, from the same man. And the obvious reading is the one most of the coverage reached for: this is pressure. An ally that wouldn’t play ball during the Iran war is getting worked over, and the working-over has two prongs.

That reading might be right. It’s worth finding out, because everything else follows from it.

So here is the test, and I’d rather run it in front of you than behind your back. The Pentagon has a stated strategic objective — allies carrying more, Washington carrying less. If both interventions can be explained as instruments of that objective, the apparent contradiction between them dissolves and there is nothing further to explain. If one of them cannot, then that objective is not sufficient to account for what this administration does, and something else is operating alongside it.

Note what the test asks for. A route, not a result. A lever that fails to move anything is still a lever, and I am not going to claim that anything Washington has threatened has actually changed a British budget line. What I want to know is whether the threat was even pointed at that outcome.

Run it on the Falklands and it holds up. In late August the Telegraph reported that senior Pentagon officials were weighing a review of Washington’s position on the islands as one of several levers to push Britain toward NATO’s 5 per cent spending target — Britain, a senior official said, was likely to get special attention, because it had so much potential. The idea wasn’t new. A leaked Pentagon memo in April had already floated using the sovereignty question as pressure on allies judged insufficiently supportive over Iran. In the same report, the Pentagon’s policy chief Elbridge Colby is quoted saying he hoped to see more from countries like the United Kingdom, while praising the progress of Germany, Poland, the Netherlands and the Scandinavian allies. The route runs end to end and you can follow every step of it: threaten something Britain cares about, aim the threat at a number in a spending review. Clumsy, ugly, and entirely legible as leverage toward the burden-shifting goal the Pentagon has written down. Whether it works is a different question, and not the one on the table.

Now run it on Ireland, and it stops.

There is no route from a united Ireland to a British defence budget. Endorsing unification does not aim at a spending line, does not threaten anything Britain could concede in a spending review, and cannot be traded for one. It breaks with a convention every American president since the Good Friday Agreement has held to, and it does so in front of the one community in these islands that was still turning up. Sinn Féin has boycotted the White House’s St Patrick’s events for two consecutive years over Gaza. The DUP and the UUP have kept engaging. Within hours, the DUP’s Gavin Robinson was saying that as unionists they were not interested in the destruction of their country by joining with the Irish Republic. Jim Allister of the TUV, who agrees with Robinson about almost nothing, said much the same and added that Northern Ireland was not a pawn in the President’s row with London. Downing Street pushed back. The Conservative leader compared it to the threats over Canada and Greenland.

The cost was immediate. Within the objective being tested, the return is nil.

There may well be other reasons an American president says such a thing — Irish-American politics, a genuinely held view, the pleasure of saying it in Dublin. He has said versions of it before, and the Taoiseach’s own office noted afterwards that it should have surprised nobody. I don’t need to rule those out. I only need one thing: none of them belongs to the Pentagon’s burden-shifting objective, which is the explanation on trial.

That asymmetry — not either intervention on its own — is the evidence. Both remarks intervene in Britain’s interests. Only one of them is recognisably pressure in pursuit of the thing Washington says it wants.

So what are you actually looking at?

Three distinguishable modes of behaviour visible in the record so far, running through the same institutional machinery, each working from a different answer to the question of what power is for. I want to be precise about the size of that claim, because a satisfying shape is the easiest thing in the world to mistake for evidence. I am not claiming three organised projects with staff, plans and a shared understanding of themselves as rivals. Nor do the three rest on equal weights of evidence: the first is documented across an institutional paper trail, the second inferred from a pattern of presidential interventions, the third built on one major speech and what has happened around the man who gave it. They differ sharply in how project-like they are, and that difference turns out to be the whole argument about what lasts.

And before any of it, here is what would prove it wrong.

The three-mode reading fails if the modes stop costing each other anything. If every intervention, whatever its surface, turns out on inspection to serve a single objective, then there is one project and I have mistaken internal friction for internal difference. What follows is that test being run. Watch for the frictions — three of them, each running against one of the others. If they aren’t there, or if they disappear from here, the honest move is to drop this reading rather than defend it.

The first mode treats power as something to be reallocated against a changed map. That is the 2026 National Defense Strategy, published in January, which prioritises the homeland and deterring China, describes itself as focused engagement rather than isolation, and states plainly that as American forces concentrate on the homeland and the Indo-Pacific, allies elsewhere will take primary responsibility for their own defence with more limited American support. This one genuinely is a project. It has a document, an author, a budget and an institutional home. It has been tested in public: Colby sat in front of the Senate Armed Services Committee in March and was given a rough afternoon by senators from both parties, the Republican chairman among them, who does not accept that the Russia problem can be delegated to Europe.

Two things need saying about it, and they pull in different directions.

The first is that the underlying problem does not depend on Colby. An ageing continent, a rising China and a finite American budget will still be there whoever holds that desk. A future administration could reach completely different policy conclusions from the same structural facts. The problem outlasts the doctrine, and it is the problem, not the doctrine, that is durable.

The second is that the handover the strategy describes is not happening cleanly. European defence spending reached 864 billion dollars in 2025 on SIPRI’s figures, up 14 per cent in a single year, the fastest growth of any region in the world. Taken alone that looks like the plan working, and it may well be part of the plan working. But spending figures tell you what was spent, not why, and the political economy underneath them is doing something stranger than compliance. The analysis of it I find hardest to argue with is Foreign Policy’s, from January: the European constituencies most willing to fund rearmament are the ones that still believe NATO can be salvaged, and the ones that have concluded America is now an adversary are the least willing to pay for replacing it. Poland and the Baltics, who most want the Americans to stay, spend most. Italy — which spent 1.54 per cent of GDP in 2024, reached 2 per cent in 2025 largely by reclassifying what counted, and whose finance minister has been reluctant to use the EU’s escape clause — is among the least willing, held back by a domestic tradition that reads defence spending as a cause of conflict rather than a response to it.

That is Foreign Policy’s evidence. Here is my reading of it, and you can take it or leave it: if the pattern holds, the countries best placed to replace American cover are the ones still trying to keep it, and the money is arriving in the wrong places on the wrong logic.

There is the first friction. Burden-shifting is met by hedging rather than compliance, which weakens the leverage the second mode depends on.

And there is a complication I’d rather put in front of you than tidy away. The same strategy document that makes the serious argument about China also commits the Pentagon to guaranteeing American access to key terrain from the Arctic to South America, Greenland and the Panama Canal named specifically. That isn’t the whole of Colby’s realism. The President’s priorities are in the same document. These modes do not run in clean parallel lines. They share a building, and they bleed into each other’s paperwork.

The second mode treats power as personal. Not merely transactional, though transaction is its commonest form — the Falklands remark is the instrument used for leverage, and the Ireland remark is the same instrument used for preference. What makes them one mode is not that both are deals. It is that in both, the relationship between two states is being handled as an extension of the man currently holding the office: his grievances, his enthusiasms, his sense of who has been fair to him. That is why it is the least portable of the three. There is no document to inherit and no institution to carry it. It survives only by imitation, if some future leader decides to do it the same way.

And there is the second friction, already on the page. The Dublin remarks complicated relations with the two Northern Irish parties still willing to engage with Washington — the ones any American administration needs if it wants anything at all in that part of the world.

The third mode is the one that gets the least attention and has the clearest route to portability, because it isn’t about policy at all.

In February 2025, at the Munich Security Conference, the Vice President told an audience of European governments that the firewall keeping parties like the AfD out of coalition politics was itself the democratic failure — that there was no room for firewalls. He met the AfD’s co-leader privately the same day, nine days before a German federal election. The German Chancellor and defence minister condemned it within hours.

Notice what kind of act that is. It is not a policy position taken inside an agreed frame, and it is not improvised. It was written, argued and delivered to the people whose frame it disputes, in their own hall. Colby argues about how power should be distributed. The President extracts from whoever is in front of him. This is an argument about who counts as a legitimate participant in democratic politics at all.

It has intellectual antecedents that long predate this administration. That genealogy is a separate piece and I’m not going to lean on it here, because nothing I could show you in the space available would establish the link rather than assert it.

What can be shown is that something is being treated as inheritable. Vance came out of the Republican midterm convention in Dallas last week with delegates chanting a number that isn’t his yet, and the President praising his performance while stopping short of an endorsement. The Secretary of State is a serious rival, the contest is genuinely open, and its subject is openly described as what the movement becomes once its founder has gone. What is actually being fought over — the argument about legitimacy, an electoral coalition, or simply the keys — is not settled, and I’m not going to pretend it is. But note what nobody is contesting. Nobody holds a contest over an instinct.

And there is the third friction. Munich complicated relations with the governments whose cooperation the first mode’s European strategy requires.

Three frictions, then. The test I set out at the top has been run, and the reading survives it — for now, and only for as long as the frictions keep showing up.

Here is the shape of it, and then I’ll stop.

Two of these men are playing the game. They disagree sharply about how to win it — one wants the resources moved, the other wants a better deal from whoever is across the table — but both are contesting outcomes on a pitch whose markings they accept. The third is not arguing about the score. He is arguing about where the lines go, who is allowed on the field, and which side the referee is entitled to book.

That is why the administration looks incoherent. It isn’t confusion, and it isn’t a hidden plan being executed badly. The machinery is coherent enough to operate. The people operating it disagree about what the machinery is for, and nothing requires them to agree in order to keep using it.

Five years from now, the strategic problem will still be real and somebody will still be trying to solve it. The particular way this President relates to the world will most likely have gone with him.

Nobody remembers the score. Everybody lives with the rules.

Sources and acknowledgements

The Dublin remarks are reported by RTÉ, the Independent and the Press Association, all of 12 September. The Falklands lever comes from the Telegraph’s report of 28 August, and the earlier Pentagon memo from Reuters. The 2026 National Defense Strategy was published on 23 January and is available in full through USNI News; Colby’s testimony on it was given to the Senate Armed Services Committee on 3 March. The European spending figures are SIPRI’s, published in April. The rearmament paradox, and the Italian figures, are from Foreign Policy, 28 January. Vance’s Munich speech and the Weidel meeting are of 14 February 2025. The Dallas convention is reported by the Associated Press and PBS.

The piece builds on earlier work here on where structural frameworks stop generating prescriptions, and on the load-bearing test applied to claims that look sturdier than they are. Among the analysts whose thinking runs alongside it: Adam Tooze on why the specific figure beats the general claim, and Lawrence Freedman on the distance between a strategy document and what a state actually does.

On the evidential line. The underlying behaviours are documented. Their grouping into three distinguishable modes is my analytical inference. That they are three separate projects is not claimed. That the third has the clearest route to portability is inference from the succession contest, not a demonstrated fact — and what that contest is actually about remains open.

A volunteer named only as Neil was photographed and put on social media as a traitor.

He told the BBC what his crew had actually been doing. They had been tasked to Langstone Harbour to light the beach - not for anyone in a dinghy, but for the safety of the protesters, some of whom had gone into the water. He said he would like the people calling him a traitor to know that he isn’t one.

The whole argument is inside that. The crew being abused had been called out to keep the people abusing them from drowning.

Since the Portsmouth landing and the disorder after it, the Royal National Lifeboat Institution has increased security at its sites and issued guidance to volunteers and staff on handling difficult encounters with the public - including, where appropriate, not wearing the charity’s branding. On Wednesday a small anti-migrant gathering was held at the RNLI College in Poole. The charity has said it will carry on rescuing anyone in trouble at sea, which is the only sentence available to it, because that sentence is the entire organisation.

Two things belong alongside that.

The first is a number the charity publishes itself: Channel crossing incidents accounted for 1.2 per cent of RNLI launches in 2024. Whatever the argument about small boats is, the institution absorbing it spends ninety-nine launches in a hundred on something else.

The second is that donations rose significantly over the same period, while some supporters contacted the charity to withdraw. Both are reported. I am not turning that into two halves of the country squaring up - the reporting establishes two things happening at once, not that the givers were answering the protesters. It does mean the simpler story, Britain turns on the lifeboats, is wrong.

There is an academic term that looks like it fits, and the interesting thing is where it stops fitting.

Christopher Achen and Larry Bartels called it blind retrospection: people punishing whoever is nearest for pain nobody nearby caused. The illustration everyone remembers is a run of shark attacks off New Jersey in 1916 and the coastal counties said to have taken it out on Woodrow Wilson.

Be careful with that one. Anthony Fowler and Andrew Hall checked every fatal shark attack in American history against county returns from 1872 to 2012, and found no systematic electoral effect anywhere. The New Jersey result weakened under alternative specifications, and the town-level version rested on a couple of small places. Achen and Bartels accepted that a typical shark attack would not be expected to move a presidential election, so the general claim is gone. What they still defend is the historical reading of New Jersey in 1916, and Fowler and Hall still reject the evidence for it. That argument is open. The drought and flood findings have held up better.

So I am not going to claim the RNLI is a case of blind retrospection. I am using it as a contrast, because the contrast is sharper than the analogy.

The mechanism it describes needs a government in it. It is a theory about what an electorate does to an incumbent at a ballot box - punishing whoever is standing for re-election for the weather, because that is what an electorate can punish. The RNLI is not a government. You cannot vote it out. Nobody at the college gates in Poole will ever get the chance.

Which means this is not that mechanism with one part modified. It is a different situation that shares a shape: a felt harm, an attribution of blame, and a target chosen for something other than responsibility. What has gone missing is not a detail. It is the entire apparatus by which anger is supposed to turn into accountability.

Andy Burnham is the intermediate rung, a comparator rather than an argument. A mayor is elected, is a public official, can be voted against, and talks about immigration. He sits near the decision-making world while holding none of the decisions that matter here, and he is nowhere near the water.

The RNLI is the inverse, and this is what the piece actually turns on. Nobody is closer to the event than a lifeboat crew. They are in the Channel, at night, pulling people out of it - which is precisely why they are visible enough to shout at. And almost nobody is further from the decision. Proximity to a consequence is not proximity to a cause. The two get confused because only one of them can be seen from the beach.

Set the three in a line.

Ukraine, where Britain and Europe hold partial agency. Weapons, sanctions, money, a seat at the table. Compromised and expensive and argued over, but real. Anger about Ukraine has a government to go to, and going to it is a coherent act.

The Gulf and the Red Sea, where the agency is far weaker and far more indirect than the effects being felt. Britain has had warships in the Red Sea and has struck Houthi targets, so this is not nothing. But when the Saudi crown prince rang Washington twice this month asking for American strikes against the Houthis, and Trump declined and sent intelligence and targeting support instead, no European government was making that call. There is still a government here to be angry at. The anger just cannot buy much.

Then the RNLI, which is nearer the consequence than any government and further from the cause than almost anyone. The lifeboat did not start the crossings, does not decide who crosses, sets no policy and is not standing for election. Each step down that line removes another layer of legitimate attribution. Not of complaint as such - the RNLI can be held to account for how it does its own job, like anything else - but of responsibility for the thing people are actually angry about. The last step removes the machinery of complaint altogether.

Now the part I cannot settle, and should not pretend to.

I have shown that an organisation without policy agency has become a target. I have not shown why this target rather than another, and the two people closest to it disagree about that. The local MP called it a misunderstanding - frightened people lashing out at the wrong ones. The charity’s own chief executive called it “deliberate misinformation” about the charity - its purpose and its tasking. Those are not the same diagnosis. One says the anger arrived at the wrong address by accident. The other says the address was changed.

That matters, because if it is the second, proximity is not the explanation. Recasting a rescue service as a participant in the thing it rescues people from is work, and work gets done on purpose. What the evidence establishes is that the charity’s chief executive believes false claims about its purpose and tasking have been circulated. What it does not establish is who that serves, or why this institution and not another. I am not claiming an answer. I am saying that a piece which quietly picked the tidier mechanism would be doing the thing it accuses everyone else of.

What would tell the difference is countable. If this runs on proximity and reachability, targeting should track visibility - how identifiable an organisation is, how locally present, how unlikely to retaliate - with no relationship to how close it sits to the actual decision. If instead targeting tracks the work of recasting, it should follow whichever institutions have been most heavily rewritten in the preceding months, whatever their visibility. Those two are testable. A third is not, and worth marking as such rather than letting it pass as evidence: if anger tracked power, the protest would have been somewhere other than a lifeboat college in Dorset. That is an observation. People can protest at a symbol while knowing exactly where the decisions get made.

Which brings me to the opposite error, because a piece about misdirected blame has no business misdirecting it in turn.

Putin’s war is a substantial driver of displacement across Europe, and one of several things pushing energy costs up. Trump holds real decision-making power over the Gulf, and demonstrated it this month: asked for American strikes against the Houthis, he said no. Both men hold vastly more of the relevant agency than a volunteer lighting a beach in Hampshire. That is a comparison, not a theory of causation. Migration into Europe has causes reaching well beyond two men, and anyone offering you two names as the full explanation is selling the same simplification from the other end.

But the comparison is the point. Somewhere between the people who make these decisions and a volunteer in a yellow jacket, the anger stopped travelling and settled on what it could reach.

So the argument is not that people should be less angry. The cost of living is real, the sense that nobody you can reach is in charge of it is substantially correct, and telling people their anger is irrational is both wrong and useless. That reflex has failed for a decade.

The argument is about aim. In this case, at least, the anger has landed on people who cannot change the policy, directed by people who mostly cannot reach anyone who can. That is not a moral failing. It is what happens when the distance between where a decision gets made and where it gets felt grows longer than anger can travel.

The crews will go out anyway. They have said so, and on the night in question they went out for the protesters.

The least the rest of us can do is be clearer than they need to be about who is responsible for the water.

The RNLI material - the volunteer labelled a traitor and his account of the Langstone Harbour tasking, the security guidance including the advice on branded clothing, the small gathering at the RNLI College in Poole, chief executive Peter Sparkes on misinformation about the charity’s purpose and tasking, and the local MP’s description of the abuse - comes from PA wire copy carried by multiple outlets on 10 and 11 September, with the volunteer’s account given to the BBC. The donations reporting is the Guardian. The 1.2 per cent launch figure is the RNLI’s own published data for 2024.

Blind retrospection is Christopher Achen and Larry Bartels, most fully in Democracy for Realists. The reassessment is Anthony Fowler and Andrew Hall in the Journal of Politics, with published replies from both sides; the disaster and relief literature divides on whether punishment is indiscriminate or conditioned on partisanship and the quality of relief. Europe’s weak position in the Gulf and the Red Sea was set out in The Price-Takers, 12 September, and the Saudi calls to Washington are reported by Axios, Reuters and Al-Monitor.

On evidence: the RNLI events are documented. The shark-attack finding is disputed and is described as disputed rather than used as support; the claim that the framework does not extend to a non-electoral target is my argument rather than theirs. Andy Burnham appears as a comparator, not on the strength of any specific statement. Why this target rather than another is not established here, and two rival accounts are left standing. The agency comparison involving Trump and Putin is a comparison against a lifeboat crew, not a complete account of what causes migration.

On Thursday the Crown Prince of Saudi Arabia rang the President of the United States twice in one day. The first call told him the Houthis were advancing on Mocha. The second asked for American airstrikes. He did not get them. He got intelligence and targeting support, and the two hundred or so American personnel already in the country stayed in their non-combat role.

Nobody in Europe was on either call.

By Friday the Houthis held the whole of Yemen’s Red Sea coast, Mocha included, and most reporting puts them on Perim Island, in the Bab al-Mandeb itself. The same day Saudi Arabia shut down Petroline - twelve hundred kilometres of pipe running from the Eastern Province across to Yanbu on the Red Sea, built to move oil out of the Gulf without it passing through Hormuz, at a capacity of up to seven million barrels a day. A drone strike closed it. The drones came from Iraq, not from Yemen. Hold onto that; it comes back at the end.

There is a word for Europe’s position in all of this.

In economics a price-taker is a firm too small to move the market. It does not set the price of what it buys or what it sells. It accepts whatever the market hands it and builds its entire operation around that fact. The price is not its decision. Its only decision is what it does once the price arrives.

That is exactly Europe’s position on the oil. Not on the war - a war is not a market, and nobody is a price-taker in one.

Europe is not absent from the region. It has ships in the Red Sea, money across the Gulf, embassies everywhere that matters. What it lacks is influence over what Washington, Riyadh, Tehran or the Houthis decide next - which is a political weakness, and a separate one. Set that against Ukraine, where European weapons and European sanctions buy a seat at the table - a compromised seat, bought at a price Europe is still arguing about paying, but a real one. Here there is no seat. There is an invoice.

Brent touched a hundred and eight dollars on Thursday, the same day as the phone calls, its highest since May. It settled back to around a hundred and four on Friday - close to nine per cent up on the week, and more than fifty-five per cent up on the year. Bloomberg Economics runs a model for exactly this. At a hundred and ten dollars a barrel it puts roughly a point on euro-area annual inflation and takes roughly six tenths off growth. At a hundred and seventy, roughly double both.

You will have seen the bigger numbers. A hundred and fifty to two hundred dollars has been named repeatedly since March - Macquarie, Eurasia Group, UBS as recently as this week, Jason Bordoff at Columbia. Every one of those figures carries a condition: a sustained, full closure of the Strait of Hormuz lasting quarters rather than weeks. Seven months into this war, that has not happened. The base case those same analysts hold sits nearer eighty to a hundred and ten, and the EIA’s September outlook still forecasts around ninety for the second half of the year - a forecast that may already be out of date. The tail is not impossible. It has a specific condition attached to it, and the condition has not been met.

A cost-of-living shock, then, landing on a continent where the far right is already gaining ground.

Here is the conclusion that writes itself, and the first half of it is true. The gains are real and they are documented. A governing party heading into a winter of rising prices and falling growth is in serious trouble, and anyone telling you otherwise is selling something. If you have watched European politics for a decade and your instinct is that this shock pushes Europe right, that instinct has earned its keep.

What I dispute is the second half, and the reason is that the mechanism underneath is better evidenced than the conclusion drawn from it.

Inflation punishes incumbents. Not conservatives. Not progressives. Incumbents.

That is a counted pattern rather than a law, and here is the count. Twenty twenty-four was the first year on record, going back to nineteen-oh-five, in which every governing party across the developed world lost vote share - all of them, through a synchronised post-pandemic inflation shock, and regardless of which direction the government facing it happened to face.

Be careful about what that does and does not say, because I am about to ask it to hold a lot. It is strong on incumbency and silent on direction. It tells you the sitting government loses. It does not tell you who gains. That silence is the gap the tidy story fills in without being asked to, and filling it is the move I am objecting to.

So here is what would show me wrong, and it is countable. If incumbency is not the operative variable, you would expect challengers of the reactionary right to gain under this shock while challengers of other kinds, running against reactionary incumbents, fail to. Count that across the next year of European elections. If it holds, the mechanism I am describing does not exist and the tidy story was right all along.

One observation relevant to that count has already happened, and it did not go the tidy way.

Hungary voted on the twelfth of April. Peter Magyar’s Tisza party took 53.18 per cent against Fidesz’s 38.61, ending sixteen years of Orban. It was fought explicitly on cost of living and corruption, against an incumbent campaign built on a war-or-peace security frame with public backing from Trump and Vance.

The incumbent who was punished was the reactionary. But be precise about the beneficiary, because the imprecise version dresses a weaker argument as a stronger one. Tisza is not a party of the left. It is pro-European, it is broadly centre-right, and its position on immigration is harder than the word centrist suggests. Hungarian voters did not reject the right. They rejected one right-wing incumbent in favour of a different offer - more European, less authoritarian, more plausibly competent.

The anger did not disappear. The vessel changed.

The second observation arrives tomorrow. Sweden votes, and as of this weekend the polling is genuinely split - Verian had Kristersson’s governing bloc ahead 52.7 to 44.6, Demoskop had the centre-left opposition ahead 50.6 to 47.3. Cost of living and energy supply sit at the top of the issue polling alongside immigration. By the time you read this it may be settled. I am writing before it is, which means you can check my working against a result I did not have.

The useful question is not which way each country leans. It is who is holding the bill when it lands, and how much authority they have left to argue about it. France governs without a majority. Spain governs as a PSOE minority. Germany runs a grand coalition whose main achievement is continuing to exist. Three different politics, one shared condition: constrained incumbent authority, about to be handed a price it did not set and cannot refuse.

Which brings back the drones. They came from Iraq. The instinctive read - it must be the Houthis, they are closest and loudest - is the read that failed on this same pipeline in 2019, when an attack the Houthis claimed was later traced by American intelligence to southern Iraq. The obvious answer was wrong then, on this specific piece of infrastructure. Proximity is not evidence.

Being a price-taker does not stop you reading the market. The reading is the part you own.

So read this one properly. The claim that a cost-of-living shock has one political destination is not a description of the mechanism. It is one of the inputs to it. The reactionary is the most available vessel for that anger, not the only one - and availability is not a fact of nature. It gets built, partly by everyone who insists it is inevitable.

The price is set elsewhere. What it buys is not.

Reuters, Axios and Al-Monitor report the Trump-MBS calls and the limited American response. Al Jazeera, OilPrice.com and the Iraqi prime minister’s office are the sources for the Petroline shutdown and the Iraqi origin of the drones; the Wall Street Journal for the American attribution of the 2019 attack on the same pipeline. Brent figures from market reporting for the tenth and eleventh of September. Inflation and growth estimates from Bloomberg Economics’ SHOK model; the base case from the EIA’s September outlook; the higher scenarios from Goldman Sachs, Macquarie, Eurasia Group, UBS and Jason Bordoff at Columbia. Hungarian results are official returns via electionguide.org. Swedish polling is Verian and Demoskop.

Builds on the project’s standing work on the global incumbent-punishment wave, and on Where the Framework Stops and Which Way They’re Facing. Debts to Adam Tooze on energy and inflation, Lawrence Freedman on the Gulf, and Cas Mudde on why a shared stance need not mean a shared politics.

On evidence: the event chain and the figures are documented. The seven million barrel figure is Petroline’s capacity rather than its current throughput, which is lower. Reporting that Iran’s Revolutionary Guards instructed the Houthis to escalate against Saudi Arabia is sourced but disputed, and is not used. That inflation punishes incumbents is documented. That the political beneficiary is therefore undetermined is inference, resting on one wave and one result - a pattern and a single relevant count, not a law. The Swedish result was unknown at the time of writing.

On Friday, on a stage in Birmingham, Nigel Farage and Jordan Bardella signed a document. Reform UK has published it, and it is worth reading yourself, because it is printed in two columns — English on the left, French on the right — and the columns do not say the same thing.

The clause that matters is the last one. Britain intercepts boats that set out from French beaches and crossed into British waters, and returns the people aboard to France. Then, in English: “France will ordinarily accept these persons”, and will send them home under French deportation law.

The French column has no acceptance clause. It moves straight to France returning those people to their countries of origin. A French reader is not told that France has agreed to take anybody. A British reader is told France has agreed to take them. Usually.

Farage was asked about that word at the press conference. He said it was legalese.

Sit with that, because a memorandum of understanding is made of nothing else. And the harder question does not appear to have been put to him at all: the word he was waving away does not exist in the other column.

It is the document’s only commitment about what France does when Britain hands somebody back. The English version hedges it. The French version leaves it out.

I am not going to tell you this was deliberate, and I am not going to offer you a tidy alternative either. Bilingual texts produced at speed go wrong in several dull ways. What matters is that the discrepancy sits in the one sentence a real agreement would be argued over. Whatever put it there, that is not the thing it is being sold as.

The reaction in France arrived inside a day, and landed precisely where the French text is silent. Edouard Philippe, a former prime minister now standing for Horizons in next April’s presidential election, called the pact a U-turn for the National Rally, and said of Bardella agreeing to take back migrants intercepted in the Channel: “that one we didn’t see coming.” Gabriel Attal, another former prime minister, called it submission to foreign interests. Philippe also asked the question nobody has answered — whether Marine Le Pen agrees. She is the party’s presidential candidate, restored to eligibility by the Paris appeal court in July. Bardella is its president. A signature from one is not a settled position from the other.

Put the two columns down, though. There is something larger in this document than a translation problem.

Everything in it is a promise about what two future governments will do. Clause five is different. It describes a physical operation: interception of small boats inside British territorial waters, removal of everyone aboard, return to France. It is the mechanism behind what Reform calls Operation Fortress.

That act has been tried. Not proposed. Tried.

In 2021 Priti Patel announced turnaround tactics for the Channel, and this was not rhetoric. Parliament legislated the express powers in the Nationality and Borders Act. The Home Office bought specialist maritime equipment. Border Force officers began training. The Royal Marines ran trials at Weymouth to establish whether it could be done.

The answer came back no. Not no in principle — no in that water. James Heappey, then armed forces minister, told the Commons Defence Committee on 12 July 2022 that the Ministry of Defence had advised against the tactic from the outset; that the trials, together with an assessment of the water and the threat, concluded it was inappropriate; and that government dropped it because professional mariners in the Royal Navy made the case against it compellingly. To collide deliberately with another vessel in a shipping lane that busy would be, in his words, “a breach of the saving life at sea convention”.

The policy had already written its own limit into itself. Its operational guidance, disclosed in the litigation, told commanders not to attempt a turnaround unless reasonably confident the boat could reach French shores unaided. Heappey told the same committee that overloaded small craft in the Channel are each treated as a vessel in distress and escorted to Britain, because the obligation to save life at sea requires it. Set those two beside each other and the policy very nearly cancels itself.

Then it was withdrawn, and the sequence is worth having straight. The Ministry of Defence took primacy over Channel operations on 14 April 2022. Six days later a military order notified the Home Office that the MoD’s joint commander had never been given permission to authorise turnaround tactics. On 24 April the Government Legal Department confirmed the policies under litigation were withdrawn, and the High Court hearing listed for 3 May never happened.

So here is the claim, and here is what would show it wrong. The memorandum’s preamble is aimed at the wrong obstacle. It asserts, before it sets out a single commitment, that international law — the European Convention on Human Rights named specifically — must not stop sovereign nations securing their borders. But no court ever constrained this policy, because the policy never reached a court. It was gone before the hearing.

I am not going to claim the military order was written on the strength of the Weymouth trials; the public record does not say so. What it does say is that the professionals asked to carry out the operation concluded it could not be done safely, that the authorisation was never given, and that it was never done.

The Convention does appear in this story, though. It appears inside the Home Office’s own policy document, as a limit the Home Office wrote for itself.

Two things would falsify this reading. If either party publishes an operational answer to the maritime problem — any method by which boats can be intercepted and turned in the Dover Strait without breaching the duty to assist a vessel in distress, using whatever vessels, tactics or cooperation they think have changed since 2022 — then they are doing the work rather than performing it, and I am wrong. Or if a fresh professional maritime assessment reverses the old one, the central fact loses its force. Neither requires an election. Both are checkable now.

There is a test worth applying to any political claim: remove the stated premise and see whether the position still stands. The premise here is that human rights law is the barrier. So suppose a Reform government left the Convention on its first morning.

The Refugee Convention would still be there. Non-refoulement is a principle of the Refugee Convention, which is a separate treaty, and leaving one does not release you from the other.

The duty to render assistance to a vessel in distress would still be there. That is not a human rights instrument at all. It is the law of the sea, older than any of this, and on the armed forces minister’s own account it is why these boats get escorted in rather than turned around.

And the Weymouth finding would still be there. That one was never a legal objection. It was mariners saying the manoeuvre could not be carried out safely in that water.

Take the premise out and the policy does not move. Which tells you the premise is not load-bearing for the policy. It is load-bearing for the politics.

I teach people to read conditions outdoors. Most of that job is teaching them to recognise what a place is going to do before it does it, and to hear it when somebody who knows that place says no. The hardest part is the bit where you accept that conditions are not negotiating with you. When the people whose job is that water say a manoeuvre cannot be carried out without risking lives, that is not an opinion you can win an argument against. You can override it. You cannot refute it.

France refused this from its own side at the time. Gerald Darmanin, then interior minister, called turning boats back at sea a red line: “France will not accept any practice that breaks maritime law.” The English column purports to answer him: a future National Rally government pre-commits to accept what he refused. The French column does not contain that commitment at all. Take the English text at its word and that is one half of the 2021 failure addressed. The half the Royal Marines found off Weymouth it does not mention at all.

There is a comparison available, though it is not like for like. Britain and France already have a returns arrangement: Operation Hillmore, the one-in, one-out pilot running since September 2025, a ratified treaty between two sitting governments, signed off by the European Commission, carrying the reciprocal legal route that made France agree to it. By the first of May it had returned six hundred and six people and admitted five hundred and eighty-eight the other way, against nearly seventeen thousand small boat arrivals in the seven months to March. The reporting put the return rate at around three and a half per cent.

That does not prove an interception scheme would fail; they are different mechanisms answering different questions. What it shows is the distance between what the machinery produces with a treaty behind it and the certainty sold on a stage in Birmingham, where Farage said the pact would stop the boats forever.

Which is rather the point, and this part is inference, so hold it as that. Bardella said before signing that the two of them could be heading their governments tomorrow, and were preparing accordingly. A document with numbered clauses and two signatures performs exactly that — whether or not a single clause of it could survive contact with a Force 5 and a lawyer.

The strange thing is that the document knows where its problem is. Somebody drafted that preamble. Somebody decided this text needed to say, before anything else, that international law must not constrain a sovereign nation’s borders. That is not decoration. That is a party looking straight at the obstacle and naming it.

They named a court.

The obstacle was the sea.

You can withdraw from a convention. You can pass the Act, buy the kit, train the crews and legislate the powers. Britain did all of that, four years ago, and then went out on the water and found out.

The sea is not a signatory.

Sources and acknowledgements

The memorandum was signed in Birmingham on 4 September 2026 and published in full, in both languages, by Reform UK; the two texts of clause five are compared directly from it, and Farage’s “legalese” answer is from the press conference the same day. The 2021-22 history draws on James Heappey’s evidence to the Commons Defence Committee of 12 July 2022, the pushback policy’s own operational guidance as disclosed in the judicial review brought by Freedom from Torture, Care4Calais, Channel Rescue and the PCS union, Gerald Darmanin’s statements of September 2021, the withdrawal sequence of 14 to 24 April 2022, and Robert Jenrick’s written answer that November. The Hillmore figures are Benjamin Haddad’s to a French parliamentary committee, as at 1 May 2026, against Home Office arrival figures to 31 March 2026.

This builds on earlier work here on the load-bearing test, and on Which Way They’re Facing, which asked where Reform is pointing domestically. This one asks the same question about the water.

The texts, the trials, the ministerial evidence and the withdrawal sequence are documented. What the military order of 20 April rested on is not, and I have said so in the piece rather than filling the gap. Why the two language versions differ is likewise not established, and I have not claimed to know. Reading the memorandum as a performance of governing seriousness rather than an operational plan is inference from the document’s own contents and its signatories’ own framing, and is offered as that.

Reform UK’s conference runs at the NEC in Birmingham today and tomorrow. It opened under a Channel 4 News investigation, broadcast on Thursday evening, in which two senior figures were filmed discussing how half a million pounds of foreign money might be routed into the party. That money was discussed, not taken. What did change hands was smaller and real: three polls the party commissioned, costing £32,500, paid for by a US company. Both men have stood down pending an internal inquiry. The Liberal Democrats have written to the Metropolitan Police and the Electoral Commission. Labour says it will do the same. The Met has not opened an investigation; it says it is assessing what it has been sent. Reform denies wrongdoing.

Two days earlier, Zack Polanski told the Camden New Journal he intends to stand in Holborn and St Pancras — the seat Keir Starmer vacated on Tuesday after eleven years, six weeks after he stopped being Prime Minister.

You could file those as two stories about two parties having very different weeks. They are not the same story; their mechanics could hardly be less alike. They are two instances of the same problem, and it is a problem neither party can route around.

Which way are they facing.

## The question

A party growing from nothing grows on a flank. The people who join first joined precisely because the mainstream was intolerable, and early on they are the whole engine — more committed, louder, more useful than anyone else available. The problem arrives when the party stops being able to grow that way and starts needing people who are not like that at all. It has to decide, then, who it is answering when it speaks.

Not what it believes. Who it is answering. A party can hold the same platform for two years and change entirely by changing the direction it turns to address.

This is not a law of political development and nothing here establishes it as one. It is a recognisable problem, and two British parties are visibly inside it at the same time.

## Reform: asked in public, answer withheld

Reform’s flank has a name, an office and a Member of Parliament. Restore Britain was launched by Rupert Lowe as a pressure group on 30 June 2025, announced as a party on 13 February 2026, and registered with the Electoral Commission on 20 March. Lowe was elected as a Reform MP in 2024 and had the whip withdrawn in March 2025; Reform alleged threats against its then chairman and referred its own MP to the police, and Lowe says he was pushed out for pressing mass deportations harder than Farage would go. Restore’s deportation paper sets a target of half a million or more voluntary departures and between 150,000 and 200,000 enforced removals a year. Its platform would outlaw the burqa, sharia courts, and halal and kosher slaughter.

Its campaigners are more direct than its website. Canvassing in Makerfield in June, an unnamed Restore campaigner told *EU Reports* that Reform were ex-Conservatives doing the bare minimum on illegal migration, and that Restore, by contrast, believed in “ethnic homogeneity and a homeland for our own particular ethnic group”. That is one campaigner speaking for himself, and he should be kept separate from the party. Restore’s own site does not use that language. It says native British births will fall below half by 2030 and native Britons will be an absolute minority by 2070. What can be said is that a demographic argument and an ethnic one were running side by side in the same campaign, and only one of them was in writing.

The fork opened on 3 August. After Restore’s candidate finished ahead of the Conservatives in the Greater Manchester mayoral by-election, Lowe made a public offer of an electoral pact. The terms reported were not vague: mass deportations, reversing mass immigration, a burqa ban, a referendum on capital punishment, and “demographic security” for native Britons. A price, itemised, in public.

Farage’s answer came hours later and named nobody. It committed to no talks, did not mention Lowe, said Reform would welcome anyone who shared its aims, and made clear he was not walking away from the people already around him. Read generously, a leader keeping his options open. Read structurally, a party asked what it will pay for its flank and declining to say.

A month later something else happened, and the fact and the reading of it are two different objects. The fact: presented on Thursday with footage of its own head of policy and the leader’s communications adviser discussing foreign money, Reform called the investigation “yet another establishment attack on Reform UK”. That is documented, and it is the party’s own wording.

The reading is mine, and the seam should be visible. A party under that kind of scrutiny has two audiences available — the people who already believe every institution is corrupt, and everyone else it needs in order to govern. The establishment-attack frame is addressed to the first. One line of crisis rhetoric cannot carry a claim about settled strategy, and I am not asking it to. It is a data point about direction, and it arrived a month after the same party declined to answer the same question when Lowe asked it politely.

## The Greens: the question becomes visible without being asked

The tempting move here is to hand the Greens a left-hand Restore and let the symmetry close. It does not survive contact with the evidence, and the reason it does not is the most interesting thing in this piece.

The Greens have no Restore. There is no breakaway to their left competing for their voters and demanding a price. Nobody has handed Polanski an itemised bill. What exists instead is a set of occasions on which the direction of address became briefly legible in a party that does not usually state it.

Be honest about what that sentence is doing. These are not four instances of one phenomenon, and grouping them is my move rather than something the events perform by themselves. One of them is not a Green decision at all. What they have in common is not their type. It is that on each occasion you could see, for a moment, which way the party was turned.

Start with the one that is not a decision. In the Gorton and Denton by-election in February, George Galloway’s Workers Party announced it would not stand, saying the decision was in the interests of the working class and that “Labour and Reform must lose”. The Workers Party had taken nearly as many votes as the Greens in that seat in 2024, and John Curtice had already described the contest as one where the two were competing for the same voters. The Greens won it.

That tells us almost nothing about Green strategy. The agency was entirely elsewhere. The Greens did not negotiate it, did not reciprocate it, and were under no obligation to refuse a gift that was never offered to them directly. What it tells us is about the ground rather than the party: there is an electorate the Greens and the Workers Party are both reaching for, and one of them has already shown it will stand aside to make sure that electorate is not split. That is worth knowing before a second favour arrives.

The other three are decisions, or something close to one.

Take March first, and take it slowly, because it is the one most easily overread. At spring conference on 28 March, held on Zoom with around a thousand members present out of a membership above two hundred thousand, Motion A105 — “Zionism is Racism” — was never voted on. Brought by the Greens for Palestine group and proposed by Lubna Speitan, it had gathered more than 330 co-proposers, a record for a Green conference motion. Opponents first tried to rule it out of order on the grounds that its call for a single state contradicted existing two-state policy. Supporters then successfully fast-tracked it up the agenda. In the closing stages, a run of no-confidence motions in the chair was introduced, each of which the party’s constitution required to be debated separately, and the clock ran out. Platform failures compounded the delays. The motion was not defeated. It was deferred, and it returns at the October conference.

Those are the facts and they are worth keeping clean, because they do not point where a hurried reading would send them. This was not a leadership suppressing its flank. The obstruction came from the motion’s opponents; its supporters had the numbers to move it up the agenda. Polanski is reported to oppose the motion, has not moved against it publicly, and under the party’s internal governance has no authority to keep it off the floor.

Here is my inference and it is separate from all of that. A leader who cannot prevent the question being asked is in a materially different position from one who has chosen which way to face. Reform’s problem is that Farage will not answer. The Greens’ problem may be that Polanski cannot.

In May, before the Makerfield by-election, the party split openly over how hard to campaign against Andy Burnham. Caroline Lucas argued for restraint, on the grounds of Burnham’s position on electoral reform. Jenny Jones argued the party should always give voters a Green option. That is the fork stated in its purest form, by two people who have both spent their lives in the party: is the point to be present, or to be relevant?

And now Camden. Polanski has said he intends to stand; he has not been selected, and local members choose. There are reported splits among Green activists over whether the candidate should instead be explicitly and primarily a Gaza candidate. That is not a personality dispute. In 2024 Andrew Feinstein, standing chiefly on Gaza, took nearly nineteen per cent in this seat and the Green candidate took over ten.

Reform has been asked the question once, in public, and has not answered. The Greens have not been asked so directly. They have simply kept ending up in positions where the answer becomes visible anyway.

## What the Workers Party is doing in this argument

It is not the Greens’ flank, and calling it that would be the flattening this piece exists to argue against. Its role here is narrower, and it should stay narrow.

Three things, and no more. It cleared a path for the Greens in February. It has documented working relationships well outside conventional Green politics: in April 2026 it co-promoted a rally outside the Ukrainian embassy in London with Unity News Network, run by David Clews, who spoke at conferences of the neo-Nazi group Patriotic Alternative in 2022 and 2023, with the Workers Party’s own deputy leader billed as a speaker. And its leader has spent 2026 in what he calls exile in Russia, having attended Putin’s Victory Day parade in Moscow.

From which one thing follows, and only one. Electoral cooperation from this direction is not evidence of ideological convergence, and it should not be read as any. But it is not free either, and a party that accepts it twice has made a choice the second time that it did not make the first.

## Shared structure, not shared substance

Restore Britain and the Workers Party are not the same thing. One is ethnonationalist, the other socialist, and on substance they are not neighbours and never will be.

Cas Mudde describes populism as a thin-centred ideology: not a programme but a structural stance, a virtuous people against a corrupt elite, which attaches to almost any host politics. That is Mudde’s, and it explains how two movements can share a stance and share nothing else.

What follows is mine. They reach for the same words. In a 2024 interview, Galloway said he no longer wanted to hear himself described as a leftist, because the word now meant refugees welcome. Then: “We are the patriots. It’s the globalists who are the traitors”. Polanski, appealing to disillusioned Labour voters, says “we need to reclaim patriotism”, and argues that the genuinely unpatriotic thing in Britain is inequality.

Same contested word, and the difference between the two uses is the whole test. One claims the word and names a category of traitors. The other claims the word and names a condition. My proposition is that a thin stance becomes dangerous at the point an enemy is fitted to it — not Mudde’s claim, and not a settled one. It is testable, which is the only reason to keep it. Watch which of those two moves a party makes when it is under pressure and losing, because that is when an enemy is cheapest to reach for.

There is a tempting alternative explanation running through psychology rather than structure — that people at both edges are contrarians by disposition, or worse, that they have taken leave of reason. Set the second aside: it is unfalsifiable, since calm and fury both read as confirmation. The first is more respectable and probably backwards. Contrarianism is more likely the story people tell about themselves after arriving somewhere for ordinary reasons than the engine that took them there.

## What would prove this wrong

The argument rests on a trade-off: that intensifying the committed constituency and broadening the address pull against each other, so that a party has to choose. If that trade-off does not exist, there is no fork and I am describing a shape that is not there.

Start with what will not settle it, because the obvious test is the one that fails. In 2024 the Gaza-forward independent and the Green candidate took a combined 29.3 per cent of Holborn and St Pancras between them. It is tempting to say that if a Green candidate clears that figure the two constituencies were never separate. That does not follow. A by-election result carries too much else — Labour’s condition, turnout, the quality of the candidate, whatever has happened in the world by polling day. One number cannot separate my explanation from four others, and I am not going to pretend otherwise.

What can bear weight is direction, repeated, and one scheduled event.

Direction first. Reform is around a quarter of the national vote and Restore three to four per cent. The trade-off predicts that over several quarters those two lines move against each other rather than together. A single quarter proves nothing. A sustained parallel rise across a run of them is real evidence against me, because that is the pattern the trade-off says should not happen, and it is much harder to explain away than a single result.

Then October. Motion A105 returns to Green conference. That is a rare thing in this kind of analysis: a dated occasion on which a party is forced to face one way or the other in public, rather than an inference drawn from an outcome. Whatever the party does with it, and whatever Polanski does or cannot do, the answer will be observable rather than reconstructed. If the motion passes and the Greens’ standing outside the party is undisturbed through the quarter that follows, the trade-off did not operate at the point where it was most exposed. That would not prove it never operates. It would mean I cannot locate it where I said it was, which is close enough to being wrong to say so plainly.

And the outright killer, for either party: sustaining both at once. If a party can intensify its committed constituency and extend beyond it across a run of contests, without one costing the other, then there is no fork, and everything above is a pattern I brought with me.

One asymmetry belongs on the page rather than in a footnote. Reform is polling around a quarter of the vote, has an itemised offer from an ethnonationalist party sitting on the table, and is facing police referrals. Nothing on the Green side is at that scale. The structural question is shared. The stakes are not, and any account that treats them as equivalent is doing the flattening.

## The turn

Reform’s conference will produce a speech today and another tomorrow. Camden Greens will hold a selection meeting. In October a deferred motion comes back to a conference floor. None of those events will announce itself as the moment a party chose.

But somebody is going to turn, slightly, toward one part of the room and away from another. That is what the decision looks like from outside. It is not a manifesto and it will not be in a press release.

Watch which way they turn.

**Sources.** Reform donations investigation: Channel 4 News / Verbatim, broadcast 3 September 2026; reported ITV News, 3 September 2026. Reform conference: NEC Birmingham, 3–5 September 2026. Polanski candidacy: Camden New Journal, reported 3 September 2026. Restore Britain chronology: Electoral Commission registration 20 March 2026. Makerfield campaigner quote: *EU Reports*, 22 June 2026. Lowe pact offer and terms: reported 3 August 2026. Workers Party withdrawal in Gorton and Denton: party statement, 1 February 2026, reported *The London Economic*. Green spring conference and Motion A105: 28 March 2026, online; motion deferred to autumn conference. Polanski’s reported position and the limits of leadership authority over conference motions: *Financial Times*. Makerfield campaigning split: *The Guardian*, 18 May 2026. Holborn and St Pancras 2024: Feinstein 18.9%, Green 10.4%. Galloway interview: Russell Brand, 2024, via *International Socialism*. Workers Party / Unity News Network rally: *Searchlight*, April 2026. Polling averages: Politico poll of polls, early summer 2026.

NOTHING FALLS DOWN

Some political claims hold the roof up. Some are bolted to nothing. The difference is testable, and it does not run between left and right.

On 12 February 2025, the American defence secretary Pete Hegseth told European allies that Ukraine would not be joining NATO. Later the same day, Nigel Farage said Ukrainian membership was probably essential — his words were hedged, a guess, something he would have thought looking ahead — and he repeated it that evening on GB News.

Eight months earlier, in a Panorama interview with Nick Robinson broadcast on 21 June 2024, the same man had said: “We provoked this war.” He meant the eastward expansion of NATO and the European Union. When the criticism came he restated the framing rather than withdrawing it. Rishi Sunak called it completely wrong. Lord Robertson, who ran NATO from 1999 to 2003, said Farage was parroting the Kremlin line, and added that saying we provoked Russia is like saying buying a burglar alarm provokes burglars.

Two positions, eight months apart, pointing in opposite directions. Nothing in the public record I have been able to examine shows him reconciling them, and I am not going to pretend that is a scandal. It is a symptom, and the interesting question is what of.

If you want to know whether a wall is holding the roof up, you do not stand and look at it. You work out what would happen if it were not there.

That is the whole method. Take a political claim out of the position built around it and see what moves.

Here is what would show this reading wrong, and I want it before the evidence rather than after. If a claim is load-bearing, removing it should make some specific position unreachable — not weaker, unreachable. Show me that the positions I say survive the removal actually collapse without it, and the reading fails. Three things sit outside the test, because they are different arguments and I am not making them. This is not about anyone’s sincerity. It is not about anyone’s motives. And a claim that turns out not to hold up a policy is not thereby empty — it may be doing explanatory or moral or tribal work I have not measured. I am testing one thing only.

Start with the case that passes, because a test you only ever run to destruction is not a test.

Stop the War’s 2025 annual general meeting resolved to continue calling for Britain to leave NATO, which it described in the same sentence as an expansionist, imperialist alliance. The claim is not decoration there. It is the subordinate clause doing the justifying.

Now remove it. Delete the proposition that NATO is an instrument of imperial expansion and see which positions you can still reach. Opposition to British troops in Ukraine — still reachable, from pacifism alone. Opposition to rising defence spending — still reachable, from the welfare argument, which stands on its own. But leaving NATO is gone. Not unreachable by any argument at all — non-alignment, disarmament and sovereignty all get you there. Unreachable within the argument they actually made. Strike that clause and the resolution stops justifying its own demand.

The wall is holding something up.

Their March 2026 resolution then restated the same priority while explicitly registering the American turn — a British government seeking a privileged place in the order Washington is imposing, American backing for far-right parties in Europe. They noticed. The framework took those facts in and produced a reading of them. That is what a working claim does. You can disagree with every word, and I do, but you are disagreeing with an argument, and an argument can be met.

Now run the identical procedure on Farage.

Remove “the West provoked this war” and see what moves. Ukraine’s membership is probably essential — unchanged. Britain has more than done its bit for the alliance, which is what he said arriving at Davos in January 2026, pushing back at an American president claiming the opposite — unchanged. He would vote against British troops in Ukraine — unchanged, and it never needed the provocation claim, because cost and capacity get you there on their own.

Nothing falls down.

That does not tell you the claim is doing nothing. It tells you it is not doing this. What it is for remains open, and I would rather leave it open than fill the gap with a motive I cannot evidence.

There is a third result, and it is the one that makes this a test rather than a verdict.

For years the Green Party of England and Wales held that NATO was not a sustainable mechanism for maintaining peace and that a Green government would take Britain out of it. A neighbouring claim to Stop the War’s rather than the same one — theirs turns on imperial expansion, this on whether the alliance could keep the peace at all — but the same structure, load-bearing the same way. Then in March 2023, at spring conference, the membership voted it down. It is on the conference record: a voting paper instructing the policy committee to delete the existing chapter outright and replace it with one acknowledging NATO’s role in letting member states respond to threats. The Scottish Greens did not follow.

That is a wall being tested and moved. Not by an interviewer, not by a columnist — by a room of members who had held the claim for decades and decided the invasion had changed what it rested on. And it has stayed contested since: conference material this spring described the party moving to a more nuanced position while a competing amendment proposed keeping withdrawal as a medium-term objective. The argument is live, which is what an argument being load-bearing looks like from the inside.

So what travels, and what doesn’t?

Two days after Peter Hitchens argued in the Mail on Sunday that the war had been provoked, his column reappeared on the English-language site of International Affairs, the official journal of Russia’s foreign ministry. He did not write it for them, and the republication tells you about the republisher’s editorial appetite rather than anything about him.

Look at what made the journey. The claim did, and so did much of the history he built it on. What did not survive is the provenance — the version that ran there is cut short and credits him to the wrong newspaper. The portable core crosses. The things that let a reader check it, and check him, are what fray at the edges.

None of which is new, and better to grant it early. Left and right have converged against foreign wars repeatedly — Kennedy and Sanders alongside Ron Paul over Iraq, the far-left Nation alongside the Chicago Tribune over intervention in 1941. The pattern is old, it recurs, and its recurrence is evidence of nothing on its own. What is worth noticing is that the claim doing the travelling and the claim doing the work are no longer the same claim.

All of which matters this week rather than in the abstract.

Elbridge Colby has been setting out what he calls NATO 3.0 since February, and a six-month review of American force posture in Europe has been running through the summer, designed to move fast and irreversibly towards Europe taking primary responsibility for its own conventional defence. Read the shape of it: the extended nuclear deterrent continues, the conventional load shifts. Narrower than withdrawal, and the thing actually on the table.

Then on 28 August the Telegraph reported that Pentagon officials were weighing American support for British sovereignty in the Falklands as pressure on Andy Burnham over defence spending. Handle that carefully. Single anonymous official, whose actual words were that nothing is being ruled out — weaker than the headlines built on them. Every outlet carrying it is carrying the Telegraph, which is reach, not corroboration.

Burnham faces the Commons this week, into all of that.

So here is the question.

If your account of why this war began is load-bearing — and I have spent several hundred words arguing that it is — then the policy that follows is part of the argument and not a separate preference that can be set aside while the premise stays. When Washington starts pricing the alliance and your answer is the answer you already had, that is your framework working. It is not the world agreeing with you.

And the argument is being had. The TUC carried a Ukraine solidarity motion in 2023 that Stop the War had lobbied delegates to oppose. The UCU overturned its earlier position in 2024 and affiliated to the Ukraine Solidarity Campaign. UNISON affiliated the same year. The NEU passed a solidarity motion in 2025, after several years in which such motions did not reach the floor — that last part I have from the Ukraine Solidarity Campaign’s own account of the procedural history rather than from conference records, and it is weaker evidence than the votes themselves.

None of which means the left is converging on anything. It means the argument has been running for four years, in rooms that get a fraction of the attention the same argument gets online.

Which leaves the phrase you and Farage have in common. The test says only that it holds nothing up on his side of the argument. What it is doing there instead, I do not know, and I am not going to invent it.

You already know how to check this.

Take the wall out. See what falls.

Farage’s Panorama remarks were broadcast 21 June 2024 and reported by Reuters, the Associated Press and the Guardian; the Sunak and Robertson responses come from contemporaneous coverage. His conditional support for Ukrainian NATO membership was given 12 February 2025, the day Hegseth ruled it out, and repeated on GB News that evening. Stop the War’s positions are quoted from its published AGM resolutions. The Green Party’s decision was taken at spring conference, March 2023, on voting paper B03. Hitchens’s column ran in the Mail on Sunday, 26 April 2025; the republication appeared at International Affairs two days later. The union record draws on TUC Congress reports and the Ukraine Solidarity Campaign. Colby’s NATO 3.0 framing dates from February 2026.

This piece builds on Where the Framework Stops and Britain Has Already Answered, and on the project’s standing distinction between convergence and coordination.

On evidence: the dated statements are documented. The counterfactual reading built on them is my inference, with its falsification condition stated in the body. The Green Party motion and its passage are on the conference record; the account of the debate around it is from contemporaneous reporting. The NEU procedural history rests on the Ukraine Solidarity Campaign’s account rather than on conference records. The Falklands report is single-sourced and treated as an uncorroborated claim throughout.

At around ten past eight on the evening of Friday 28 August, a Russian drone hit a warehouse in the village of Myla, in the Bucha district of Kyiv Oblast. The strike started a fire. What killed people was what happened next: the ammunition in the warehouse began to detonate, and went on detonating long enough that a stretch of the Kyiv-Chop highway had to be closed. At least thirty-seven died and forty-two were injured, four of them children. The heaviest damage was to a care home for elderly and disabled residents beside it. Among the dead was Taras Didych, who ran the local community and who was there because he had gone to help.

Ukraine has opened criminal proceedings over alleged official negligence. Zelensky said plainly that ammunition is not supposed to be stored beside houses and that those who allowed it will answer for it. The defence ministry ordered an audit of every site where weapons and ammunition are held. It was the second disaster of this kind near Kyiv in under two months.

That was three days after the director of the CIA left Moscow, and on the same Friday that American officials briefed Zelensky on what had been said there and on a proposal for a summit.

Laid out in sequence, the week reads like a design. A CIA director flies to Moscow in secret for the first time since November 2021, meets the heads of the SVR and the FSB, and does not meet Putin. The president denies the obvious reading of the trip on a radio show. Days before, thirty-odd governments meet in Kyiv and hand over missile technology. Days after, Washington is reported to be weighing pressure on an ally’s territorial position.

Three different things get collapsed into one when a week looks like this, and it is worth pulling them apart before going further.

There is central design: someone intends the shape as a whole and works to produce it. There is adjustment: actors reading each other and moving accordingly, which happens constantly between allies and requires no author. And there is independent response: separate institutions arriving at similar decisions because they face similar constraints, with no contact needed at all.

Only the first is what people mean when they say a plan. The second and third get mistaken for it all the time, because from the outside all three produce the same thing — a pattern.

So I am not telling you nobody coordinated anything. Washington is openly pressing London and London is openly responding. Kyiv is asking for specific things and being partly answered. Adjustment is plainly happening and pretending otherwise would be its own kind of story.

I am telling you something narrower. Nothing in the public evidence requires central design to explain what you saw. The sequence can be accounted for by adjustment and independent response, without anybody having intended the combination.

That is a smaller claim than “there was no plan,” and it is the one the evidence can carry. You don’t have to prove a conspiracy didn’t happen. You only have to show it isn’t needed.

Here is what would break it. Not evidence that these governments talk to each other; they do, constantly, and that isn’t what I’m contesting. What would break it is evidence that the shape was intended — someone who wanted Britain squeezed while Moscow was courted and worked to produce that combination, rather than pursuing each thing for its own reasons and finding they landed in the same week. A document, an account, a person who says that is what they were doing. And separately: if Ratcliffe went for reasons unconnected to the constraints below, then what looks like convergence isn’t convergence at all.

Now the constraints. Note that these are not one set of numbers. They are several different ledgers running the same kind of sum.

In an interview released on Sunday 23 August, the day before the Kyiv summit, Zelensky put figures on the record that are normally held back. Ukraine received 675 American Patriot interceptors across 2023. Across 2025 it received 364. The projection for 2026 is 264. He asked for at least 300 for the winter; Ukraine’s own air force had asked for 360. He gave three years rather than four, so read it as three points, not a curve. The direction is not in doubt. Demand rising, supply falling, said out loud, with the arithmetic already done for you.

Separately, at the Coalition of the Willing meeting the following day, the European Commission approved 6.1 billion euros of defence procurement — air and missile defence, ammunition, radars — drawn from the ninety-billion-euro Ukraine Support Loan. Real money rather than communiqué language. It is not, though, a payment against the twenty-seven billion Zelensky named as Ukraine’s 2026 defence funding gap. Different institution, different mechanism, different account. The coalition’s own statement carried no new figure at all. The gap and the disbursement are both real. They do not cancel.

Separately again, the United States has been managing a shortage of its own since it went to war with Iran. On 17 August the Navy awarded Raytheon 22.9 billion dollars over seven years to take Tomahawk production from sixty missiles a year to more than a thousand, firming up a framework agreed in February; earlier the same month the Pentagon signed deals to ramp up Patriot and THAAD interceptor components. Tomahawks are not interceptors and none of this reaches Ukraine this winter. Different ledger. Same sum.

Three ledgers. One shape.

So: the Moscow visit. What is documented is the date, the aircraft, the route through Riga, the two men he met, that Putin was not among them, and that Peskov confirmed it afterwards, calling intelligence-channel contact a positive phenomenon while the wider relationship stayed where it was. What rests on two sources speaking to a single outlet is the proposal of a Trump-Putin-Zelensky summit, the first time Ratcliffe has been personally involved in the diplomatic push over Ukraine rather than in intelligence liaison. What is genuinely unresolved is why he went: the New York Times reports he made no threats about consequences, CNN reports he carried a warning against action on NATO territory and pressed Russia to cut its support to Iran, and Trump called the trip semi-routine and dismissed the speculation before adding that something might come of it. I am not going to reconcile those. Either the administration is downplaying a real message, which would be ordinary, or it does not have one version of what the trip was for.

With that on the table, here is my reading, offered as reading and not as reporting.

The American calculation is the more visible of the two. Set the interceptor figures beside the Raytheon contract and you get a country applying pressure it is currently short of the means to sustain. That is a reason to open a channel, and it is inference from published figures rather than from anything anyone has said.

The Russian calculation is a further step out, and rests on one newspaper’s sourcing. According to the Washington Post’s account, the Kremlin had concluded that the Iran war has left America too stretched to intervene. If that is right, the gap was not only real but noticed, and a conversation becomes more attractive than a test. If it is wrong, the first inference stands and the second falls. They are not the same claim and I would rather you could see the join.

Then, on Saturday, the same shape appeared from a different building. Kaupo Rosin runs Estonia’s foreign intelligence service. He told the Telegraph that Russia is in a very bad place, that unhappiness with the war is rising through every layer of society up to the oligarchs, and that Putin has only bad options: continuing risks internal pressure eventually reaching his own position, stopping risks looking weak. He also said what a less careful man would have left out. He sees no street uprising forming. He knows of no organised effort to remove Putin. He cannot say whether any of this changes what Putin decides.

Tallinn did not need Langley to tell it this. When two services with no obligation to agree describe the same shape within four days, that is a form of evidence — and it is evidence about Russia’s condition, not about anybody’s plan.

Constraint is becoming visible inside the Russian system too, which is a weaker and more useful claim than saying Moscow has reached the same diagnosis. On 29 August, Zelensky’s deputy chief of staff Pavlo Palisa said Russia’s political leadership had ordered its forces to the administrative boundaries of Donetsk and Luhansk by the end of December, and that the military leadership came back asking for March 2027. The same day, the Russian cabinet extended the fuel export ban for direct producers to the end of September. The week before, Putin signed a decree letting the state take temporary control of infrastructure it judges inadequately protected. None of that is collapse. All of it is a system doing arithmetic it does not enjoy.

Britain has less room to refuse the burden-shift than the domestic conversation assumes — less room, specifically, to decline replacing American conventional capability in Europe on Washington’s timetable. On Monday, Andy Burnham made Kyiv his first foreign trip as prime minister and approved release of classified information on British components of the Storm Shadow so Ukraine can build the capability itself. On Thursday, Elbridge Colby was in Brussels telling allies to move faster on replacing what Washington has provided, saying on the record that he hoped to see more from countries like the United Kingdom. Britain spends around 2.3 per cent of GDP on defence, its June plan leaves core spending near 2.7 per cent to the end of the decade, the alliance target is 5 per cent by 2035, and Healey’s first budget on 28 October is not expected to commit to 3 per cent by 2030. Those figures are as visible in Washington as they are in the Treasury. There is also a Telegraph report that an unnamed senior American official would not rule out the Falklands as a lever, following an internal Pentagon email reported by Reuters in April and played down at the time. Single-sourced, and not load-bearing. Colby said it on the record.

You will be offered a tidier account of this week than mine, and it will be more satisfying, because it will have an author. Someone decided that Britain would be squeezed while Moscow was courted. I understand the appeal and I am not disputing the pattern. The pattern is there.

But a pattern is what you get when a lot of people can see the same kind of sum and each of them can count. An outcome that looks designed is not evidence that it was designed, and the habit of reading backwards from a shape to an intention is the single most reliable way to be wrong about how power actually moves.

No instruction has been produced. No decision point has been named. Nobody has claimed the shape.

You can see a pattern without inventing someone to have drawn it.

Learn to do the sums.

SOURCES AND ACKNOWLEDGEMENTS

The Patriot interceptor figures come from Zelensky’s interview released on 23 August, reported by the Associated Press and carried widely. The twenty-seven billion funding gap and the Storm Shadow decision come from the Coalition of the Willing meeting in Kyiv on 24 August. The 6.1 billion euro approval is the European Commission’s own announcement under the Ukraine Support Loan. Details of the Moscow visit draw on Washington Post, CNN, New York Times, Axios and Kremlin statements carried by Tass. The Estonian assessment is Kaupo Rosin’s Telegraph interview. Myla casualty figures come from Kyiv Oblast authorities and the Prosecutor General’s Office; the Palisa remarks from the Institute for the Study of War. British spending figures are the government’s June investment plan and NATO’s published shares.

The piece builds on this project’s earlier work on interceptor stocks as finite inventory rather than background condition, and on individually defensible decisions aggregating into a position nobody has defended publicly. It relies throughout on the distinction between convergence of function and coordinated intent, which this project has had to correct in its own drafting more than once.

On evidence: the visit, the meeting, the figures and Colby’s remarks are documented. The summit proposal is well corroborated but traces to two sources at one outlet. The Falklands claim is single-sourced and anonymous. The two inferences about the Moscow visit are separated in the text and rest on different evidence. The central claim is not that no coordination occurred, but that none was needed to produce what you saw.