Whitehall's war against the future

Dominic Cummings is the former director of Vote Leave, former chief adviser in Downing Street, and the man most likely to tell you, with no apparent pleasure, exactly which official in which committee killed the thing you wanted to build. He arrives not as a Westminster memoirist but as a diagnostician. The post-1945 order — UN, NATO, WTO, WHO, IMF, the European project — was built for a world that no longer exists, and his frame for the whole conversation is brutally simple: the institutions and the ideas gradually drift out of alignment with reality, and then you have crisis. We are, on this account, somewhere in the gap between the drift and the crisis.

He starts with the technology because that is where the gap is widest. The people who predicted the most success for machine learning have turned out to be the most accurate predictors of the future, the straight lines on the graphs have stubbornly kept being true, and the political world is doing what it always does, which is practise deliberate blindness to the whole thing. Stack the exponentials together — frontier AI, democratised biological engineering, models improving month on month, and what he calls completely crackers agencies regulating all of it — and you get a state of the world that Westminster treats as a fourth-order junior-minister hobby. Technologically it is increasingly China and California that dominate, and Europe, he says flatly, is not in the game: a mix of stagnation and anti-growth bureaucracy dedicated to Leninist centralism in Brussels. Britain’s one accidental piece of luck is that, through sheer inertia, it has not yet adopted every EU regulation and so has not quite shot itself in both feet the way Brussels has.

Through Little Dorrit and the Circumlocution Office, through Northcote-Trevelyan, through the room in summer 1914, Cummings builds the case that the rot is structural and old: by Cummings’s reckoning, 1795 Whitehall was better at procurement than 2025 Whitehall by a massive, massive factor. The Cabinet Office is now the centrepiece. The two things everybody at its founding agreed would be a disaster if it ever happened are now, he says, literally its official functions. Cabinet itself has become a Potemkin process; the real decisions are taken by some director-general or task force, and everyone on the other side of the Number 10 door knows the old system is fake.

The mechanism is everywhere. Officials write memos saying legal advice forbids the sensible thing; ask to see the legal advice and there is no document. Very few MPs have ever hired or fired anyone or built anything, so they cannot grip the machine even when they want to. And the machine has a worldview: in the Cabinet Office, Cummings claims, it is explicit that you cannot talk about personal responsibility, because that is bullying, or fascism. The whole apparatus is designed to programme the prime minister psychologically so that he does not even know what his own powers are — a huge amount of theatre, the Friday box of appointments, a steady drip of “you just can’t do that.” Calum presses on whether this is simply what modern democracies are, and gets the counterintuitive optimism in return: in practical terms it is far easier to do real regime change in Britain than in America or anywhere in Europe, if anyone wanted to.

The prescription is that science and technology must become a fundamental aspect of the prime minister’s job — a top-three priority embedded in economy, security and institutional reform, not a fourth-order issue handed to a junior minister. Britain’s aerospace past demonstrates why. Britain genuinely had frontier aerospace ideas, the Barnes Wallis lineage, the engineers, the possibility — and from the sixties Whitehall shut down the entire way of thinking that says Britain might produce frontier things itself. Tom names the law of the whole episode here: when technology comes up against an ideological commitment from the governance class, technology loses. The idea of Britain building something genuinely futuristic, Cummings says, brings out an allergic reaction in Whitehall, and the fact that it works only makes them more determined to stop it. There is a great deal of talent here and a great many things that could actually be done. The people responsible for budgets and power are actively hostile to doing them.

The episode explores

— Why the post-1945 order drifted out of alignment with reality, and what happens in the gap before the crisis arrives

— The straight lines on the AI graphs that kept being true, and why the people who predicted the most success have been the most accurate

— Democratised biological engineering, exponentially improving models, and the completely crackers agencies meant to be regulating all of it

— Why China and California dominate the frontier, Europe is not in the game, and Britain’s only luck is the EU regulations it was too inert to copy

— Little Dorrit, the room in summer 1914, and the claim that 1795 Whitehall beat 2025 Whitehall at procurement by a massive factor

— How the Cabinet Office became the exact two things everyone agreed at its founding would be a disaster

— Why “I take full responsibility” in Parliament now means “I take zero responsibility”

— The memos that cite legal advice forbidding the sensible thing — and the legal advice that turns out not to exist as a document

— Why talking about personal responsibility inside the Cabinet Office gets reclassified as bullying, or fascism

— The Friday box and the theatre that programmes a prime minister not to know what his own powers are

— Why science and technology has to be a top-three prime-ministerial priority rather than a junior-minister hobby

— Barnes Wallis, the British space plane, and Whitehall’s standing view since the seventies that very big, very futuristic projects are Britain’s out

— A state so broken it has failed four separate times to restart sewage monitoring

Dominic Cummings writes on AI, science, procurement and state capacity at his Substack.