The man warning you about idiots forgot to check his own work

Mark Manson is half right about intellectuals. The other half is him doing the thing.

Mark Manson has published an essay arguing that intellectuals are, in his phrasing, f*cking idiots, and the awkward thing about reading it as a psychologist is that he is right for about two-thirds of the way down the page. The argument he is right about is old, and true, and worth saying again. Clever people build models of the world. The models are useful. Then, somewhere along the line, the model stops being a tool the person uses and becomes a room the person lives in, and once you live somewhere you defend it, and you stop noticing that the world outside the window has quietly stopped matching the floor plan.

He tells this through Malcolm Caldwell, the Marxist academic who admired Pol Pot enough to fly to Cambodia and offer notes, and was shot in Phnom Penh for his trouble. He tells it through Robert McNamara, whose statistics proved America was winning Vietnam right up to the helicopters on the embassy roof. Good cases. Cleanly chosen. Reality always wins, Manson says, and I am not going to stand here and argue that it doesn’t.

I am going to argue with something else. The essay is not a description of the trap. It is a worked example of it. Manson builds a model, the model is intellectuals are idiots, and then he spends the back third of the piece defending that model the way Caldwell defended his: by collecting the evidence that flatters it and waving off the evidence that doesn’t. He has written a sermon against a sin and delivered it from inside the sin. I don’t think he knows. That is the part worth slowing down for, because not knowing is the whole mechanism. It is not a footnote to his argument. It is his argument, happening to him, live, while he types.

Watch the voice change

Here is the tell. For most of the essay Manson is the cool one, the man at the next table who can see through everyone else’s pretty model. Then he reaches the climate activists, the ones who glue themselves to roads, and the voice changes. It stops being diagnostic and starts being personal. These people, he tells us, are not really worried about the climate at all. They are empty inside. They are probably just angry at their parents. They are acting out, in his framing, unresolved attachment stuff.

And then, in almost the same breath, he reassures us that the panic is unwarranted anyway, because the marginal cost of energy is heading towards zero, technological innovation is exponential, and carbon capture is about to become economical.

Read that twice. He has just made three confident predictions about how energy markets and technology will behave over the coming decades. That is a model. It is an elegant, untested, unsourced, lovely-on-paper model, and it is the precise species of thing he spent two thousand words mocking other people for trusting. He doesn’t catch it. He can’t catch it, because catching it would mean turning the instrument around, and the instrument he is holding only points at other people. That is not Manson being ironic. That is irony arriving uninvited and sitting down at his table.

The psychoanalysis is the worse half. ‘They’re just angry at mum and dad’ is not an analysis. It is contempt wearing a lab coat. I do this for a living. I sit with people’s actual attachment histories, the real ones, the ones that take months to surface and do not fit on a placard, and I can tell you that you cannot read a stranger’s childhood off the fact that they are annoying you on a Tuesday. What Manson has done is build a tidy little model of the activist mind, a model whose single function is to let him dismiss the activist without engaging the argument. Which is, word for word, the thing he says idiots do. He even names it. Shut down the discourse rather than engage with it. Then he does it, in print, to people he finds tiresome.

The therapy claim, which is where I put my coffee down

Then he gets to my profession, and this is the part that made me stop reading and go back to the top to check I had not misjudged him.

Manson tells you that trained clinical therapists, on average, do no better than a random person you might talk to, and that more training does not improve a therapist’s results. The implied conclusion, and he does not work very hard to keep it implied, is that clinical psychology is an expensive way of buying something you could get from a kind stranger on a bus.

This is what a real research finding looks like after it has been through a wood chipper. There is genuine literature under there. The finding that the therapeutic modality, the brand of therapy, matters far less than people assume is real and well replicated; it even has a nickname—the Dodo bird verdict—after the bird in Alice who declares that everyone has won and all must have prizes. The finding that the relationship between therapist and client predicts outcome better than the technique does is also real. Manson has taken those two findings, pressed them flat, and read off the sentence ‘therapy barely works.’

That is not what the literature says. The foundational work here is Smith and Glass, who pooled hundreds of controlled trials and found an effect size of about 0.85, which in plain terms means the average person who had therapy ended up better off than roughly eighty per cent of comparable people who had none (Smith & Glass, 1977). Now, in the interest of not doing the exact thing I am accusing Manson of, I will tell you the part that complicates my case. That figure has been challenged. A later reanalysis recoded a sample of the same studies and put the effect closer to 0.30 to 0.50 (Matt, 1989). Fine. Take the gloomy end. Take 0.30. An effect of 0.30 is still real, still replicated, still clinically meaningful. It is not ‘barely works.’ It is ‘works, modestly, and honest people should argue about the size.’ Those are different sentences. Manson needed the first one, so the second one did not make the edit.

The alliance research sharpens the point rather than blunting it. Pooling 295 studies and more than thirty thousand patients, the correlation between the strength of the therapeutic alliance and the outcome of the therapy sits at about r = 0.28 (Flückiger et al., 2018). Manson reads ‘the relationship matters more than the technique’ as a debunking. It is the opposite of a debunking. It is the field locating its own active ingredient. The relationship is not the embarrassing absence of the science. The relationship is the science. A man arguing that the experts missed the point has, in this instance, missed the point the experts spent forty years and thirty thousand patients establishing.

And while we have the bonnet up: the claim that diversity training makes people more racist is sourced to a think-tank report rather than peer-reviewed work, on a question the actual research treats as genuinely unsettled. The figure that seventy-five per cent of corporate training makes employees worse traces back, when you follow it, to an infographic on a training vendor’s blog. A man writing an essay about the danger of building your worldview on bad data has built a chunk of his essay on bad data. I am not saying that to score a point off him. I am saying it because it is the point. It is the entire point. He is in the room he is describing and he cannot see the wallpaper.

The one model he never tests

Every model in the essay gets walked out to the edge of the cliff and shown the drop. Caldwell’s. McNamara’s. Ehrlich’s. Every model except one. His own.

‘Intellectuals are idiots’ is itself a model, and it is a flattering one, which is the first thing that should make a careful person suspicious of it. It is flattering because of where it puts the man holding it. If you have decided that the credentialled, the studied, and the peer-reviewed are the real dupes, then you, the plain-talking unaffiliated fellow with the Substack, are automatically the clear-eyed one. The model costs its owner nothing. It asks him to give up no comfort, abandon no belief, sit with no doubt. And it pays him, in return, the exact thing the essay says models illegitimately pay people: a sense of being right, a side to be on, an identity. He even describes that transaction. He just exempts the model performing it.

Look, too, at the line-up. The idiots in the essay are a Marxist, a population-panic environmentalist, a diversity trainer, and some road-gluing climate kids. They lean, every one of them, the same way. A smoke detector that only goes off in other people’s houses is not a smoke detector. It is a mirror that the owner has decided to call a window, and the decision is the comfortable part, and the comfort is the warning sign.

Here is the contrarian move that was sitting right there, and that Manson walked past. The most dangerous intelligent idiot is not the professor. The professor can at least be peer-reviewed, contradicted, heckled at a conference, embarrassed in a journal. The most dangerous one is the popular communicator who has decided that not being an intellectual is the same thing as being immune. Because nobody peer-reviews a Substack post. ‘I’m just being honest’ is the most frictionless model ever built. It updates for nothing. It feels like humility and functions like armour. It is the gentleman’s club with the membership requirements removed, and the man at the door insisting he isn’t in a club.

Keep the two-thirds that’s true

None of this means bin the essay. Hold your opinions loosely. Get close enough to the consequences of your ideas that reality gets a vote before it gets a veto. Spend less of your life watching faces on a screen and more of it in front of faces in a room. All true. All worth pinning above the desk. The sharpest line in the whole package is not even Manson’s; it belongs to a commenter, Vass Tsokov, who points out that the failure was never intelligence on its own. It was intelligence with the feedback loop cut. Prestige, applause, and institutional reward wrap a clever mind in insulation, and an insulated mind drifts from reality while feeling more certain by the year.

Sit with that one a moment longer than Manson did, though, because it has his name on it. What is a large engaged audience, a bestselling reputation, and an algorithm that pays out for confident contrarian takes, if not prestige and reward wrapping a mind in insulation? Manson has feedback loops. They are excellent. They count restacks, subscriber growth, and time on page. They do not, anywhere in the dashboard, count whether he got the therapy literature right. He will be rewarded for this essay whether it is true or not, and that sentence is not my cleverness, it is just his own essay read back to him with the name changed.

So read him. He writes well, and the warning is real, and two-thirds of it is the best version of itself. Just take the warning and apply it to the man who wrote it, because he forgot that step, and the forgetting is the most honest thing on the page. Reality always wins. It has not got around to this essay yet. Give it time. Reality is thorough, and it is patient, and it does not read the bestseller list before deciding who is next.

References

Flückiger, C., Del Re, A. C., Wampold, B. E., & Horvath, A. O. (2018). The alliance in adult psychotherapy: A meta-analytic synthesis. Psychotherapy, 55(4), 316–340.

Matt, G. E. (1989). Decision rules for selecting effect sizes in meta-analysis: A review and reanalysis of psychotherapy outcome studies. Psychological Bulletin, 105(1), 106–115.

Smith, M. L., & Glass, G. V. (1977). Meta-analysis of psychotherapy outcome studies. American Psychologist, 32(9), 752–760.