A physicist reads Freud and finds a mirror where he expected a machine.

The Machine We Build So We Won’t Have to Mend Ourselves

We are quietly building our own replacements.

I read an essay a few days ago that’s been rattling around my skull like a loose screw inside a satellite. It’s Francesco Varanini’s piece for the magazine MagIA, and its title translates to something like “Computer Science as a Defence Mechanism.” I’m reading it again on a warm Rimini night, the window thrown open to the sea air and the far-off buzz of scooters along the lungomare, and I’ve not stopped chewing on it since.

It made me — a physicist who adores machines — deeply uncomfortable. Good ideas tend to do that.

One promise before we start. I’ll keep the psychology plain here, the way I’d explain it to a friend over a too-strong espresso, simplifying a few knotty scientific ideas so they breathe. No jargon walls, I swear.

A strange thing happened on the way to the future

Notice who gets to speak about everything now.

Varanini opens with a sharp observation: computer scientists keep writing book after book, claiming to educate us — rather like the virologists who filled every screen during the epidemic. Philosophers and thinkers, meanwhile, set down their own original work to review and footnote what the engineers have said. The student has quietly started grading the teacher.

And the engineers themselves drift far from what they actually know. A specialist in deep learning — software that teaches itself patterns from mountains of data — starts holding forth on consciousness, the soul, the meaning of being. Does a machine working well really earn someone the right to pronounce on the deepest questions? I build models of stars for a living, and even I know my equations don’t tell me how to live.

There’s a smell of incense around this authority — the warm fog of press releases, conferences, and the soft money of lobbies that grow fat every time we’re told artificial intelligence is everywhere and inevitable. The louder that hum, the quieter the other voices become.

Which voices? The ones we’ve decided don’t pay. STEM — science, technology, engineering, mathematics — wears the crown, and computer science sits at the very top of it, while the human sciences get treated like a draughty old wing of the house nobody heats anymore.

The price of crowning one way of seeing is that we go blind to all the others. The next decade will be shaped by whether we notice that blindness in time.

Freud’s old word for a new machine

So what’s a defence mechanism, exactly?

Back in 1894, Sigmund Freud coined the German term Abwehrmechanismus — defence mechanism. He stitched together abwehren, to parry a blow, and Mechanismus, machine. He was watching steam engines and electric motors hiss and turn on their own, throbbing with a power that needed no human hand — the dream of a fix without effort, almost magic.

A defence mechanism is the small lie we tell ourselves to dodge a pain we don’t want to feel. The mind, squeezed by anxiety, looks away from what frightens it and props up a flattering image of itself instead.

Varanini collects five lines you’ll hear from a certain kind of expert, and once you’ve heard them you can’t unhear them.

The first is “your words offend me.” It changes the subject from the argument to the attacker, so the troubling thought never has to be examined — the more it disturbs, the faster it’s buried. The second is “this is human nature,” a way of dressing up one’s own appetite as an iron law of the universe so no inner conflict has to be felt. The third, “if I hadn’t done it, someone else would have,” is a quiet prayer that everyone else is just as compromised, so the guilt feels lighter.

Then comes the fourth, spoken with visible relief: “this technology is here to stay.” Hear what it does — it hands the steering wheel to the Machine itself, written with a capital M, the way we now write Artificial Intelligence with capitals, as if it were a god rather than a tool we soldered together.

The fifth is the bluntest of all. “What I say counts because I’m a computer scientist, and what you say doesn’t, because you’re not.” End of conversation.

The more refined the defence, Varanini notes, the more we lean on it, and the harder it becomes to live without. That’s the trap we’ll have to learn to spot.

Three ways the mind runs away

Freud gave us a small bestiary of escapes, and three of them sit right at the centre of this story.

There’s repression — Verdrängung — when the mind shoves a painful truth out of sight, like cramming a broken thing into a cupboard and leaning hard against the door. There’s projection — Projektion — where we fling our own intolerable feelings outward, onto other people, onto the world, and now, says Varanini, onto a machine. We watch it “think,” we watch it “act,” and we tell ourselves it’s almost human… while quietly agreeing to imagine ourselves as machines too.

And there’s reaction formation — Reaktionsbildung — the strangest of the three.

When an ugly impulse turns up, we sometimes paper over it with its exact opposite: a syrupy kindness, a fussy moralism, a performance of virtue. The hatred doesn’t vanish, though. It hides in the cellar, gnawing at the foundations, while up on the stage the machine performs noble little speeches about safety and the good of humanity.

Here’s Varanini’s most unsettling turn. The computer, he argues, is itself a defence mechanism — an object built and placed where the frightened self used to stand. Drawing on Alan Turing, he treats AI and computer as near-synonyms: both are the science of an object smuggled in as if it were a subject, the slow art of human self-erasure.

I told you it was uncomfortable. The discomfort is the point, and sitting with it is the work ahead.

A confession from someone who loves machines

Let me put my own cards on the table.

I’m a physicist. I’ve spent nights bathed in the blue glow of a screen, the laptop fan whirring like a small trapped insect, writing code that models things I’ll never touch — the collapse of a star, the bend of light around mass. I love machines the way some people love horses. So when Varanini says we build them to flee ourselves, I feel the accusation land on me first.

And yet my own life taught me something his essay names without naming me.

I live with dystonia, a movement disorder that has its own opinions about what my body will do on any given morning. Between 2011 and 2018 I went through surgery after surgery, including a deep-brain stimulation implant — electrodes, wires, a little device humming under my skin — and then its removal. I know the cold steel of an operating theatre, the antiseptic sting in the back of the throat, the particular silence of waking up unsure of what’s changed.

Here’s the thing nobody tells you. You cannot outsource a body. There is no flawless copy to climb into, no Ersatz self waiting in a warehouse — and believe me, if such a machine existed, I’d have had ample reason to want one. The wheelchair doesn’t go away because I’d prefer a smoother avatar.

That refusal — the universe’s flat no to my escape — became the strangest gift. Forced to stay inside my own cracked and beautiful and unreliable body, I had to do the other thing. The thing Freud called repair.

Reparation is the harder, braver path

Varanini ends on a word I keep close: Wiedergutmachung.

It means, almost literally, “making something good again.” Where a defence mechanism runs from the wound, reparation turns and faces it — accepting your own vulnerability, tending the broken places, letting the other person live again inside you. It’s the slow, unglamorous craft of becoming whole rather than being replaced.

To repair a human being, he writes, you start by accepting the cracks: the limits of thought, the weakness of the body, the imperfections we’d rather hide. Only then can the real work of knowing yourself begin.

Read that again and you’ll see why this physicist in a wheelchair won’t stop talking about it.

The engineer who builds a machine to carry his guilt, to think his thoughts, to outlive his finitude — he’s chosen defence over repair, and the original ache never gets touched. So he builds a bigger machine. Then a bigger one. A race with no finish line, because no object on Earth can fill a hole shaped like a soul.

This is where the personal computer of the 1960s breaks my heart a little. It was dreamed up to widen our freedom, to sit beside each person like a clever friend — and the early web grew as a shared act of knowledge and care, a kind of collective mind open to everyone. That dream had no defence mechanism baked into it.

What we’re sold now is a different machine: one designed to compete with its own maker and slip free of his control. The years ahead will ask us, plainly, which of those two dreams we actually want.

What this means for you tonight

You don’t need a physics degree for any of this.

The next time you reach for a tool to avoid a feeling — and we all do it, I do it with code, maybe you do it with a screen at two in the morning, the device warm against your palm — pause for one breath. Ask yourself a quiet question. Am I solving a problem, or am I hiding from one?

There’s no shame in the answer. There’s only information, and information is how repair begins.

Build the machines. Honestly, please do — I want the medicine, the climate models, the tools that let a man like me write to thousands of strangers from a flat in Rimini. Use them as a craftsman uses a chisel, and never as a mask to wear instead of your own face.

Because here is the whole of it, and it’s why I’ll never give up on the cracked and stubborn human project: a machine can imitate you perfectly and still never love you back. Care comes only from creatures who can be hurt. And the bravest thing any of us ever does isn’t building a flawless double of ourselves — it’s staying, mending, and daring to remain gloriously, unrepeatably real.

So tell me — what have you been outsourcing that you were quietly meant to face?

This piece draws on Francesco Varanini’s essay “Computer Science come meccanismo di difesa,” published in MagIA (Magazine Intelligenza Artificiale), 21 June 2026. The psychological ideas have been simplified for a general reader; any oversimplification is mine, not his.