Have We Lost the Instinct That Built Bridges?: Brunel, Musk and Vanishing Margin

Cross the River Tamar by train into Cornwall and you ride across an argument that has been settled for 167 years. The Royal Albert Bridge at Saltash still carries the load, exactly as Isambard Kingdom Brunel intended when Prince Albert opened it in 1859 — built by a man with no computer, no stress model, and no way of knowing what would one day run across it, who simply made it far stronger than anything of his era could conceivably require. Every time I make that crossing I find myself thinking about the man most often called his modern heir, and about how completely he has inverted the one principle that kept the bridge standing.

The heir, of course, is Elon Musk. The comparison is half right, which is more than most comparisons manage — and the half where it fails is the half that should worry us.

Begin with what the two men genuinely share, because it is not trivial. Neither could ever bear to supply a mere component. Brunel’s Great Western was never just a railway; it was a complete system — a chosen gauge, the stations, the tunnels, and a steamship waiting at Bristol to carry the passenger onward to New York. Musk’s enterprises have the same totalising shape: SpaceX builds the engines, the rockets, the launch site and the Starlink constellation above them; Tesla the car, its battery packs and increasingly its own cells, the charging network and the software. Both wanted the entire stack, and distrusted anyone who told them a piece of it belonged to someone else.

They share, too, a near-reckless appetite for ruin. Tesla and SpaceX both came within days of collapse in 2008: Musk poured in the last of his fortune, split it between the two rather than let either die, watched the fourth Falcon 1 reach orbit after three straight failures, and closed a Tesla round in the final hour before payroll would have bounced. Brunel lived the same way. The Great Eastern, the largest ship in the world, was a commercial catastrophe that bankrupted a succession of her owners — and then became the only hull afloat big enough to lay the first lasting transatlantic cable in 1866. Both men ran a portfolio of long-odds bets and ate the failures — a stalled prototype here, an abandoned line there — for the rare payoff that changed the world. And both understood spectacle as capital: Brunel floated the great Saltash truss into place before some thirty thousand silent spectators, directing it by flag signal; Musk livestreams a launch and moves a market with a single post.

Nowhere is the kinship tighter — or more instructive — than in a single shared failure, separated by 175 years. Brunel’s South Devon atmospheric railway of 1847 ran a sealed pipe between the rails with a piston towing the train, pumping stations drawing the air out ahead of it so that ordinary atmospheric pressure could push it forward. It was abandoned within a year: the continuous leather valve that sealed the pipe froze in the winter cold, dried and cracked in the salt sea air, and finally began to shred — and the pumping stations burned coke so wastefully that the whole system cost far more to run than ordinary steam. (Legend blames the rats that supposedly nibbled the tallow-greased leather; historians think they were a footnote, not the cause.) Brunel swallowed the loss. Musk’s Hyperloop is the same wager with the geometry reversed — put the pod inside a near-vacuum tube to abolish air resistance — and it foundered on the grown-up version of the identical problem. Hyperloop One raised more than $450 million, never won a single contract to build a working line, and shut down at the end of 2023; the Boring Company’s one demo tunnel was quietly turned into a car park. Maintaining a pressure differential inside a very long tube, against leaks and heat and cost, turns out to defeat the optimiser exactly as it defeated the Victorian. The physics does not care which century you attempt it in. There is even a closing irony: Hyperloop One’s remaining patents were handed to DP World of Dubai — the dream did not so much die as emigrate to the Gulf.

The lesson sits in what each man did with that defeat. Both lost the tube. But Brunel expected to be surprised, and built the rest of his work — the bridges, the hulls, the tunnels — with margins fat enough to survive the surprises he could not predict. That is the difference the comparison keeps circling back to.

So far, a flattering likeness. But here the resemblance breaks, on one fault line, and the fracture matters more than the parallels. Brunel was an over-builder. His instinct was to insure against the load he could not foresee — to pour iron and cost into margins that looked wasteful to his contemporaries and look like genius to ours, because the bridge is still there. Musk’s governing creed is the precise inversion: strip the mass, delete the part, “the best part is no part,” iterate to the very edge of failure and rebuild after it breaks. It is a brilliant philosophy. It has put more rockets in the sky and more electric cars on the road than any rival, faster and cheaper than anyone thought possible. And it is, at its core, a bet that the future will resemble the present closely enough that optimising for today’s known loads is safe.

That bet is no longer Musk’s alone. It has become the dominant engineering creed of the age — in our software, our supply chains, our power grids, our just-in-time everything. We strip redundancy because redundancy is expensive and, in any ordinary year, unused. We optimise relentlessly for the average case and quietly assume the catastrophic one will not arrive on our watch. It is a magnificent way to build when the world behaves. It is a catastrophic way to build when it does not — and the events that destroy systems live precisely in the tail of the distribution that the optimiser has trained himself to ignore.

This is what Brunel understood in his bones and what we have very nearly forgotten. He did not build for the load in front of him; he built for the load he could not yet name. The margin he paid for looked like waste right up until the moment it was the only thing standing between a structure and its failure — and because he paid it, the moment of failure never came. We have inherited his cathedrals of iron and concrete and concluded, wrongly, that they were over-engineered. They were correctly engineered for a world that surprises you. We are the ones who have been getting it wrong, mistaking the absence of disaster for proof that the insurance was unnecessary.

There is an aesthetic loss folded into the engineering one. Brunel cared, genuinely and at cost, that a thing was beautiful — the proportion of a span, the line of a portal, the experience of crossing the Tamar — because he was building for a century, not a product cycle. The relentless optimiser rarely shares that reverence, and it shows: a Brunel structure was made to be admired in a hundred years; a great deal of modern hardware is designed to be superseded in five. Permanence and beauty turn out to be the same discipline wearing two faces, and we have abandoned both for speed.

None of this is to diminish what Musk has achieved, which is staggering, nor to pretend Brunel’s appetite for risk came free — his margins ruined backers and broke his own health, and the over-builder’s faith always carries a bill. It is only to say that the man we have crowned as Brunel’s successor has inherited his nerve and his vision while discarding the one habit that made his work outlast him. Musk builds trajectories meant to bend the future. Brunel built monuments meant to survive it. And as we remake our entire civilisation in the optimiser’s image — leaner, faster, stripped of every margin a balance sheet can find — it is worth asking, on the long ride across the Tamar, whether we have quietly stopped building the things that are still here in 167 years.

Brunel would have admired the nerve enormously. I am not at all sure he would have signed off on the tolerances.