What Is a Human Life Worth?

Somebody, at some point, has to decide what a person was worth. A court awarding damages. An insurer pricing a policy. A government deciding how much a safety regulation is allowed to cost. A military planning an operation. A hospital allocating a treatment that not everyone can receive. None of these institutions particularly wants the job, and none of them can avoid it, because refusing to produce a number is itself a decision that produces a number.

The interesting part is not that the numbers are wrong. Of course they are wrong. The interesting part is that each institution arrives at a completely different answer, each answer is internally coherent, and none of them is compatible with any of the others. A life can be worth a projected salary, a retrieval risk, a strategic calculation, a research asset, a procurement line item, or everything a family has, and all of those valuations can be applied to the same species in the same century without anybody noticing the contradiction.

What follows is not an argument that we should stop calculating. We cannot stop. It is an argument about what the calculations are actually measuring, which is almost never what they claim to be measuring.

The Fund That Had to Produce a Number Per Person

The September 11th Victim Compensation Fund is the cleanest example available, because the whole process happened on paper, in public, with a named individual responsible for it.

Congress created the fund eleven days after the attacks. In exchange, the families gave up their right to sue the airlines. Kenneth Feinberg was appointed Special Master and given the task of deciding what each life lost that day was worth. Working with accountants, his team reviewed pay stubs, tax forms, insurance policies and social security statements, and built models to calculate the earnings each person would have accumulated over the rest of a working life.

So the award had two components. Economic loss, which was mostly projected future earnings. And non-economic loss, for pain and suffering, which was set at a presumed figure for everybody, with an additional amount for a spouse and each dependent.

Look at what that structure actually says. The part of the award that varied between one life and another was, overwhelmingly, the earnings part. Pain and suffering was flat, because there was no way to measure whose suffering was greater and no institution would survive trying. So the only thing that could differentiate one life from another, in a document produced by the United States government, was money the person had not yet earned.

Then the edges of the model got strange in both directions. Feinberg treated those earning above the 98th percentile as though they were at the 98th percentile, on the grounds that projecting the earnings of extremely high earners was too speculative to attempt. Cantor Fitzgerald, which lost more employees than any other firm, produced a report arguing that this systematically understated what its traders would have gone on to make. In the later reauthorised fund, Congress capped the annual income that could be considered at $200,000.

Those caps are the most revealing part of the whole thing. They exist because the model, followed honestly to its conclusion, would have produced numbers that were politically unbearable. A bond trader's family would have received tens of millions and a dishwasher's family would have received a fraction of that, and everyone involved understood that publishing those two figures side by side would say something about the country that the country did not want said. So the formula was truncated. Not because the maths was wrong, but because the maths was working exactly as designed and the result was intolerable.

Now put a teacher through it.

A teacher's projected earnings are modest and easily calculated. That is the number that goes in. What does not go in is thirty years of children, some of whom became something because of her, a few of whom would not have made it through without her, none of whom appear in any actuarial table. What does not go in is the colleague she talked out of quitting. What does not go in is that in twenty years a former pupil will teach a class in a way she would recognise.

That is not a failure of imagination on the fund's part. It is a structural impossibility. Legacy has no counterparty. Nobody was ever going to pay her for it, so it has no price, so it cannot enter a model built out of prices. The fund did not measure what her life was worth. It measured what a labour market had priced her remaining output at, and then handed that number to her family as though it were an answer to the question they were actually asking.

Everyone involved knew this. Feinberg has said as much repeatedly since. The families knew it. The fund still had to produce a figure, because the alternative was decades of litigation, and a wrong number arriving in 2003 was better than a differently wrong number arriving in 2015.

That is the real defence of the whole apparatus. Not that it is accurate. That it is faster than the alternative.

The Man Who Became a Landmark

At around 27,900 feet on the north-east ridge of Everest, inside a small limestone alcove, a body lay for roughly thirty years in bright green climbing boots.

He died in the storm of May 1996, one of three Indo-Tibetan Border Police officers who did not come down. The route passes directly by the alcove. Every expedition from the Tibetan side walked past him. Climbers began using him as a waypoint, and the alcove acquired a name: Green Boots Cave. He was not memorialised, not marked, and not moved. He simply became part of the description of the route.

The reason he stayed is a calculation, and it is one of the few genuinely honest ones in this essay. Recovering a body from above 8,000 metres is close to the hardest physical task in mountaineering. It requires multiple people working at altitudes where the body is actively dying, for hours, dragging a frozen weight over terrain where a fall kills everyone attached to it. People have died attempting recoveries. So the mountain enforces a rule that no committee had to write: a climber who dies up there stays where he is, because moving him risks living climbers.

Everyone accepts this. It is not disputed, not scandalous, and not something anyone campaigns against. It is the most widely agreed valuation of human life in this entire essay, and it is agreed precisely because nobody had to write it down.

Here is the part that keeps returning to me, though.

He may have kept people alive.

At that altitude, orientation kills. Fatigue, oxygen deprivation, whiteout, and a route that looks identical in every direction. A fixed, unmistakable, unambiguous marker at a known height is not a morbid curiosity. It is navigational infrastructure. Climbers descending in deteriorating conditions have known where they were because of him. A man who could not be saved became, without anyone deciding it, a reason other people were.

So the same body carries two valuations that have nothing to do with one another. One is a risk calculation that says retrieval costs more than recovery is worth. The other is an accidental utility that nobody priced, nobody predicted, and nobody could have argued for out loud. And these two valuations point in the same direction only by coincidence. Had he been retrieved in 1997, the calculation would have been just as correct, and the second value would simply never have existed.

Then there is the detail that makes the whole thing worse, and better.

For thirty years, the identification was disputed. He was widely believed to be Tsewang Paljor, and for decades that is the name attached to the story. A competing account held that it was Dorje Morup, another officer from the same expedition. Reporting from mid-2026 indicates that Indian authorities, using DNA evidence, have identified him as Morup and are planning a recovery.

Sit with what that means. The most photographed body on the most photographed mountain on earth, passed by hundreds of people, filmed, written about, turned into a waypoint on a route description, and for thirty years we were not certain whose son he was. Two families in Ladakh, each with a reason to believe it might be theirs. One of them was right, and neither of them knew.

He was assigned a function before he was assigned a name.

A Scale at Which Nobody Is Counted at All

Hiroshima belongs in this essay, but it does something different from everything else in it, and the difference is the point.

The 9/11 fund valued individuals. It got the valuation wrong in specific, traceable ways, but it valued them one at a time, with a file per person. Everest values one body against several living bodies. Both operate at a scale where a human being is still the unit.

The decision to use atomic weapons on Japanese cities did not operate at that scale. The reasoning that was offered, at the time and since, weighed projected casualties from an invasion against casualties from the bombing. Both sides of that comparison are aggregates. Estimates against estimates. And crucially, one side of the ledger consisted of people who did not yet exist as casualties, while the other consisted of people who were about to.

Nobody working that problem had a file per person. Nobody could have. The people in those cities were never individuated in the calculation at all, and those who died slowly, of radiation, over the following months and years, were not fully in the model either, because the effects were not understood well enough to be in it.

I am not going to adjudicate the decision here, and I would encourage you to notice how strongly you want me to. Every reader arrives at that paragraph with a position already loaded, and the moment this essay takes a side, it stops being about valuation and becomes an argument about the bomb, which is a different piece.

What matters for this argument is narrower and, I think, harder. Above a certain scale, the practice of valuing individual lives is not performed badly. It stops being performed. The unit changes from the person to the estimate, and once the unit has changed, there is no mechanism by which any particular person's particular life can enter the reasoning. Not because the people deciding were monsters, but because there is no version of that decision made person by person. The scale forbids it.

This is the failure mode nobody has solved. It is the same reason a single identifiable child trapped in a well commands more resources than a famine, and the same reason a statistic cannot make anybody cry. The machinery that lets humans value a life needs one life to point at.

Value Without Any Valuation at All

In 1951, a thirty-one-year-old woman was treated for cervical cancer at Johns Hopkins. During the procedure, tissue was taken from her without her knowledge and without her consent. She died that year. She was buried in an unmarked grave.

Her cells did not die. They divided, and kept dividing, and became the first immortalised human cell line. HeLa cells went into the polio vaccine, into cancer research, into genetic mapping, into work on COVID-19 vaccines. They have been shipped to laboratories on every continent for seventy years. Companies built products on them and sold those products for large amounts of money. In 2021 her descendants sued one of those companies, and in 2023 they settled on confidential terms, on what would have been her 103rd birthday.

Henrietta Lacks is in this essay because she breaks the pattern in the most useful way.

Nobody undervalued her. Nobody performed a calculation and got a low number. The point is that no valuation event ever occurred. She was not weighed against anything, not priced, not compensated at too low a rate. She was simply not part of the transaction. Value flowed out of her body for seventy years through a process that had no place in it where her worth could have been considered, because she had never been asked, and a person who is never asked does not appear anywhere in the accounting.

That is a distinct failure from the fund's. The fund used the wrong measure. This used no measure. And of the two, the second is far more common, because it is invisible. A bad number can be criticised, appealed, capped, argued about in a Cantor Fitzgerald report. An absent number cannot be, because there is nothing there to point at.

Note also what the eventual settlement measured. Not what she was worth. It measured what a company had made from her, seventy years later, once litigation became viable. Even the correction ran through the only channel available: money that had already changed hands elsewhere.

The Calculation That Nobody Performed, and Then Everybody Performed

In 2004 and early 2005, in Al Anbar province in western Iraq, American Marines were being killed by IEDs. An IED is an Improvised Explosive Device, a homemade bomb built from whatever is available, usually old artillery shells, buried in a road or hidden at the roadside and triggered by a wire, a radio signal or a pressure plate. Improvised does not mean weak. They got steadily more powerful throughout the war, and that matters to everything that follows.

The Marines were driving Humvees. The Humvee was designed in the 1980s as a general-purpose military truck, not a fighting vehicle. It sits low, and its floor is flat. The response to the bombs was to add steel plating and thicker glass, producing the up-armoured version known as the M1114. That helps against fragments and bullets. It does much less against a large explosion directly underneath, because armour does not change the geometry, and a flat floor close to the ground takes a blast square on and passes it into the crew.

A vehicle designed for exactly this problem already existed. An MRAP, Mine Resistant Ambush Protected, has a V-shaped hull that deflects a blast outwards rather than absorbing it, and a crew compartment sitting much higher off the ground so the explosion has further to travel. They weigh up to forty tons and cost several times what a Humvee costs. The design was neither new nor experimental. The Marine Corps had already recognised the track record of mine-protected vehicles from conflicts in sub-Saharan Africa and elsewhere.

In February 2005, then-Brigadier General Dennis Hejlik, who had commanded in western Iraq since the previous June, signed an Urgent Universal Needs Statement requesting 1,169 MRAPs. That document is the emergency channel, the one a field commander uses to bypass normal multi-year procurement. He wrote that the Marines could not continue to take serious and grave casualties from bombs when a solution was commercially available.

It was briefed to the Marine Corps Executive Safety Board in March 2005 and the Executive Off-Site in May, which means the entire Marine Corps executive body, including the Commandant, considered MRAP-type vehicles. Commandant Michael Hagee chose the armoured Humvee instead, with the advice of his executives and with full knowledge of Hejlik's request. The command at Quantico responsible for deciding what to buy then did not pursue the request, and never completed the process. The Inspector General later found that the simultaneous decision to buy up-armoured Humvees appeared to have deterred that command from finishing the job, while Hagee told investigators he had not intended that to happen.

The armoured Humvees proved incapable of blunting the increasingly powerful explosives. MRAPs were eventually declared the Pentagon's top acquisition priority by Defense Secretary Robert Gates, and arrived in Iraq in quantity from May 2007, roughly two years after the request.

Here is the crucial thing, and it is the opposite of what the story looks like.

Nobody weighed the price of a vehicle against a Marine's life. There is no meeting where that sum was performed, no memo where someone concluded a man was worth less than the equipment. What happened was that a cheaper, faster-to-field option was chosen for reasons of availability and momentum, and an urgent request then died inside a process where no single person had the job of killing it. In 2008 the Inspector General concluded that officials had abandoned the request, and found no criminal negligence by anyone involved in reviewing it.

That is the same shape as Henrietta Lacks. Not a bad valuation. An absent one. And an absent number cannot be appealed, because there is nothing there to argue with. It took two years and a Defense Secretary to overturn a decision that nobody, individually, had made.

And then, years later, everybody performed the calculation at once.

By 2011 the Pentagon's MRAP office estimated the vehicles had saved as many as 40,000 lives, ten thousand in Iraq and thirty thousand in Afghanistan, across a programme costing close to fifty billion dollars. In 2012 two economists, Chris Rohlfs and Ryan Sullivan, published an analysis in Foreign Affairs arguing the opposite. Using restricted Defense Department data, they found that medium-armoured vehicles were highly effective in units facing heavy combat, saving roughly one infantry life for every seven vehicles bought at a cost of one to two million dollars per life, but that heavily protected vehicles like the MRAP saved no more lives than the medium ones despite costing around three times as much. They also took apart the 40,000 figure, arguing it counted every soldier present during an attack on an MRAP as a life saved, which assumes that in a Humvee everyone aboard would have died, when for long stretches of the war US forces faced sixty or more attacks a day, the overwhelming majority of which produced no casualties at all.

The rebuttals came fast, and from people with better data. Classified combat records reportedly showed that between 2005 and 2011 the average number of troops killed per IED attack in Iraq was fourteen times higher in Humvees than in MRAPs, and nine times higher in Afghanistan. Franz Gayl, the civilian Marine Corps employee whose whistleblowing had triggered the original investigation, cited Inspector General figures putting the ratio at four or five killed in armoured Humvees for every one killed in an MRAP struck by a comparable device. One rebuttal went further and estimated that had the Pentagon acted promptly on the 2005 field requests, 1,609 lives might have been saved.

Meanwhile an internal Marine Corps case study concluded that the MRAP had been the cheaper vehicle all along, once you counted Humvees wearing out in two to three years, armour shortening that further, the future medical costs of casualties, force replacement, and operational effectiveness.

So: same war, same vehicles, same question. One study says fifty billion dollars was wasted. Another says a delay cost sixteen hundred men. A third says the expensive option was cheaper from the start. All of them show their working.

And in the academic version of the argument, the estimates ran from roughly one million to twenty-five million dollars per life saved, measured against a cost-effectiveness threshold of seven and a half million. There it is, written down in a defence economics paper. A figure above which saving an American soldier is deemed not worth the money. Not implied. Not inferred from behaviour. A threshold, with a number attached, that a life either clears or does not.

This is a third failure mode, and it may be the bleakest one in this essay.

The fund used the wrong measure. Henrietta Lacks had no measure at all. Here, the right measure exists, everybody agreed on the units, the data was collected, economists were hired, live-fire trials were run, peer review was applied, and the answer is still unknown. Because the whole thing rests on a counterfactual nobody can observe: how many of those men would have died in the other vehicle, in a war that did not happen.

Every figure from forty thousand to zero is a guess about that war.

The honest conclusion is not that the Pentagon got its sums wrong. It is that the sum cannot be done, even in principle, even afterwards, even by people with access to classified casualty databases and no particular reason to lie. Which raises an uncomfortable question about all the other numbers in this essay, produced under time pressure, in public, by people who had no choice but to produce something.

The Price of a Life Not Yet Ended

Everything so far concerns a life that has finished. The valuation happens afterwards, over a body or a file or an absence, and whatever number comes out, the person it describes is beyond being affected by it.

There is another category, and it is stranger, because the valuation happens while the person is still alive and still capable of finding out what they were worth.

Militaries are unusual among institutions in that they publish their prices. Death gratuities, service life insurance, disability tariffs. But the more revealing number is not what a state pays when a soldier dies. It is what a state pays to persuade someone to accept the risk in the first place, because that price is set in a market, and markets move.

Across the war in Ukraine, Russia has recruited heavily through contract payments: a lump sum on signing, on top of monthly pay, on top of regional authorities adding their own supplements to hit local quotas. Those figures have been raised repeatedly since 2022, and the regional top-ups have varied enormously depending on how badly a given area needed men. Which means something quite specific is observable here that is observable almost nowhere else. There is a running market rate for a man agreeing to risk his life, it is published, and it has gone up, because supply got tighter.

Nothing else in this essay does that. The 9/11 fund was a single settlement with a fixed formula. Everest enforces one rule and never revisits it. Here the number is live, and it responds to demand rather than to anything about the men it is offered to. The price of a life, on that market, has nothing to do with whose life it is.

Be careful with what that number actually is, though. A payment for accepting risk is not a payment for a death. It is closer to an insurance premium in reverse: the state buys a probability, not an outcome. Which is its own quiet admission, because the arrangement only functions if both parties agree not to look too closely at what that probability actually is.

Then there is conscription and mobilisation, where no price is offered at all. The state simply asserts a claim on a portion of someone's remaining life. Not a purchase, not a settlement, no negotiation, and no point anywhere in the process where the person's own valuation of their own life could enter. That is closer to Henrietta Lacks than it first appears. Value extracted through a procedure with no slot for consent, and therefore no slot for worth.

But the part that keeps returning to me is the third one, and it is the hardest to write about, because there is no number for it anywhere.

A life can be destroyed without ending.

A man comes home with both legs gone and forty years still to live. A woman comes home intact and cannot sit in a room with a door behind her. Someone comes home and is recognisably not the person who left, and the family has to work out what to call that, because there is no word for grieving somebody who is sitting at the table.

Every compensation system in the world is built around a binary. Alive, or not. Systems handle death reasonably well, because a death is a discrete event with a date on it. What they handle badly is a life that continues in a form the person would never have chosen, because pricing that requires you to state what fraction of a person remains, and no institution will write that sentence down. So instead there are tariffs. A percentage for a limb. A percentage for an eye. Schedules that quietly imply a whole person is the sum of their parts, because the alternative is admitting there is no formula at all.

And there is a fourth group, who are not dead, not injured, and not counted anywhere.

Ukraine maintains a register of missing persons. It holds tens of thousands of names. Some are prisoners of war, some have died and not been identified, some are civilians deported or disappeared, and for a great many of them nobody knows which. Their families exist in a state that has no administrative category. They cannot bury anyone. They frequently cannot claim anything, because a claim requires a death certificate and a death certificate requires a death nobody can confirm. They wait, sometimes for years, holding a life open on the possibility of a return.

That is not a valuation at all. It is the total suspension of one. And it leads directly to the last thing this essay has to say, because a woman in Kharkiv with no answer and a woman at a Commonwealth headstone in 1921 are doing the identical thing, a century apart, for the identical reason.

The Ghost Soldier

There is one more valuation, and it runs in the opposite direction to everything above.

Somewhere there is a woman standing at a grave that contains a man she has never met. Her son went missing and was never recovered, and she comes here instead. She knows he is not underneath. She comes anyway, and she brings him what she would have brought him.

Across the Commonwealth war cemeteries, headstones for unidentified dead carry an inscription chosen by Rudyard Kipling, who was working for the war graves commission and whose own son had gone missing at Loos: a soldier of the Great War, known unto God. More than two hundred thousand of them. Each one a full headstone, the same size and shape as every named one, in a system that deliberately refused to distinguish between a general and a private.

Look at what has happened there. An institution that could not establish a single fact about a man, not his name, not his nationality, not reliably even which army he fought in, granted him the identical marker it granted everyone whose entire biography it knew. Zero information in, full value out. That is the exact inverse of the compensation fund, which held complete files on every victim and used them to produce different numbers.

And Britain went further, burying one unidentified man in Westminster Abbey and treating him as the nation's, specifically so that any grieving family could stand there and believe it might be theirs. The anonymity was not an unfortunate limitation. It was the mechanism. He is valuable to everyone precisely because he could be anyone's.

That is the thing the formulas cannot reach, and it is not sentimentality. It is a different operation entirely. A family assigns total value to a person on the basis of no data at all. No earnings history, no projected output, no verification that the body in front of them is even the right body. One unnamed man becomes a brother to one family, a son to another, a cousin to a third, and none of them are wrong, because the valuation was never derived from facts about him in the first place.

It also, finally, resolves Green Boots. Two men who could not be identified. One was given a function by necessity, because the mountain does not permit sentiment. The other is given a name by love, because a mother does not require evidence. Same absence of identity. Opposite response.

Six Answers, None Compatible

So here is where we end up.

A life is worth its projected earnings, capped where the results become embarrassing. A life is worth less than the lives required to retrieve it, and possibly more once it becomes a landmark. A life is worth nothing individually at the scale where lives are counted in estimates. A life can generate seventy years of value while never entering the ledger at all. A life is worth whatever a recruitment market currently requires, or nothing at all if the state can simply take it, and there is no number of any kind for a life that continues in a form nobody would have chosen. And a life is worth everything, unconditionally, to people holding no information about it whatsoever.

Every one of those is a working system. Each is defensible inside its own frame. And no two of them can be reconciled, which means the question in the title does not have an answer, only a set of answers determined entirely by who is asking and what they need the number for.

I do not think that incoherence is a problem to be solved. I think it is the honest description of where we are, and I distrust any framework that resolves it, because resolving it would require deciding that one of those six is the real one.

What I would say is this. The formulas are all wrong, and they are all necessary, and they will go on being both. The fund had to pay somebody something. The mountain has to keep living climbers alive. Somebody has to decide which vehicle to buy before the next convoy leaves. Those numbers will keep being produced, and they will keep failing to contain the thing they claim to contain.

And underneath all of it, people go on assigning full value with no method at all. At unmarked graves. To bodies that might not be theirs. To a man in an alcove at 27,900 feet whose name we established thirty years late.

Nobody has ever needed a calculation to grieve properly. That may be the only part of this that was never broken.