On the fifth of August a federal helicopter lifted a hundred and twenty-two wild horses off a stretch of western Colorado. On the range next to it, a thousand more were being driven into pens that same week. Now look at the ground behind them. Every horse still out there has more grass, and less competition for it. And a herd with more grass produces more foals, and more of them live through the winter. The roundup meant to shrink the herd is the thing that helps it grow back. How does a programme end up built around the one tool that works against it?

Start with the grass, because everything here runs on it.

A stretch of open public range carries only so much grass. When a herd sits near that limit, competition is fierce: mares are in poorer condition, fewer foals are born, and fewer survive their first winter. The land does the limiting, quietly and constantly.

Now take a thousand horses off it. That is what the Bureau of Land Management set out to do the same week, on that range beside it — a stretch called East Douglas, in Rio Blanco County, Colorado. The bureau is the federal agency that runs this land, and its helicopter is the one you watched in the opening. Every horse still there now has more to eat. Better-fed mares carry more foals to term. More of those foals survive. The population climbs back faster than it would have on its own.

Biologists call this compensatory reproduction, and it is not something critics invented to attack the agency. In two thousand thirteen the National Academy of Sciences reviewed this exact programme. It found that the way the programme is run is itself helping the population grow: taking horses off the range can raise the growth rate.

So the tool is not neutral. Used alone, it empties the range and then helps refill it. That raises the obvious question: why is it still the main tool? That is one of three rules in this machine. The other two were written by Congress.

The answer starts with what the programme has already piled up.

Across ten western states the bureau now holds more than sixty-four thousand animals off the range, fed at public expense, most for the rest of their lives. That costs about a hundred million dollars a year, and in twenty twenty-four close to two-thirds of the programme's budget went to catching animals and keeping them.

Those hundred and twenty-two horses from the opening are part of it. Catching them cost a hundred and eighty-seven thousand dollars. Holding them until somebody takes them will cost four million over the next two years — for one roundup, in one county.

By the agency's own estimate, removing one horse and keeping it for life comes to about forty-eight thousand dollars — for one animal the range would have fed for nothing.

And the range keeps filling. This March the bureau counted about eighty-five thousand animals still out there — twelve thousand more than the year before. And the roundups never stopped that year.

There is an obvious answer here, and you have probably already thought of it. Don't catch them — stop them from being born. Congress has been telling the bureau to do exactly that since twenty twenty-two, and the money for it is there. Almost nothing happens. The vaccine has a price of its own, and we will come to it. First, the two rules that keep the helicopters flying whatever the vaccine does.

The first of those two written rules is a number. Start with the target, because the programme is not managing a herd — it is closing a gap to a number somebody wrote down. That number has a name — the appropriate management level — and it sets how many animals each stretch of range is allowed to hold. Add up every stretch and you get about twenty-five thousand nationwide. Against eighty-five thousand counted, that leaves a "surplus" of roughly sixty thousand horses, and closing that gap is what the whole machine exists to do. The bureau set most of those limits in the eighties and nineties, and has barely moved them since. The National Academy of Sciences examined how those limits were set and found they do not rest on consistent data about what the land can actually carry. So the arithmetic driving everything rests on a thirty-year-old guess.

There is a second rule, and it explains why every horse taken off the range stays off it.

Almost everyone believes the nineteen seventy-one law banned killing wild horses. It did not. Signed by a Republican president and defended by both parties ever since, the act protects them and bans selling them off to be turned into commercial products. But it also gives the Interior Secretary — the cabinet official in charge of federal land — power to order the removal, or the destruction, of excess animals.

The bureau last used that authority in nineteen eighty-two. Six years later Congress made sure of it: from nineteen eighty-eight, and in most years since, it has attached a line to the spending bill forbidding the bureau to spend money on putting healthy horses down. It did not repeal the power. It refused to pay for using it.

That closed one exit. In two thousand four Congress opened another, then did the same thing to it. A senator from Montana placed a paragraph inside a three-thousand-page spending package — never introduced on its own, never voted on separately. It ordered the bureau to sell off the older horses, and the ones nobody had taken, with no conditions on what the buyer did with them.

The House voted three times to undo it, once without a single objection. The Senate never took it up. So the paragraph is still law, switched off each year by the same budget line that switched off the other power — under Republican and Democratic Congresses alike. Those sales never happened. There were fourteen thousand animals in holding when that paragraph passed; today there are sixty-four thousand.

And this is where the argument sticks rather than settles. Wild-horse advocates call the roundups cruel: a captured horse loses everything that made it wild. A county commissioner in Rio Blanco, who used to run a ranch herself, says she has watched horses starve on stripped ground, and a pen is kinder. Both are describing something real, and neither position has moved in fifty years.

What that leaves is a permanent authority nobody will fund, and a temporary ban nobody will make permanent. Neither camp wants that vote on the record, because neither is sure of winning it.

When a system cannot use its exit, it improvises one. From twenty nineteen the bureau paid a thousand dollars to anyone who adopted a horse. An investigation by wild-horse advocates, confirmed by The New York Times, followed what happened next. Families adopted the maximum of four horses each and took the money. As soon as the payment cleared, the horses turned up at livestock auctions where slaughter buyers shop. At least eight hundred and forty went that way, and a federal court struck the programme down in March of twenty twenty-five. So look at the three rules together — two of them written by Congress, and one written by the grass. The first fixes how many horses have to come off the range. The second leaves capture as the only method Congress will pay for. And the third means that taking them off is what makes the number climb back, which puts you back at the first rule with a bigger gap to close than before. That is the loop.

Three designs could break that loop, and every one has a price.

The first is the obvious one from earlier — treat the birth rate instead of the herd size. There is a vaccine that stops a mare from conceiving for a year or two, and it wears off on its own. It is fired from a dart gun. A foal not born is forty-eight thousand dollars not spent.

The trade-off is distance. The same mares have to be found and treated again, and again. On open range with no roads, getting a dart to one specific mare can cost more than catching her.

And there is a scale problem underneath. To turn the population around you have to reach most of the breeding mares, every year or two, across ten states. Done herd by herd, that is not a smaller operation than the roundups — it is a permanent one. The helicopter flies for a week. This would fly forever. And it does nothing about the horses that already exist: it slows next year's foals, while the sixty-four thousand in pens still eat every day.

The second design leaves the tools alone and goes after the target itself. Survey what the range carries now, and set the number from the land instead of from the nineteen eighties. Its trade-off is that a recount cuts both ways, and nobody knows in advance which way. Some ranges would support more horses than the old figure allows; others would support fewer than are there today, which would mean bigger removals, not smaller. And the deeper cost is that it is not really a survey. The same grass is shared with private cattle, under a separate rulebook with its own quotas — and the ranchers who hold those quotas. A recount done with current data immediately becomes an argument about whose share shrinks — which is why the cheapest technical fix in this story has sat undone for thirty years.

The third design needs no new law. The nineteen seventy-one act already permits destroying excess animals, and Australia does exactly this with the wild horses in its Kosciuszko National Park — culling from helicopters, quickly and cheaply.

The price here is not money. The law permits it — but the public anger that produced that law was about exactly this picture: wild horses shot and hauled off western ranges. The animal at the centre of it is a national symbol. In Australia every cull brings lawsuits and public fury. In America the price has already been measured: in most years since nineteen eighty-eight, under Congresses of both parties, the money has been withheld rather than the authority allowed.

So there are three designs on the table: treat the births instead of the numbers, recount the target, or switch on the exit that already exists. None of them is free.

Nothing here required anybody to be a villain. Congress protected an animal the country loves, then declined, year after year, to decide what to do with the animals that protection produced. The bureau chases a target it did not set, using the one tool Congress has never refused to pay for. And the pressure that made removal the only method anyone will pay for comes from the same people pressing for the contraception the agency is slow to deploy. Meanwhile the machine keeps running: about eighty-five thousand horses on the range and climbing, sixty-four thousand in pens and climbing, a hundred million dollars a year, and a helicopter over Colorado in August. So when this comes round again, and it comes round every budget cycle, two questions are worth having ready. Is the target in that budget a measurement, or an inheritance? And if two-thirds of the money goes to a tool the National Academy of Sciences says can push the growth rate up, why is it still the main one?