Blog

The form is not complicated. In most places on earth, two people who want to be married walk into a civil registry office, fill in their details, sign where indicated, and walk out again. It is paperwork. It is boring. Somebody behind a desk stamps it and calls the next number.

Fabiana Banzer and Scarlett Rocha filed exactly that form, in the Bolivian city of Santa Cruz, and were refused. Then they spent three years and ten months making somebody accept it.

They finally succeeded last month, and when Banzer wrote about it publicly, the phrasing she chose was striking for how plain it was. After a legal fight of three years and ten months, she said, they had finally been able to access civil marriage as two women. Not won. Not broke through. Accessed. The word you would use about a service everybody else already has.

That is the whole shape of this story, and it is worth getting right from the beginning, because it has been widely described in ways that are not quite accurate.

Nothing was legalised here. Bolivia’s constitution still says what it said before. Article 63 defines marriage as a union between a woman and a man, and no legislature changed a word of it. What happened is narrower and, in some ways, stranger than a law changing. Two specific people went to court, twice, and obtained rulings that applied to them. The civil registry then recorded their marriage. It is, as far as the public record shows, the first same-sex civil marriage on the books in the country. It did not make the next couple’s marriage automatic, and everyone involved has been careful to say so.

The couple are both 29. They met through mutual friends in Santa Cruz around four years ago. Banzer works as a psychologist. Rocha is a neuroscientist, born in the United States and raised between both countries, which turns out to matter to why they wanted marriage specifically rather than the alternative on offer.

Because there was an alternative, and this is the distinction that most of the coverage flattened.

Bolivia has recognised what it calls free unions between same-sex partners since 2020. That precedent was set by David Aruquipa and Guido Montaño, who fought their own long battle and became the first same-sex couple in the country to have a civil union registered. It was a genuine landmark and it opened a genuine door. Since then, according to figures reported from the civil registry, around 148 same-sex civil unions have been recorded.

It is worth pausing on what the 2020 case cost, too, because the pattern repeats. Aruquipa and Montaño had been together for years before they began trying to register anything, and their own process ran long and public and unpleasant before the registry gave way. They were not asking for marriage. They were asking for the lesser category, and even that had to be litigated. When people say the door opened in 2020, what they mean is that two people held it open with their shoulders for a very long time and then a small number of other couples were able to walk through.

So Banzer and Rocha were not shut out of every form of recognition. They could have had a civil union. They chose to spend nearly four years pursuing something else instead, and understanding why is most of the point.

A civil union and a marriage are not the same instrument. The union carries fewer protections. It sits slightly to one side of the ordinary legal furniture that everybody else’s relationship sits inside, which means that in any situation where somebody has to interpret your status, you are relying on their goodwill rather than on the obvious. For a couple who might one day move countries, that gap is not abstract. Marriage travels in a way that a locally defined union may not, and Rocha in particular wanted a status that would still mean something if they relocated. Representatives of the human rights organisations involved put it more simply, describing marriage as carrying a higher degree of social and legal recognition.

The portability question deserves more than a passing mention, because it is the least romantic and most practical reason couples in this position fight for marriage rather than settle for a union. A marriage certificate is a document that other countries’ systems already know how to read. A locally defined civil union frequently is not. It can mean that a status recognised at home becomes a shrug at a border, or in a hospital, or on an immigration form, or in front of an official deciding whether you count as next of kin. For a couple where one partner holds ties to another country, that is not a theoretical inconvenience. It is the difference between having your relationship recognised automatically and having to explain it, with paperwork, to a stranger, at the worst possible moment.

There is also the part that has nothing to do with law at all. A civil union is a category that exists because somebody decided you could not have the ordinary one. Choosing the ordinary one is a statement about wanting to be inside the same sentence as everybody else, not adjacent to it.

The legal route they took is the genuinely interesting part, and it was not a Bolivian law.

Their lawyers, working with the NGO IGUAL Bolivia, built the case on Advisory Opinion 24/17 of the Inter American Court of Human Rights, issued back in 2017. That opinion holds that states in the region should recognise same-sex families on the same terms as any other family. Bolivia is bound by the regional human rights system, and international human rights obligations can be treated as taking precedence over conflicting domestic provisions. So the argument was not that Bolivian law permitted this. The argument was that a document written years earlier, for an entire region, already required it, and that the registry had no standing to refuse.

They won by pointing at something that had been sitting there the whole time.

That advisory opinion is worth understanding, because it explains why a case like this can succeed in a country whose own constitution points the other way. It was issued at the request of a member state, and it set out the court’s reading of the obligations the regional human rights convention already imposed. Its conclusion was that the protections owed to a family cannot be withheld on the basis of the sex of the people in it, and that states should extend the existing institution rather than invent a lesser parallel one for some couples.

It is an opinion rather than a statute, and it does not repeal anything by itself. What it does is give a lawyer in any country bound by that convention something concrete to hold up in front of a judge. Courts in the region have reached for it repeatedly since, and in more than one country the practical route to recognition ran through a courtroom citing that text rather than through a parliament voting on a bill. Banzer and Rocha’s case belongs to that pattern. The instrument was regional, the argument was old, and the only new thing was two people willing to spend four years pressing it.

That did not make it quick. They filed their application in late 2022 and were turned down. What followed was two separate constitutional protection actions, the legal mechanism known locally as an amparo, and years of administrative grinding in between. Banzer has described the period in terms that are noticeably free of triumph. There were so many obstacles, she said, and so much discrimination.

Rocha’s description of it is quieter and somehow worse. They had many doubts and a great deal of uncertainty, she said, because the process simply went on so long.

That is the cost that does not show up in the headline. Nearly four years is not four years of dramatic courtroom scenes. It is four years of waiting for a response, of being told to resubmit, of a hearing being scheduled and then moved, of having your relationship treated as a legal question that other people are still thinking about. It is four years in which you cannot plan properly, because the thing you are planning around might not happen.

And even winning did not mean it was over. Earlier this year, a constitutional tribunal ruled in their favour and ordered the civil registry to issue the licence within ten days. It then took roughly five more months before the registration actually happened. A court had said yes, a deadline had been set, and the paperwork still sat there through most of a year.

When it finally went through, on a Friday last month, they married in white dresses with more than a hundred friends and family in the room. Banzer’s description of the day is the least legal sentence in this entire story. They had not expected it to be so beautiful, she said. It was really, really great. They danced until late.

After three years and ten months of arguing about whether a form could be filed, the thing itself turned out to be an ordinary wedding. Which was, of course, the entire request.

The question now is what this means for anyone else, and the honest answer is that nobody is certain yet.

Legally, the rulings applied to Banzer and Rocha. The registry recorded their marriage because it was ordered to record theirs. Another couple walking into another office on Monday morning is not guaranteed anything. What they have is a precedent, which in practice means an argument that has now worked once and a set of filings that can be reused.

There are early signs it is being used. A lawyer involved in the case has reported that in the weeks after the wedding, somewhere between five and ten couples had already approached them about doing the same thing.

The useful comparison is the civil union. After Aruquipa and Montaño won their case in 2020, it took roughly three years before registration became broadly available to other couples rather than a one-off. Then the numbers started accumulating, and now there are nearly 150. If marriage follows the same curve, the practical effect of what happened last month will not be visible for a while, and it will arrive as an administrative shift rather than an announcement.

Aruquipa himself, who has more standing than most to comment, framed it as continuity rather than a beginning. He suggested it should be understood as a historical throughline, from the first recognised union to the first civil marriage, with courts progressively widening what is available rather than any single moment changing everything.

That framing is worth borrowing, partly because it is accurate and partly because it is a corrective. Stories like this get reported as a country flipping a switch. Bolivia did not flip a switch. Its constitution is unchanged, its legislature has not acted, and it remains one of a handful of South American countries without legislative recognition of same-sex marriage. What actually happened is that two people used an existing regional human rights instrument to compel one office to process one form, and that outcome is now written down somewhere other people’s lawyers can find it.

That is slower and less satisfying than a law passing. It is also, on the evidence, how a great deal of this has actually moved across the region. Somebody applies. Somebody refuses. Somebody spends years on it. The next couple’s version is shorter.

There is a temptation to make this sound larger than it is, and it is worth resisting, because the restraint is what makes it land. Banzer and Rocha did not set out to change a country. They were two people in their twenties who met through friends, stayed together, and wanted the same legal arrangement that millions of people obtain without thinking about it. The extraordinary part was not the ambition. It was the amount of time and argument and public exposure required to get something entirely unremarkable.

They are married now. They have the certificate, which after everything is just a piece of paper with their names on it, filed in a registry in Santa Cruz alongside everybody else’s.

Somewhere in that same system, five or ten other couples are currently starting their own applications, armed with a set of arguments that did not exist four years ago. Some of them will be refused too, at first. But they will be refused with a precedent in hand, which is not nothing, and their wait will probably be shorter.

That is how it tends to go. The first one takes nearly four years. The rest take a little less each time, until eventually somebody walks into an office, signs a form, and walks out again without it being a story at all.

Jay Jones was laid off near the end of 2023, and his daughter was about to be born. That is the part of the story most people skip past, but it is the part that explains everything that came after. He was not a researcher with a grant or a journalist with an assignment. He was a man with a newborn on the way, a severance clock ticking, and a laptop open to a job board, doing exactly what tens of millions of people do every single day.

Then the recruiters started messaging him. And something about them was off.

Not dramatically off. Not the obvious, badly spelled, too-good-to-be-true kind of off that anyone would catch. These were polished. They had headshots and job titles and company names you recognised. But Jones, who had spent his career around verification and fraud work, kept noticing small things that did not sit right. A recruiter whose entire digital existence was one LinkedIn profile and nothing else. A corporate hiring manager emailing from a free consumer address. A company that claimed a decade of operation but whose web domain had been registered a few weeks earlier.

So he did what almost nobody does. Instead of blocking the account and moving on, he opened a spreadsheet.

What happened over the next two and a half years is genuinely hard to hold in your head. By his own public accounting, Jones has now documented and had removed tens of thousands of job listings he says were never real. The number he has attached to his campaign is 57,952. His own website puts the running total higher, past 58,500. A petition he filed cited 55,308 as of the middle of this year. Those figures do not contradict each other so much as they show you what kind of number this is. It is not a finding. It is an odometer. Every time he goes back to work, it climbs.

Alongside the listings, he says he has flagged more than 7,000 fake recruiter profiles, and disrupted over a thousand hijacked accounts belonging to real people whose identities had been quietly taken over and repointed at job seekers.

The thousand-plus hijacked accounts deserve their own moment, because they are the strangest part of all this. These were not invented people. They were real professionals, with real careers and real connections, whose accounts had been taken over and quietly redirected. The photograph stays. The work history stays. The decade of genuine endorsements from genuine former colleagues stays. Only the purpose changes. Someone who has spent fifteen years building a professional reputation becomes, without knowing it, the friendly face on an operation aimed at people in trouble. Most of them found out only because a stranger with a spreadsheet noticed and filed a report.

All of this was, for a long time, one man’s private obsession. It stopped being private this summer, when the state of Texas got involved.

The Attorney General’s office there opened a formal investigation into LinkedIn over what it describes as fake and misleading job opportunities advertised to people who were paying for the privilege. The legal instrument is a Civil Investigative Demand, which is a compulsory request for documents, and the office has asked the company to hand over internal material about its advertising, its marketing, its verification practices, and specifically what it tells subscribers about job listings.

The money angle is what gives the investigation its teeth. A Premium Career subscription runs about $40 a month, and a Premium Business one around $70. The state’s position is that people are paying those fees on the strength of a promise about the quality of the job market they are buying access to, and that a meaningful share of what they find there may be inactive, already filled, or not attached to any real hiring intention at all. Attorney General Ken Paxton framed it simply, saying the company has a duty to actually deliver the service it advertises to people who are paying for it.

It is important to be precise about what this is, because it is easy to read the headline and conclude that something has been proven. Nothing has. This is an investigation, not a finding. No charges have been filed and no lawsuit has been brought. A Civil Investigative Demand is the state asking questions, and the company is entitled to answer them.

LinkedIn, for its part, has not conceded any of it. The company says its goal is to help job seekers find their next role and that its policies require posted jobs to be authentic and accurately represented. On the fraud side specifically, it says its automated systems catch and remove the overwhelming majority of spam and scam content before any user ever sees it, and that it proactively blocks the large majority of fake accounts it detects. Those are real numbers and they are almost certainly true in percentage terms. The uncomfortable arithmetic is that a platform with billions of pieces of content can block 99 percent of something and still leave an enormous absolute quantity standing.

The state’s release also points to something broader than any one company. It notes that independent studies have estimated ghost listings may account for somewhere between a fifth and a third of online job adverts generally. That is an estimate, and estimates in this area vary widely depending on who is counting and what they count. But it matches the lived experience of nearly everyone who has looked for work recently, which is part of why this story travelled the way it did.

It is also worth separating two things that get tangled together, because the investigation sits exactly on the seam between them. Outright fraud, the invented recruiter with the stolen photograph, is one problem. The ghost job is a different one, and in most places it is entirely legal. A real company with a real page posts a real-looking vacancy it has no current intention of filling. Sometimes that is a firm keeping a pipeline warm for a role it expects to open next quarter. Sometimes it is a policy requiring a position to be advertised externally when an internal candidate has already been chosen. Sometimes a listing is simply left up long after the seat was filled, because nobody’s job description includes taking it down. None of that is a crime, and none of it involves a scammer. It still means a person spent an evening writing a cover letter into a void.

For the applicant refreshing their inbox, those two problems are indistinguishable. The silence feels identical. That overlap is part of why the numbers in this area are so contested, and why any estimate of how many listings are not real depends enormously on what the person counting decided to count.

What makes Jones’s work land, though, is not the totals. It is the texture.

There was a recruiter he found operating under the name Linda Anthony. Same name, same photograph, multiple accounts, each one claiming a different title in a different place. One version was a contract recruiter at a firm in one state. Another was a contract specialist at the same firm, in a different state entirely. Same face, looking out of every profile. Jones described the experience of pulling on that particular thread as one rabbit hole opening into another.

More troubling than the invented people were the borrowed companies. Jones documented fake vacancies posted against the genuine pages of large, well-known employers, including a major carmaker and a large financial services firm. When a listing sits on a real company’s page, it arrives wrapped in that company’s logo, its follower count, its verification marks, its entire accumulated credibility. A job seeker doing everything right, checking the employer, looking for the badge, can still walk straight into it. And in some cases the companies themselves have said plainly that they do not hire for the roles being advertised in their name.

He also mapped out what he alleges are networks of shell company pages, clusters of fabricated employers posting hundreds of listings between them, all funnelling applications away from the platform and toward systems nobody can see. His allegations about what those systems are for are serious. He says they harvest personal information, in some cases push malware, and in some cases extract money outright.

Here is where you have to be careful, and it is worth saying so directly. Those are Jones’s characterisations, drawn from his own documentation. They have not been established in court. He is one person, working from open sources, following patterns. His work has drawn coverage from major news organisations and he now speaks to law enforcement and corporate security audiences about it, which suggests professionals take him seriously. But his numbers are his numbers, and they should be read that way.

What is harder to dispute is who these operations aim at. The people most likely to answer a suspicious recruiter are the people who cannot afford to ignore any message at all. Someone recently laid off. Someone new to a country and unfamiliar with how hiring normally works. Someone in low-paid care work, applying to a dozen places a week. The tactics Jones catalogues include asking for sensitive personal documents before anyone has been hired, moving the conversation off the platform almost immediately, and requesting fees for equipment or training. Each of those only works on somebody who is already stretched thin enough to say yes.

Jones has turned a good deal of what he learned into practical advice, and it is unglamorous in a way that makes it more useful. Look at whether the recruiter exists anywhere else on the internet. A person who has genuinely been hiring for fifteen years leaves a trail. If the single LinkedIn profile is the whole of them, that is a signal. Look at the email domain. Someone recruiting for a major corporation should not be writing to you from a free consumer address. Look up when the company’s web domain was registered, because a business claiming twenty years of history with a domain three weeks old is telling on itself. Check whether the vacancy also appears on the employer’s own careers site, because the good ones almost always do.

His favourite technique is the simplest. Ask an unscripted question. Real scams run on templates, and a fraudulent recruiter working from a script has nothing prepared for a candidate who wants to know about the benefits package, or who is leaving the role, or how the team has grown in the last year. The conversation either stalls or goes strange, and you have your answer in about ninety seconds.

He is candid that reporting is where the frustration lives. He has described the process as constant friction, opening case after case on the same networks, watching pages come down and near-identical ones go up. That is a familiar shape to anyone who has worked in platform safety. It is not necessarily evidence of indifference. It is evidence that removal is cheap for the platform and even cheaper for the people building the fakes.

And that, in the end, is what the state investigation is actually probing at. Not whether fraud exists on the internet, which is not in dispute anywhere. The question is narrower and sharper. If a company sells a paid product on the promise of a better hiring marketplace, what does it owe the people who pay for it about the condition of that marketplace. Nobody has answered that yet. The investigation may answer it, or it may quietly conclude, or it may end in a settlement nobody reads. Those are all real possibilities.

What has already changed is smaller and more concrete. Tens of thousands of listings are down that would otherwise still be up. Thousands of invented recruiters no longer exist. Over a thousand people got their own professional identities back without ever knowing they had lost them. That happened because one person, at the worst possible moment in his own working life, with a baby coming and no income, decided that the pattern he was noticing was worth writing down.

He could have done the entirely reasonable thing. Block, report, move on, keep applying. Almost everyone does. The gap between that and what he actually did is not talent or resources. It is just the decision to keep a record.

Jones now works in fraud prevention professionally, which is the tidy ending, and it is a real one. But the better ending is less tidy. The spreadsheet is still open. The count is still moving. Every so often the total on his site ticks up again, and somewhere a person who was about to send their passport scan to a stranger does not.

If you have applied into silence recently and quietly concluded that the problem was you, this is worth sitting with. Sometimes it was you, and sometimes the market was just brutal. But some meaningful share of those listings were never going to answer anybody, because there was nobody on the other end to answer. That is not a comfort exactly. It is just true, and one man spent two and a half years proving it one line at a time until a state government picked up the file.

There is a ridge in northwestern Morocco with a line running across it. You can see it from the air and you can see it from the ground, a bank of earth and stone cutting off the southern approach to a spur of high land, the kind of feature that any walker would clock as man-made and then stop thinking about. It sits about a hundred kilometres inland from the Atlantic, in Khémisset province, above a village whose houses you can see in the background of the photographs the archaeologists have been taking.

An archaeologist named Armand Ruhlmann wrote it down in the 1930s. He even said what he thought it was, a prehistoric fortified spur, a settlement that had sealed itself off by walling across the neck of the ridge. And then, in his own words and everybody else’s afterwards, the dating of it was very uncertain.

So it sat there. For close to ninety years, the deep history of this part of the world was written without it. Not because anyone hid it or lost it or argued about it. Because nobody went back and asked the question. The wall was a known feature of a known place, sitting in the published record, and the simple follow-up of finding out when it was built did not get done until a few years ago.

When it finally did get done, it produced what the research team describes as the oldest defensive system known anywhere in North Africa outside the Nile Valley.

The site is called Oued Beht, and to understand why the wall matters you have to understand what the same team had already found underneath it. Because this is the second surprise at a place that had already delivered a big one.

Until recently, the standard account of North Africa in the fourth and third millennia BC was a blank with an apology attached. Egypt was a civilisation. The rest of the continent north of the Sahara was assumed to be thinly occupied, marginal, the edge of somebody else’s story, a place that received influences rather than generating them. Part of that assumption was genuinely about evidence, and part of it was about who had done the digging, which for most of the twentieth century meant a small number of European researchers working in a colonial frame with a limited set of questions.

Then the Oued Beht Archaeological Project, a joint effort between Morocco’s National Institute of Archaeological and Heritage Sciences, Italy’s National Research Council, the international association ISMEO and the University of Cambridge, started working the ridge systematically. The first big publication landed a couple of years ago, and it described a farming society dated to roughly 3400 to 2900 BC occupying nine or ten hectares of the ridge, which made it the largest agricultural settlement known from that period anywhere in Africa outside Egypt.

The detail in that report is what makes it feel like a real place rather than a data point. Surface survey across nearly twenty hectares recovered close to twenty thousand artefacts, sixteen thousand of them pottery. Some of that pottery was slipped and burnished, a small percentage painted dark on light. There were hundreds of grinding tools and dozens of polished stone axes, with evidence that they were being made on site rather than brought in. Sieving the fill of one pit produced domesticated barley, wheat and peas, alongside wild olives and pistachios. The animal bones were goat, sheep, cattle and pig, with butchery marks on them.

And there were the pits. Deep bell-shaped pits, cut into the ridge at something like one for every ten square metres of ground. That is an extraordinary density, and the most plausible reading of them is storage, most likely grain. Which means surplus. Which means a community producing more than it ate, holding it, and presumably having opinions about who got to open which pit.

The other thing that report argued was that this community was not isolated. The painted pottery and several of the tool types have real parallels in southern Iberia at the same period, across the strait. Not identical, not a copy, but the kind of resemblance that comes from two regions that knew about each other. The team’s framing was that this was a two-way relationship, not a periphery absorbing culture from a centre.

That was the first surprise. The second is the wall, and it arrives from a completely different era.

In the autumn of 2023, in the project’s third season, the team put trenches into Ruhlmann’s old feature and into the ground around it, and sent a drone up to photograph the whole spur from above. The excavation was not quick, the analysis took the best part of two years, and the results were published a few weeks ago in the journal Libyan Studies, open access, with more than twenty named authors and the project’s three co-directors, Giulio Lucarini, Cyprian Broodbank and Youssef Bokbot, among them.

What the trenches showed is that the line across the ridge is not one thing. It is several things stacked on top of each other, built at wildly different times, which is exactly why it had resisted an easy answer for ninety years and exactly why nobody could date it by looking at it.

The stonework you can see on the surface is the youngest part. That wall, the visible one, was built in the Islamic middle ages, somewhere around the eleventh to thirteenth centuries AD, along with a cemetery. People came back to this ridge two thousand years after the prehistoric settlement had gone, looked at the same natural bottleneck, and reached the same conclusion about where to build a barrier.

Underneath it is the part that rewrites the map. A broad ditch and an earthen embankment, with the remains of an earlier wall phase built of daub and stone, running across the southern approach to the spur. Radiocarbon dating on seeds and charcoal puts that construction in the early to middle second millennium BC. Roughly three and a half to four thousand years ago. The Bronze Age.

That is a period nobody had documented at Oued Beht at all. The site was known for its Final Neolithic farming phase and nothing else. The 2023 season produced around two dozen new radiocarbon dates, spanning four distinct periods of use on the ridge, and seventeen of those are the ones that carry the fortification argument. The team’s position is that the second millennium BC activity is now established beyond reasonable doubt at a place where it was not previously known to exist.

The trench that did most of the work went into the feature the team labels Wall B1, and what came out of it reads less like a wall and more like a biography. The excavators describe a complex accumulation of walling, embankment and cut features rather than a single structure, with an earlier phase built of daub and limestone sealed beneath a later stone build. The earlier phase is the one the Bronze Age dates attach to, landing in a bracket somewhere around the nineteenth to seventeenth centuries BC depending on which samples you weight. The later stone build sits on top of it and is not prehistoric at all.

There is a term for the shape of the thing, éperon barré, which translates roughly as barred spur and describes a settlement that uses a natural promontory for three of its sides and builds across the fourth. It is a form that turns up all over prehistoric Europe, because it is the obvious solution when the ground offers you most of a defence for free. Ruhlmann used exactly that term for Oued Beht in the 1930s. He had the right label and no way to test it, which is a fair summary of what a great deal of early twentieth century archaeology could and could not do.

The drone work matters more than it sounds. An oblique photograph from the air shows the barrier as a single continuous line across the neck of the spur in a way no ground-level view can, and a vertical one lets the team map the trenches, the earlier excavations and the run of the wall onto the same frame. The published figures show both, with the village of Aït Siberne behind the ridge in the oblique shot. Ninety years of uncertainty and part of the resolution came from being able to get twenty metres off the ground cheaply.

Which is why they are willing to make the claim they make: that a probable ditch, embankment and crowning wall of that date is the oldest documented defensive system in North Africa outside the Nile Valley.

It is worth being careful with that claim, and the team is careful with it too. It is a statement about the documented record, in a region that has been surveyed far less thoroughly than Europe or the Near East. Something older may well exist and simply has not been dug, or has been dug and never dated, which is precisely the failure this whole story is about. Read it as the strongest thing currently on the table rather than as a final ranking.

What makes it more than a trophy is the timing. In the second millennium BC, across the strait in southeastern Iberia, the society archaeologists call El Argar was doing something very specific. Fortified hilltop settlements, metalworking, a social order with visible ranks and unequal burials, running from roughly 2200 to 1550 BC. It is one of the best known Bronze Age cultures in western Europe.

And at the same time, on a ridge in Morocco, people were digging a ditch and throwing up an embankment to close off the only easy way into their settlement.

The team is careful not to turn that into a direct cause. Nobody is claiming Iberians came over and taught anyone to build a rampart. The honest version is more interesting anyway: two societies on either side of a narrow sea, in the same centuries, both deciding that their settlements needed defending. That suggests a shared set of pressures, or at least a shared set of anxieties, in a western Mediterranean that was clearly more connected and more crowded than the old story allowed.

Nobody knows what those pressures were. That is the open question the paper leaves standing, and it is open mostly because the archaeology of the Maghreb in the second millennium BC is so thin. You cannot say who the wall was built against when you have almost no other sites of the same date to compare it with. Raiding, competition over stored grain, pressure over grazing land, a shift in climate, an internal dispute about who controlled the pits. All plausible, none demonstrated.

What you can say is that somebody decided it was worth the labour. A ditch and bank across a ridge is a serious piece of communal work. It takes organisation, a reason, and enough people who agree on the reason to keep digging. Communities do not build that on a whim, and they do not build it in a landscape where nothing ever happens.

The sequence, once you lay it out, is the thing that stays with you. A large farming settlement on the ridge around 3400 to 2900 BC, storing grain in hundreds of pits, trading ideas with people across the water. Then a gap in the record. Then, a millennium or so later, people fortifying the same spur in the Bronze Age. Then another long silence. Then medieval communities coming back, building a stone wall on top of the old embankment and burying their dead nearby. Then a French archaeologist in the 1930s writing the feature down, correctly identifying what it probably was, and having no way to date it. Then nothing for ninety years.

Lucarini’s own summary of it is that the ridge kept its secret for nearly a century and that the discovery forces everyone to look at this landscape with fresh eyes, which is a mild way of saying that a lot of published assumptions about Bronze Age North Africa were built on an absence of evidence that turned out to be an absence of looking.

That is the part of this worth carrying away, and it is not really about Morocco. Archaeology has a bias toward the new: new site, new trench, new season, new announcement. The thing sitting in a 1930s report with an unresolved question attached to it does not generate funding or headlines, and the question quietly stops being asked. It becomes furniture. Ruhlmann was not wrong about this wall. He said what it probably was, said he could not prove when, and moved on, exactly as a careful researcher should. The failure came afterwards, in the decades where everyone read his report and nobody followed it up.

Oued Beht is now one of the more closely studied prehistoric sites in northwestern Africa, and the project is continuing. There will be more seasons, more dates, and probably more surprises, because the ridge has produced two already and there is no obvious reason to expect it has run out. The medieval phase alone is barely explored. The gap between the Neolithic settlement and the Bronze Age fortification is still a gap.

But the wall has an age now. It took drones, radiocarbon dating, a team from four countries and two field seasons of preparation to get it, and all of that was in service of a question anyone could have asked in 1936 and nobody did.

The crew at Sardis was doing the least glamorous work an archaeologist ever does. They were clearing a Roman colonnaded avenue in western Turkey, lifting the slabs that had made up the road surface, recording them, moving them aside. Paving is paving. You log it and you keep going.

Then one of the slabs turned over, and there was a face on the underside of it.

Not a carving in relief, not a fragment. A colossal marble figure of a young man, over seven feet tall when he was whole, broken into two big pieces, lying face down in the roadbed where somebody had dropped him roughly fifteen hundred years ago and paved right over him. His feet are still missing. A few smaller fragments have not turned up. But the face, the hair, the folds of the clothing and the veins on the back of one hand came out of the ground looking almost the way the sculptor left them, because the Romans who used him as building rubble had unknowingly done the single best thing anyone could have done for him. They put his face against the dirt and sealed it there.

The team that found him works for the Archaeological Exploration of Sardis, a joint Harvard and Cornell project that has been digging the same enormous ruined city for more than sixty years. The field director is Nicholas Cahill, professor of Greek and Roman art and archaeology at the University of Wisconsin-Madison, and his public reaction to the find was not the usual polished institutional calm. He said the team was astounded by it, and also a bit perplexed. That second word is the interesting one, and it is the reason this statue is still generating argument months after it came out of the ground.

Here is the sequence, because it matters and a lot of the coverage blurred it. The statue was lifted from the pavement during the dig season, in late May. It sat under study and protection while the team worked out how to move something that heavy without losing more of it. Conservators raised the pieces properly in mid-July and took them to a conservation laboratory. The discovery was announced publicly in early August, which is when most of the world heard about it. So this is not a statue that came out of the ground last week. It came out of the ground at the start of the season, and the announcement came after the specialists had enough time with it to be confident about what they were looking at, which is how this is supposed to work.

What they are looking at, as far as anyone can currently say, is a young man carved somewhere in the range of 470 to 450 BC. That puts him in the early Classical period, right at the hinge where Greek sculpture stopped making stiff, symmetrical, smiling figures and started making bodies that look like they contain a skeleton. And the dating is not a guess from style alone. It comes from the anatomy. The hands are rendered with visible veins. The facial structure has the kind of observed detail that belongs to the fifth century and not before it. Whoever made this knew exactly what the best sculptors of his era were doing.

That distinction is worth a moment, because it is the whole basis of the date. For roughly two centuries, Greek and Anatolian sculptors made large standing figures to a formula. Feet together or one slightly forward, arms locked at the sides, a fixed pleasant expression, hair in neat beaded rows, the body treated as a pattern rather than an organism. Then, within about a generation, that formula broke. Sculptors began putting weight on one leg, letting hips and shoulders answer each other, carving muscle that reads as muscle and skin that reads as skin over bone. The change is dramatic enough that specialists can usually place a piece within a few decades just by looking at how the body is understood. On this statue, the understanding is late. The hands in particular give it away, because nobody carving in the archaic manner bothered with the veins on the back of a hand.

Which is what makes the rest of him so strange. Because from the neck up and from the shoulders down, he is dressed and posed like a man from a hundred years earlier.

His hair is long and elaborately arranged in a way that had gone out of fashion generations before he was carved. He stands bolt upright, weight even, in the old archaic manner rather than the relaxed, shifted stance that had just become the thing to do. And his clothing is the part nobody can place at all. He wears a thick himation, a heavy draped mantle, over a thin chiton, which is ordinary enough. Underneath both of those is a third garment marked with distinctive horizontal stripes or folds, and that garment has no known parallel in any comparable sculpture from Greece or from anywhere in Anatolia in that period. It is not a style that survives anywhere else in stone. Cahill’s reading is that the combination is deliberate: a sculptor working at the technical frontier of his own moment, who chose to dress his subject in the local traditional clothing of a culture that was, by then, politically finished.

To understand why that choice is loaded, you have to know whose city this was.

Sardis was the capital of Lydia, and Lydia was rich in the way that becomes a figure of speech. Its last king was Croesus, whose name still gets used as shorthand for money in languages spoken thousands of miles from where he lived. The Lydians are generally credited with the early use of coined money, and the gold that funded them came out of the river running through their own valley. The expedition has excavated the remains of a gold refinery in the city. For a couple of centuries Sardis was one of the wealthiest places on earth.

That ended in 546 BC, when Cyrus the Great took the city and folded Lydia into the Persian Empire. Sardis did not shrink. It became the capital of the empire’s western province, an administrative anchor for Persian power in Anatolia, sitting at the end of the royal road that ran clear across to the Persian heartland. It stayed important. It simply stopped being Lydian in the way it had been.

So the young man was carved by someone living in a Persian provincial capital, decades into Persian rule, using cutting-edge Greek technique, and wearing Lydian clothes that were already old-fashioned when his grandparents were born. And there is, by the team’s account, not one visible Persian element in him. Not in the dress, not in the pose, not in the ornament.

You can build several stories out of that, and this is exactly where a responsible account has to slow down. It is tempting to read the statue as a quiet act of cultural defiance, a conquered city insisting on its own past in stone. It is equally possible that the old dress was simply what you wore in this kind of sculpture, the way a formal portrait today might put someone in clothing nobody wears to the office. Archaic styling persisted in religious images across the ancient Mediterranean long after it had vanished from everyday life, precisely because sacred things are supposed to look old. The honest position is that the statue records a choice, and the motive behind the choice is not recoverable from the marble.

The same caution applies to the biggest question of all, which is who he is.

In his right hand he holds what researchers believe is a quince, a hard golden fruit that grew in the region and carried weight in local ritual. A figure holding a fruit like that could be a worshipper bringing an offering to a deity. He could also be the deity receiving it. He might be a specific young man whose family paid for a monument at a sanctuary. He might be a generic type. The team has been careful not to pick one, and that restraint is the most trustworthy thing in the whole announcement. Monumental sculpture of this period is rare anywhere in the Mediterranean, as Cahill has pointed out, and a piece this large and this unusual does not come with a label. Anyone telling you confidently that this is a god, or a prince, or a named individual, is telling you something the evidence does not currently support.

What is not in doubt is how unlikely his survival was.

Sardis kept going for a thousand years after he was carved. It was Hellenistic, then Roman, then late antique and Byzantine. Earthquakes flattened parts of it more than once. Every time a city like that rebuilds, the older city becomes a quarry. Marble is expensive to cut and transport and there is a broken statue lying right there. So at some point in the Roman period, builders laying out a grand colonnaded avenue at the entrance to the city took a five-century-old sculpture that had already been snapped in two, laid the pieces face down, and used them as road surface.

It is a small humiliation and it is the only reason we have him. Marble left standing in the open gets weathered into blankness, or burned for lime, which was the usual fate of ancient sculpture in later centuries. Marble sealed face down under a road gets none of that. No rain, no wind, no sun, no lime kiln. The carved surface was pressed against packed earth and essentially left alone. When the slab came up, the details that had been protected were the details that mattered most.

It is also worth picturing what the avenue itself was. A Roman colonnaded street is not a lane. It is a wide processional axis running through the centre of a city, lined on both sides with columns holding up covered walkways, with shops behind the colonnades and a paved carriageway down the middle. Cities across the eastern Roman world built them as civic statements, and they needed enormous quantities of cut stone to do it. A single avenue could swallow the dismantled remains of several earlier buildings and everything that had stood in front of them. The statue was not singled out for insult. He was simply one more piece of usable marble in a construction project that needed a great deal of it, and the crew laying the road almost certainly gave the carved side no thought at all beyond noticing that it was uneven and should therefore go downwards.

Getting him there in the first place had been a serious undertaking too. A figure over seven feet tall in solid marble weighs a great deal, and in the fifth century BC it had to be quarried, roughed out, hauled by ox cart or by water, finished by hand with iron tools and abrasives, and then set on a base without dropping it. That is months of skilled labour and real money. Whoever commissioned him was not making a casual gesture, which is another reason the question of who he represents keeps nagging at people. Nobody spends that kind of effort on nobody.

The project that found him has been at this since 1958, when George Hanfmann of Harvard and Henry Detweiler of Cornell set up the modern expedition. Before them, Howard Crosby Butler of Princeton dug there from 1910 until his death in 1922, uncovering the Temple of Artemis and more than a thousand tombs. The modern expedition has run continuously for over six decades under Turkey’s culture ministry, through the long directorship of Crawford Greenewalt Jr. and then Cahill’s from 2008, with something like seven hundred people from more than a hundred institutions passing through it. They have excavated a synagogue, a bath and gymnasium complex, rows of Byzantine shops, Lydian houses and the royal tumuli out on the plain. Sixty-eight seasons of steady, unspectacular work, and then a routine morning on a Roman road produces the most striking single object the site has yielded in a long time.

That is the part of archaeology that never makes it into the documentaries. Almost nothing is found by digging where you think treasure is. Things are found because someone is patiently recording a road surface nobody expects anything from, and one of the stones is heavier than it should be.

The statue is now in a conservation laboratory. The work of stabilising the marble and joining the broken sections is being carried out under the supervision of Turkey’s General Directorate of Cultural Heritage and Museums, and when it is finished, he goes on display at the archaeological museum in Manisa, the provincial capital near the site. That is the resolution. He does not go to a foreign museum and he does not disappear into a storeroom. He gets put back together and stood up, about twenty miles from the road he spent fifteen centuries under, in a building where anyone can walk in and look at him.

There will be a scholarly publication eventually, and it will run to a great deal more than the early announcement did. Specialists will argue about the striped garment, about whether the workshop was local or imported, about whether the archaism is religious convention or something more pointed. Some of those arguments will not be settled in our lifetime. The evidence is one damaged figure and a great deal of silence around him.

But the thing worth holding onto is simpler than any of that. Around two and a half thousand years ago, in a city that had recently lost its independence, somebody with real skill spent a long time carving a young man in clothes that nobody had worn for a century, holding a piece of fruit. The statue broke. The pieces were used to fill a hole in a road. Traffic went over him, then the city died, then the road was buried, then the whole valley was farmland, and for fifteen hundred years nobody on earth knew he existed.

And then a crew lifted a paving slab on an ordinary working morning, turned it over, and he was still there, looking back up.

The last song of a busker’s set is usually decided by the busker. Hanny Junee, a South Korean singer and guitarist who plays on the stone in central London, has a habit of handing that choice to whoever is still standing there. On this particular evening the crowd had two options in front of them, a song by The Script or a song by Coldplay, and they went with Coldplay. He started playing Viva la Vida. Somewhere in the middle of it a man in sunglasses and a blue baseball cap stepped forward out of the audience and began singing along with him, close enough to share the microphone, confident enough to take a verse.

Junee’s account of the next ninety seconds is the reason this story travelled. He has said he was thinking, while it was happening, that this stranger had a really good voice, great energy, a nice presence, and that he sounded remarkably like Chris Martin. And then he talked himself out of it, on the entirely reasonable grounds that Chris Martin does not wander around London singing on street corners with people he has never met. So he carried on, enjoying the duet on its own terms, treating the man as a gifted passerby. The song ended. The man left. And then people in the crowd came over to tell Junee who he had just been singing with.

It was Chris Martin. He had not introduced himself, had not made a moment of it, had not taken the microphone or addressed the crowd, and had walked off without mentioning his name.

That is the whole event, and it is worth separating what is solidly established from what rests only on Junee’s own telling, because they are not the same thing. What is independently confirmable is that the encounter happened and was filmed, and that the video spread fast enough to be picked up within a day by the BBC, CNN, MusicRadar, Guitar World, Guitar.com and a long tail of music and radio outlets. Multiple reports place it in Trafalgar Square, in central London, in the past week or so, though at least one account puts the evening on a different day of the week than another, which is the sort of small discrepancy that shows up whenever a story reaches print through several hands at once. The identification of Martin is not seriously disputed by anyone, and the footage is clear, but it should be said plainly that no statement from Coldplay or from Martin himself has appeared confirming it. Nobody has denied it either.

What rests on Junee alone is everything interior. The crowd vote, the two songs on offer, and above all the reasoning he says he went through in the moment, none of that is visible in a video and none of it has a second witness. It comes from him, in interviews given afterwards, and it should be read as his account rather than as established record. It also happens to be the most charming part of the story, which is a good reason to be careful with it rather than a reason to skip the caveat.

There is a detail that gets lost in most of the retellings, and it is the one that makes the whole thing feel less like a fairy tale and more like something that nearly did not happen. This was the final song of Junee’s set. If the crowd had voted the other way, he would have played The Script, finished, packed the guitar into its case, and gone home. Martin, on whatever errand had him crossing that part of the city in a cap and sunglasses, would have heard a song he had no particular reason to join and kept walking. The entire encounter turned on a show of hands from strangers who had no idea what they were deciding.

It also helps to understand what busking in that part of London actually involves, because the romantic version and the real version are some distance apart. Street performance in Westminster is licensed, not spontaneous. Performers apply to the council, agree to standard conditions covering amplification levels, pitch locations and time limits, and can lose the licence if they breach them. Trafalgar Square in particular sits under its own layer of rules, being managed space rather than ordinary pavement. A busker working there is not a person who wandered out with a guitar. They are someone who applied, waited, got approved, and then showed up to compete for attention against traffic noise, tour groups, pigeons and the general indifference of a city where everybody is on their way somewhere else.

The economics are unglamorous. A busker is paid entirely in the decisions of people who were not planning to stop. Most of them do not. The skill is not only musical, it is the ability to hold a crowd that has no obligation to stay and no ticket invested in staying, and to keep doing it for hours in weather that is frequently against you. Inviting the remaining listeners to vote on the last song is a small, clever piece of that craft. It gives people who stopped for free a stake in what happens next, and it usually means the set ends with the crowd at its largest rather than drifting away song by song.

That is the context Martin walked into, and there is reason to think he knew exactly what he was doing. This is not the first time he has turned up somewhere small in a disguise. A couple of years ago he appeared at a karaoke lounge in Las Vegas in an oversized suit, an obviously fake wig and a pair of unflattering glasses, sang a ballad from the album Coldplay were about to release, and only pulled the wig off at the end while the room lost its mind. That one was a planned shoot for a music video, so it is not a perfect comparison. But the instinct behind it, the wish to sing in a room where nobody is looking at you yet, is visible in both. The difference is that in Las Vegas he eventually revealed himself, and in London he did not.
It is also not the first time Martin has done something small and unbilled in public. He has been reported joining in with fans, helping a stranger stage a proposal, and generally turning up in situations far below the scale he normally operates at. None of that is remarkable on its own. What is remarkable, in this instance, is the restraint of leaving without saying anything. The easy version of this story is the one where the famous man reveals himself, the crowd erupts, the phones come up and everyone gets the clip they came for. He chose the version where the person he sang with did not find out until he was already gone, which meant that for the duration of the song, the duet was entirely honest.

The song they ended up sharing carries its own weight here. Viva la Vida came out in 2008 and became Coldplay’s first number one single in both Britain and the United States, the first British single to manage that double since Rod Stewart in 1971. It won Song of the Year at the Grammys. Martin has said the melody arrived around two in the morning and the skeleton of the song took about ten minutes, after which the band spent an enormous amount of time trying arrangement after arrangement before the church bell sound finally made it work. The title comes from a Frida Kahlo painting, the phrase written across a watermelon in the last picture she finished. Lyrically it is sung by a ruler who has lost everything, looking back from the far side of his own downfall, which is a strange and specific thing for a stadium anthem to be about and part of why it has outlasted almost everything else from that year. It has passed three billion streams on Spotify.

So the song a crowd of strangers voted for, without knowing there was anything at stake, was the one the man standing among them had written seventeen years earlier at two in the morning. He then sang it in public without his name attached to it, next to someone who did not know who he was, and the person he sang it with has said the thing he keeps coming back to is that the connection was purely musical and had nothing to do with fame. Which is exactly the point. For the length of one song, Martin got to be a man with a good voice who joined in, judged on nothing but whether he could hold the harmony. Very few people in his position ever get that again.
There is something worth sitting with in the disguise, too, because it is not much of one. A cap and sunglasses will not fool anyone who is looking. What they do is buy a few seconds of not being looked at, which in a crowded square is enough. People in a city are running a constant, unconscious sorting process on the faces around them, and almost nobody is expecting the answer to come back famous. Junee ran that same process in real time, got an implausible result, and rejected it, which is what any of us would have done. The reason the story lands is that he was not starstruck and he was not fooled. He simply applied ordinary reasoning to an extraordinary situation and reached the wrong conclusion for entirely sensible reasons.

The footage that carried all this around the world was not, as far as can be established, shot by Junee. Reporting credits it to a member of the audience, named in one radio outlet’s report as Beatrice Tiritiello, who said afterwards that she had simply been filming the performance and could not believe what she had ended up capturing. That matters more than it sounds, and not only for the obvious reason that the person who pressed record owns what they recorded. It is the most ordinary element of the whole story. Somebody stopped, liked what she heard, held up a phone the way people do, and by accident became the only reason anyone else knows this happened. Junee posted about it himself afterwards, and has spoken about it since, but the record of the moment exists because a stranger was paying attention.

What happens to Junee now is the part nobody can answer yet, and it would be dishonest to pretend otherwise. Viral moments do not reliably convert into careers. Street performers have been briefly famous before and gone back to the same pitch a fortnight later with the same open case in front of them. There is a well-documented experiment in which a world-class violinist played in a train station and was almost entirely ignored, which is usually cited as proof that people miss greatness when it is not framed for them. This story is the inverse of that and slightly more hopeful. A crowd did stop. They stayed long enough to be given a vote. And one of the most successful songwriters alive listened to a busker for long enough to decide he wanted to sing with him, which is a form of recognition that does not depend on anybody’s algorithm.

Junee’s own framing of it has been consistent and, to his credit, not especially grand. He has described it as amazing and said he still cannot quite believe it, and he has kept the emphasis on the music rather than on the celebrity, which is a harder thing to do than it looks when you have just been handed the most attention of your life. He has not, as of now, turned it into an announcement of anything. He was a working street musician before the cap and sunglasses appeared at the front of his crowd, and he is a working street musician still, with a substantially longer list of people who know his name.

The reason this story keeps getting sent between people is not really that a famous man sang in the street. It is the shape of the thing. Someone gave away a small decision to a crowd. The crowd chose, at random, the one option that opened a door. And the person who walked through the door was so unassuming about it that the only person in the square who did not know what was happening was the man it was happening to.

He played the song he was asked to play. He shared it with a stranger who sounded, he thought, quite a lot like Chris Martin. Then he found out, from people who had been standing there the whole time, that he had been right about the voice and wrong about everything else, and that the last song of the night had turned out to be the one he will be asked about for the rest of his life.

There is a finished film that a major studio decided was worth more to it unreleased than released, and for three years that decision held. The cast had been paid, the animation was done, the test screenings had gone well, and none of it mattered, because the accounting said the movie was more useful as a write-down than as a product. Then the people who made it refused to let it go quietly, a small distributor with almost no leverage bought it, and this year it finally opened in cinemas. It is sitting at 96 percent on Rotten Tomatoes across close to 300 reviews, audiences handed it an A on the exit polls that studios actually care about, and depending on which set of numbers you believe, it has taken somewhere between roughly $65 million and more than $100 million worldwide.

The film is Coyote vs. Acme, and its premise is almost too neat given what happened to it. Wile E. Coyote sues the Acme Corporation for selling him decades of defective products. A company is held to account for the damage its own catalogue caused. That idea did not start as a movie. It started as a short piece by the writer Ian Frazier, published in The New Yorker at the start of the 1990s, written as a mock legal brief on behalf of one coyote against one manufacturer. Frazier has said he typed it on a manual typewriter, mailed it off in an envelope, and then had essentially nothing to do with it for the next thirty years.

It helps to know how unusual the underlying idea is. Warner Bros. has been making Looney Tunes since the 1930s, and the studio’s own history with those characters is a long argument between the people who loved them and the people who owned them. Chuck Jones built the coyote and the road runner around a set of self-imposed rules: the road runner never leaves the road, the coyote is never hurt by anything except his own machinery and gravity, and no dialogue is ever needed. Every single failure is his own purchase decision coming back to meet him. Frazier’s joke worked because he took those rules seriously and asked the only logical follow-up question, which is why nobody ever sued.

Warner Bros. began developing it as a feature in the late 2010s. Dave Green, who had made Earth to Echo and a Teenage Mutant Ninja Turtles sequel, came on as director at the end of 2019. Will Forte, John Cena and Lana Condor were cast, with Eric Bauza voicing the animated characters, and the live-action shooting happened in New Mexico in the spring of 2022. The budget landed at around $70 million. What Green was building was a hybrid, cartoon characters inside a photographed world, which is one of the hardest things to make work and one of the few things in cinema that reliably delights people when it does. His team studied the original Chuck Jones shorts frame by frame, and also studied Buster Keaton, because Wile E. never speaks and everything he feels has to arrive through his body.

Then came 2023, and a period in Hollywood that people who work in it still talk about with a specific kind of anger. Studios had discovered that a completed, unreleased film could be treated as a loss and written down against tax, and that in certain cases this produced a better immediate outcome on paper than releasing it. Warner Bros. had already done this with Batgirl and with a Scooby-Doo holiday film. Coyote vs. Acme was next. The film had been penciled in for a summer 2023 release, and that date quietly went to Barbie instead. In November of that year the studio confirmed it would not be releasing the completed movie at all, and reporting at the time put the write-down at about $30 million. The reasoning that circulated was that marketing the film properly would have cost a further $40 million or so, money the studio did not want to spend on it. Both of those figures come from trade reporting of the period rather than from any public filing, and they should be read that way.

The wider context matters too, because this was not one bad decision by one company. Across that period several studios were merging, restructuring debt or being sold, and finished content became a line item that could be moved around for tax advantage in a way that would have been unthinkable a decade earlier. Batgirl was the case that made headlines first, a film reported to have cost around $90 million and shelved before anyone outside the building saw it. Writers and directors started asking, in public, whether a completed film was now a liability rather than an asset. Coyote vs. Acme arrived in the middle of that conversation and became its emblem, partly because of the timing and partly because there is something almost unbearably on the nose about a corporation quietly disposing of a movie about a corporation being held liable for what it sold.

Understand what that means in practice for the people involved. Several hundred artists had finished a film. Animators had spent years on shots nobody outside the building would ever see. A director had made the movie he had signed on to make, and it had tested well, and he was told it would be destroyed in a spreadsheet. There is no version of that phone call that is not brutal.

The reaction was louder than the studio expected. Cartoonists, animators, industry unions and a very large number of ordinary people who grew up on those shorts made enough noise that the decision became a reputational problem rather than a quiet line item. Within days, the studio reversed far enough to say it would let the film be shopped to buyers. That is not the same as releasing it, and it did not end the fight. Warner Bros. reportedly wanted something in the region of $75 to $80 million for it, which is a strange price to attach to an asset you have just publicly declared worthless, and for well over a year nothing happened. Green and the people who had worked on the film kept talking about it publicly, kept screening it privately where they could, and kept the story alive at a point where the easiest thing in the world would have been to stop.

The buyer, when it finally arrived, was not one of the obvious streaming giants. It was Ketchup Entertainment, an independent distributor most people had never heard of, which acquired worldwide rights in early 2025 for a reported $50 million. Ketchup had form here. It had already picked up and released another shelved Warner Bros. Looney Tunes project, The Day the Earth Blew Up, and had done respectably with it. Taking on a second orphaned Looney Tunes feature at that price was still a serious bet for a company that size, made on the theory that the audience the studio had written off was real.

The film opened in North America at the end of last month. It took about $15.5 million in its first weekend across roughly 3,500 screens, finishing second, which was the best opening Ketchup had ever had. Reviews were close to unanimous. Rotten Tomatoes currently shows 96 percent positive from 288 critics, with an audience figure of 95 percent, and the critical summary talks about a film that is equal parts heart and mischief. The CinemaScore came back an A. Green posted after the opening numbers came in, and the line people quoted back at him was short and not remotely modest. We won.

The box office is where you need to be careful, because the reported figures genuinely do not agree, and blending them would be dishonest. Mid-month, ABC in Australia reported the worldwide total at about $65 million, of which nearly $50 million was North American, with international territories still opening. A few days later, trade and industry coverage including IndieWire reported that the film had crossed $100 million globally. The animation trade site Cartoon Brew, tracking it more conservatively, had the worldwide figure at around $82.8 million at roughly the same point, with the domestic total passing $50 million about three weeks into the run. The commercial tracking database The Numbers currently lists $55.3 million domestic and $21.2 million international for a worldwide total of $76.4 million. These are not small discrepancies. They reflect different reporting dates, different definitions of what counts as reported international gross, and in some cases studio estimates versus actuals. What can be said without picking a side is that the film has taken well over the $30 million that was written off, and that its North American gross alone has passed the $50 million the distributor paid for it.

That last comparison is the one everybody reaches for, and it deserves a caveat that most of the headlines skip. Cinemas keep a large share of ticket revenue, and the distributor still had to spend on prints and advertising, so passing the purchase price at the box office is not the same as being in profit. Cartoon Brew made exactly this point, calling the milestone symbolic rather than a break-even calculation. What is not in dispute is the direction of travel. Industry coverage in the past week has had the film out-earning most of Warner Bros.’ own releases this year, and heading toward profit for the company that rescued it.

There is a further twist that would have been unimaginable in 2023. Because the film is a live-action and animation hybrid with a high enough proportion of animated runtime, reporting suggests it qualifies for consideration in the Academy’s animated feature category. A movie that a studio classified as a tax loss is now in the conversation for an Oscar nomination. It went to digital platforms at the end of this month, and it is still opening in international markets, having reached Australian screens only in the past couple of weeks.

Frazier, for his part, has been gracious and slightly bemused throughout. He has said he hopes people like it and that he had absolutely nothing to do with it, and when asked about the film’s lineage he made a point of naming Tex Avery and the other people who built those original cartoons, which is the sort of credit-sharing that almost nobody does anymore. He took part in some of the promotion, including a conversation with screenwriter Samy Burch and a sketch with Will Forte, which means the man who wrote the original joke in 1990 got to see it finish its journey.

What makes this story matter beyond one film is what it proves about the assumption underneath the shelving. The assumption was that a finished movie’s value is fully knowable in advance, and that if the internal model says the marketing spend will not be recovered, the film has no value worth pursuing. Coyote vs. Acme was the clearest possible test of that assumption, because everything else about it was constant. Same movie, same cast, same 96 percent, same audience. The only thing that changed between worthless and profitable was who owned it and whether anyone was willing to put it in front of people. The model was not measuring the film. It was measuring the studio’s appetite.

It also settles a question about whether public pressure does anything. Normally the answer in this industry is no, or not quickly. In this case the noise that followed the shelving is the only reason the film was ever put up for sale, and the fact that Green and his colleagues would not stop talking about it is the only reason a buyer knew there was something to buy. Nobody won that fight in a week. It took roughly a year and a half between the reversal and the sale, and another year and a half after that before anyone could buy a ticket. The people involved spent that entire period advocating for a film they had already been paid for, with no guarantee it would ever be seen.

Green has talked about the making of it in a way that suggests he never really let himself believe in the release until it happened. He has described the hybrid process in detail, the sketch-vis artists who drew two-dimensional passes over the previs during lockdown, the puppeteers on set standing in for characters who would later be drawn, the jokes added late in animation that ended up getting the biggest laughs. At one point the choir recording the score was asked to settle an argument by vote on whether the Road Runner says meep or beep. That is the texture of a production where people cared about small things, which is exactly the kind of work that disappears without trace when a film is shelved.

The ending here is not that the system changed, because it has not. The rules that made writing off a finished film attractive are still in place, and other completed projects are still sitting in vaults with nobody making noise on their behalf. What changed is that one of them got out, and in getting out demonstrated that the accounting had been wrong on its own terms. A film valued at a $30 million loss has returned multiples of that, won over nearly every critic who saw it, and is now being discussed as an awards contender.

Three years ago a group of people were told that the thing they had built would never be shown to anyone. This year they sat in cinemas and watched strangers laugh at a spider gag that was added at the last minute. The audience got the last word, which in this business is rarer than it should be.