A team led by Thomas Evans-Soma of the Max Planck Institute for Astronomy in Heidelberg has published observations of the exoplanet WASP-121b that complicate the existing picture of how chemistry moves through an ultra-hot atmosphere. Using the James Webb Space Telescope’s Near-Infrared Spectrograph to watch the planet through a complete orbit of its host star, the team detected methane in abundance on the planet’s nightside. On the dayside, temperatures reach approximately 3,000°C, well above the point at which iron vaporises. Methane, a molecule that decomposes far below that threshold, should not survive the heat. The paper was published in Nature Astronomy in June 2025.
The detection itself is the straightforward part. The explanation is not.
What the JWST data showed
WASP-121b is a tidally locked gas giant orbiting its star in about 30 hours, close enough that one face is always turned toward the heat. The dayside atmosphere, at roughly 3,000°C, is hot enough to break apart most molecules and sustain metals in vapour form. Water, carbon monoxide, and silicon monoxide were all detected there, the last of these making WASP-121b the first planet in which silicon monoxide has been identified in any atmosphere.
The nightside, while still extreme by any ordinary measure, sits at considerably lower temperatures. It was there that the team found methane in concentrations high enough to register clearly in the spectroscopic data. Methane typically breaks apart under heat well below what the dayside experiences. Even on the nightside, the temperatures are sufficiently elevated that methane’s persistence requires an active replenishment mechanism. Without one, the molecule would deplete faster than the atmosphere could sustain it.
The circulation the models did not predict
The explanation Evans-Soma’s team proposes is vertical atmospheric circulation: strong upward winds lifting methane-rich gas from deeper, cooler layers of the nightside atmosphere to replace what is continuously being destroyed at higher altitudes. Deeper in the nightside atmosphere, where temperatures are lower and the planet’s carbon-to-oxygen ratio works in methane’s favour, the molecule forms readily. The proposed circulation carries it upward fast enough to maintain the observed concentration.
The problem is that this kind of vigorous vertical mixing was not predicted by existing atmospheric models for planets in this temperature range. The standard picture of ultra-hot Jupiter dynamics has tended to emphasise horizontal circulation, with hot gas flowing from the dayside to the nightside at high altitudes and cooler gas returning at lower levels. Strong vertical mixing on top of that, specifically on the nightside, was not part of the expected behaviour.
The authors are careful to frame this as a proposal. The methane detection is robust; the vertical circulation as its cause is the team’s interpretation, and the models will need to be tested against the new data before the picture can be called settled. In the authors’ framing, the finding indicates that existing dynamical models will likely need to be adapted to account for the degree of vertical mixing the nightside data appears to require.
What the silicon monoxide finding adds
The detection of silicon monoxide on the dayside is a separate result worth noting in its own right. Silicon monoxide has not previously been identified in any planetary atmosphere, exoplanet or otherwise. Its presence at the temperatures and pressures of WASP-121b’s dayside is consistent with the conditions that would vaporise silicate minerals from deeper layers and carry the products to detectable altitudes.
This is the kind of detail that matters when building out a coherent picture of how ultra-hot gas giants differ chemically from anything in our solar system. The JWST’s NIRSpec instrument has now made it possible to run a reasonably complete inventory of the major carbon, oxygen, and silicon compounds across a planet’s full orbital phase, which is what Evans-Soma and Gapp did here. The dayside and nightside of WASP-121b turn out to carry chemistries that are not just different in degree but qualitatively distinct from each other.
What this means for the models
Ultra-hot Jupiter dynamics are a young field, and the models built so far have been calibrated against a limited set of observations. WASP-121b is one of the most intensively studied exoplanets in this class, partly because its tight orbit and high temperature make its atmosphere accessible to spectroscopic observation from a range of instruments over many years.
The challenge the new paper poses is specific: how does an atmosphere this hot maintain the chemical gradients the data shows, and what is driving the vertical circulation that appears necessary to explain the nightside methane? Those questions will determine what adjustments atmospheric modellers need to make, and whether the circulation observed here is particular to WASP-121b’s conditions or something common across the ultra-hot Jupiter population. The paper does not resolve that question. It opens it.
In 1974 two psychologists sent an attractive young woman to stand at the end of a bridge and stop men walking across. She asked each one to fill out a short questionnaire, then tore the corner off a sheet of paper, wrote down her name and phone number, and said he could call if he wanted to talk more about the study. The only thing that changed was which bridge the men had just crossed.
On a high, swaying footbridge, nine of the eighteen men who took her number later called. On a low, solid bridge nearby, only two of sixteen did. The men who had just been frightened were far more likely to phone a stranger, and Donald Dutton and Arthur Aron argued the reason was that they had felt their hearts pounding and read it, wrongly, as attraction.
Two bridges over the same river
The study ran over the Capilano River in North Vancouver, British Columbia. The frightening bridge was the Capilano Canyon Suspension Bridge, a five-foot-wide walkway of wooden boards strung from wire cables and stretched 450 feet across the canyon. It tilted and wobbled underfoot, its handrails were low cable, and it hung about 230 feet above rocks and shallow rapids. Roughly four in ten people crossing it gripped the rail and inched along.
The calm bridge sat upriver: a heavy cedar structure, wider and firmer, only about ten feet above a small stream, with high handrails and no sway. When the researchers later asked a separate group of men how frightened they felt crossing each one, the suspension-bridge group rated their own fear at 65 on a 100-point scale. The men on the solid bridge rated theirs at 3.
So the two groups began in very different physical states. One had a body still humming with adrenaline. The other did not. Everything that came after was designed to see whether that difference leaked into how the men responded to a woman.
What the interviewer measured
The interviewer told the men she was studying how scenery affects creativity, then asked each to write a short, dramatic story about an ambiguous picture of a woman. The stories were scored from one to five for sexual content by raters who did not know which bridge each man had been on. Men from the suspension bridge wrote stories with noticeably more sexual imagery, averaging 2.47 against 1.41 for the solid bridge.
The phone number was the second, blunter measure. The researchers reasoned that curiosity about the study should be about equal across both groups, so a difference in who actually called back would reflect a difference in interest in the woman herself. That is where the gap was widest: half the men from the suspension bridge called, against roughly one in eight from the solid bridge.
One detail mattered to the interpretation. When a male interviewer ran the same routine, the effect vanished. Men leaving the scary bridge wrote no racier stories and were no more likely to call him back. The arousal only seemed to redirect toward someone the men might plausibly be attracted to.
The check the authors built for themselves
A skeptic could object that the two bridges drew different kinds of people. The suspension bridge is a tourist attraction, so its visitors might be bolder, more out-of-town, or simply in a different mood than someone strolling a quiet park. Dutton and Aron raised that objection themselves, and ran a second version to answer it.
This time they kept only suspension-bridge visitors, comparing men stopped right after crossing with men from the same crowd approached at least ten minutes later, once their nerves had settled. The freshly rattled men again wrote more sexual stories and called more often: thirteen of twenty, against seven of twenty-three for those who had cooled down.
A third study moved indoors, where the researchers could control more. Eighty male university students were told they would receive an electric shock, some led to expect a strong jolt and some a mild one. A woman they believed was another participant sat nearby. Men braced for the painful shock reported more attraction to her and again produced more sexual imagery. Crucially, the boost came from anticipating their own shock, not from worrying about hers, which argued against the idea that they were simply moved by a woman in distress.
How much can two bridges really tell us?
The studies are vivid, but they are small and old, and the famous bridge experiment carries the usual hazards of research done in the wild. The groups were not randomly assigned; they sorted themselves by which bridge they chose to cross. The phone-call measure rested on a few dozen men, the kind of sample where a handful of decisions swings the result. And the whole design worked in one narrow direction only: a female interviewer and unaccompanied young men. It says nothing about anyone outside that setup.
The interpretation is also more contested than the tidy version suggests. Dutton and Aron leaned on the idea, drawn from earlier work by Stanley Schachter, that the body produces a wash of generic arousal and the mind looks around for a label to pin on it. On a swaying bridge, the nearest available label is the woman. But the authors themselves listed other readings of their data. Perhaps fear did not get relabeled as desire so much as loosen the brakes on feelings that were already there. Perhaps the link between fear and attraction is its own phenomenon rather than a special case of some grand rule about arousal. They could not rule those out, and which of those mechanisms is at work, if any, is not something a single field study could settle.
What survives is narrower than the headline but still striking: under the right conditions, a body stirred up by one cause can color a judgment about something else entirely. The bridge did not manufacture attraction out of nothing. It seems to have nudged men who were already inclined to notice a woman into noticing her a little more, and into reading their own pounding hearts as a reason.
That is the quiet, unsettling part of the finding. We tend to trust the feeling in our chest as a verdict about whoever is standing in front of us. Sometimes it is only a verdict about how high up we are.
The bird most people would name first if asked to picture something pink is not pink when it hatches. A flamingo chick comes out of the egg in dull grey or white down, and it stays that way for a long time.
The famous rosy colour is not part of the bird when it arrives. It is built, slowly, out of what the bird eats.
The grey beginning
According to most accounts of flamingo development, a young bird needs roughly two to three years of the right diet before it carries the full adult flush.
That grey phase is not a flaw waiting to be corrected. As the British Ornithologists’ Union notes, the melanin behind that early colour strengthens the feathers and provides photoprotection and thermoregulation. The dull plumage is doing a job of its own before the pink ever arrives.
Where the colour comes from
The pink is borrowed. Flamingos, like many other birds, cannot manufacture the pigments that make them rosy, so they have to eat them. Their food, algae and small crustaceans, carries pigments called carotenoids. The bird’s liver metabolises the raw pigment, converts much of the beta-carotene into canthaxanthin, and deposits the result into newly growing feathers, skin, bill and legs.
This is why all six flamingo species end up rosy despite living in very different places. Because carotenoids cannot be synthesised by animals and must come entirely from the diet, the shade on the outside is essentially a record of what went in.
Some species go further and apply colour directly. Greater flamingos studied in Spain were found to spread pigmented secretions from the uropygial gland near the tail onto their neck, breast and back feathers, a cosmetic touch-up before breeding. As reported in Smithsonian, the more often a bird does this the pinker it gets, and the colour fades within days if it stops reapplying.
Colour as a signal worth reading
If the colour comes entirely from food, then the colour says something about the eating. Paul Rose, a behavioural ecologist at the University of Exeter, argues the pink is likely an honest signal rather than mere decoration. Rose told National Geographic, “It’s an honest signal. That pink color tells other birds that it’s healthy and fit.”
The logic is straightforward. A bird that feeds well takes in more carotenoids and shows more colour, so a brighter flush generally points to an efficient feeder. Rose put it plainly: “A healthy flamingo — which is demonstrated by its colorful feathers — is an efficient feeder.” This is a reliable proxy, not an infallible one.
That signal appears to matter in the daily scrum of flamingo life. In an observational study of captive lesser flamingos at WWT Slimbridge, Rose and colleagues scored birds on a colour scale. The brightest birds were more aggressive. The authors are careful to note only correlation, not causation, so the link between hue and behaviour is a pattern worth taking seriously rather than a settled mechanism. On colour’s social role, Rose observed that “Color plays an important role” in their social relationships.
What captivity reveals
The clearest evidence that the colour is diet rather than identity comes from birds that lose access to the right food. In captivity, where the natural carotenoid-rich prey of a salt lake is absent, flamingos fade toward pale pink or white unless their keepers intervene. Zoos now supplement the diet with formulated pellets to maintain the colour, a workaround that only exists because the colour was never built into the bird.
The pink is not a costume the flamingo was born wearing. It is information, laid down feather by feather over years, a running record of how well the bird has fed and where. Look closely at a flamingo’s colour and you are reading its meals.
In a 2012 study, researchers handed some people a plain black storage box still in its flat-pack and let others inspect the same box already assembled. Then they asked everyone to name a price they would actually pay to keep it. The people who had built the box themselves bid an average of 78 cents. The people handed the finished box bid 48 cents. Same box, same parts, but the builders wanted it 63 percent more.
Michael Norton, Daniel Mochon and Dan Ariely called this the IKEA effect, after the Swedish furniture company whose products famously arrive with some assembly required. The short version of their finding is that labor leads to love: when people put their own effort into making something, they value the result more highly than anyone else does, sometimes absurdly so.
The origami people valued like a pro’s
The boxes showed the effect existed. A second experiment showed how large it could get. This time people folded an origami frog or crane from a single sheet of paper, with only an instruction diagram to guide them. Afterward they bid on their own creation, while a separate group of bystanders bid on those same paper frogs and cranes.
The builders valued their origami at about 23 cents. The bystanders, looking at the same lopsided results, would pay about 5 cents, treating them as close to worthless crumpled paper. The makers priced their work nearly five times higher than the market did.
The stranger comparison came next. The researchers had experienced origami folders make clean, well-crafted versions, and asked another group to bid on those. Outsiders valued the expert work at about 27 cents. The amateurs valued their own fumbling attempts at 23 cents. In their own eyes, in other words, their handiwork was worth almost as much as a professional’s, a gap so small it was not statistically meaningful.
To check that the builders were not simply overbidding to avoid losing something they had made, the researchers asked another set of folders to guess what the average student would pay for their creation. Those guesses came back about as high as their own bids. The makers did not just want to keep their origami. They genuinely seemed to believe other people would see the value too.
Effort only counts when the work is finished
If sheer effort were the whole story, then any labor should raise the price. The researchers found it did not. What mattered was completing the task.
In one experiment people built small Lego models, around a dozen pieces shaped like a helicopter, a bird, a dog or a duck. As expected, builders valued their own model more than an identical one built by a partner, bidding about twice as much for it. But a third group built a model and then took it apart again. For them the premium collapsed. Undoing the work erased the love, even though Lego is designed to be rebuilt in minutes and they could have reassembled it easily.
A final experiment made the same point from the other side. Some people assembled an IKEA box completely, while others were stopped two steps short of finishing. The finishers bid $1.46 for their box. Those left with an almost-complete box, all the pieces in hand, bid 59 cents, less than half as much. The boost came from crossing the finish line, not from the minutes of effort alone. And it held whether or not people described themselves as the do-it-yourself type.
Why a wobbly box feels more valuable
The researchers tested a few obvious explanations and set them aside. The premium was not simply about owning the thing, since the people who built and then dismantled a Lego model had handled and possessed it just as long as the builders, yet valued it less. It was not about customization either, because a standard IKEA box offers nothing to personalize.
What is left, they argued, is a mix of two older ideas. One is effort justification, the long-studied tendency to value what we have sweated over. The other is a sense of competence: finishing a task, even a trivial one, gives a small jolt of having shaped the world, and the finished object becomes proof of it. A crooked origami frog is not really worth 23 cents. But it is evidence that you made something, and that evidence feels valuable.
The authors trace the same instinct outside the lab. They open with the story of 1950s instant cake mixes, which reportedly sold poorly until manufacturers required cooks to add a fresh egg, restoring a little real labor to the box. They note the story probably had several causes, not one. But it rhymes with their results, as do the businesses that charge people to stuff their own teddy bears or harvest their own dinner on a farm.
What four small studies can and cannot show
The IKEA effect is real in these experiments, but it is worth being precise about their size. These were lab studies with modest numbers of mostly student participants, and the stakes were tiny. The products were boxes, paper frogs and Lego sets, and the prices ranged from a few cents to under two dollars. The researchers measured willingness to pay through bidding games, which approximate real purchasing but are not the same as walking into a store.
The authors are careful about reach. They flag that they used small-ticket items and that whether the effect holds for expensive purchases, a renovated kitchen or a self-built deck, was not something their data could answer. They suggest it might, pointing to homeowners who price their handmade brick walkway far above what any buyer will pay for it, but that is a reasonable guess, not a result.
The mechanism, too, is an argument rather than a settled fact. Effort justification and a feeling of competence fit the pattern, and the researchers ruled out a couple of rivals, but they say plainly that more work is needed to know exactly why building something makes us prize it. What the studies pin down is the pattern itself, and its limit: the glow comes only when the work succeeds. Effort that ends in a half-built box or a model torn back to pieces buys nothing.
That limit may be the most human part. We do not fall for our labor simply because it cost us something. We fall for it when it produces a thing we can hold up and call finished, however lopsided it turns out to be.
Najib Razak walked into Kajang Prison outside Kuala Lumpur on 23 August 2022, the first Malaysian prime minister ever to be jailed, after the Federal Court upheld a 12-year sentence and a 210 million ringgit fine tied to the looting of a state investment fund he himself had chaired. Investigators had traced roughly $681 million into his personal AmBank accounts, money that prosecutors said originated in 1Malaysia Development Berhad — better known as 1MDB — the sovereign wealth vehicle Najib launched in 2009 to develop the Malaysian economy.
The fund was supposed to build power plants and finance affordable housing. Instead it became the largest kleptocracy case the U.S. Department of Justice had ever pursued, with at least $4.5 billion siphoned through a daisy chain of shell companies, offshore banks, and a Hollywood film studio.
The story is partly about a politician. It is mostly about how a single bank account became the evidentiary spine of a global financial scandal.
The fund that was supposed to build a country
Najib became prime minister of Malaysia in April 2009. Within months he had converted a Terengganu state investment outfit into 1MDB and installed himself as chair of its advisory board, with sign-off authority over major transactions. The pitch was ambitious — strategic infrastructure, partnerships with Gulf sovereign funds, a sovereign wealth approach modeled on Singapore’s Temasek.
Behind the brochures, something else was being built. Between 2009 and 2014, 1MDB raised more than $10 billion in bonds, much of it underwritten by Goldman Sachs, which collected close to $600 million in fees on three bond offerings — an extraordinary multiple of what a sovereign-linked deal would normally pay. Hong Kong regulators later fined Goldman $350 million for its role in the offerings.
Money flowed out of the fund almost as fast as it came in.
The $681 million that arrived by SWIFT transfer
In March 2013, weeks before a general election, two transfers totaling roughly $681 million landed in Najib’s personal accounts at AmBank in Kuala Lumpur. The transfers came through Falcon Private Bank in Singapore, originating from a British Virgin Islands shell called Tanore Finance Corporation.
When the Wall Street Journal first reported the deposits in July 2015, Najib’s office did not deny the money had arrived. The official explanation, repeated for months, was that the funds were a personal donation from the Saudi royal family to support Malaysia’s fight against the Islamic State.
Saudi officials never publicly confirmed the gift. U.S. prosecutors later traced the money back not to Riyadh but to 1MDB itself, routed through a chain of accounts in Switzerland, Luxembourg, and Singapore designed to obscure its origin.
The young financier at the center
The architect of the scheme, according to filings by the U.S. Department of Justice and Malaysian prosecutors, was a Penang-born financier named Low Taek Jho, known as Jho Low. He held no official position at 1MDB. He carried no government title. But he moved between Najib’s inner circle, Abu Dhabi’s sovereign wealth funds, and Hollywood producers with a fluency that prosecutors described as the operational core of the fraud.
Low spent some of the diverted money on a $250 million superyacht called the Equanimity, on Picassos and Basquiats, on a stake in EMI Music Publishing, and on financing the 2013 film The Wolf of Wall Street through Red Granite Pictures — a company co-founded by Najib’s stepson Riza Aziz. The Department of Justice would later seek to claw back the film’s profits as proceeds of crime.
Low has been a fugitive since 2016. Malaysian authorities believe he is in China. He denies wrongdoing.
How the jewelry box became the smoking gun
When Malaysian police raided properties linked to Najib and his wife Rosmah Mansor in May 2018, days after his shock election defeat to Mahathir Mohamad, the inventory read like an auction catalog. Officers seized 12,000 pieces of jewelry, 567 handbags including 272 Hermès Birkins, 423 watches, and 234 pairs of sunglasses. The total estimated value reached 1.1 billion ringgit, roughly $273 million at the time.
One pink diamond necklace alone was valued at $27 million.
The seizure became a visual shorthand for the case. Physical objects — handbags, watches, real estate — anchor abstract financial crimes in the public imagination in a way that SWIFT codes and shell-company structures never can. A spreadsheet doesn’t make the news. A wall of Birkins does.
The SRC International trial
The 12-year sentence Najib began serving in 2022 stems not from the full 1MDB case but from a smaller subsidiary called SRC International. Prosecutors proved that 42 million ringgit, about $10 million, flowed from SRC into Najib’s personal accounts between 2014 and 2015.
High Court judge Mohamad Nazlan Mohamad Ghazali convicted him on all seven charges in July 2020 — one count of abuse of power, three counts of criminal breach of trust, and three counts of money laundering. The Court of Appeal upheld the verdict in December 2021. The Federal Court, Malaysia’s highest, dismissed his final appeal on 23 August 2022.
He was driven from the courthouse straight to Kajang Prison the same afternoon.
The far larger 1MDB trial — the one dealing directly with the funds that flowed into his personal accounts — ran on for years afterward. On 26 December 2025, the Kuala Lumpur High Court convicted Najib on four counts of abuse of power and 21 counts of money laundering over more than $700 million channeled from 1MDB into his accounts, and sentenced him to a further 15 years in prison along with a fine of 11.4 billion ringgit — the new term to begin only after his current sentence ends. His lawyers filed an appeal against the conviction and sentence within days.
The global recovery effort
The U.S. Department of Justice filed civil forfeiture actions seeking to recover more than $1.7 billion in 1MDB-linked assets — paintings by Van Gogh and Monet, a Bombardier jet, penthouses in New York and London, the Equanimity yacht, and rights to The Wolf of Wall Street. By 2024, settlements and recoveries had returned more than $1 billion to Malaysia.
Switzerland, Singapore, and Luxembourg opened their own investigations. Falcon Private Bank lost its Singapore license. BSI Bank was shuttered. Several private bankers received prison sentences ranging from a few weeks to four and a half years.
Goldman Sachs settled with the U.S. government for $2.9 billion in 2020 — at the time the largest penalty ever paid under the Foreign Corrupt Practices Act. The bank’s Malaysia unit pleaded guilty. Two former Goldman bankers, Tim Leissner and Roger Ng, were convicted in connection with the scheme.
What the bank account proved
The legal turning point in the SRC case was not testimony or whistleblower documents. It was the bank account itself. Najib’s defense argued he believed the money in his AmBank accounts was a Saudi donation. The court rejected that defense partly because the volume of deposits, withdrawals, and transfers passing through the accounts — and the fact that they continued long after the supposed donation purpose had lapsed — was inconsistent with any innocent reading.
A prime minister with sign-off authority over a sovereign fund, receiving nine-figure sums from offshore shells into his personal account, then spending portions of that money on credit-card bills and political expenses, leaves a paper trail that even sympathetic judges cannot ignore.
The gradual normalization of small ethical breaches can make the larger ones feel routine to the people committing them. A donation here. A consulting fee there. A bank account that no one ever questions.
Life inside Kajang
Najib is now an inmate at Kajang Prison, the same facility that once held opposition leader Anwar Ibrahim on charges Anwar maintained were politically motivated. The symmetry is not lost on Malaysians.
In February 2024, the Pardons Board chaired by then-king Sultan Abdullah halved Najib’s sentence to six years and reduced his fine from 210 million ringgit to 50 million. He then spent more than a year arguing that a royal addendum issued alongside the pardon entitled him to serve the rest of that term at home. In December 2025 the High Court ruled the addendum was not legally enforceable, and in April 2026 Najib withdrew his appeal against that decision, closing off the house-arrest bid.
The transition from high status to incarceration describes a particular kind of disorientation — not just the loss of liberty but the collapse of the deference structures that previously shaped every interaction. For a man who spent nine years as prime minister, addressed as Yang Amat Berhormat, the transition to a numbered uniform and a 6 a.m. roll call is its own kind of sentence.
The number that stays
Of all the figures in the 1MDB case — the $4.5 billion siphoned, the $2.9 billion Goldman penalty, the $1.7 billion in U.S. asset seizures, the 1.1 billion ringgit in jewelry — the one that keeps coming back is the original $681 million that arrived in a politician’s personal bank account in March 2013.
It is a small number compared to the total fraud. It is, by the reckoning of the prosecutors who built the case, among the largest sums ever traced into the personal account of a sitting head of government anywhere in the world.
The wire transfer took seconds. The reckoning took the better part of a decade to reach him. And with a second conviction now stacked on the first — under appeal, but handed down — the years Najib has left in Kajang to think about how it ever got that far stretch well beyond the sentence that first put him there.
The strangest thing about a blue whale’s heart is not its bulk. It is how close that heart comes to stopping. On a deep foraging dive, the largest heart on the planet can slow to as few as two beats a minute, a pace so far below what physiology predicts for an animal that size that it looks, at first, like an error in the data.
It was not an error. In 2019, a team led by Stanford’s Jeremy Goldbogen, with Paul Ponganis of the Scripps Institution of Oceanography as senior author, made the first recording of a blue whale’s heart rate in the wild. What they found raised a question the whale’s size alone never could: how does a heart that big run that slow, then race back so fast, dive after dive, without faltering?
How do you record the heartbeat of the largest animal that ever lived?
Carefully, and with a long pole. Working in Monterey Bay, co-author David Cade attached a suction-cup electrocardiogram tag near the left flipper of a male blue whale estimated to be about 15 years old. The tag, roughly the size of a lunchbox, was held by four suction cups, two carrying the electrodes that picked up the electrical signal of the heart.
It stayed on for about 8.5 hours of diving before detaching and floating up to be retrieved. The team almost did not believe what came back. “We had no idea that this would work and we were skeptical even when we saw the initial data,” Goldbogen said. Reading the trace was its own task. “With a very keen eye, Paul Ponganis … found the first heart beats in the data,” he recalled. Goldbogen later told Smithsonian the whole attempt felt unlikely: “I honestly thought it was a long shot because we had to get so many things right.”
The results were published on 25 November 2019 in Proceedings of the National Academy of Sciences, under a title that named the puzzle directly: extreme bradycardia and tachycardia in the world’s largest animal. This is a single animal measured once in the field, not a survey of the species, so the pattern is best read as a first window rather than a settled rule.
What did the data actually show?
A heart swinging between two extremes on every dive. During dives that reached as deep as 184 metres and lasted up to 16.5 minutes, the heart rate typically settled to four to eight beats a minute and bottomed out at just two beats per minute. For a body this large, the predicted resting rate is closer to 15 beats per minute, so the diving heart was running far below what its size alone would suggest.
Then came the reversal. The moment the whale returned to the surface to breathe and re-oxygenate, the heart climbed to its highest recorded rate of 30 to 37 beats per minute. There was one wrinkle even at depth. During a lunge to feed, the rate briefly jumped to about 2.5 times the dive minimum, before sinking again.
Why would a heart run itself so close to stopping?
The slow pace is oxygen management. Holding a single breath through a long dive, the whale conserves its limited oxygen by dropping its heart rate and routing blood to the organs that need it most, the same dive response seen across diving mammals, scaled up to the largest body on Earth. The surface surge is the repayment, the heart racing to reload oxygen before the next plunge.
What struck the team was how little headroom remained. The surface rate sat near the whale’s estimated physiological maximum, which led them to suspect the heart may already be working at its functional ceiling. That ceiling, set by body size and the demands of lunge feeding, “can help us understand biological limits to size,” Goldbogen said. The idea that a maxed-out heart is one reason blue whales never grew larger is the team’s hypothesis, not a proven fact, and they have been clear that much more work is needed to test it.
So is the heart really the size of a small car?
Probably not. The comparison is repeated everywhere, but it does not survive a close look at an actual specimen. A preserved blue whale heart weighed roughly 400 pounds, while National Geographic noted that a Volkswagen Beetle weighs around 3,000 pounds, which makes the car framing closer to legend than measurement. A researcher’s suggestion of a small golf cart is nearer the mark, though even that overshoots, since golf carts run heavier than the heart itself. The simpler version holds: it is the largest heart of any animal, and its size is not the most interesting thing about it.
What makes the 2019 recording matter is the method as much as the numbers. One tag, one whale, 8.5 hours. The questions the data invites next are the ones that take a fleet of recordings to answer: whether the same cardiac swing holds across larger and smaller individuals, whether it shifts with feeding, depth, or age, and how close other giants of the ocean run to their own limits. The first beat was the hard part. What it measures next is still open.
NASA’s Artemis site selection team finished the most exhaustive survey of the lunar South Pole ever attempted and came back with a list that would fit on an index card: candidate regions where the first woman and the next man are meant to step out of a Starship Human Landing System sometime this decade. The Moon is roughly 14.6 million square miles of surface — approximately the area of Africa and Australia combined — and the places everyone wants to put boots down add up to a cluster of patches you could drive across in an afternoon.
The shortlist sits close to the lunar south pole. Mons Mouton Plateau. Mons Mouton itself. Peak near Cabeus B. de Gerlache Rim 2. Haworth. Malapert Massif. Two patches of Nobile rim. Slater Plain. Together they cover a sliver of geography smaller than a major metropolitan area.
Every other square mile of the Moon is, for Artemis purposes, the wrong square mile.
Why a continent shrinks to a handful of neighborhoods
The reason is sunlight, water, and line-of-sight radio — three things that almost never coexist on the same patch of lunar dirt. The South Pole is the only region on the Moon where they sometimes do.
At the pole, the Sun never climbs far above the horizon. Shadows stretch for kilometers. A crater rim a few hundred meters tall can stay lit for much of the year while the floor of the crater right next to it has not seen sunlight in roughly two billion years.
Those permanently shadowed regions are where the water hides. Temperatures inside them sit around minus 250 degrees Fahrenheit, cold enough to trap water ice delivered by ancient comet impacts and hold onto it through every lunar day since. NASA wants astronauts close enough to walk to that ice, but standing in sunlight that recharges their solar arrays. The geometry that allows both is rare.
The sites
The nine candidate regions named in NASA’s October 2024 update for an Artemis III crewed landing are Peak near Cabeus B, Haworth, Malapert Massif, Mons Mouton Plateau, Mons Mouton, Nobile Rim 1, Nobile Rim 2, de Gerlache Rim 2, and Slater Plain. The list was refined down from thirteen regions NASA had identified in 2022.
Each region spans roughly 15 by 15 kilometers. Inside each box, the actual touchdown ellipse for a Starship-class lander is smaller still, on the order of 100 meters across.
The names sound abstract. The features are not. Shackleton Crater — adjacent to the de Gerlache Rim 2 candidate region — is approximately 21 kilometers wide and 4 kilometers deep, named after the Antarctic explorer, with a rim that catches almost continuous sunlight while its floor is one of the coldest places ever measured in the solar system. Malapert Massif is a mountain about 5 kilometers high — taller than anything in the contiguous United States — that sits in near-permanent daylight and has a clear radio line of sight to Earth. Mons Mouton Plateau, the broadest of the nine candidate regions, is large enough on its own to dwarf the other eight combined.
The ridge connecting Shackleton to de Gerlache sits just outside the Artemis III shortlist but has, in a real sense, become the most valuable real estate on the Moon: solar panels mounted on a mast there would generate power for much of the year. NASA has separately identified that ridge as the focal point for the long-term Moon Base program announced earlier this year.
The water question
NASA’s Volatiles Investigating Polar Exploration Rover, VIPER, was planned to drive into those shadowed pockets and find out exactly how much ice is there and in what form before the agency canceled the rover programme in July 2024. The science it was meant to settle has not gone away. Observations from the LCROSS impact in 2009 and the SOFIA telescope in 2020 confirmed water molecules exist in lunar soils. Nobody yet knows whether you can dig a shovelful and get a cupful of water back.
That distinction matters because the entire architecture of a long-term human presence depends on it. Water at the pole means rocket propellant — hydrogen and oxygen — produced on the Moon instead of hauled up from Earth at significant cost per kilogram. It means breathable air. It means radiation shielding.
If the ice turns out to be diffuse — a few percent by mass scattered through cubic meters of regolith — the economics of a lunar base change overnight. The site selection narrows even further.
A continent’s worth of nowhere
To appreciate how small the shortlist really is, consider the rest of the Moon.
The Apollo program landed six crews between 1969 and 1972, all of them in the equatorial belt where sunlight is reliable, temperatures swing predictably, and Earth sits high in the sky for easy communication. Tranquillity Base, where Armstrong and Aldrin set down Eagle, is thousands of kilometers from the south pole.
None of the Apollo sites would work for Artemis. They have no water ice, no permanent sunlight for power, and they are too far from the resources that make a sustained base possible. The entire equatorial near side, the part of the Moon you can see from your backyard, is essentially off the table.
So is the far side, which has no direct line of sight to Earth and would require a relay satellite for every radio call home. So is the north pole, which has similar lighting geometry to the south but appears to hold less water based on remote sensing.
What remains is a thin ring around the south pole, broken up by crater walls and shadow lines, where a handful of ridges and peaks satisfy every constraint at once.
The terrain itself
Photos returned by Lunar Reconnaissance Orbiter show a landscape that looks less like the rolling gray plains of Apollo footage and more like a battlefield of stone. Slopes of 10 to 20 degrees are common. Boulders the size of cars litter the rims. Shadows are absolute — not dim, but black, because there is no atmosphere to scatter light into them.
An astronaut walking from a sunlit ridge into a shadowed crater would step from a temperature of about 130 degrees Fahrenheit into one below minus 200, with no transition. Suit heaters and coolers will work harder there than anywhere humans have ever operated.
Firefly Aerospace was contracted to deliver drones that will fly into the permanently shadowed regions ahead of crewed missions, scouting routes that humans cannot scout themselves because no sunlight reaches the ground.
The $20 billion plan
NASA’s long-range proposal calls for roughly $20 billion to build out the supporting infrastructure: pressurized rovers, surface habitats, a small nuclear fission reactor to keep things running through the lunar night, and the logistics chain to move all of it from Earth orbit to the candidate landing sites.
The reactor matters because even at the best-lit ridge, the Sun does eventually set. When it does, temperatures collapse and solar power vanishes for days at a time. A 40-kilowatt fission surface power system, currently under development with Westinghouse and other contractors, would let a base ride through those darknesses.
Three new missions announced this year are aimed at delivering the first hardware to support that permanent presence before the end of the decade.
Why the shortlist will probably get shorter
The candidate list is a planning number, not a final answer. Artemis III is funded for one landing. Artemis IV and V are funded for one each. Each mission needs a single site, chosen well before launch based on the most recent lighting models, the most recent ice surveys, and the most recent assessment of which Starship variant can actually thread the touchdown ellipse.
Mons Mouton Plateau has drawn particular attention in mission-planning discussions because of its sheer size — flat enough, broad enough, and high enough for cumulative sunlight to power a base camp across multiple Artemis missions. Sites along the de Gerlache rim offer extended periods of cumulative sunlight per year, direct line of sight to Earth, and walking distance — by lunar standards — to permanently shadowed terrain inside Shackleton.
If one of the leading parcels proves out, much of the rest of the shortlist becomes academic. The first base goes there, and everything that follows clusters within a few kilometers of the first habitat, the first rover tracks, the first cache of spare oxygen tanks.
The competition for the best parcels is already shaping international policy. China has named its own south polar candidates near Shackleton for the planned Chang’e 7 and Chang’e 8 missions. Some of the Chinese sites sit near regions still under U.S. consideration. There is no treaty that resolves who gets there first or what arriving there first means in the absence of fences.
A continent’s worth of Moon, and the part that matters is small enough to walk across before lunch. Whoever lands first on one of these polar ridges will be standing on a few hundred meters of regolith that the rest of the twenty-first century’s lunar program is going to be built around.
The shadows will be longer than anything the Apollo astronauts saw. The Sun will sit on the horizon like a spotlight on a stage. And somewhere down inside Shackleton, in the dark that has held its breath for two billion years, there is — probably — enough water to make the trip worth taking.
Miranda Kerr handed jewelry to the U.S. Department of Justice in 2017, including a heart-shaped pink diamond that had been given to her by Malaysian financier Jho Low, after federal investigators traced the gems to money allegedly siphoned from the Malaysian sovereign wealth fund 1MDB. The Australian model told prosecutors she had no idea the gifts were bought with stolen money and surrendered them voluntarily once the chain of custody was explained.
The heart-shaped pink diamond was reportedly purchased from New York jeweler Lorraine Schwartz. The rest of the haul Kerr returned — earrings, a pendant, a bracelet — pushed the total package toward what court filings described as a multimillion-dollar parcel of stones that had moved from Geneva safes to a model’s jewelry box to, finally, an evidence locker in Washington.
None of it would have been remarkable in the world of celebrity gifting if the money behind it had been clean. It wasn’t.
What 1MDB actually was
1Malaysia Development Berhad was a sovereign wealth fund set up by then–Prime Minister Najib Razak, ostensibly to drive long-term economic projects across Malaysia. U.S. prosecutors say billions of dollars were diverted out of the fund through a chain of shell companies, fake consulting fees, and forged bond documents.
The Department of Justice called it one of the largest kleptocracy cases it had ever brought. The money funded penthouses in Manhattan and London, a superyacht called the Equanimity, Picassos and Basquiats, and — improbably — much of the production budget for the 2013 Martin Scorsese film The Wolf of Wall Street, a movie about financial fraud.
At the center of the spending was Low Taek Jho, known as Jho Low, a young Penang-born financier with no formal title at 1MDB but enormous influence over how its money moved. He has denied wrongdoing and remains a fugitive, last reported to be living under Chinese protection.
How the diamond ended up around Kerr’s neck
Kerr and Low met in the early 2010s. According to filings unsealed by the DOJ’s Kleptocracy Asset Recovery Initiative, Low gave her a series of high-end pieces from Lorraine Schwartz over several months — including earrings, a heart-shaped diamond pendant, and a diamond bracelet.
The heart-shaped pink diamond was the centerpiece. Pink diamonds of that size are vanishingly rare — the Argyle mine in Western Australia, which produced most of the world’s pink diamonds, shut down in 2020, and stones in the multi-carat range routinely sell for tens of millions at auction.
Kerr stopped seeing Low and married Snapchat co-founder Evan Spiegel in 2017. By the time the DOJ contacted her, the jewelry had been sitting in her possession for several years.
The Kleptocracy Initiative
The unit that came knocking was a specialized DOJ team launched to chase stolen foreign-government funds parked in U.S. assets. Its method is unusual: rather than waiting to extradite a suspect, it files civil forfeiture suits against the assets themselves — a yacht, a painting, an apartment, a diamond — and lets the property owner contest the claim.
That structure is why Kerr’s name appeared in court documents at all. The jewelry, not the model, was the legal defendant. Returning the pieces removed her from the proceeding entirely.
Actor Leonardo DiCaprio, who starred in The Wolf of Wall Street and had also received gifts from Low, separately handed over items including artwork. Both Kerr and DiCaprio cooperated; neither was accused of any wrongdoing.
What voluntary return actually means
The phrase voluntary disclosure carries specific legal weight in U.S. white-collar practice. Federal prosecutors offer reduced penalties — sometimes full non-prosecution — to individuals and companies that hand over assets or information before being formally charged. The IRS runs its own version, and recent reporting suggests the agency is trying to revive its Voluntary Disclosure Practice after years of low participation.
The Southern District of New York, which has handled many of the largest financial-crime cases of the past decade, recently formalized its own framework. In February 2026, SDNY announced a new corporate enforcement and voluntary self-disclosure program aimed at companies that come forward early with evidence of misconduct.
The Commodity Futures Trading Commission followed in May with a similar cooperation policy for commodities-market participants. The pattern is consistent across agencies: hand things over fast, and the penalty curve bends.
Kerr’s surrender of the jewelry in 2017 was the celebrity-shaped version of the same move.
Where the diamond went
The DOJ sold the returned 1MDB assets — including the Kerr jewelry, DiCaprio’s artwork, real estate in Beverly Hills and Manhattan, and the seized superyacht — and repatriated the proceeds to Malaysia. By the most recent DOJ accounting, substantial sums have been returned to the Malaysian government in what represents one of the largest civil-forfeiture recoveries in U.S. history.
The pink heart-shaped diamond was auctioned privately. Its current whereabouts are not public.
The Equanimity, the superyacht Low commissioned, was seized by Indonesian authorities at the DOJ’s request and sold to Malaysian casino operator Genting. It now sails under the name Tranquility.
The Najib trial and the verdict that followed
Najib Razak, the former prime minister at the center of the scheme, was voted out of office in 2018 after nearly a decade in power — an election upset driven in large part by public anger over 1MDB. He was convicted in 2020 on multiple charges including criminal breach of trust, abuse of power, and money laundering, and sentenced to prison.
Najib has continued to deny knowing the money was stolen, insisting Jho Low misled him. In 2024, his sentence was reduced by Malaysia’s pardons board, a decision that drew significant domestic criticism.
Goldman Sachs, which underwrote bond issuances for 1MDB, paid billions of dollars in settlements to the Malaysian government and pleaded guilty in 2020 to violating the Foreign Corrupt Practices Act. Its Malaysian subsidiary pleaded guilty to conspiracy to bribe foreign officials.
Why the case still matters to compliance officers
1MDB has become a kind of teaching case for bank compliance teams, partly because the money moved through so many supposedly well-monitored institutions before reaching jewelers, art dealers, and yacht brokers. Modern anti–money-laundering programs lean heavily on transaction monitoring, beneficial-ownership checks, and what compliance specialists call enhanced due diligence on politically exposed persons.
The next phase, industry analysts argue, will be heavier use of AI for pattern detection. A recent Compliance Week piece describing how banks will stay ahead of financial crime in 2026 argues that real-time analytics and behavioral models are replacing the old rules-based screens that 1MDB-era transactions slipped through.
That shift creates its own problem: opaque models that flag transactions without showing their work. Financial regulators have started pressing institutions to use systems that can explain a decision in human-readable terms, a push that one Forbes Technology Council piece frames as the need for transparent, trustworthy AI in financial decision-making.
Jho Low, still missing
Low himself has never been arrested. U.S. prosecutors indicted him on money-laundering and bribery charges. Malaysia and Singapore have also issued warrants. He is widely reported to be living in Shanghai or Macau, and a 2019 Wall Street Journal book co-authored by reporters Tom Wright and Bradley Hope, Billion Dollar Whale, documented his attempts to maintain a luxury lifestyle in transit between safe jurisdictions.
In 2019, he agreed to a civil settlement with the DOJ to forfeit hundreds of millions of dollars in assets — without admitting wrongdoing — including a Bombardier jet, a Beverly Hills hotel stake, and various real estate parcels.
The detail that lingers
Kerr’s handover is, in dollar terms, a rounding error against the billions the DOJ traced. The heart-shaped diamond was one of thousands of objects that money bought before it was caught.
What it captures is the geometry of modern financial crime — stolen sovereign wealth becoming a stone, the stone becoming a gift, the gift becoming evidence, the evidence becoming a wire transfer back to the country it came from. The diamond moved through Geneva, New York, Los Angeles, and Washington before its proceeds settled, eventually, back in Kuala Lumpur.
Somewhere it is being worn again, by someone who probably doesn’t know where it has been.
A robotic spacecraft built by a Flagstaff, Arizona, startup is preparing to chase down one of NASA’s space telescopes and boost it into a higher orbit. Katalyst Space Technologies’ craft, called LINK, is set to launch later this month aboard an air-launched Northrop Grumman Pegasus XL rocket from Kwajalein Atoll in the Marshall Islands, then rendezvous with NASA’s Neil Gehrels Swift Observatory and raise its altitude.
The target is a telescope that cannot save itself. After 21 years in orbit, Swift is sinking. The spacecraft carries no thrusters of its own, and its low Earth orbit has been decaying for years as it brushes against the top of the atmosphere. A recent stretch of heightened solar activity made the problem worse, puffing the atmosphere outward, thickening the drag, and pulling Swift down faster than NASA had planned for.
A telescope built to chase explosions
Swift launched in 2004 to study gamma-ray bursts, the most powerful explosions in the universe. NASA describes it as a kind of cosmic dispatcher: when a sudden, violent event flares somewhere in the sky, Swift pins down where it happened and alerts other observatories to swing around and look. Over two decades it has fed discoveries about exploding stars, flares from distant galaxies, comets, asteroids, and even high-energy lightning on Earth.
Goddard Space Flight Center in Greenbelt, Maryland, manages the mission, with partners that include Penn State, Los Alamos National Laboratory, the UK Space Agency, the University of Leicester, and the Italian Space Agency. Losing the telescope to an uncontrolled reentry would end all of that.
A race against the clock
Rather than let Swift fall, NASA decided to try something it had never done with a science satellite. In September 2025 the agency awarded Katalyst a $30 million contract, routed through its Small Business Innovation Research program, to mount a robotic boost on a compressed schedule. NASA framed the speed itself as the point, calling the effort a way to prove it can go from concept to implementation in under a year.
“This is a forward-leaning, risk-tolerant approach for NASA,” said Shawn Domagal-Goldman, the acting director of NASA’s Astrophysics Division at the time of the award, who noted that boosting Swift is both cheaper than building a replacement and a way to extend satellite servicing to “a new and broader class of spacecraft.” The clock is real: without a boost, NASA expects Swift to reenter the atmosphere in the fall of 2026, and for the rescue to have its best chance the telescope needs to stay above roughly 185 miles, about 300 kilometers. To buy time, controllers suspended Swift’s science observations and reoriented the spacecraft to cut its drag, a change the mission had never made before.
Why an air-launched rocket
The choice of a Pegasus XL is itself a consequence of the rush. Swift sits in an unusually low-inclination orbit of 20.6 degrees, which is hard to reach from U.S. launch pads, where most small rockets are confined to steeper angles. Pegasus solves that by skipping the pad entirely: Northrop Grumman’s modified L-1011 aircraft, named Stargazer, carries the rocket aloft and releases it in midair, letting it start its climb from tens of thousands of feet and from a location chosen to match Swift’s path.
A first, if it works
NASA has been careful about what it is claiming. By the agency’s own description, a successful boost would be the first time a commercial robotic spacecraft captures a government satellite that is uncrewed, or not originally designed to be serviced in space. That qualifier matters. Robotic servicing has been demonstrated before on craft fitted with docking targets or otherwise built to be approached; Swift was not. LINK, an 880-pound spacecraft that grips its target with three robotic arms, will have to dock with an observatory that was never built to cooperate with a visiting robot, over a maneuver NASA expects to unfold across several months.
It is also, for now, an attempt rather than a sure thing, and the schedule remains soft. NASA has described the launch as targeted for later in June while cautioning that shifting solar activity could still move the timing. Whether the boost works or not, it is a test of a capability NASA says it will need as satellites pile up in orbit and more of them, someday, may be worth saving rather than discarding.
A bowhead whale calf born the year Thomas Jefferson took office could, in principle, still be alive today, drifting under the Arctic ice with a harpoon point from the 1840s lodged in its blubber. The species (Balaena mysticetus) is the longest-lived mammal on Earth, with peer-reviewed lifespan estimates exceeding 200 years, and biologists who have spent the last decade reading its genome think they are beginning to understand why a 60-tonne animal with roughly a thousand times more cells than a human almost never dies of cancer.
The latest work, published in Nature in October 2025, points to an unusually abundant DNA repair protein that lets bowhead cells fix double-strand breaks with both speed and accuracy. In a 200-year-old whale, that precision may be the difference between a body and a tumour.
The 115-year-old whale with a Victorian harpoon
The hardest evidence for bowhead longevity is physical. In May 2007, Iñupiat hunters off Utqiaġvik (then Barrow), Alaska, landed a 49-foot male bowhead with a rusted metal fragment buried in its neck. The shard was identified by John Bockstoce of the New Bedford Whaling Museum as part of a bomb lance patented in 1879 and manufactured only between roughly 1879 and 1885. The whale had been swimming around with that fragment lodged in its shoulder for at least 115 years.
Other methods agree. Aspartic acid racemisation in eye lens crystallins — proteins laid down at birth and never replaced — has yielded age estimates for individual bowheads exceeding two centuries. The technique has its critics, but the Victorian harpoons keep turning up. Multiple bowheads landed in recent decades have carried stone or iron weapon fragments older than electric light, a pattern documented in the comparative-aging literature.
Why size should kill them, but doesn’t
A bowhead has roughly a thousand times as many cells as a person, and each cell divides over a lifespan ten times longer than ours. Every cell division is a chance for a copying error. Every error is a chance for cancer. By the simplest arithmetic, every bowhead should be riddled with tumours before its first century.
This is Peto’s paradox, which describes the observation that cancer rates do not scale with body size across species. Mice get cancer. Elephants mostly don’t. Whales almost never do. The Rochester team frames the puzzle directly in their Nature paper: a bowhead has vastly more cells than a human but is not highly cancer-prone. Something about being big and long-lived has forced evolution to invent better defences, and each large animal seems to have invented its own.
Elephants carry multiple copies of the tumour suppressor gene TP53, where humans carry one. Naked mole rats appear to have cellular mechanisms that suppress tumour growth. The bowhead, it now seems, took a different route entirely.
The CIRBP discovery
In October 2025, a team led by Vera Gorbunova and Andrei Seluanov at the University of Rochester, with first authors Denis Firsanov and Max Zacher and collaborators in Alaska, the UK and Korea, published findings that pointed at a single protein doing a lot of the work. The protein is called CIRBP — cold-inducible RNA-binding protein — and bowheads make it at roughly 100 times the level seen in other mammals. As Gorbunova explained in the University of Rochester announcement, several DNA repair proteins were elevated in bowhead cells, but CIRBP stood out because it was present at 100-fold higher levels than in other species.
What CIRBP does is striking. When a DNA strand breaks — and in any large animal, strands break constantly, from radiation, from oxidative stress, from the simple mechanics of replication — the cell can repair the damage in two ways. The fast pathway, called non-homologous end joining, slaps the loose ends back together and often loses or scrambles a few letters in the process. The slower one, homologous recombination, uses an intact copy of the sequence as a template and rebuilds the break with high fidelity. The Rochester team found that bowhead CIRBP boosts both pathways — making end joining more accurate and homologous recombination more efficient — so breaks get repaired faithfully rather than papered over with errors.
The result, in laboratory tests on bowhead fibroblasts, is a cell that fixes double-strand breaks with what the Rochester group described as remarkable precision and lower mutation rates than cells of other mammals. When the team inserted bowhead CIRBP into human cells, DNA repair improved. When they did the same to fruit flies, the flies lived longer and tolerated radiation better. Heidi Ledford’s news coverage in Nature framed the result as a possible mechanism for the species’ near-immunity to cancer.
What “unique gene variants” actually means here
The phrase “unique gene variants” gets thrown around loosely in popular coverage, so it is worth being careful. CIRBP itself is not a bowhead invention. Humans have it. Mice have it. Almost all mammals do. What is unusual in the bowhead is the expression level — how much of the protein the cell actually produces — along with a small number of sequence differences and a second protein, RPA2, also present at high levels. The Rochester team also flagged differences in genes governing the cell cycle and tumour suppression, though the headline finding is the CIRBP-driven repair quality.
The work has not yet been replicated by independent groups, and the leap from bowhead cells showing efficient DNA repair in laboratory conditions to this explaining bowhead longevity of up to 200 years remains an interpretation rather than a settled fact. But it is a plausible interpretation, and it lines up with the 2015 comparative genomics work that had already flagged the bowhead’s DNA damage response as unusually robust.
The cold matters
One of the more elegant aspects of the CIRBP story is that the protein’s name describes its trigger. Cold-inducible RNA-binding protein is produced in larger quantities when a cell’s temperature drops below normal. Human cells make a little more of it during mild hypothermia. Bowhead cells, sitting at the core temperature of an animal that spends its entire life in water hovering near freezing, appear to make it constantly and in vast quantities. The Rochester group showed that simply chilling cells by a few degrees ramps up CIRBP production.
This raises an awkward and interesting question: is bowhead longevity partly a side effect of living in the Arctic? If so, the warming of polar seas is not just a habitat problem for the species but potentially a longevity problem. A bowhead in 2-degree water may not produce CIRBP the way its great-grandparents did in water that was routinely colder. No one has measured this yet. It is a hypothesis hanging in the air, waiting for someone to test it.
What this might mean for humans — and what it almost certainly won’t
Coverage of the Rochester paper has been breathless. Smithsonian magazine framed it as a possible route to extending healthy human lifespan, and the Rochester press materials openly floated the idea that lifestyle interventions like cold exposure might modestly boost human CIRBP production.
The honest answer is that nobody knows whether a CIRBP-based therapy could extend human lifespan, and there are good reasons for caution. Human cells already make CIRBP. Upregulating it in a person is not the same as engineering a whale-grade repair system, because the repair pathway depends on dozens of other proteins working in concert, and on cellular machinery that may not respond to extra CIRBP the way a bowhead cell does. Gorbunova herself has been clear that the immediate medical target is cancer resistance, not radical life extension.
Still, the gap between species is provocative. A human has a median lifespan around 80 years. A bowhead, more than 200. The variation across mammals is enormous, and it is increasingly clear that much of it is written in genes that govern how cells respond to damage rather than in any single gene responsible for aging.
The whale that watched the steamship era
Consider what a 200-year-old bowhead has lived through. The oldest individuals swimming today were calves when the Crimean War ended. They were juveniles during the American Civil War. They survived the commercial whaling fleets that nearly wiped them out — the Alaska Department of Fish and Game estimates the global bowhead population collapsed from around 50,000 to fewer than 3,000 by the early twentieth century — and they have lived to see their species protected, the Bering-Chukchi-Beaufort stock climbing back above 12,000 in the most recent survey.
A bowhead born in 1820 has, if it is still alive, seen the Arctic ice retreat in ways no previous generation of its species ever experienced. The same DNA repair that lets it reach 200 also means its cells carry an unusually faithful record of two centuries of environmental change — a kind of biological logbook written in the slow, accurate hand of an unusually well-resourced repair crew.
What comes next
The Rochester team and collaborators are now trying to engineer the bowhead version of CIRBP into mouse cells and see whether it produces the same repair fidelity in a different cellular background. If it does, the next step is small-animal lifespan studies, which take years. If those work, human trials are at least a decade off, and probably longer.
In the meantime, the whales themselves remain the experiment. The Iñupiat subsistence hunt, monitored by the North Slope Borough and the International Whaling Commission, continues to produce the occasional carcass with an old harpoon point or a lens crystallin that aspartic acid racemisation places in the eighteenth century. Each one is a data point in an experiment that started before the Napoleonic Wars and has not yet ended.
Somewhere under the ice off Point Barrow tonight, an animal is exhaling through a blowhole it first used when Andrew Jackson was alive. Its cells are quietly correcting the damage of another day. The harpoon, if there is one, is older than the lightbulb.
In 2002 two astronomers at Johns Hopkins University, Karl Glazebrook and Ivan Baldry, averaged the light of more than 200,000 galaxies and first announced that the collective color of the universe was a pale turquoise. Within weeks they took it back. A mistake in their color software had skewed the result, and the corrected answer was a milky off-white that the public soon nicknamed cosmic latte.
The color was never the point of the research. It started as a footnote to a serious survey of the light galaxies give off, and it became, briefly, one of the most talked-about findings in astronomy. Newspapers ran with the idea that the cosmos had a single average shade, and home decorators took an interest in a color drawn from every galaxy at once. The retraction, when it came, was almost as widely reported as the original claim.
How you average the color of the universe
The two were studying the cosmic spectrum, built from the Australian Two-degree Field Galaxy Redshift Survey, which had measured the spectra of more than 200,000 galaxies. By combining many tens of thousands of those spectra, they produced a single volume-averaged, luminosity-weighted result, in effect the average color of light emitted by the nearby universe. Their real aim was to read the history of star formation out of that combined light, since young blue stars and old red stars leave different fingerprints in it.
That blend is why the spectrum landed where it did. It is a mixture of young, hot, blue populations and old, cool, red ones, and the average of the two is neither strongly blue nor strongly red. The paper noted that the resulting color is not the kind a single glowing object would produce, which was the first hint that turning a spectrum into a single perceived shade is less straightforward than it sounds.
On a whim, Glazebrook decided to work out what that spectrum would look like to a human eye, using the standard color-matching functions that describe how people perceive different wavelengths. Integrating the cosmic spectrum through those functions returned a blue-green that the team announced as “a few percent greener than pale turquoise.” Announced at a January 2002 meeting, the idea that the universe had an average color, and that the color was greenish, caught the public imagination at once.
The white-point mistake
Not long after the turquoise claim made news, Mark Fairchild of the Munsell Color Science Laboratory at the Rochester Institute of Technology got in touch. Fairchild had, almost literally, written the book on the subject, the 1998 text “Color Appearance Models.” He and his colleagues found that the free software the astronomers had used had set its “white point” incorrectly. The white point is the light that the eye treats as plain white, and it shifts with surroundings, a little yellow under household bulbs, a little blue on some screens. The wrong setting had tinted the whole result green.
A color scientist at the same laboratory, Francis Imai, put the problem in everyday terms. “Tungsten lights, for example, can make the white point a little bit yellowish,” he explained. “Some monitors have a bluish white point. Your visual system tries to adapt, so that you assume that color is white, but it’s actually yellowish or bluish.” The freeware had effectively chosen the wrong reference for white, and the error pulled the whole result toward green.
When the white point was fixed for someone viewing the light in a dark room, the color came out beige. Adjusted for daylight it turned faintly red, and for indoor lighting it shifted blue. Baldry made the dependence vivid: “if you look at all the light in the universe from a room that has a red neon light, then it may appear turquoise,” he said. “But that’s not a standard perspective.”
Glazebrook did not hide from the mistake. “It’s our fault for not taking the color science seriously enough,” he said. “I’m very embarrassed. I don’t like being wrong, but once I found out I was, I knew I had to get the word out.”
The astronomers posted their spectrum data online and invited anyone to recompute the color. From a flurry of suggested names, cosmic latte stuck, chosen for the milky off-white shade.
What the color actually means
The “color of the universe” is not a property of space the way temperature or density is. It is a perceptual quantity, and the turquoise blunder showed exactly why that matters: the same spectrum reads as beige, red or blue depending only on the lighting you imagine yourself standing in. There is no neutral place to view all the galaxies from, so there is no single true color, only a color for a stated set of assumptions.
It is also an average, luminosity-weighted across one survey’s galaxies at modest distances, not a tally of every photon the cosmos has ever released. The galaxies in the 2dF sample are relatively nearby in cosmic terms, so the figure describes the light of the present-day universe rather than the redder glow of its distant, earlier reaches. And it was, by the authors’ own account, a lighthearted aside to a paper about how fast stars have formed across cosmic time. The beige is real enough, but it describes a calculation and a convention, not a wall you could paint to match the sky.
The naming, too, was informal. Cosmic latte was a popular choice rather than a scientific term, and it appears in galleries of color values more often than in the astronomical literature.
A beige worth keeping
What lingers is less the color than the correction. Two researchers made a vivid, widely repeated claim, learned within weeks that they had got it wrong, and said so in plain language rather than letting the prettier answer stand.
The universe, averaged and viewed in the dark, is roughly the color of milky coffee. The more durable lesson is in how quickly that beige replaced the turquoise once someone checked the color science.
The quote has a clean, precise quality that most observations about fear do not. Reid Wiseman posted it to Twitter in October 2014, a few days after completing the first spacewalk of Expedition 41 aboard the International Space Station: “While #spacewalking I realized something: I used to think I was scared of heights but now I know I was just scared of gravity.”
He was floating outside the station at roughly 400 kilometres above Earth, working with European Space Agency astronaut Alexander Gerst for six hours and thirteen minutes on cooling pump relocation and external robotics work. The spacewalk was conducted on 7 October 2014. At that altitude, with nothing below and the planet curving away at the edge of peripheral vision, Wiseman noticed the absence of something he had expected to find: the fear he thought he had.
The quote has resurfaced in the months around the Artemis II mission, for which Wiseman serves as commander. On 1 April 2026, he launched as part of the first crewed mission to the Moon since Gene Cernan commanded Apollo 17 in December 1972. The insight is more than a decade old. It reads differently now, from a trajectory that takes Wiseman and his crew past the Moon and back.
The distinction itself
The difference between fearing heights and fearing gravity sounds like wordplay until you hold it carefully. Both terms are commonly used to describe the same cluster of behaviour: the discomfort on balconies, the reluctance to stand near the edge of a cliff, the particular unease of a glass floor above a long drop. The behaviours are consistent. What Wiseman is questioning is the diagnosis.
A fear of heights, understood as a response to elevation and exposure, locates the problem in the altitude. But what most people are responding to in those situations is not the height in any abstract sense. It is the fall. It is gravity’s action on the body in the event of a slip. It is the trajectory the body would follow if the ledge gave way, and the irreversibility of what that trajectory leads to. The height is the setup. Gravity is the mechanism.
In microgravity, the mechanism has been removed. There is no directional pull toward a surface that recedes infinitely. There is no trajectory the body would follow if it lost contact with the station. An astronaut outside the ISS is not high in the sense that a person on a balcony is high, because the relationship between position and consequence that produces the fear of heights has been cancelled. Wiseman was in the most extreme version of elevation a human can occupy, and the thing he expected to feel was not there.
He worked backward from that absence to what it meant. The fear he had carried was not of heights. It was of gravity.
What reclassification does
This kind of revision is harder than it sounds. Most people who believe they fear heights have arranged something around that belief: the avoidance patterns, the decisions about which windows to stand near, the self-description offered when explaining a refusal. The label has become load-bearing. It organises behaviour.
What Wiseman found was that the label was describing a proxy rather than the cause. The altitude was consistently present when the fear appeared, so the altitude got named. But the consistent presence of altitude was itself a consistent presence of gravity’s consequences. The two variables had never been separated. Spacewalking separated them.
The implication is not that people who fear heights are wrong about their fear. The fear is real and the behaviour it produces is real. What shifts is the precision of the account. Someone who understands their discomfort as being about gravity’s consequences, rather than about altitude per se, has a different description of where the sensitivity lives. That description may or may not change anything practical about what they avoid. But it is more accurate, and accuracy in these matters has a certain value even when it is not immediately actionable.
Wiseman did not overcome a fear. He reclassified one. The distinction matters because overcoming implies confronting the same situation repeatedly until the response diminishes. Reclassification involves recognising that the situation you have been avoiding is not quite the thing you thought it was.
Twelve years and a lunar trajectory
Wiseman is 50 years old, a retired United States Navy captain and test pilot. As of April 2026, he is the commander of a mission carrying him further from Earth than any human since the Apollo programme. His crew includes NASA astronauts Christina Koch and Victor Glover and Canadian Space Agency astronaut Jeremy Hansen.
Artemis II is a circumlunar flyby, not a landing. The crew will loop around the Moon and return. The distances involved, the communication delays, the duration, make any analogy between this and an ISS spacewalk look quite small. The environment Wiseman occupied in 2014 was far more controlled than the one he entered in April 2026.
The 2014 insight does not become more or less true in this context. It is not predictive of how Wiseman or anyone else experiences the Artemis II mission specifically. But the Artemis II frame does something to the quote: it makes the scale of what he has learned to not be afraid of more visible. The man who discovered in 2014 that his fear was of gravity, not altitude, is now travelling to a distance at which Earth’s gravity is a much smaller presence.
There is a version of this story that treats it as an arc of overcoming, each mission more extreme than the last, the fear perpetually shrinking. That is not what Wiseman is describing. He is describing a single clarification, arrived at empirically, during six hours of work outside a space station. The rest followed from that, or did not.
The question the observation leaves open
Wiseman’s observation raises a question that has no clean answer: when we say we fear something, how often are we naming the right thing?
The fear of heights is a durable shorthand, but it collapses a set of variables that could in principle be separated. The height. The exposure. The lack of handholds. The awareness of what would happen if the body fell. The mental simulation of the fall itself. These are not identical. They would, if separated, suggest different things about where the sensitivity lives. Most people never encounter the conditions that would separate them. Wiseman did.
There is no general lesson here that applies predictably from person to person. What seems clear is that the precision of Wiseman’s account is a function of the specificity of the test. An environment that removed gravity as a variable allowed him to see that gravity was the variable that mattered. Ordinary environments, which almost always include gravity as a background constant, do not offer the same diagnostic clarity.
He has not said whether the reclassification changed anything else. He has said that he now knows which word to use.
The fireball was visible in broad daylight. Roughly 200 witnesses across Georgia and South Carolina saw it streak across the sky that afternoon, bright enough that more than 160 of them reported it to the American Meteor Society within a few hours, and bright enough to be picked up by an orbiting satellite. The object, by NASA’s initial estimate, had been roughly three feet in diameter and weighed over 2,000 pounds when it hit the upper atmosphere. Most of that mass burned up as friction and pressure decelerated it through increasingly dense layers of air. Approximately 27 miles above Wake Forest, Georgia, the remaining bulk exploded with energy equivalent to roughly 20 tons of TNT, sending multiple sonic booms across the region. One specific piece, no larger than a cherry tomato, continued on a trajectory that intersected with the roof of a single-family home in McDonough. The homeowner — who has chosen to remain anonymous through the entire subsequent media cycle — was inside at the time. He later described the sound to local news as something like a gunshot.
According to Astronomy.com’s coverage of the analysis by University of Georgia researchers, the homeowner subsequently provided fragments of the rock to Scott Harris, a planetary geologist in UGA’s Department of Geology, who used optical microscopy and electron microscopy on approximately 23 grams of recovered material. Harris’s analysis identified the rock as an ordinary chondrite with low metal content (L) — one of the most common types of stony meteorite, but also one of the oldest classes of material that survives intact from the early solar system. The age determination, based on standard radiometric methods used in meteoritics, placed the rock’s formation at approximately 4.56 billion years ago. Earth itself, by the best current estimates, formed approximately 4.54 billion years ago. The piece of rock that had just punched through the homeowner’s living room ceiling was, by a margin of about 20 million years, older than the planet he had been standing on his entire life.
What 20 million years older actually means
The arithmetic is small in proportional terms — 20 million years is roughly 0.4 percent of Earth’s age, less than half of one percent. The intuitive impact is substantially larger. For the entire history of the human species, every solid object any human being has ever stood on, picked up, lived inside, or built anything from has been, in the deepest sense, younger than the planet. The crust of Earth is younger than the planet’s core; the rocks of the oldest mountains are younger than the formation of the planet; even the meteor showers that hit the early Earth and contributed to its current composition were already part of the solar system’s general inventory at the time the planet was assembled. To be in physical contact with a piece of material that was floating around the early solar system before Earth had finished accreting is to be touching something that exists in a temporal category most humans never encounter and most physical objects available to ordinary experience do not occupy.
Where the rock came from
Per CNN’s coverage of Harris’s analysis, the McDonough Meteorite’s specific origin can be traced with some confidence. The rock belongs to a family of ordinary chondrites that originated in the main asteroid belt between Mars and Jupiter — the same region from which most meteorites that reach Earth’s surface ultimately come. Harris’s analysis suggests that the parent body of the McDonough fragment was substantially larger than the rock that eventually hit the Georgia house, and that the parent body broke up in a major collision approximately 470 million years ago. The fragments from that collision were scattered across the asteroid belt, with some pieces eventually perturbed by gravitational interactions into Earth-crossing orbits. The specific piece that became the McDonough Meteorite had been drifting through space on such an Earth-crossing trajectory for hundreds of millions of years before the geometry finally aligned for it to intersect with Earth’s orbit at the specific moment on 26 June 2025 when the planet was passing through its path.
The probability of any specific Earth-crossing fragment hitting any specific house at any specific time is, of course, vanishingly small. Approximately 17,000 meteorites of various sizes reach Earth’s surface each year, by current estimates; the great majority fall on oceans, ice fields, deserts, and uninhabited terrain. House strikes are rare enough that individual cases become news. The 1954 Hodges meteorite in Alabama famously struck Ann Hodges directly while she was napping on her couch — the only well-documented case in modern history of a human being physically struck by a meteorite. The McDonough homeowner missed becoming the second such case by approximately 14 feet.
What the impact actually felt like
As reported by Gizmodo’s coverage of Harris’s reconstruction of the impact, the homeowner heard three distinct sounds in essentially the same instant. Harris explained the acoustic structure: “One was the collision with his roof, one was a tiny cone of a sonic boom, and a third was it impacting the floor all in the same moment. There was enough energy when it hit the floor that it pulverised part of the material down to literal dust fragments.” Months after the impact, the homeowner was reportedly still finding specks of meteoritic dust scattered around his living room. The cherry-tomato-sized fragment that punched through the roof had been travelling fast enough at the moment of impact that part of it was reduced to powder on contact with the floor — even after losing the vast majority of its initial atmospheric-entry energy during the explosive disintegration at 27 miles altitude.
What happens to the rock now
Per CBS News’s coverage of the broader research plans, Harris and his collaborators at Arizona State University have submitted the name “McDonough Meteorite” to the Meteoritical Society’s Nomenclature Committee for formal inclusion in the Meteorite Bulletin — the standard catalogue of confirmed meteorite falls. Harris noted that this is the 27th meteorite ever recovered in Georgia and the sixth where the fall was actually witnessed. The frequency of such events appears to be increasing in the modern era of dense surveillance: “This is something that used to be expected once every few decades and not multiple times within 20 years.” A paper detailing the meteorite’s full composition and atmospheric entry dynamics is being prepared, and fragments of the rock will eventually be displayed at the Tellus Science Museum in Cartersville, Georgia.
The homeowner, by all available indications, has been left to deal with the practical questions on his own — including whether his standard homeowner’s insurance policy covers damage from a space rock that predates the planet on which the house is built. The deeper feature of the event, beyond the household consequences, is what the rock itself represents. A fragment of the solar system older than the planet you are standing on, travelling on a trajectory that has been gradually converging with Earth’s orbit for hundreds of millions of years, finally arrives, traverses the atmosphere in a few seconds at hypersonic speed, breaks apart, and one specific piece reaches the ground in a particular spot that happens to have a house on it. The probability is astronomically small. The probability, distributed across enough houses and enough centuries, is not zero. On the afternoon of 26 June 2025, the geometry came together for one specific house in McDonough, Georgia. The rock that hit it was older than Earth.
Take two pairs of entangled photons that were generated separately and never interacted. Perform a single joint measurement on one photon from each pair, and the two leftover photons — strangers until that instant — come out entangled with each other. Boeing has now pulled off that maneuver, called entanglement swapping, on a compact payload built to survive a rocket launch rather than sit on a temperature-controlled optical bench.
The company announced on June 18 that its Q4S satellite system demonstrated high-fidelity entanglement swapping during ground testing, then passed environmental qualification and moved into final spacecraft integration ahead of a planned 2027 launch and a one-year orbital demonstration. According to SpaceNews, Boeing ran the swapping tests for more than a year before beginning integration. The question the mission exists to answer is not whether the physics works — it does — but whether it survives the constraints of a spacecraft: power budgets measured in watts, thermal swings of hundreds of degrees, and vibration loads that would shatter a laboratory optical bench.
What Entanglement Swapping actually does
Entanglement swapping is the protocol that makes a real quantum network possible, and the mechanics are worth slowing down on. Two pairs of entangled photons are generated independently. A joint measurement is then performed on one photon from each pair, and the remaining two photons — which never interacted — become entangled with each other.
The effect is, in essence, quantum teleportation of entanglement itself. Without it, the maximum distance over which entangled photons can be reliably shared is limited by fiber loss and atmospheric attenuation, capping any terrestrial quantum link at a few hundred kilometers. With swapping, those links can be chained, node to node, into something that spans a continent.
The theoretical foundation traces back decades. A foundational Nature Photonics review identified photonic systems as a natural carrier for quantum information, owing to their high transmission speed and low noise. Photons do not decohere easily in transit. They simply get lost.
That loss is the engineering problem. Repeater architectures built on entanglement swapping are the proposed solution, with a review of quantum communication in Nature Photonics framing such repeaters as the central architectural choice for scaling quantum networks beyond metropolitan distances.
Why space changes the calculus
Fiber attenuates photons exponentially with distance. Vacuum does not. A satellite in low Earth orbit looking down through the atmosphere only has to push photons through roughly 10 kilometers of dense air at each end of the link — a far gentler optical path than 1,000 kilometers of glass.
This is the geometric argument that has driven nearly every serious proposal for global quantum networking toward space. China’s Micius satellite demonstrated satellite-to-ground quantum key distribution in 2017. Boeing’s Q4S goes a step further by aiming to test the swapping protocol itself in orbit, which is what a true quantum repeater node would need to perform.
According to Boeing, the Q4S mission is intended to test quantum networking protocols under real space-mission constraints, moving beyond laboratory demonstrations.
The qualification matters. Laboratory demonstrations of entanglement swapping have existed for years. None have flown.
The engineering challenge of flight hardware
Single-photon detectors typically need cryogenic cooling. Entangled photon sources require precise temperature control and vibration isolation. Optical alignment tolerances are measured in micrometers. Compressing all of that onto a spacecraft bus is the kind of problem that consumes engineering careers — and Boeing’s partner HRL Laboratories has described fitting the core capability into a roughly briefcase-sized, 15-kilogram space-grade subassembly.
Boeing’s earlier work points toward how the company built toward this moment. NASA, the University of Illinois and Boeing had already shown that a payload in orbit could generate pairs of photons that remained entangled in microgravity. That experiment validated the source. Q4S now has to validate the network primitive built on top of it.
The broader race for quantum infrastructure
Boeing is moving in a field that is starting to attract serious commercial interest beyond research institutions. Cisco has publicly identified quantum networking as a strategic direction for its long-term roadmap, signaling that the traditional networking industry now treats this as more than a physics curiosity.
Recent ground-based progress has accelerated the timeline. A team at Heriot-Watt University reported teleporting entanglement across two linked quantum networks in late 2025, joining two smaller networks into a single reconfigurable eight-user system and showing that a network-of-networks architecture works at lab scale.
The geopolitical stakes are difficult to ignore. China has invested heavily in quantum communication infrastructure. The European Union has funded a quantum communication initiative aimed at deploying a secure network across member states. The United States has been slower to commit at the federal level, leaving room for companies like Boeing to define what American space-based quantum capability looks like.
What Q4S will actually prove
The on-orbit mission has a narrow scientific goal: demonstrate that the entanglement-swapping protocol Boeing has tested on the ground performs at acceptable fidelity when subjected to launch, orbital thermal cycling, radiation exposure and the mechanical realities of a satellite bus.
Fidelity is the figure of merit. Quantum protocols degrade gracefully — entanglement is not binary — and the question is whether enough quality survives the trip to support useful applications. Boeing has said it will share technical results for peer review, a notable choice for a payload funded entirely by company research-and-development money rather than a government contract that would typically mandate disclosure.
That choice signals where Boeing thinks the value lies. Q4S is not a product. It is a credentialing exercise — a way to show that the company can build flight-qualified quantum hardware in a field where the customer base of the future likely includes national security agencies, financial institutions, and whoever ends up operating the backbone of any quantum internet.
The 2027 context
The Q4S launch will arrive in a year already crowded with significant space milestones. Commercial cargo programs are maturing — China has shortlisted four commercial launch providers for its low-cost Qingzhou cargo spacecraft, whose first full-scale mission is tentatively scheduled for January 2027. Artemis hardware is moving toward crewed lunar flight. Multiple commercial space stations are targeting initial elements.
Against that backdrop, a roughly briefcase-sized payload doing physics experiments may sound modest. The implications are not. If quantum networking matures into infrastructure, it would reshape the security architecture of global communications, enable distributed quantum computing across continents, and provide timing references precise enough to redefine geodesy and navigation.
What separates a footnote from a foundation comes down to the orbital numbers: the fidelity the swapping protocol holds in space against its ground-test baseline, whether the payload survives the full mission rather than failing on a single component, and whether Boeing follows Q4S with a constellation or a partnership once governments decide they need flight-qualified quantum nodes. The company has not said what comes after. Funding the work itself and publishing the results suggests it is positioning for the contracts that follow.
For now, the hardware sits in a clean room, two photon sources and a measurement stage compressed into something that weighs about as much as a large suitcase, waiting for a ride to orbit. The photons it will entangle up there have never met and never will, in any ordinary sense. Whether that fragile connection holds through launch vibration, vacuum, and the long thermal day-night cycle of low Earth orbit is the one thing a workbench can never tell you — and the only thing the 2027 flight is built to find out.
At its record top speed, NASA’s Parker Solar Probe could cover the distance to the Moon in a little over half an hour. The same trip took the Apollo 11 crew roughly three days, launching on 16 July 1969 and slipping into lunar orbit on 19 July. Put the two numbers side by side and the gap is startling, but it is worth understanding what the comparison does and does not show.
The arithmetic is simple. The Moon sits about 384,400 kilometres away on average. At Parker’s record speed of 692,000 kilometres per hour, a craft moving in a straight line would clear that gap in about 33 minutes.
The fastest thing we have ever built
Parker Solar Probe holds the record for the fastest human-made object. On 24 December 2024, during its closest pass to the Sun, it reached 692,000 kilometres per hour, around 430,000 miles per hour, while skimming about 3.8 million miles above the solar surface. That figure is logged by Guinness World Records as the highest speed any spacecraft has ever reached.
It is a hard number to picture, which is why the Moon comparison is useful. Half an hour to the Moon is the kind of figure that makes the speed real.
Why it goes so fast
Parker does not reach that speed by burning fuel. It reaches it by falling.
The probe is on a long, stretched orbit that carries it repeatedly close to the Sun, and a series of flybys past Venus has tightened that orbit over the years. Each time it drops toward the Sun, the Sun’s gravity pulls it faster, the way a skateboarder gains speed dropping into a bowl. The 692,000 figure is the speed at the very bottom of that drop, at the closest point to the Sun. For most of its orbit it travels far slower. Parker has made the dive repeatedly, edging nearer with each Venus flyby, and it has reset its own speed record more than once as its path has tightened.
So the record speed is a momentary peak, reached at the Sun and pointed nowhere near the Moon.
The half-hour trip is a thought experiment, not a route anyone could fly.
Why the comparison is not quite fair to Apollo
The other half of the contrast deserves the same care, because Apollo 11 was not slow by accident.
The Saturn V could have pushed its crew toward the Moon harder than it did. It did not, because the goal was never raw speed. Apollo flew an efficient trajectory that spent fuel sparingly, and, crucially, it had to arrive slowly enough to be captured into orbit around the Moon rather than hurtle past it. A crewed ship that needs to stop at its destination has a very different job from a probe flung sunward with nothing to slow down for.
Read that way, the comparison is not old technology against new. It is a spacecraft accelerated by the Sun’s gravity against a crewed vehicle taking the careful, controlled path that keeping people alive demands.
What it actually shows
Held loosely, the number still earns its place. It is a vivid way to feel two things at once: how fast the Sun can throw an object that falls close enough, and how far away the Moon really is, given that a three-day journey once counted as one of the great feats of the century.
The exact figure shifts a little depending on which of Parker’s record-breaking passes you use and where the Moon is in its orbit, but the shape of it holds. Roughly half an hour, against three days.
The face you walked into the room with this morning was being read, by every person who looked at you, in approximately one hundred milliseconds — about a tenth of a second, faster than any conscious thought. The reading was not primarily about the features that took your DNA twenty years to assemble. The bones of your face, the colour of your eyes, the shape of your jaw, the symmetry of your features — these all contributed to the judgement, but they were not, on the available evidence, the only ingredients of it. A substantial part of how attractive any individual stranger perceived you to be, in the brief moment before they had any other information about you, was almost certainly whether you appeared to be having a good time. A genuine smile, animated expression, and the visible signs of a person enjoying being where they are produce, in the moment of first impression, attractiveness boosts that can rival what carefully selected clothing, makeup, or grooming can plausibly achieve.
According to a 2014 paper by Jessika Golle, Fred Mast, and Janek Lobmaier published in Cognition and Emotion titled “Something to smile about: The interrelationship between attractiveness and emotional expression”, the social effect of smiling on stranger-rated attractiveness has been replicated across enough independent studies and across enough cultural contexts to be considered one of the more robust findings in social psychology. Smiling faces are rated more attractive than neutral faces of the same person, across both sexes, across multiple age ranges, and across essentially every population that has been examined. The neuroimaging literature has identified the underlying mechanism: when human observers see a smiling face, the medial orbitofrontal cortex — a region of the brain involved in reward processing and the assignment of positive value to stimuli — shows a substantially stronger response than when the same observer sees a neutral face. The smiling face is, in a literal sense, registering in the observer’s brain as more rewarding to look at.
What kind of smile actually works
The specific type of smile matters substantially. As reported by a peer-reviewed PMC review of smile research and the broader perception literature, the most extensively studied distinction in the smile literature is between what researchers call Duchenne smiles and non-Duchenne smiles. The Duchenne smile, named after the 19th-century French neurologist Guillaume Duchenne who first characterised it, activates two distinct sets of facial muscles: the zygomatic major, which pulls up the corners of the mouth, and the orbicularis oculi, which creates the small crinkles around the outer corners of the eyes. Both muscle activations occur involuntarily when a person is genuinely amused or pleased. The non-Duchenne smile activates only the mouth corners and is the kind of polite, social smile produced on demand for photographs or to acknowledge a casual greeting.
The two types of smile are perceived very differently by observers, even though most people cannot articulate exactly what the difference is when they see one. Duchenne smiles register as genuine, warm, trustworthy, and substantially more attractive. Non-Duchenne smiles register as polite at best, performative or insincere at worst, and produce far smaller boosts to attractiveness ratings. The discrimination is automatic and is operating in the observer’s brain at the level of unconscious pattern recognition rather than conscious analysis. This is the underlying reason why the standard advice to “smile more in photos” often produces mediocre results in practice: the smiles are non-Duchenne. The viewer’s brain registers the difference even when the viewer cannot consciously identify what is wrong with the expression.
Why the effect is so large
Per a 2023 study in Evolutionary Human Sciences examining how smiling expressions interact with facial features in attractiveness judgements, the magnitude of the smiling-on-attractiveness effect is large enough to substantially affect outcomes in real-world social contexts. Across studies, happy facial expressions consistently produce the highest attractiveness ratings of any expression tested; angry expressions produce the lowest. The effect appears in both sexes but is particularly pronounced for ratings of female faces, where smiling can shift attractiveness ratings by amounts comparable to the difference between an averagely-rated face and a substantially above-average face. Even male facial masculinity — which interacts in complex ways with attractiveness preferences — is moderated by smiling: smiling reduces the avoidance response some women show toward strongly masculinised male faces.
The evolutionary and social explanation for this effect is, in the relevant literature, fairly direct. A smile is one of the few human expressions that is broadly cross-cultural — recognised across nearly every population that has been tested — and that consistently signals friendliness, low threat, willingness to engage, and positive emotional state. For an observer making rapid social judgements about whether to approach a stranger, the smile is a substantial piece of information: it indicates that the encounter is likely to go well, that the person on the other side is in a positive frame of mind, and that initial overtures will probably be reciprocated. The attractiveness boost is, in this account, the observer’s brain doing what brains do — bundling the information about the smile into a single quick-summary judgement about whether this stranger is worth engaging with. The summary judgement, in shorthand, comes back as “attractive.”
What the framework does not claim
Intellectual honesty requires noting the limits of the smiling-as-attractiveness effect. The research does not show that smiling can substitute for every other physical or social factor in attractiveness ratings. As covered by a 2025 review at All About Psychology on the broader smile-attractiveness literature, static physical features — facial structure, body shape, height, age, grooming — still contribute substantially to first-impression attractiveness ratings, and the smile effect is best understood as a substantial modifier of these underlying factors rather than as a replacement for them. A smiling face is more attractive than the same face not smiling. A smiling person is generally more approachable than a neutral one. Whether the smiling person is, in absolute terms, more attractive than a non-smiling person with substantially different underlying physical features is a question the research does not consistently answer.
What the research does support is a more specific and more practically useful claim. For any given person, in any given moment, the version of that person who appears to be genuinely enjoying themselves is going to be perceived as more attractive than the same person looking flat, distracted, or unhappy. The variable that each individual has the most direct control over, in the seconds before a first impression forms, is not their face or their body — those are mostly fixed at the moment of the encounter. The variable that is actually within their control is their expression, their visible emotional state, the degree to which they appear genuinely engaged with what is happening around them. The research consistently suggests that this variable, despite being the one most readily modifiable, is one of the larger factors that the observer’s brain weights when assigning the first-impression attractiveness rating. Looking like a person who is glad to be where they are turns out to be, by every available measure, a substantial part of what other people are responding to when they form their first impression of you.
The face you saw in the mirror this morning was not the face anyone else saw when they looked at you over breakfast. The two are related but not identical. Most human faces are slightly asymmetric — the eyes are at slightly different heights, the smile pulls more strongly on one side than the other, the hair parts on one side, freckles sit in specific positions that do not perfectly mirror across the vertical axis. When you look in a mirror, you see one version of your face, with left and right reversed relative to what other people see. When other people look at you, they see the opposite. You have, over the course of your life, become deeply familiar with one of these two versions. They have become familiar with the other. Neither version is wrong. Both are, however, surprisingly different from each other in ways that affect how each party perceives the same underlying face.
According to a 1977 paper by Theodore Mita, Marshall Dermer, and Jeffrey Knight published in the Journal of Personality and Social Psychology, the experimental setup that established this phenomenon was straightforward. The researchers took a frontal photograph of each of 37 women — undergraduate psychology students at the University of Wisconsin-Milwaukee — and printed each photograph twice: once as a true image (showing the face as a camera or another person would see it), and once as a mirror-flipped image (showing the face as the subject herself would see it in a bathroom mirror). The subjects were then shown the two versions side by side and asked which they preferred. The same pairs were shown to the subjects’ friends and lovers, who were asked the same question. The results, in both Study 1 and a subsequent Study 2, were striking and consistent. The subjects themselves preferred the mirror-flipped version of their own face. Their friends and lovers preferred the true version.
Why this happens
The mechanism underlying the effect is one of the better-established findings in social psychology: the mere-exposure effect, first proposed by the University of Michigan psychologist Robert Zajonc in 1968. As described in Gizmodo’s coverage of the mere-exposure mechanism behind the mirror preference, the mere-exposure effect is the phenomenon by which repeated exposure to a stimulus increases a person’s preference for that stimulus. The more often you encounter something, the more comfortable you become with it, and the more positively you evaluate it on subsequent encounters. The effect operates across an enormous range of contexts. People develop preferences for songs they have heard repeatedly, for faces they have seen frequently in public, for words that appear often in their reading, for foods they have eaten before. It does not require any explicit reasoning or evaluation; it works at the level of automatic familiarity processing.
The mirror-image preference is a particularly clean demonstration of the effect because the two stimuli being compared — the true image and the mirror image of the same person’s face — are objectively almost identical. The asymmetries are small. The lighting is the same. The expression is the same. The only difference is which side of the face is on which side of the photograph. The subject has, over the course of her life, spent many thousands of hours looking at one specific version. Her friends and lovers have spent many thousands of hours looking at the opposite version. When the two parties are then asked which version “looks more like” the person in question, or which version is more attractive, each party prefers the version they have been seeing all along. Neither is wrong. Each is simply applying the mere-exposure effect to the version of the face they have the most experience with.
What this implies about self-perception
Per the Mita et al. paper’s broader theoretical discussion, the finding has implications that extend well beyond the specific experimental setup. The general principle is that the version of yourself that you have internalised as canonical — the face you recognise as belonging to you, the face you adjust before going out, the face whose minor blemishes and asymmetries you have catalogued in detail — is a version that essentially nobody else has ever seen. Every photograph anyone else has ever taken of you. Every memory anyone else has formed of your face. Every impression you have made on a stranger across a room. All of these have been based on the opposite version of the face from the one you have been carrying around in your own head.
This is the underlying psychological mechanism behind the common experience of looking at a photograph of yourself and finding it subtly wrong. The face in the photograph is the true image — the version other people have been seeing all along — but it does not match the mirror image you have spent your life looking at. The mismatch produces the small but real sense of discomfort that most people experience when seeing themselves in photographs, particularly photographs they did not pose for and in which they had no opportunity to adjust their expression before the shutter clicked. The photographs are not unflattering. The photographs are accurate. Your face is mirroring back at you, several centimetres off from where you expected it to be.
The selfie problem
The modern environment has complicated the picture in ways that the Mita team could not have anticipated in 1977. As covered by Goody Feed’s analysis of how mirror-image preferences apply to selfie photography, most smartphone cameras now display selfies in the front-facing camera as a mirror image of the user — because that is what the user is accustomed to seeing — and then save the resulting photograph in one of two ways, depending on the phone model and the specific app being used. Some phones save the mirror-flipped version (the version the user saw in the viewfinder). Others save the true version (the version other people would see). The user is, accordingly, receiving inconsistent feedback about which version of their face is “really” them. A selfie that looked fine in the camera viewfinder may appear subtly wrong when reviewed afterwards, because the photo has been flipped to the true orientation. The same selfie viewed by a friend may look slightly strange to the user but entirely normal to the friend, who has been seeing that orientation of the face all along.
The broader implication, beyond the specific case of selfies, is that nearly every modern human is now navigating a world in which both versions of their own face are routinely available — the mirror version every morning in the bathroom, the true version in every photograph, video, and reflective non-mirror surface. The two versions are different enough that the brain registers the discrepancy. The version each person has learned to recognise as their own is, by all available evidence, the one that other people have not seen, except in the rare case of looking at the person directly while standing on the opposite side of a mirror. Everyone else has been seeing the other one. Your face, the one in your head, is essentially the reverse of the one the world has been looking at for as long as you have had it.
On 7 March 2025, a jury at the Old Bailey returned guilty verdicts against three Bulgarian nationals — Katrin Ivanova, Vanya Gaberova and Tihomir Ivanchev — for conspiring to spy for Russia from a seaside guesthouse in the English town of Great Yarmouth. Three more members of the same cell, including ringleader Orlin Roussev, had already pleaded guilty. In May, Justice Nicholas Hilliard handed down sentences totalling more than 50 years. The longest, for Roussev, was 10 years and 8 months.
The cell called themselves the Minions. Their handler, prosecutors told the court, was a fugitive Austrian executive believed to be living in Moscow under the protection of Russian intelligence: Jan Marsalek, the former chief operating officer of the collapsed German payments company Wirecard.
A guesthouse in Great Yarmouth
Great Yarmouth is a faded Norfolk resort on the North Sea coast, the kind of place that smells of vinegar and diesel. It is roughly 130 miles northeast of London. Roussev, a 47-year-old Bulgarian who had lived in the UK since 2009, ran his operation out of a former guesthouse there.
When Counter Terrorism Command officers raided the building in February 2023, they found what police described as a “treasure trove” of surveillance equipment. Cameras hidden in pens, neckties, sunglasses, a fake rock, a Coke bottle. One camera was sewn inside a plush Minion toy from the Despicable Me films. There were Wi-Fi jammers, GPS spoofers, eavesdropping devices, vehicle trackers. Roussev, in his Telegram messages, referred to the space as his “Indiana Jones warehouse.”
The Crown Prosecution Service called it “spying on an almost industrial scale” — one of the largest foreign intelligence operations ever prosecuted in a British court.
Six operations across five countries
The trial focused on six assignments the cell carried out between 2020 and 2023 in the UK, Germany, Austria, Spain and Montenegro. The targets were chosen with care.
The most prominent was Christo Grozev, the Bulgarian-born lead Russia investigator at Bellingcat, who had identified the GRU officers behind the 2018 Novichok poisoning of Sergei Skripal in Salisbury and the 2020 poisoning of Alexei Navalny. The cell followed Grozev from Vienna to a journalism conference in Valencia. In Telegram messages later read out in court, Roussev and Marsalek discussed robbing him, kidnapping him to Russia, or killing him outright.
“Learning only in retrospect that foreign agents have been monitoring my movements, communications and home, surveying my loved ones over an extended period — has been terrifying, disorientating and deeply destabilizing,” Grozev said in a statement read during the four-day sentencing hearing.
A second target was Roman Dobrokhotov, editor-in-chief of the Russian investigative outlet The Insider, who had worked with Grozev on the Skripal and Navalny cases. A third was Bergey Ryskaliyev, a former Kazakh politician living under asylum in the UK. A fourth was the Russian dissident Kirill Kachur. A fifth involved surveillance on Ukrainian soldiers being trained at the US Army’s Grafenwöhr base in Bavaria, with a view to tracking them after they returned to the front. A sixth involved a plan to drop fake pigs’ blood by drone on the Kazakh embassy in London to fake a protest — a piece of theatre intended to curry favour with Kazakh intelligence.
The Minions and their code names
The cell took its operational vocabulary from action movies. Roussev called himself Jackie Chan. His deputy, Biser Dzhambazov, was Mad Max — or, on lighter days, Jean-Claude Van Damme. The junior operatives were the Minions, after the yellow sidekicks who serve the supervillain Gru in the animated Despicable Me franchise. They were paid handsomely. Roussev received more than €200,000 to fund the activities. Sums of up to €1m were discussed in Telegram chats — a measure, prosecutors argued, of the value Moscow placed on the work.
The cell’s internal life was tangled. Dzhambazov, a 44-year-old who held down a day job as a medical courier while claiming to friends he was an Interpol officer, was in simultaneous relationships with two of the women he recruited. Katrin Ivanova, his long-term partner, worked as a laboratory assistant. Vanya Gaberova, 30, was a beautician who had left her partner Tihomir Ivanchev for Dzhambazov. When officers raided the addresses in February 2023, they found Dzhambazov in bed with Gaberova rather than at home with Ivanova.
Both women told the jury they had been deceived. The jury did not believe them. Justice Hilliard told Gaberova at sentencing she had known exactly what she was doing: “You found what you were doing exciting and glamorous, as demonstrated by the film you took of yourself wearing surveillance glasses in Montenegro.”
The fugitive in Moscow
The figure at the other end of Roussev’s Telegram channel was, according to prosecutors, Jan Marsalek — the 44-year-old former COO of Wirecard AG. Until June 2020, Marsalek had been one of the most powerful executives in European fintech. Wirecard was the only German technology company in the DAX index. Its market capitalisation briefly exceeded Deutsche Bank’s. Then €1.9 billion in claimed cash balances were found not to exist, and the company filed for insolvency. Marsalek boarded a private jet from Bad Vöslau airfield in Austria and disappeared. Investigators have placed him in Moscow ever since.
At the Old Bailey, prosecutors said Marsalek had acted as the go-between linking the Bulgarians to Russian intelligence services, principally the GRU. A selfie of Marsalek in a Russian military uniform was recovered from Roussev’s phone. The Telegram exchanges between the two men ran to thousands of messages, dealing with logistics, payments and targets. The exact technical claim — that the encrypted channel itself traced cleanly back to Marsalek in Moscow — is the prosecution’s account of what the digital evidence showed; it has not been independently audited outside the court process.
Marsalek remains on the Bundeskriminalamt’s wanted list and is the subject of an Interpol notice. He has not responded publicly to the espionage allegations. Wirecard’s former CEO Markus Braun, awaiting verdict in his Munich fraud trial, has denied wrongdoing and said he too was deceived by Marsalek.
Outsourced espionage
The Met’s Counter Terrorism Command described the case as a marker for a newer way of running intelligence work. Commander Dominic Murphy said it “highlights a relatively new phenomenon whereby espionage is being ‘outsourced’ by certain states.” Rather than deploying career officers under diplomatic cover — which became sharply harder after the mass expulsions that followed the Salisbury attack and the 2022 invasion of Ukraine — Russian services have been contracting out surveillance to non-Russian nationals who can move freely across European borders.
The Bulgarians had no formal training. Roussev had run a small signals-intelligence consultancy. Dzhambazov drove medical samples between hospitals. Gaberova ran a beauty salon in north London. None of them held a Russian passport. All of them could fly to Vienna, Valencia or Podgorica without raising flags. They learned tradecraft on the job — and, the trial showed, they were reasonably good at it. They tailed Grozev across two countries without being noticed. They sustained the operation for nearly three years.
The analysis published by War on the Rocks after the verdicts argued that the Minions case maps onto a wider pattern of Kremlin proxy operations across Europe since 2022 — arson attacks in Poland and the UK, parcel-bomb plots routed through DHL hubs, sabotage of rail lines in Germany. Cheap, deniable, run through intermediaries paid in cryptocurrency or cash.
What the messages contained
What made the trial unusual, beyond its scale, was how much of the operational chatter ended up in front of a jury. Roussev had archived almost everything. The court saw photographs of targets, video files, surveillance logs, and the day-to-day Telegram traffic between Roussev and his handler — banter about gadgets, gossip about the Minions’ love lives, and detailed operational planning.
In one exchange about Grozev, the two men discussed waiting for him at his Vienna apartment. In another, they speculated about whether they could lure him to a country with weaker rule of law. Marsalek, prosecutors said, signed off on payments and relayed instructions he attributed to Russian intelligence officers. When Roussev was arrested, he told police he had never done anything for any government. “No James Bond activity on my end, I guarantee you,” he said in interview. The Indiana Jones warehouse, and his messages calling himself Q, told a different story.
The sentences
On the morning of 12 May 2025, Justice Hilliard delivered sentences totalling more than 50 years across the six defendants. Roussev got 10 years and 8 months. Dzhambazov got 10 years and 2 months. Ivanova, 9 years and 8 months. Ivanchev, 8 years. Gaberova, after mitigation citing depression, panic disorder and the coercive nature of her relationship with Dzhambazov, was sentenced to 6 years, 8 months and 3 weeks. Ivan Stoyanov, the mixed martial arts fighter who had pleaded guilty early, received 5 years and 3 weeks. All six face deportation to Bulgaria after release.
Roussev was also ordered to pay a confiscation order of £180,768 in ill-gotten gains by 11 November 2025. The maximum sentence for conspiracy to spy under the Official Secrets Act is 14 years; the early pleas from Roussev, Dzhambazov and Stoyanov drew the standard credit. The mitigation arguments at the Old Bailey turned on whether the Bulgarians were ideological operatives or, as one defence barrister put it, “a world away from a classic spy.” The judge sided with the prosecution. “All were motivated by money,” he said.
The man who isn’t there
The trial answered the question of who the Minions were. It did not answer the question of who, exactly, gave Jan Marsalek his instructions, or what he was doing in Austria in the years before Wirecard collapsed, or how a fintech executive came to be running a surveillance network across the EU from a Moscow safe house. Those questions belong to investigators in Munich, Vienna and Berlin, and to the joint inquiries that outlets including Der Spiegel, ZDF and The Insider have continued to publish.
What the Old Bailey did establish, in admissible evidence, is that a former Norfolk guesthouse with mould on the walls and a Minion plushie on the shelf was, for nearly three years, a node in the European architecture of Russian intelligence. The cell was paid in euros. Its members shopped at Lidl. Its leader called himself Jackie Chan. And the man they reported to — the man whose face appears in a Russian military uniform on the ringleader’s phone — has, as of June 2026, still not been seen in public outside Russia since the evening of 19 June 2020, when he boarded a private jet in Bad Vöslau and vanished.
The six convicted Bulgarians began their sentences in May 2025. Marsalek’s room in Moscow, wherever it is, is still warm.
Somewhere beneath a mountain in southern Albania, in a region called Vromoner near the small town of Leskovik on the Greek border, a body of warm mineral water has been sitting in complete darkness for an indeterminate but very long time, with no human aware of its existence, no light reaching its surface, and no possibility of being entered by anyone except via a hundred-metre vertical drop through a limestone shaft. In 2021, a team of Czech speleologists became the first humans to make that drop. They were led by Marek Audy, a veteran Czech cave explorer who had spent several years investigating the underground hydrology of the Albanian-Greek mountain borderlands. The team had been tracking high-temperature springs and tall columns of steam that emerged from cracks in the surface limestone — the visible exhaust of a hidden hot-water system that had to be coming from somewhere below. Following the steam back to its source, the speleologists located a vertical shaft they later named Atmos, approximately a hundred metres deep, leading downward into a previously unmapped cavern. At the bottom of the shaft, in the darkness, was a thermal lake.
According to Radio Prague International’s coverage of the discovery and subsequent confirmation, the 2021 initial exploration produced a preliminary map of the cave system, drawn with the equipment the team had on hand at the time, that strongly suggested the underground lake was extraordinarily large. The team needed more precise instruments to confirm this. Mapping the full dimensions of a body of water in complete darkness, at the bottom of a hundred-metre vertical shaft, with limited working space and toxic hydrogen-sulphide air, required a mobile LIDAR scanner — a piece of equipment that the team did not own and could not initially afford. The Czech Neuron Foundation, a scientific funding organisation supported by private benefactors, provided approximately one million Czech crowns to purchase the necessary equipment. The expedition returned to the cave system in late 2024 with the LIDAR scanner and a team of 18 researchers. The measurements they took confirmed the lake’s full dimensions: 138.3 metres long, 42 metres wide, with a perimeter of 345 metres and a total volume of approximately 8,335 cubic metres of hot mineral water.
How big the lake actually is
The numbers translate into more intuitive comparisons that the team has used in subsequent communications. As reported by Euronews Albania’s coverage of the announcement, the lake’s volume is equivalent to approximately 3.5 Olympic-sized swimming pools. The cavern containing the lake is approximately three times the size of the main hall of the National Theatre in Prague. The previous record-holder for the world’s largest underground thermal lake was substantially smaller; the Lake Neuron measurements nearly double the previous record. The cavern is large enough that the LIDAR scanner had to be moved to multiple positions to capture the full geometry, and parts of the lake remain partially unexplored — the team plans to return with sonar equipment to measure the underwater portion of the cavern, which extends below the visible waterline and whose depth is not yet known.
The water itself is the unusual feature of the find. The lake is saturated with hydrogen sulphide — the gas responsible for the distinctive rotten-egg smell that hangs in the air throughout the cavern and that is toxic at high concentrations. The hydrogen sulphide, on contact with the atmosphere at the lake’s surface, oxidises into sulphuric acid. This is also why the cavern that contains the lake exists at all. Most caves in karst topography are formed by carbonic acid — slightly acidic rainwater dissolving limestone over millions of years. The Albanian cavern was formed by a fundamentally different process: sulphuric acid, generated by the oxidising hydrogen sulphide rising from the lake, has been gradually dissolving the limestone into soft gypsum from below. The technical term is sulphuric acid speleogenesis. It produces caves that are larger, more chemically active, and substantially more dangerous to explore than the carbonic-acid caves that dominate most karst regions.
What was down there before anyone got there
The lake is, in essential terms, an ecosystem operating in complete darkness, fuelled by chemistry rather than sunlight, sealed off from the surface biosphere for an unknown but presumably substantial period of time. As described in the Jerusalem Post’s coverage of the confirmed measurements, the researchers expect the lake to host extremophile microorganisms — bacteria and archaea adapted to high temperatures, no light, and high concentrations of hydrogen sulphide and dissolved minerals. Communities of this kind have been documented in other thermal and sulphurous cave systems, where they form the base of food webs that are entirely independent of photosynthesis, drawing their energy instead from chemical reactions between dissolved minerals and the surrounding water. Whether the Lake Neuron community contains microbial species new to science is a question the research team plans to investigate in subsequent expeditions.
The broader scientific significance of the find runs in several directions. The cave hydrology is connected, at some unknown depth, to the surface springs that originally led the speleologists to look for a hidden source. Mapping the connection between the underground lake and the surface water system will substantially improve hydrologists’ understanding of how groundwater moves through the karst terrain of the western Balkans. The presence of a previously unknown body of thermal water of this size raises questions about how many other comparable underground lakes might exist beneath similar mountain ranges around the world, where the geological conditions for sulphuric-acid speleogenesis are present but no surface evidence has yet been investigated. The Czech team’s discovery is, in technical terms, a sample of one — but it is a sample of one that is unusually large and that suggests the broader population may be larger than has previously been recognised.
The exploration itself
The story is also, separately, a story about contemporary speleology — a science that retains an unusually high proportion of physical risk and discovery for an activity that takes place in the 21st century. As covered by Monitor Magazine’s coverage of the team’s broader work, descending a hundred-metre vertical shaft in a karst cave system, breathing air contaminated with hydrogen sulphide, with no possibility of rescue if anything goes wrong, is not work that most scientific fields require their practitioners to do. The Czech speleological tradition, which has been mapping the underground of central and south-eastern Europe for several generations, sits at the intersection of mountaineering, scientific exploration, and a particular kind of physical patience that the broader scientific establishment has largely outgrown. Marek Audy and his team’s discovery of Lake Neuron is, by every available measure, the most significant single find of their careers. It is also a reminder that the planet still contains substantial unmapped territory, sitting in complete darkness beneath mountains that humans have lived alongside for thousands of years, waiting for someone to find a vertical shaft and decide to go down it.
Casey Harrell can now talk for twelve hours at a stretch. Two years ago, he could not talk at all. The ALS he was diagnosed with several years earlier had progressively destroyed the motor neurons controlling the muscles of his face and throat, leaving him with the symptoms doctors call dysarthria and, eventually, anarthria — the inability to speak intelligibly even with substantial effort. Harrell still knew what he wanted to say. The neural commands his brain was sending toward the muscles of his lips, tongue, jaw, and larynx were intact. The muscles themselves no longer responded. What the BrainGate clinical trial team at UC Davis did, in a July 2023 surgery that took several hours, was implant four small arrays of microelectrodes into the left precentral gyrus of Harrell’s brain — the region of the speech motor cortex that ordinarily sends those commands to those muscles. The arrays contain 256 electrodes between them. The electrodes record the firing patterns of individual neurons in real time. Machine-learning software running on a nearby computer translates the patterns into the words that Harrell, despite no longer being able to produce them with his mouth, is still trying to say.
According to UC Davis Health’s June 2026 announcement of the latest results, the system now operates at approximately 99 percent accuracy across a working vocabulary of 125,000 words — essentially every word an English-speaking adult would ordinarily need. Harrell’s average speaking rate via the device is approximately 56 words per minute, comparable to slow natural conversation. He has used the system at home, without researchers present, for more than 3,800 hours over a 400-day continuous-use period — averaging roughly five hours of daily use, with maximum single sessions of approximately twelve hours. He has produced more than 183,000 sentences and approximately two million words. His synthesised voice is modelled on recordings of his own pre-ALS voice. When he first heard it during the early trial phase, his own comment was: “It feels a lot like me.”
What changed between 2024 and 2026
The BrainGate team’s first publication on Harrell’s case, in the New England Journal of Medicine in August 2024, established that the basic approach worked — that brain activity recorded from the speech motor cortex could be decoded into fluent speech with high accuracy, in a setting with researchers present and a constrained working session. The June 2026 paper documents something fundamentally different: the same patient continuing to use the same system, but now in an independent home setting, operated by his own care team rather than by researchers, for thousands of hours over multiple years. As reported by ScienceAlert’s coverage of the latest results, this transition — from a proof-of-concept demonstration in a research lab to a sustained, independent communication tool used as part of daily life — is the genuinely novel achievement that the BrainGate team is reporting.
David Brandman, the UC Davis neurosurgeon who implanted the device and co-led the research, framed the shift directly: “For years, BCIs have been proof-of-concept devices that lived in highly controlled research labs. This work shows that we may have crossed a threshold, by empowering a person with paralysis to speak on his own terms.” The practical difference is substantial. A device that requires researchers to be physically present every time it is used can demonstrate scientific feasibility but cannot become part of a person’s actual life. A device that the patient’s regular home care team can connect every morning and disconnect every evening, with no specialist intervention required, can. Harrell has been doing the latter for two years.
What the system actually does
The fundamental insight underlying the BrainGate speech BCI is that the brain continues to send speech commands even after the muscles can no longer execute them. When Harrell attempts to speak, the neurons in his speech motor cortex fire in patterns that are essentially identical to the patterns they would have produced if he could still move his speech muscles normally. The implanted microelectrodes record those firing patterns from approximately 256 individual neurons. The decoding software, trained on Harrell’s specific neural signatures during initial calibration sessions, identifies which phonemes he is trying to produce and assembles them into words. Per The Next Web’s analysis of the BRAND platform that powers the decoding, the system uses a custom machine-learning architecture optimised for real-time speech reconstruction, with continuous adaptation to changes in the neural signal over time.
One of the more remarkable aspects of the system is its inclusion of a deliberate “privacy mode,” in which Harrell can disable data recording when he does not want his neural activity captured or used to train future versions of the decoder. The feature exists because the technology raises real concerns about the boundary between intended speech and unspoken thought. Brain-computer interfaces, by their nature, record neural activity that the user may not have chosen to externalise. The BrainGate team has been careful to design the system so that the user controls what gets recorded and what does not.
What this means for the broader field
The Harrell case is part of a broader convergence in the BCI field that has been gathering pace over the past several years. As covered by a Clinical Trials Arena analysis of the BCI clinical trial landscape, the BrainGate consortium — which has been running clinical trials of brain-computer interfaces since the early 2000s, and now spans Brown University, Massachusetts General Hospital, the VA, and UC Davis — is one of several parallel efforts working on the same general problem. Elon Musk’s Neuralink, Synchron, Paradromics, and the UCSF and Stanford BCI groups have all produced demonstrations of varying scope and ambition. The BrainGate Harrell case is, at present, the most extensively documented sustained-use BCI result in the published literature, but it sits within a broader trajectory in which the basic technology is moving from research curiosity to clinical reality.
For the approximately 30,000 Americans currently living with ALS, the practical implications of the Harrell case are limited in the short term. The technology remains experimental, requires invasive surgery, and is not yet commercially available outside the clinical trial setting. Scaling the approach to a second patient, a tenth patient, a hundredth patient will require substantial work on hardware durability, surgical standardisation, software adaptation across individual neural signatures, regulatory approval, and the long list of practical and economic questions that any new medical device must answer before becoming generally available. None of these are insurmountable. None of them are immediate. Harrell himself is now able to communicate, work, and participate in family and social life in ways that were not available to him two years ago. The question of how soon the same becomes true for other ALS patients, for stroke survivors with severe communication impairment, for people with locked-in syndrome, and for the broader population of patients with severe paralysis is now a question of engineering, manufacturing, and policy rather than of fundamental scientific possibility. The threshold the BrainGate team is now claiming to have crossed, in Brandman’s framing, is the one between proof-of-concept and tool. Whether the rest of the field follows, and how quickly, will define the next decade of BCI research.
Login
