Science

sleeping-pills-stop-the-brain’s-system-for-cleaning-out-waste

Sleeping pills stop the brain’s system for cleaning out waste


Cleanup on aisle cerebellum

A specialized system sends pulses of pressure through the fluids in our brain.

Our bodies rely on their lymphatic system to drain excessive fluids and remove waste from tissues, feeding those back into the blood stream. It’s a complex yet efficient cleaning mechanism that works in every organ except the brain. “When cells are active, they produce waste metabolites, and this also happens in the brain. Since there are no lymphatic vessels in the brain, the question was what was it that cleaned the brain,” Natalie Hauglund, a neuroscientist at Oxford University who led a recent study on the brain-clearing mechanism, told Ars.

Earlier studies done mostly on mice discovered that the brain had a system that flushed its tissues with cerebrospinal fluid, which carried away waste products in a process called glymphatic clearance. “Scientists noticed that this only happened during sleep, but it was unknown what it was about sleep that initiated this cleaning process,” Hauglund explains.

Her study found the glymphatic clearance was mediated by a hormone called norepinephrine and happened almost exclusively during the NREM sleep phase. But it only worked when sleep was natural. Anesthesia and sleeping pills shut this process down nearly completely.

Taking it slowly

The glymphatic system in the brain was discovered back in 2013 by Dr. Maiken Nedergaard, a Danish neuroscientist and a coauthor of Hauglund’s paper. Since then, there have been numerous studies aimed at figuring out how it worked, but most of them had one problem: they were done on anesthetized mice.

“What makes anesthesia useful is that you can have a very controlled setting,” Hauglund says.

Most brain imaging techniques require a subject, an animal or a human, to be still. In mouse experiments, that meant immobilizing their heads so the research team could get clear scans. “But anesthesia also shuts down some of the mechanisms in the brain,” Hauglund argues.

So, her team designed a study to see how the brain-clearing mechanism works in mice that could move freely in their cages and sleep naturally whenever they felt like it. “It turned out that with the glymphatic system, we didn’t really see the full picture when we used anesthesia,” Hauglund says.

Looking into the brain of a mouse that runs around and wiggles during sleep, though, wasn’t easy. The team pulled it off by using a technique called flow fiber photometry which works by imaging fluids tagged with fluorescent markers using a probe implanted in the brain. So, the mice got the optical fibers implanted in their brains. Once that was done, the team put fluorescent tags in the mice’s blood, cerebrospinal fluid, and on the norepinephrine hormone. “Fluorescent molecules in the cerebrospinal fluid had one wavelength, blood had another wavelength, and norepinephrine had yet another wavelength,” Hauglund says.

This way, her team could get a fairly precise idea about the brain fluid dynamics when mice were awake and asleep. And it turned out that the glymphatic system basically turned brain tissues into a slowly moving pump.

Pumping up

“Norepinephrine is released from a small area of the brain in the brain stem,” Hauglund says. “It is mainly known as a response to stressful situations. For example, in fight or flight scenarios, you see norepinephrine levels increasing.” Its main effect is causing blood vessels to contract. Still, in more recent research, people found out that during sleep, norepinephrine is released in slow waves that roll over the brain roughly once a minute. This oscillatory norepinephrine release proved crucial to the operation of the glymphatic system.

“When we used the flow fiber photometry method to look into the brains of mice, we saw these slow waves of norepinephrine, but we also saw how it works in synchrony with fluctuation in the blood volume,” Hauglund says.

Every time the norepinephrine level went up, it caused the contraction of the blood vessels in the brain, and the blood volume went down. At the same time, the contraction increased the volume of the perivascular spaces around the blood vessels, which were immediately filled with the cerebrospinal fluid.

When the norepinephrine level went down, the process worked in reverse: the blood vessels dilated, letting the blood in and pushing the cerebrospinal fluid out. “What we found was that norepinephrine worked a little bit like a conductor of an orchestra and makes the blood and cerebrospinal fluid move in synchrony in these slow waves,” Hauglund says.

And because the study was designed to monitor this process in freely moving, undisturbed mice, the team learned exactly when all this was going on. When mice were awake, the norepinephrine levels were much higher but relatively steady. The team observed the opposite during the REM sleep phase, where the norepinephrine levels were consistently low. The oscillatory behavior was present exclusively during the NREM sleep phase.

So, the team wanted to check how the glymphatic clearance would work when they gave the mice zolpidem, a sleeping drug that had been proven to increase NREM sleep time. In theory, zolpidem should have boosted brain-clearing. But it turned it off instead.

Non-sleeping pills

“When we looked at the mice after giving them zolpidem, we saw they all fell asleep very quickly. That was expected—we take zolpidem because it makes it easier for us to sleep,” Hauglund says. “But then we saw those slow fluctuations in norepinephrine, blood volume, and cerebrospinal fluid almost completely stopped.”

No fluctuations meant the glymphatic system didn’t remove any waste. This was a serious issue, because one of the cellular waste products it is supposed to remove is amyloid beta, found in the brains of patients suffering from Alzheimer’s disease.

Hauglund speculates it could be possible zolpidem induces a state very similar to sleep but at the same time it shuts down important processes that happen during sleep. While heavy zolpidem use has been associated with increased risk of the Alzheimer disease, it is not clear if this increased risk was there because the drug was inhibiting oscillatory norepinephrine release in the brain. To better understand this, Hauglund wants to get a closer look into how the glymphatic system works in humans.

“We know we have the same wave-like fluid dynamics in the brain, so this could also drive the brain clearance in humans,” Haugland told Ars. “Still, it’s very hard to look at norepinephrine in the human brain because we need an invasive technique to get to the tissue.”

But she said norepinephrine levels in people can be estimated based on indirect clues. One of them is pupil dilation and contraction, which work in in synchrony with the norepinephrine levels. Another other clue may lay in microarousals—very brief, imperceivable awakenings which, Hauglund thinks, can be correlated with the brain clearing mechanism. “I am currently interested in this phenomenon […]. Right now we have no idea why microarousals are there or what function they have” Hauglund says.

But the last step she has on her roadmap is making better sleeping pills. “We need sleeping drugs that don’t have this inhibitory effect on the norepinephrine waves. If we can have a sleeping pill that helps people sleep without disrupting their sleep at the same time it will be very important,” Hauglund concludes.

Cell, 2025. DOI: 10.1016/j.cell.2024.11.027

Photo of Jacek Krywko

Jacek Krywko is a freelance science and technology writer who covers space exploration, artificial intelligence research, computer science, and all sorts of engineering wizardry.

Sleeping pills stop the brain’s system for cleaning out waste Read More »

peeing-is-contagious-among-chimps

Peeing is contagious among chimps

Those results supported the initial hypothesis that chimps tended to urinate in sync rather than randomly. Further analysis showed that the closer a chimp was to another peeing chimp, the more likely the probability of that chimp peeing as well—evidence of social contagion. Finally, Onishi et al. wanted to explore whether social relationships (like socially close pairs, evidenced by mutual grooming and similar behaviors) influenced contagious urination. The only social factor that proved relevant was dominance, with less-dominant chimps being more prone to contagious urination.

There may still be other factors influencing the behavior, and more experimental research is needed on potential sensory cues and social triggers in order to identify possible underlying mechanisms for the phenomenon. Furthermore, this study was conducted with a captive chimp population; to better understand potential evolutionary roots, there should be research on wild chimp populations, looking at possible links between contagious urination and factors like ranging patterns, territory use, and so forth.

“This was an unexpected and fascinating result, as it opens up multiple possibilities for interpretation,” said coauthor Shinya Yamamoto, also of Kyoto University. “For instance, it could reflect hidden leadership in synchronizing group activities, the reinforcement of social bonds, or attention bias among lower-ranking individuals. These findings raise intriguing questions about the social functions of this behavior.”

DOI: Current Biology, 2025. 10.1016/j.cub.2024.11.052 (About DOIs).

Peeing is contagious among chimps Read More »

life-is-thriving-in-the-subsurface-depths-of-earth

Life is thriving in the subsurface depths of Earth

Nitrospirota is an archaeal phylum that’s particularly common in the terrestrial subsurface. Some species of nitrospirota are capable of oxidizing ammonia, while others can reduce it to nitrite, which is used by phytoplankton and also defends against pathogens in the human stomach, mouth, and skin.

Proteobacteria is a bacterial phylum that’s especially abundant in the terrestrial and marine subsurface. Some proteobacteria live in deep ocean trenches, and oxidize carbon monoxide (which contributes to global warming and depletes ozone). Bacteria also common in the marine subsurface include Desulfobacteria and Methylomirabilota. Desulfobacteria reduce sulfates, and other sulfate-reducing bacterias have already shown they can be used to help clean up contaminated soil. Methylomirabilota help control methane levels in the atmosphere by oxidizing methane.

Something unexpected that caught Ruff’s attention was how total diversity went up with depth. This was surprising because less energy is available at deeper levels of the subsurface. For archaea, diversity went up with the increase in depth in terrestrial environments but not marine environments. The same happened with bacteria, except in marine instead of terrestrial environments.

Much of what lies far below our feet still eludes us. Ruff suggests that single-cell microbes in even deeper, yet unexplored levels of the subsurface may have adapted to the absence of energy by slowing down their metabolisms so drastically that it could take decades, even centuries, for them to divide just once.

If there really are microbes that manage to live longer than humans with this survival tactic, it is possible similar species might be hiding on planets such as Mars, where the surface has long been blasted by radiation.

“Understanding deep life on Earth could be a model for discovering if there was life on Mars, and if it has survived,” Ruff said in a press release.

Maybe future technology could retrieve samples several kilometers below the Martian surface. Until then, keep digging.

Science Advances, 2024. DOI: 10.1126/sciadv.adq0645

Life is thriving in the subsurface depths of Earth Read More »

has-trump-changed-the-retirement-plans-for-the-country’s-largest-coal-plants?

Has Trump changed the retirement plans for the country’s largest coal plants?


A growth in electricity demand is leading to talk of delayed closures.

A house is seen near the Gavin Power Plant in Cheshire, Ohio. Credit: Stephanie Keith/Getty Images

This article originally appeared on Inside Climate News, a nonprofit, non-partisan news organization that covers climate, energy, and the environment. Sign up for their newsletter here.

There is renewed talk of a coal power comeback in the United States, inspired by Donald Trump’s return to the presidency and forecasts of soaring electricity demand.

The evidence so far only shows that some plants are getting small extensions on their retirement dates. This means a slowdown in coal’s rate of decline, which is bad for the environment, but it does little to change the long-term trajectory for the domestic coal industry.

In October, I wrote about how five of the country’s 10 largest coal-fired power plants had retirement dates. Today, I’m revisiting the list, providing some updates and then taking a few steps back to look at US coal plants as a whole. Consider this the “before” picture that can be judged against the “after” in four years.

Some coal plant owners have already pushed back retirement timetables. The largest example, this one from just before the election, is the Gibson plant in Indiana, the second-largest coal plant in the country. It’s set to close in 2038 instead of 2035, following an announcement in October from the owner, Duke Energy.

But the changes do not constitute a coal comeback in this country. For that to happen, power companies would need to be building new plants to replace the many that are closing, and there is almost no development of new coal plants.

That said, there have been some changes since October.

As recently as a few months ago, Southern Co. was saying it intended to close Plant Bowen in Georgia by 2035 at the latest. Bowen is the largest coal plant in the country, with a summer capacity of 3,200 megawatts.

Southern has since said it may extend the plant’s life in response to forecasts of rising electricity demand. Chris Womack, Southern’s CEO, confirmed this possibility when speaking at a utility industry conference in November, saying that the plant may need to operate for longer than previously planned because of demand from data centers.

Southern has not yet made regulatory filings that spell out its plans, but this will likely occur in the next few weeks, according to a company spokesman.

In October, I reported that the Gavin plant in Ohio was likely to get a 2031 date to retire or switch to a different fuel once the plant’s pending sale was completed. The person who shared that information with me was involved with the plans and spoke on condition of anonymity because the sale was not final.

Since then, the prospective buyer of the plant has said in federal regulatory filings that it has no timetable for closing the plant or switching to a different fuel. The plant is changing hands as part of a larger deal between investment firms, with Lightstone Holdco selling to Energy Capital Partners, or ECP. Another company, coal exporter Javelin Global Commodities, is buying a minority share of the Gavin plant.

I went back to the person who told me about the 2031 retirement date. They said forecasts of rising electricity demand, as well as the election of Trump, have created enough uncertainty about power prices and regulations that it makes sense to not specify a date.

The 2031 timeline, and its abandonment, makes some sense once you understand that the Biden administration finalized power plant regulations last spring that gave coal plant operators an incentive to announce a retirement date: Plants closing before 2032 faced no new requirements. That incentive is likely to go away as Trump plans to roll back power plant pollution regulations.

Gavin’s sale is still pending. Several parties have filed objections to the transaction with the Federal Energy Regulatory Commission, arguing that the sellers have not been clear enough about their plans.

An ECP spokesman said the company has no comment beyond its filings.

Other than the changes to plans for Bowen and Gavin, the outlook has not shifted for the rest of the plants among the 10 largest. The Gibson and Rockport plants in Indiana still have retirement dates, as do Cumberland in Tennessee and Monroe in Michigan, according to the plants’ owners.

The Amos plant in West Virginia, Miller in Alabama, Scherer in Georgia, and Parish in Texas didn’t have retirement dates a few months ago, and they still don’t.

But the largest coal plants are only part of the story. Several dozen smaller plants are getting extensions of retirement plans, as Emma Foehringer Merchant reported last week for Floodlight News.

One example is the 1,157-megawatt Baldwin plant in Illinois, which was scheduled to close this year. Now the owner, Vistra Corp., has pushed back the retirement to 2027.

A few extra years of a coal plant is more of a stopgap than a long-term solution. When it comes to building new power plants to meet demand, developers are talking about natural gas, solar, nuclear, and other resources, but I have yet to see a substantial discussion of building a new coal plant.

In Alaska, Gov. Mike Dunleavy has said the state may build two coal plants to provide power in remote mining areas, as reported by Taylor Kuykendall of S&P Global Commodity Insights. Flatlands Energy, a Canadian company, has also talked about building a 400-megawatt coal plant in Alaska, as Nathaniel Herz reported for Alaska Beacon. These appear to be early-stage plans.

The lack of development activity underscores how coal power is fading in this country, and has been for a while.

Coal was used to generate 16 percent of US electricity in 2023, down by more than half from 2014. In that time, coal went from the country’s leading fuel for electricity to trailing natural gas, renewables, and nuclear. (These and all the figures that follow are from the US Energy Information Administration.)

The United States had about 176,000 megawatts of coal plant capacity as of October, down from about 300,000 megawatts in 2014.

The coal plants that do remain are being used less. In 2023, the average capacity factor for a coal plant was 42 percent. Capacity factor is a measure of how much electricity a plant has generated relative to the maximum possible if it was running all the time. In 2014, the average capacity factor was 61 percent.

Power companies are burning less coal because of the availability of less expensive alternatives, such as natural gas, wind, and solar, among others. The think tank Energy Innovation issued a report in 2023 finding that 99 percent of US coal-fired power plants cost more to operate than the cost of replacement with a combination of wind, solar, and batteries.

The Trump administration will arrive in Washington with promises to help fossil fuels. It could extend the lives of some coal plants by weakening environmental regulations, which may reduce the plants’ operational costs. It also could repeal or revise subsidies that help to reduce the costs of renewables and batteries, making those resources more expensive.

I don’t want to minimize the damage that could be caused by those policies. But even in extreme scenarios, it’s difficult to imagine investors wanting to spend billions of dollars to develop a new coal plant, much less a fleet of them.

Photo of Inside Climate News

Has Trump changed the retirement plans for the country’s largest coal plants? Read More »

fire-destroys-starship-on-its-seventh-test-flight,-raining-debris-from-space

Fire destroys Starship on its seventh test flight, raining debris from space

This launch debuted a more advanced, slightly taller version of Starship, known as Version 2 or Block 2, with larger propellant tanks, a new avionics system, and redesigned feed lines flowing methane and liquid oxygen propellants to the ship’s six Raptor engines. SpaceX officials did not say whether any of these changes might have caused the problem on Thursday’s launch.

SpaceX officials have repeatedly and carefully set expectations for each Starship test flight. They routinely refer to the rocket as experimental, and the primary focus of the rocket’s early demo missions is to gather data on the performance of the vehicle. What works, and what doesn’t work?

Still, the outcome of Thursday’s test flight is a clear disappointment for SpaceX. This was the seventh test flight of SpaceX’s enormous rocket and the first time Starship failed to complete its launch sequence since the second flight in November 2023. Until now, SpaceX has made steady progress, and each Starship flight has achieved more milestones than the one before.

On the first flight in April 2023, the rocket lost control a little more than two minutes after liftoff, and the ground-shaking power of the booster’s 33 engines shattered the concrete foundation beneath the launch pad. Seven months later, on Flight 2, the rocket made it eight minutes before failing. On that mission, Starship failed at roughly the same point of its ascent, just before the cutoff of the vehicle’s six methane-fueled Raptor engines.

Back then, a handful of photos and images from the Florida Keys and Puerto Rico showed debris in the sky after Starship activated its self-destruct mechanism due to an onboard fire caused by a dump of liquid oxygen propellant. But that flight occurred in the morning, with bright sunlight along the ship’s flight path.

This time, the ship disintegrated and reentered the atmosphere at dusk, with impeccable lighting conditions accentuating the debris cloud’s appearance. These twilight conditions likely contributed to the plethora of videos posted to social media on Thursday.

Starship and Super Heavy head downrange from SpaceX’s launch site near Brownsville, Texas. Credit: SpaceX

The third Starship test flight last March saw the spacecraft reach its planned trajectory and fly halfway around the world before succumbing to the scorching heat of atmospheric reentry. In June, the fourth test flight ended with controlled splashdowns of the rocket’s Super Heavy booster in the Gulf of Mexico and of Starship in the Indian Ocean.

In October, SpaceX caught the Super Heavy booster with mechanical arms at the launch pad for the first time, proving out the company’s audacious approach to recovering and reusing the rocket. On this fifth test flight, SpaceX modified the ship’s heat shield to better handle the hot temperatures of reentry, and the vehicle again made it to an on-target splashdown in the Indian Ocean.

Most recently, Flight 6 on November 19 demonstrated the ship’s ability to reignite its Raptor engines in space for the first time and again concluded with a bullseye splashdown. But SpaceX aborted an attempt to again catch the booster back at Starbase due to a problem with sensors on the launch pad’s tower.

With Flight 7, SpaceX hoped to test more changes to the heat shield protecting Starship from reentry temperatures up to 2,600° Fahrenheit (1,430° Celsius). Musk has identified the heat shield as one of the most difficult challenges still facing the program. In order for SpaceX to reach its ambition for the ship to become rapidly reusable, with minimal or no refurbishment between flights, the heat shield must be resilient and durable.

Fire destroys Starship on its seventh test flight, raining debris from space Read More »

a-solid-electrolyte-gives-lithium-sulfur-batteries-ludicrous-endurance

A solid electrolyte gives lithium-sulfur batteries ludicrous endurance


Sulfur can store a lot more lithium but is problematically reactive in batteries.

If you weren’t aware, sulfur is pretty abundant. Credit: P_Wei

Lithium may be the key component in most modern batteries, but it doesn’t make up the bulk of the material used in them. Instead, much of the material is in the electrodes, where the lithium gets stored when the battery isn’t charging or discharging. So one way to make lighter and more compact lithium-ion batteries is to find electrode materials that can store more lithium. That’s one of the reasons that recent generations of batteries are starting to incorporate silicon into the electrode materials.

There are materials that can store even more lithium than silicon; a notable example is sulfur. But sulfur has a tendency to react with itself, producing ions that can float off into the electrolyte. Plus, like any electrode material, it tends to expand in proportion to the amount of lithium that gets stored, which can create physical strains on the battery’s structure. So while it has been easy to make lithium-sulfur batteries, their performance has tended to degrade rapidly.

But this week, researchers described a lithium-sulfur battery that still has over 80 percent of its original capacity after 25,000 charge/discharge cycles. All it took was a solid electrolyte that was more reactive than the sulfur itself.

When lithium meets sulfur…

Sulfur is an attractive battery material. It’s abundant and cheap, and sulfur atoms are relatively lightweight compared to many of the other materials used in battery electrodes. Sodium-sulfur batteries, which rely on two very cheap raw materials, have already been developed, although they only work at temperatures high enough to melt both of these components. Lithium-sulfur batteries, by contrast, could operate more or less the same way that current lithium-ion batteries do.

With a few major exceptions, that is. One is that the elemental sulfur used as an electrode is a very poor conductor of electricity, so it has to be dispersed within a mesh of conductive material. (You can contrast that with graphite, which both stores lithium and conducts electricity relatively well, thanks to being composed of countless sheets of graphene.) Lithium is stored there as Li2S, which occupies substantially more space than the elemental sulfur it’s replacing.

Both of these issues, however, can be solved with careful engineering of the battery’s structure. A more severe problem comes from the properties of the lithium-sulfur reactions that occur at the electrode. Elemental sulfur exists as an eight-atom ring, and the reactions with lithium are slow enough that semi-stable intermediates with smaller chains of sulfur end up forming. Unfortunately, these tend to be soluble in most electrolytes, allowing them to travel to the opposite electrode and participate in chemical reactions there.

This process essentially discharges the battery without allowing the electrons to be put to use. And it gradually leaves the electrode’s sulfur unavailable for participating in future charge/discharge cycles. The net result is that early generations of the technology would discharge themselves while sitting unused and would only survive a few hundred cycles before performance decayed dramatically.

But there has been progress on all these fronts, and some lithium-sulfur batteries with performance similar to lithium-ion have been demonstrated. Late last year, a company announced that it had lined up the money needed to build the first large-scale lithium-sulfur battery factory. Still, work on improvements has continued, and the new work seems to suggest ways to boost performance well beyond lithium-ion.

The need for speed

The paper describing the new developments, done by a collaboration between Chinese and German researchers, focuses on one aspect of the challenges posed by lithium-sulfur batteries: the relatively slow chemical reaction between lithium ions and elemental sulfur. It presents that aspect as a roadblock to fast charging, something that will be an issue for automotive applications. But at the same time, finding a way to limit the formation of inactive intermediate products during this reaction goes to the root of the relatively short usable life span of lithium-sulfur batteries.

As it turns out, the researchers found two.

One of the problems with the lithium-sulfur reaction intermediates is that they dissolve in most electrolytes. But that’s not a problem if the electrolyte isn’t a liquid. Solid electrolytes are materials that have a porous structure at the atomic level, with the environment inside the pores being favorable for ions. This allows ions to diffuse through the solid. If there’s a way to trap ions on one side of the electrolyte, such as a chemical reaction that traps or de-ionizes them, then it can enable one-way travel.

Critically, pores that favor the transit of lithium ions, which are quite compact, aren’t likely to allow the transit of the large ionized chains of sulfur. So a solid electrolyte should help cut down on the problems faced by lithium-sulfur batteries. But it won’t necessarily help with fast charging.

The researchers began by testing a glass formed from a mixture of boron, sulfur, and lithium (B2S3 and Li2S). But this glass had terrible conductivity, so they started experimenting with related glasses and settled on a combination that substituted in some phosphorus and iodine.

The iodine turned out to be a critical component. While the exchange of electrons with sulfur is relatively slow, iodine undergoes electron exchange (technically termed a redox reaction) extremely quickly. So it can act as an intermediate in the transfer of electrons to sulfur, speeding up the reactions that occur at the electrode. In addition, iodine has relatively low melting and boiling points, and the researchers suggest there’s some evidence that it moves around within the electrolyte, allowing it to act as an electron shuttle.

Successes and caveats

The result is a far superior electrolyte—and one that enables fast charging. It’s typical that fast charging cuts into the total capacity that can be stored in a battery. But when charged at an extraordinarily fast rate (50C, meaning a full charge in just over a minute), a battery based on this system still had half the capacity of a battery charged 25 times more slowly (2C, or a half-hour to full charge).

But the striking thing was how durable the resulting battery was. Even at an intermediate charging rate (5C), it still had over 80 percent of its initial capacity after over 25,000 charge/discharge cycles. By contrast, lithium-ion batteries tend to hit that level of decay after about 1,000 cycles. If that sort of performance is possible in a mass-produced battery, it’s only a slight exaggeration to say it can radically alter our relationships with many battery-powered devices.

What’s not at all clear, however, is whether this takes full advantage of one of the original promises of lithium-sulfur batteries: more charge in a given weight and volume. The researchers specify the battery being used for testing; one electrode is an indium/lithium metal foil, and the other is a mix of carbon, sulfur, and the glass electrolyte. A layer of the electrolyte sits between them. But when giving numbers for the storage capacity per weight, only the weight of the sulfur is mentioned.

Still, even if weight issues would preclude this from being stuffed into a car or cell phone, there are plenty of storage applications that would benefit from something that doesn’t wear out even with 65 years of daily cycling.

Nature, 2025. DOI: 10.1038/s41586-024-08298-9  (About DOIs).

Photo of John Timmer

John is Ars Technica’s science editor. He has a Bachelor of Arts in Biochemistry from Columbia University, and a Ph.D. in Molecular and Cell Biology from the University of California, Berkeley. When physically separated from his keyboard, he tends to seek out a bicycle, or a scenic location for communing with his hiking boots.

A solid electrolyte gives lithium-sulfur batteries ludicrous endurance Read More »

here’s-what-nasa-would-like-to-see-spacex-accomplish-with-starship-this-year

Here’s what NASA would like to see SpaceX accomplish with Starship this year


Iterate, iterate, and iterate some more

The seventh test flight of Starship is scheduled for launch Thursday afternoon.

SpaceX’s upgraded Starship rocket stands on its launch pad at Starbase, Texas. Credit: SpaceX

SpaceX plans to launch the seventh full-scale test flight of its massive Super Heavy booster and Starship rocket Thursday afternoon. It’s the first of what might be a dozen or more demonstration flights this year as SpaceX tries new things with the most powerful rocket ever built.

There are many things on SpaceX’s Starship to-do list in 2025. They include debuting an upgraded, larger Starship, known as Version 2 or Block 2, on the test flight preparing to launch Thursday. The one-hour launch window opens at 5 pm EST (4 pm CST; 22: 00 UTC) at SpaceX’s launch base in South Texas. You can watch SpaceX’s live webcast of the flight here.

SpaceX will again attempt to catch the rocket’s Super Heavy booster—more than 20 stories tall and wider than a jumbo jet—back at the launch pad using mechanical arms, or “chopsticks,” mounted to the launch tower. Read more about the Starship Block 2 upgrades in our story from last week.

You might think of next week’s Starship test flight as an apéritif before the entrées to come. Ars recently spoke with Lisa Watson-Morgan, the NASA engineer overseeing the agency’s contract with SpaceX to develop a modified version of Starship to land astronauts on the Moon. NASA has contracts with SpaceX worth more than $4 billion to develop and fly two Starship human landing missions under the umbrella of the agency’s Artemis program to return humans to the Moon.

We are publishing the entire interview with Watson-Morgan below, but first, let’s assess what SpaceX might accomplish with Starship this year.

There are many things to watch for on this test flight, including the deployment of 10 satellite simulators to test the ship’s payload accommodations and the performance of a beefed-up heat shield as the vehicle blazes through the atmosphere for reentry and splashdown in the Indian Ocean.

If this all works, SpaceX may try to launch a ship into low-Earth orbit on the eighth flight, expected to launch in the next couple of months. All of the Starship test flights to date have intentionally flown on suborbital trajectories, bringing the ship back toward reentry over the sea northwest of Australia after traveling halfway around the world.

Then, there’s an even bigger version of Starship called Block 3 that could begin flying before the end of the year. This version of the ship is the one that SpaceX will use to start experimenting with in-orbit refueling, according to Watson-Morgan.

In order to test refueling, two Starships will dock together in orbit, allowing one vehicle to transfer super-cold methane and liquid oxygen into the other. Nothing like this on this scale has ever been attempted before. Future Starship missions to the Moon and Mars may require 10 or more tanker missions to gas up in low-Earth orbit. All of these missions will use different versions of the same basic Starship design: a human-rated lunar lander, a propellant depot, and a refueling tanker.

Artist’s illustration of Starship on the surface of the Moon. Credit: SpaceX

Questions for 2025

Catching Starship back at its launch tower and demonstrating orbital propellant transfer are the two most significant milestones on SpaceX’s roadmap for 2025.

SpaceX officials have said they aim to fly as many as 25 Starship missions this year, allowing engineers to more rapidly iterate on the vehicle’s design. SpaceX is constructing a second launch pad at its Starbase facility near Brownsville, Texas, to help speed up the launch cadence.

Can SpaceX achieve this flight rate in 2025? Will faster Starship manufacturing and reusability help the company fly more often? Will SpaceX fly its first ship-to-ship propellant transfer demonstration this year? When will Starship begin launching large batches of new-generation Starlink Internet satellites?

Licensing delays at the Federal Aviation Administration have been a thorn in SpaceX’s side for the last couple of years. Will those go away under the incoming administration of President-elect Donald Trump, who counts SpaceX founder Elon Musk as a key adviser?

And will SpaceX gain a larger role in NASA’s Artemis lunar program? The Artemis program’s architecture is sure to be reviewed by the Trump administration and the nominee for the agency’s next administrator, billionaire businessman and astronaut Jared Isaacman.

The very expensive Space Launch System rocket, developed by NASA with Boeing and other traditional aerospace contractors, might be canceled. NASA currently envisions the SLS rocket and Orion spacecraft as the transportation system to ferry astronauts between Earth and the vicinity of the Moon, where crews would meet up with a landing vehicle provided by commercial partners SpaceX and Blue Origin.

Watson-Morgan didn’t have answers to all of these questions. Many of them are well outside of her purview as Human Landing System program manager, so Ars didn’t ask. Instead, Ars discussed technical and schedule concerns with her during the half-hour interview. Here is one part of the discussion, lightly edited for clarity.

Ars: What do you hope to see from Flight 7 of Starship?

Lisa Watson-Morgan: One of the exciting parts of working with SpaceX are these test flights. They have a really fast turnaround, where they put in different lessons learned. I think you saw many of the flight objectives that they discussed from Flight 6, which was a great success. I think they mentioned different thermal testing experiments that they put on the ship in order to understand the different heating, the different loads on certain areas of the system. All that was really good with each one of those, in addition to how they configure the tiles. Then, from that, there’ll be additional tests that they will put on Flight 7, so you kind of get this iterative improvement and learning that we’ll get to see in Flight 7. So Flight 7 is the first Version 2 of their ship set. When I say that, I mean the ship, the booster, all the systems associated with it. So, from that, it’s really more just understanding how the system, how the flaps, how all of that interacts and works as they’re coming back in. Hopefully we’ll get to see some catches, that’s always exciting.

Ars: How did the in-space Raptor engine relight go on Flight 6 (on November 19)?

Lisa Watson-Morgan: Beautifully. And that’s something that’s really important to us because when we’re sitting on the Moon… well, actually, the whole path to the Moon as we are getting ready to land on the Moon, we’ll perform a series of maneuvers, and the Raptors will have an environment that is very, very cold. To that, it’s going to be important that they’re able to relight for landing purposes. So that was a great first step towards that. In addition, after we land, clearly the Raptors will be off, and it will get very cold, and they will have to relight in a cold environment (to get off the Moon). So that’s why that step was critical for the Human Landing System and NASA’s return to the Moon.

A recent artist’s illustration of two Starships docked together in low-Earth orbit. Credit: SpaceX

Ars: Which version of the ship is required for the propellant transfer demonstration, and what new features are on that version to enable this test?

Lisa Watson-Morgan: We’re looking forward to the Version 3, which is what’s coming up later on, sometime in ’25, in the near term, because that’s what we need for propellant transfer and the cryo fluid work that is also important to us… There are different systems in the V3 set that will help us with cryo fluid management. Obviously, with those, we have to have the couplers and the quick-disconnects in order for the two systems to have the right guidance, navigation, trajectory, all the control systems needed to hold their station-keeping in order to dock with each other, and then perform the fluid transfer. So all the fluid lines and all that’s associated with that, those systems, which we have seen in tests and held pieces of when we’ve been working with them at their site, we’ll get to see those actually in action on orbit.

Ars: Have there been any ground tests of these systems, whether it’s fluid couplers or docking systems? Can you talk about some of the ground tests that have gone into this development?

Lisa Watson-Morgan: Oh, absolutely. We’ve been working with them on ground tests for this past year. We’ve seen the ground testing and reviewed the data. Our team works with them on what we deem necessary for the various milestones. While the milestone contains proprietary (information), we work closely with them to ensure that it’s going to meet the intent, safety-wise as well as technically, of what we’re going to need to see. So they’ve done that.

Even more exciting, they have recently shipped some of their docking systems to the Johnson Space Center for testing with the Orion Lockheed Martin docking system, and that’s for Artemis III. Clearly, that’s how we’re going to receive the crew. So those are some exciting tests that we’ve been doing this past year as well that’s not just focused on, say, the booster and the ship. There are a lot of crew systems that are being developed now. We’re in work with them on how we’re going to effectuate the crew manual control requirements that we have, so it’s been a great balance to see what the crew needs, given the size of the ship. That’s been a great set of work. We have crew office hours where the crew travels to Hawthorne [SpaceX headquarters in California] and works one-on-one with the different responsible engineers in the different technical disciplines to make sure that they understand not just little words on the paper from a requirement, but actually what this means, and then how systems can be operated.

Ars: For the docking system, Orion uses the NASA Docking System, and SpaceX brings its own design to bear on Starship?

Lisa Watson-Morgan: This is something that I think the Human Landing System has done exceptionally well. When we wrote our high-level set of requirements, we also wrote it with a bigger picture in mind—looked into the overall standards of how things are typically done, and we just said it has to be compliant with it. So it’s a docking standard compliance, and SpaceX clearly meets that. They certainly do have the Dragon heritage, of course, with the International Space Station. So, because of that, we have high confidence that they’re all going to work very well. Still, it’s important to go ahead and perform the ground testing and get as much of that out of the way as we can.

Lisa Watson-Morgan, NASA’s HLS program manager, is based at Marshall Space Flight Center in Huntsville, Alabama. Credit: ASA/Aubrey Gemignani

Ars: How far along is the development and design of the layout of the crew compartment at the top of Starship? Is it far along, or is it still in the conceptual phase? What can you say about that?

Lisa Watson-Morgan: It’s much further along there. We’ve had our environmental control and life support systems, whether it’s carbon dioxide monitoring fans to make sure the air is circulating properly. We’ve been in a lot of work with SpaceX on the temperature. It’s… a large area (for the crew). The seats, making sure that the crew seats and the loads on that are appropriate. For all of that work, as the analysis work has been performed, the NASA team is reviewing it. They had a mock-up, actually, of some of their life support systems even as far back as eight-plus months ago. So there’s been a lot of progress on that.

Ars: Is SpaceX planning to use a touchscreen design for crew displays and controls, like they do with the Dragon spacecraft?

Lisa Watson-Morgan: We’re in talks about that, about what would be the best approach for the crew for the dynamic environment of landing.

Ars: I can imagine it is a pretty dynamic environment with those Raptor engines firing. It’s almost like a launch in reverse.

Lisa Watson-Morgan: Right. Those are some of the topics that get discussed in the crew office hours. That’s why it’s good to have the crew interacting directly, in addition to the different discipline leads, whether it’s structural, mechanical, propulsion, to have all those folks talking guidance and having control to say, “OK, well, when the system does this, here’s the mode we expect to see. Here’s the impact on the crew. And is this condition, or is the option space that we have on the table, appropriate for the next step, with respect to the displays.”

Ars: One of the big things SpaceX needs to prove out before going to the Moon with Starship is in-orbit propellant transfer. When do you see the ship-to-ship demonstration occurring?

Lisa Watson-Morgan: I see it occurring in ’25.

Ars: Anything more specific about the schedule for that?

Lisa Watson-Morgan: That’d be a question for SpaceX because they do have a number of flights that they’re performing commercially, for their maturity. We get the benefit of that. It’s actually a great partnership. I’ll tell you, it’s really good working with them on this, but they’d have to answer that question. I do foresee it happening in ’25.

Ars: What things do you need to see SpaceX accomplish before they’re ready for the refueling demo? I’m thinking of things like the second launch tower, potentially. Do they need to demonstrate a ship catch or anything like that before going for orbital refueling?

Lisa Watson-Morgan: I would say none of that’s required. You just kind of get down to, what are the basics? What are the basics that you need? So you need to be able to launch rapidly off the same pad, even. They’ve shown they can launch and catch within a matter of minutes. So that is good confidence there. The catching is part of their reuse strategy, which is more of their commercial approach, and not a NASA requirement. NASA reaps the benefit of it by good pricing as a result of their commercial model, but it is not a requirement that we have. So they could theoretically use the same pad to perform the propellant transfer and the long-duration flight, because all it requires is two launches, really, within a specified time period to where the two systems can meet in a planned trajectory or orbit to do the propellant transfer. So they could launch the first one, and then within a week or two or three, depending on what the concept of operations was that we thought we could achieve at that time, and then have the propellant transfer demo occur that way. So you don’t necessarily need two pads, but you do need more thermal characterization of the ship. I would say that is one of the areas (we need to see data on), and that is one of the reasons, I think, why they’re working so diligently on that.

Ars: You mentioned the long-duration flight demonstration. What does that entail?

Lisa Watson-Morgan: The simple objectives are to launch two different tankers or Starships. The Starship will eventually be a crewed system. Clearly, the ones that we’re talking about for the propellant transfer are not. It’s just to have the booster and Starship system launch, and within a few weeks, have another one launch, and have them rendezvous. They need to be able to find each other with their sensors. They need to be able to come close, very, very close, and they need to be able to dock together, connect, do the quick connect, and make sure they are able, then, to flow propellant and LOX (liquid oxygen) to another system. Then, we need to be able to measure the quantity of how much has gone over. And from that, then they need to safely undock and dispose.

Ars: So the long-duration flight demonstration is just part of what SpaceX needs to do in order to be ready for the propellant transfer demonstration?

Lisa Watson-Morgan: We call it long duration just because it’s not a 45-minute or an hour flight. Long duration, obviously, that’s a relative statement, but it’s a system that can stay up long enough to be able to find another Starship and perform those maneuvers and flow of fuel and LOX.

Ars: How much propellant will you transfer with this demonstration, and do you think you’ll get all the data you need in one demonstration, or will SpaceX need to try this several times?

Lisa Watson-Morgan: That’s something you can ask SpaceX (about how much propellant will be transferred). Clearly, I know, but there’s some sensitivity there. You’ve seen our requirements in our initial solicitation. We have thresholds and goals, meaning we want you to at least do this, but more is better, and that’s typically how we work almost everything. Working with commercial industry in these fixed-price contracts has worked exceptionally well, because when you have providers that are also wanting to explore commercially or trying to make a commercial system, they are interested in pushing more than what we would typically ask for, and so often we get that for an incredibly fair price.

Photo of Stephen Clark

Stephen Clark is a space reporter at Ars Technica, covering private space companies and the world’s space agencies. Stephen writes about the nexus of technology, science, policy, and business on and off the planet.

Here’s what NASA would like to see SpaceX accomplish with Starship this year Read More »

heroes,-villains,-and-childhood-trauma-in-the-mceu-and-dcu

Heroes, villains, and childhood trauma in the MCEU and DCU

They also limited their study to Marvel and DC characters depicted in major films, rather than including storylines from spinoff TV series. So Wanda Maximoff/The Scarlet Witch was not included since much of her traumatic backstory appeared in the series WandaVision. Furthermore, “We omitted gathering more characters from comic books in both Marvel and DC universes, due to their inconsistency in character development,” the authors wrote. “Comic book storylines often feature alternative plot lines, character arcs, and multiverse outcomes. The storytelling makes comic book characters highly inconsistent and challenging to score.”

With great power…

They ended up watching 33 films, with a total runtime of 77 hours and 5 minutes. They chose 19 male characters, eight female characters, and one gender-fluid character (Loki) as “subjects” for their study, applying the ACE questionnaire to their childhoods as portrayed in the films.

The results: “We found no statistically significant differences between heroes and villains, Marvel and DC characters, or men and women and ACE scores,” said Jackson. “This means that characters who were portrayed as having difficult childhoods were not more likely to be villains. This study somewhat refutes the idea that villains are a product of their experiences. Based on the films we watched, people chose to be heroes and that was what made the difference—not their experiences.”

Notably, Black Widow had the highest ACE score (eight) and yet still became an Avenger, though the authors acknowledge that the character did some bad things before then and famously wanted to wipe out the “red” in her ledger. She “represents resilience of characters who have experienced trauma,” the authors wrote, as well as demonstrating that “socio-ecological resilience, including access to social relationships and supportive communities, can play a mitigating role in the effect of ACEs.” The Joker, by contrast, scored a six and “wreaked havoc across Gotham City.”

Heroes, villains, and childhood trauma in the MCEU and DCU Read More »

two-lunar-landers-are-on-the-way-to-the-moon-after-spacex’s-double-moonshot

Two lunar landers are on the way to the Moon after SpaceX’s double moonshot

Julianna Scheiman, director of NASA science missions for SpaceX, said it made sense to pair the Firefly and ispace missions on the same Falcon 9 rocket.

“When we have two missions that can each go to the Moon on the same launch, that is something that we obviously want to take advantage of,” Scheiman said. “So when we found a solution for the Firefly and ispace missions to fly together on the same Falcon 9, it was a no-brainer to put them together.”

SpaceX stacked the two landers, one on top of the other, inside the Falcon 9’s payload fairing. Firefly’s lander, the larger of the two spacecraft, rode on top of the stack and deployed from the rocket first. The Resilience lander from ispace launched in the lower position, cocooned inside a specially designed canister. Once Firefly’s lander separated from the Falcon 9, the rocket jettisoned the canister, performed a brief engine firing to maneuver into a slightly different orbit, then released ispace’s lander.

This dual launch arrangement resulted in a lower launch price for Firefly and ispace, according to Scheiman.

“At SpaceX, we are really interested in and invested in lowering the cost of launch for everybody,” she said. “So that’s something we’re really proud of.”

The Resilience lunar lander is pictured at ispace’s facility in Japan last year. The company’s small Tenacious rover is visible on the upper left part of the spacecraft. credit: ispace Credit: ispace

The Blue Ghost and Resilience landers will take different paths toward the Moon.

Firefly’s Blue Ghost will spend about 25 days in Earth orbit, then four days in transit to the Moon. After Blue Ghost enters lunar orbit, Firefly’s ground team will verify the readiness of the lander’s propulsion and navigation systems and execute several thruster burns to set up for landing.

Blue Ghost’s final descent to the Moon is tentatively scheduled for March 2. The target landing site is in Mare Crisium, an ancient 350-mile-wide (560-kilometer) impact basin in the northeast part of the near side of the Moon.

After touchdown, Blue Ghost will operate for about 14 days (one entire lunar day). The instruments aboard Firefly’s lander include a subsurface drill, an X-ray imager, and an experimental electrodynamic dust shield to test methods of repelling troublesome lunar dust from accumulating on sensitive spacecraft components.

The Resilience lander from ispace will take four to five months to reach the Moon. It carries several intriguing tech demo experiments, including a water electrolyzer provided by a Japanese company named Takasago Thermal Engineering. This demonstration will test equipment that future lunar missions could use to convert the Moon’s water ice resources into electricity and rocket fuel.

The lander will also deploy a “micro-rover” named Tenacious, developed by an ispace subsidiary in Luxembourg. The Tenacious rover will attempt to scoop up lunar soil and capture high-definition imagery of the Moon.

Ron Garan, CEO of ispace’s US-based subsidiary, told Ars that this mission is “pivotal” for the company.

“We were not fully successful on our first mission,” Garan said in an interview. “It was an amazing accomplishment, even though we didn’t have a soft landing… Although the hardware worked flawlessly, exactly as it was supposed to, we did have some lessons learned in the software department. The fixes to prevent what happened on the first mission from happening on the second mission were fairly straightforward, so that boosts our confidence.”

The ispace subsidiary led by Garan, a former NASA astronaut, is based in Colorado. While the Resilience lander launched Wednesday is not part of the CLPS program, the company will build an upgraded lander for a future CLPS mission for NASA, led by Draper Laboratory.

“I think the fact that we have two lunar landers on the same rocket for the first time in history is pretty substantial,” Garan said. I think we all are rooting for each other.”

Investors need to see more successes with commercial lunar landers to fully realize the market’s potential, Garan said.

“That market, right now, is very nascent. It’s very, very immature. And one of the reasons for that is that it’s very difficult for companies that are contemplating making investments on equipment, experiments, etc., to put on the lunar surface and lunar orbit,” Garan said. “It’s very difficult to make those investments, especially if they’re long-term investments, because there really hasn’t been a proof of concept yet.”

“So every time we have a success, that makes it more likely that these companies that will serve as the foundation of a commercial lunar market movement will be able to make those investments,” Garan said. “Conversely, every time we have a failure, the opposite happens.”

Two lunar landers are on the way to the Moon after SpaceX’s double moonshot Read More »

researchers-use-ai-to-design-proteins-that-block-snake-venom-toxins

Researchers use AI to design proteins that block snake venom toxins

Since these two toxicities work through entirely different mechanisms, the researchers tackled them separately.

Blocking a neurotoxin

The neurotoxic three-fingered proteins are a subgroup of the larger protein family that specializes in binding to and blocking the receptors for acetylcholine, a major neurotransmitter. Their three-dimensional structure, which is key to their ability to bind these receptors, is based on three strings of amino acids within the protein that nestle against each other (for those that have taken a sufficiently advanced biology class, these are anti-parallel beta sheets). So to interfere with these toxins, the researchers targeted these strings.

They relied on an AI package called RFdiffusion (the RF denotes its relation to the Rosetta Fold protein-folding software). RFdiffusion can be directed to design protein structures that are complements to specific chemicals; in this case, it identified new strands that could line up along the edge of the ones in the three-fingered toxins. Once those were identified, a separate AI package, called ProteinMPNN, was used to identify the amino acid sequence of a full-length protein that would form the newly identified strands.

But we’re not done with the AI tools yet. The combination of three-fingered toxins and a set of the newly designed proteins were then fed into DeepMind’s AlfaFold2 and the Rosetta protein structure software, and the strength of the interactions between them were estimated.

It’s only at this point that the researchers started making actual proteins, focusing on the candidates that the software suggested would interact the best with the three-fingered toxins. Forty-four of the computer-designed proteins were tested for their ability to interact with the three-fingered toxin, and the single protein that had the strongest interaction was used for further studies.

At this point, it was back to the AI, where RFDiffusion was used to suggest variants of this protein that might bind more effectively. About 15 percent of its suggestions did, in fact, interact more strongly with the toxin. The researchers then made both the toxin and the strongest inhibitor in bacteria and obtained the structure of their interactions. This confirmed that the software’s predictions were highly accurate.

Researchers use AI to design proteins that block snake venom toxins Read More »

meta-takes-us-a-step-closer-to-star-trek’s-universal-translator

Meta takes us a step closer to Star Trek’s universal translator


The computer science behind translating speech from 100 source languages.

In 2023, AI researchers at Meta interviewed 34 native Spanish and Mandarin speakers who lived in the US but didn’t speak English. The goal was to find out what people who constantly rely on translation in their day-to-day activities expect from an AI translation tool. What those participants wanted was basically a Star Trek universal translator or the Babel Fish from the Hitchhiker’s Guide to the Galaxy: an AI that could not only translate speech to speech in real time across multiple languages, but also preserve their voice, tone, mannerisms, and emotions. So, Meta assembled a team of over 50 people and got busy building it.

What this team came up with was a next-gen translation system called Seamless. The first building block of this system is described in Wednesday’s issue of Nature; it can translate speech among 36 different languages.

Language data problems

AI translation systems today are mostly focused on text, because huge amounts of text are available in a wide range of languages thanks to digitization and the Internet. Institutions like the United Nations or European Parliament routinely translate all their proceedings into the languages of all their member states, which means there are enormous databases comprising aligned documents prepared by professional human translators. You just needed to feed those huge, aligned text corpora into neural nets (or hidden Markov models before neural nets became all the rage) and you ended up with a reasonably good machine translation system. But there were two problems with that.

The first issue was those databases comprised formal documents, which made the AI translators default to the same boring legalese in the target language even if you tried to translate comedy. The second problem was speech—none of this included audio data.

The problem of language formality was mostly solved by including less formal sources like books, Wikipedia, and similar material in AI training databases. The scarcity of aligned audio data, however, remained. Both issues were at least theoretically manageable in high-resource languages like English or Spanish, but they got dramatically worse in low-resource languages like Icelandic or Zulu.

As a result, the AI translators we have today support an impressive number of languages in text, but things are complicated when it comes to translating speech. There are cascading systems that simply do this trick in stages. An utterance is first converted to text just as it would be in any dictation service. Then comes text-to-text translation, and finally the resulting text in the target language is synthesized into speech. Because errors accumulate at each of those stages, the performance you get this way is usually poor, and it doesn’t work in real time.

A few systems that can translate speech-to-speech directly do exist, but in most cases they only translate into English and not in the opposite way. Your foreign language interlocutor can say something to you in one of the languages supported by tools like Google’s AudioPaLM, and they will translate that to English speech, but you can’t have a conversation going both ways.

So, to pull off the Star Trek universal translator thing Meta’s interviewees dreamt about, the Seamless team started with sorting out the data scarcity problem. And they did it in a quite creative way.

Building a universal language

Warren Weaver, a mathematician and pioneer of machine translation, argued in 1949 that there might be a yet undiscovered universal language working as a common base of human communication. This common base of all our communication was exactly what the Seamless team went for in its search for data more than 70 years later. Weaver’s universal language turned out to be math—more precisely, multidimensional vectors.

Machines do not understand words as humans do. To make sense of them, they need to first turn them into sequences of numbers that represent their meaning. Those sequences of numbers are numerical vectors that are termed word embeddings. When you vectorize tens of millions of documents this way, you’ll end up with a huge multidimensional space where words with similar meaning that often go together, like “tea” and “coffee,” are placed close to each other. When you vectorize aligned text in two languages like those European Parliament proceedings, you end up with two separate vector spaces, and then you can run a neural net to learn how those two spaces map onto each other.

But the Meta team didn’t have those nicely aligned texts for all the languages they wanted to cover. So, they vectorized all texts in all languages as if they were just a single language and dumped them into one embedding space called SONAR (Sentence-level Multimodal and Language-Agnostic Representations). Once the text part was done, they went to speech data, which was vectorized using a popular W2v (word to vector) tool and added it to the same massive multilingual, multimodal space. Of course, each embedding carried metadata identifying its source language and whether it was text or speech before vectorization.

The team just used huge amounts of raw data—no fancy human labeling, no human-aligned translations. And then, the data mining magic happened.

SONAR embeddings represented entire sentences instead of single words. Part of the reason behind that was to control for differences between morphologically rich languages, where a single word may correspond to multiple words in morphologically simple languages. But the most important thing was that it ensured that sentences with similar meaning in multiple languages ended up close to each other in the vector space.

It was the same story with speech, too—a spoken sentence in one language was close to spoken sentences in other languages with similar meaning. It even worked between text and speech. So, the team simply assumed that embeddings in two different languages or two different modalities (speech or text) that are at a sufficiently close distance to each other are equivalent to the manually aligned texts of translated documents.

This produced huge amounts of automatically aligned data. The Seamless team suddenly got access to millions of aligned texts, even in low-resource languages, along with thousands of hours of transcribed audio. And they used all this data to train their next-gen translator.

Seamless translation

The automatically generated data set was augmented with human-curated texts and speech samples where possible and used to train multiple AI translation models. The largest one was called SEAMLESSM4T v2. It could translate speech to speech from 101 source languages into any of 36 output languages, and translate text to text. It would also work as an automatic speech recognition system in 96 languages, translate speech to text from 101 into 96 languages, and translate text to speech from 96 into 36 languages—all from a single unified model. It also outperformed state-of-the-art cascading systems by 8 percent in a speech-to-text and by 23 percent in a speech-to-speech translations based on the scores in Bilingual Evaluation Understudy (an algorithm commonly used to evaluate the quality of machine translation).

But it can now do even more than that. The Nature paper published by Meta’s Seamless ends at the SEAMLESSM4T models, but Nature has a long editorial process to ensure scientific accuracy. The paper published on January 15, 2025, was submitted in late November 2023. But in a quick search of the arXiv.org, a repository of not-yet-peer-reviewed papers, you can find the details of two other models that the Seamless team has already integrated on top of the SEAMLESSM4T: SeamlessStreaming and SeamlessExpressive, which take this AI even closer to making a Star Trek universal translator a reality.

SeamlessStreaming is meant to solve the translation latency problem. The baseline SEAMLESSM4T, despite all the bells and whistles, worked as a standard AI translation tool. You had to say what you wanted to say, push “translate,” and it spat out the translation. SeamlessStreaming was designed to take this experience a bit closer to what human simultaneous translator do—it translates what you’re saying as you speak in a streaming fashion. SeamlessExpressive, on the other hand, is aimed at preserving the way you express yourself in translations. When you whisper or say something in a cheerful manner or shout out with anger, SeamlessExpressive will encode the features of your voice, like tone, prosody, volume, tempo, and so on, and transfer those into the output speech in the target language.

Sadly, it still can’t do both at the same time; you can only choose to go for either streaming or expressivity, at least at the moment. Also, the expressivity variant is very limited in supported languages—it only works in English, Spanish, French, and German. But at least it’s online so you can go ahead and give it a spin.

Nature, 2025.  DOI: 10.1038/s41586-024-08359-z

Photo of Jacek Krywko

Jacek Krywko is a freelance science and technology writer who covers space exploration, artificial intelligence research, computer science, and all sorts of engineering wizardry.

Meta takes us a step closer to Star Trek’s universal translator Read More »

is-humanity-alone-in-the-universe?-what-scientists-really-think.

Is humanity alone in the Universe? What scientists really think.

News stories about the likely existence of extraterrestrial life, and our chances of detecting it, tend to be positive. We are often told that we might discover it any time now. Finding life beyond Earth is “only a matter of time,” we were told in September 2023. “We are close” was a headline from September 2024.

It’s easy to see why. Headlines such as “We’re probably not close” or “Nobody knows” aren’t very clickable. But what does the relevant community of experts actually think when considered as a whole? Are optimistic predictions common or rare? Is there even a consensus? In our new paper, published in Nature Astronomy, we’ve found out.

During February to June 2024, we carried out four surveys regarding the likely existence of basic, complex, and intelligent extraterrestrial life. We sent emails to astrobiologists (scientists who study extraterrestrial life), as well as to scientists in other areas, including biologists and physicists.

In total, 521 astrobiologists responded, and we received 534 non-astrobiologist responses. The results reveal that 86.6 percent of the surveyed astrobiologists responded either “agree” or “strongly agree” that it’s likely that extraterrestrial life (of at least a basic kind) exists somewhere in the universe.

Less than 2 percent disagreed, with 12 percent staying neutral. So, based on this, we might say that there’s a solid consensus that extraterrestrial life, of some form, exists somewhere out there.

Scientists who weren’t astrobiologists essentially concurred, with an overall agreement score of 88.4 percent. In other words, one cannot say that astrobiologists are biased toward believing in extraterrestrial life, compared with other scientists.

When we turn to “complex” extraterrestrial life or “intelligent” aliens, our results were 67.4 percent agreement, and 58.2 percent agreement, respectively for astrobiologists and other scientists. So, scientists tend to think that alien life exists, even in more advanced forms.

These results are made even more significant by the fact that disagreement for all categories was low. For example, only 10.2 percent of astrobiologists disagreed with the claim that intelligent aliens likely exist.

Is humanity alone in the Universe? What scientists really think. Read More »