This “black hole universe” idea uses everyday physics to rethink cosmic origins. Instead of starting with expansion and tracing back to a mysterious singularity, the model looks at what happens when a large mass collapses under gravity, similar to how stars become black holes. But unlike the sharp singularities predicted by classical physics, this model uses quantum mechanics to show that the collapse doesn’t go on forever.

The bounce is driven by the quantum exclusion principle, which prevents identical particles, such as fermions, from being squeezed into the same quantum state. As a result, the collapse hits a limit and then reverses, causing an outward bounce. According to the research team, “the bounce is not only possible – it’s inevitable under the right conditions.”

The model explores a spherical collapse of mass M with an initial size χ_* inside a curved region defined by k ≡ 1/χₖ² ≤ 1/χ_²*. The material inside is treated as a perfect fluid that shifts from pressureless dust (P = 0) to a stable energy density ρG over time. This leads to a bounce at: R_B = (8πGρG / 3)^−1/2

After the bounce, the universe goes through rapid expansion, with the pressure P(ρ) acting like an inflation potential—similar to how standard cosmology describes early universe inflation and today’s dark energy.

The model also predicts a small but noticeable positive curvature in space: −0.07 ± 0.02 ≤ Ωₖ

Interestingly, the bounce happens inside the gravitational radius r_S = 2GM, which acts like a cosmological constant Λ from within. Outside, it still looks like a normal Schwarzschild black hole.

Future missions such as Euclid could test the predicted curvature. Other projects, like Arrakihs, will study faint features including stellar halos and satellite galaxies, possibly tied to ancient compact objects like black holes that made it through the bounce.

In this view, the Big Bang wasn’t the birth of everything—it was the start of a new cycle inside a black hole formed in a larger universe. As The Conversation puts it: “We are not witnessing the birth of everything from nothing, but rather the continuation of a cosmic cycle.”

Source: The Conversation, American Physical Society

This article was generated with some help from AI and reviewed by an editor. Under Section 107 of the Copyright Act 1976, this material is used for the purpose of news reporting. Fair use is a use permitted by copyright statute that might otherwise be infringing.

Source

Hope you enjoyed this news post.

News posts... 2023: 5,800+ | 2024: 5,700+ | 2025 (till end of June): 2,864

RIP Matrix | Farewell my friend  

Using a highly advanced version of the OSIRIS (short for Outdoor Scene and InfraRed Image Simulation) sim software, the team recreated a phenomenon called vacuum four-wave mixing. In this process, the electromagnetic fields from three strong laser pulses polarize the virtual particles in the vacuum, causing photons to bounce off one another—resulting in a fourth laser beam.

“This is not just an academic curiosity – it is a major step toward experimental confirmation of quantum effects that until now have been mostly theoretical,” said Professor Peter Norreys from Oxford’s Department of Physics.

What makes this work timely is the global rollout of multi-Petawatt laser systems that can generate extremely strong electromagnetic fields. Facilities like Vulcan 20-20 in the UK, ELI in Europe, and SHINE and SEL in China, along with the OPAL (optical parametric amplifier line) dual-beam laser in the U.S., are expected to hit the power levels needed to see these rare quantum effects in actual experiments.

To make their simulations more accurate, the researchers used a semi-classical numerical solver based on the Heisenberg-Euler Lagrangian. This approach allowed them to model two major quantum vacuum effects and check their results against known predictions for vacuum birefringence—a phenomenon where light splits or shifts as it passes through a strong electromagnetic field.

They tested both plane-wave and Gaussian laser pulses, and found their outputs matched well with existing theories. For the four-wave mixing case, they used three Gaussian beams and were able to track the formation of the fourth beam over time. The simulation also showed a bit of astigmatism—where the output beam wasn’t perfectly shaped—and gave clear measurements of how long the interaction lasted and how big the affected area was.

“Our computer program gives us a time-resolved, 3D window into quantum vacuum interactions that were previously out of reach,” said lead author Zixin Zhang, a doctoral student at Oxford. “By applying our model to a three-beam scattering experiment, we were able to capture the full range of quantum signatures, along with detailed insights into the interaction region and key time scales.”

The team compared their results with simpler models and past data to make sure everything checked out. These tools are expected to help scientists design real-life experiments, with more control over laser timing, shape, and direction.

Professor Luis Silva, co-author from Instituto Superior Técnico and Visiting Professor at Oxford, said: “A wide range of planned experiments at the most advanced laser facilities will be greatly assisted by our new computational method implemented in OSIRIS. The combination of ultra-intense lasers, state-of-the-art detection, cutting-edge analytical and numerical modelling are the foundations for a new era in laser-matter interactions, which will open new horizons for fundamental physics.”

The simulation tool may also help in the search for new particles, such as axions and millicharged particles, which are considered strong candidates for dark matter.

Source: Oxford University, Nature

This article was generated with some help from AI and reviewed by an editor. Under Section 107 of the Copyright Act 1976, this material is used for the purpose of news reporting. Fair use is a use permitted by copyright statute that might otherwise be infringing.

Source

Hope you enjoyed this news post.

News posts... 2023: 5,800+ | 2024: 5,700+ | 2025 (till end of June): 2,864

RIP Matrix | Farewell my friend  

If you want to get into stargazing in 2025, there’s no better place to start than viewing a meteor shower. Meteor showers, or shooting stars, happen when Earth’s orbital path crosses a path of debris left by a comet and that material burns up in the Earth’s atmosphere. Watching a meteor shower is one of the most accessible ways to engage with the night sky.

The next showers of the year are the Southern Delta Aquariids—which peak for about a week at the end of July—and the Perseids—which will peak overnight on August 12–13. Both showers will be active until August 23.

These are just two of nine major meteor showers that will grace skies in 2025, and details of when they will appear in the northern hemisphere are listed below—so mark your calendar for these.

How to Watch a Meteor Shower

You don’t need any special equipment to see a meteor shower—in fact, using devices like binoculars or telescopes actually prevents you from seeing meteors, because they travel too fast to be seen through the lenses of such equipment. All you need are your eyes, a dark sky with little to no moonlight, and a location that’s away from excess light, as moonlight and light pollution can wash out shooting stars.

Note that the moon appears (rises) and disappears (sets) in the night sky at different times depending on what time zone you are in. All moonrise/moonset times in this piece are for the eastern US. You can use tools like Time and Date’s moonrise/moonset calendar or this tool from the US Naval Observatory to check the precise moonrise/moonset times in your exact location.

You should allow your eyes about half an hour to adjust to the darkness. If you need to use a flashlight while outside, use one with red light instead of white to preserve your night vision.

Each meteor shower is named after its radiant, or the constellation that the shower appears to come from. A meteor shower’s radiant usually needs to be above the horizon before you can see the meteors. You don’t need to look directly at the radiant to see meteors; shooting stars will be visible throughout the entire sky once the radiant has risen.

If you need help finding a shower’s radiant, you can use an app like Stellarium, which can also tell you when the radiant will be above the horizon in your exact location. If you really want to maximize the number of meteors you’ll see, you should watch the sky when the shower’s radiant reaches its highest point in the sky. However, you don’t need to wait until the radiant is at its highest to enjoy the show—as long as the radiant is above the horizon, you should be able to see plenty of shooting stars.

The Next Big Meteor Showers

The Southern Delta Aquariids (July–August)

The Southern Delta Aquariids are active from about July 12 to August 23, producing peak activity for about a week centered on July 29–30. Unlike some other meteor showers, the Southern Delta Aquariids don’t have a sharp peak. Instead, the number of meteors per hour gradually increases and then slowly decreases during the period of activity.

Fortunately, the last week and a half of July will yield perfect viewing conditions: The new moon falls on July 23, so there will be little to no moonlight on the nights surrounding this date—plus, in the US the moon will set by midnight until the first couple nights of August, so there is ample opportunity to observe this meteor shower under moonless skies.

On July 26, the approximate start date for the Southern Delta Aquariids’ peak activity, the moon will be about 4 percent illuminated and will set around 10 pm. During the midpoint of the Southern Delta Aquariids’ peak, on July 29–30, the moon will be about 24 percent illuminated and set at about 11 pm. Toward the end of the Southern Delta Aquariids’ peak, on the night of August 2–3, the moon will be about 70 percent full and will set around 1 am.

In other words, the best time to view the shower will probably be in the lead-up to its peak rather than afterward. After the peak, observing conditions will suddenly deteriorate in the first week of August, when the moon will be almost full and won’t set until the early morning hours.

The Southern Delta Aquariids’ radiant is the constellation Aquarius, which rises around 10 pm local time and reaches its highest point in the sky around 3 am.

Although this meteor shower typically yields meteors that are somewhat dimmer and do not have persistent trains, this meteor shower is still worthwhile: You’ll be able to see about 25 shooting stars per hour in ideal viewing conditions.

The Perseids (July–August)

The Perseids are active from about July 17 to August 23, peaking overnight on August 12–13.

The Perseids are one of the strongest and brightest meteor showers of the year, producing 100 to 150 meteors per hour under dark skies. However, the number of meteors drops off sharply after the peak.

The Perseids often yield bright fireball meteors, and about a third of the Perseids have persistent trains. Shooting stars in this shower are also known for being particularly colorful: Most have a green or bluish color, but these meteors can produce yellow, red, purple, or pink hues as well.

Unfortunately, the Perseids will peak just a few nights after the full moon. On August 12–13, the moon will be about 86 percent illuminated and will rise around 10 pm in the US, so viewing conditions will be poor.

However, the Perseids are quite bright, so it might still be worth it to catch this meteor shower. If you do plan on watching, find a spot where the moonlight is blocked out as much as possible.

The Perseids appear to radiate from the constellation Perseus, which rises around 11 pm local time and will be highest in the sky just before dawn.

Coming Up Later in 2025

The Orionids (September–November)

The Orionids are active from around September 26 to November 22. According to the stargazing website Starwalk, the Orionids have a gradual peak due to the angle at which Earth crosses the path of this trail of comet debris. The Orionids produce peak activity for about a week, centered on the night of October 21–22.

The Orionids typically yield about 20 to 25 meteors per hour during their peak and are known for being particularly bright—many of the Orionids are fireball meteors. Like the Eta Aquariids, the Orionids are also debris left behind by Halley’s Comet.

The Orionids appear to radiate from the constellation Orion, which rises around 11 pm local time and is highest in the sky just before dawn.

The new moon falls on the night of October 21–22, and in the surrounding week the moon will set well before midnight, so you will have perfect viewing conditions to see this meteor shower.

The Leonids (November–December)

The Leonids are active from about November 3 to December 2. They have a sharp peak, producing the most meteors overnight from November 16 to November 17, according to the American Meteor Society. Other organizations, however, predict that this shower will peak from November 17 to November 18. During the Leonids’ peak, you can expect to see about 15 meteors per hour under dark skies.

Although the Leonids produce fewer meteors than many other of the major meteor showers, they are known for producing fast-moving, bright, fireball meteors.

The Leonids peak just before the new moon—on the morning of November 18, the moon will be just 6 percent illuminated and in the eastern US won’t rise until around 5 am on November 17, so you’ll have ample time to see this meteor shower under perfect viewing conditions.

The Leonids’ radiant is the constellation Leo, which rises around midnight local time and is highest in the sky around dawn.

The Geminids (December)

The Geminids are active from about December 4 to December 17, peaking overnight from December 13 to December 14. They have a sharp peak, so the night of the 13th is the best time for skywatching.

The Geminids are the most spectacular meteor shower of the year. In addition to boasting up to 120 or even 150 meteors per hour during its peak, this meteor shower is also the brightest and most colorful of the year.

The Geminids are bright, slow-moving meteors that often have yellow tones, but they can be a range of other colors, including green, blue, white, red, or orange. And unlike most meteors, which are caused by comet debris, the Geminids are the remnant of an asteroid.

The night that the Geminids peak, their radiant, the constellation Gemini, will be above the horizon all night and will reach its highest point around 2 am local time, so meteors will be visible almost the whole night.

That same night, the moon will be about 32 percent illuminated and will rise around 1:30 am in the eastern US, so if you watch this shower shortly after midnight, the moonlight won’t interfere with your viewing experience.

The Ursids (December)

The Ursids are active around December 17 to December 26, peaking in the early morning hours of December 22. This meteor shower is less active than others, typically yielding about 10 meteors per hour; however, viewing conditions will be perfect for skywatching. The moon will set at approximately 6 pm in the eastern US on the 21st, so no moonlight will interfere with this meteor shower.

Even though the Ursids typically produce the most meteors just before dawn, when its radiant, the Little Dipper (or Ursa Minor), is highest in the sky, you will be able to see meteors throughout the entire night during this shower’s peak. In northern latitudes the Ursids’ radiant is above the horizon all night.

Showers to Look Out for Next Year

The Quadrantids (January)

The Quadrantids take place in December and January and peak during the first week of the year. This meteor shower has a sharp peak, meaning that most of its activity occurs in a narrow window of time. The Quadrantids typically produce many fireball meteors—that is, meteors that are very bright—with up to 120 meteors per hour during the shower’s peak.

The Quadrantids’ radiant is the constellation Quadrans Muralis, though the International Astronomical Union no longer recognizes this group of stars as a constellation. In its place is the constellation Boötes, which is next to the Big Dipper.

The Lyrids (April)

The Lyrids are active in the second half of April, with their peak lasting around three nights. You can expect to see about 15 to 20 meteors per hour under ideal viewing conditions during the Lyrids’ peak. Under optimal viewing conditions, the stargazing website Earth Sky notes, about a quarter of Lyrids shooting stars produce persistent trains—lingering streaks of light that are the result of gases being ionized as the meteors enter Earth’s atmosphere.

The Lyrids’ radiant, the constellation Hercules, rises well before midnight, so meteors are visible all night, but are most likely to be seen just before dawn, when the radiant reaches its highest point in the sky.

The Eta Aquariids (May)

The Eta Aquariids are active approximately from mid-April to the end of May. This meteor shower does not have a sharp peak: Elevated activity lasts about a week, with activity peaking for one night in the first week of May.

In the northern hemisphere, the Eta Aquariids are a medium-strength shower that produces about 10 to 30 meteors per hour. According to the American Meteor Society, many of these meteors produce persistent trains. One other thing that makes the Eta Aquariids extra special is that these meteors are actually remnants of the famous Halley’s Comet.

The Eta Aquariids’ radiant, the constellation Aquarius, appears very low in the sky in the northern hemisphere in April and won’t start to peak above the eastern horizon until after 2 am local time. However, meteors from this shower are still visible even when the radiant is just below the horizon.

Source

Hope you enjoyed this news post.

News posts... 2023: 5,800+ | 2024: 5,700+ | 2025 (till end of June): 2,864

RIP Matrix | Farewell my friend  

A 51-year-old man showed up at a hospital in Germany looking as though he was wasting away, with swelling and tenderness in his ankles and knees. Then, his heart stopped.

Doctors were able to resuscitate him. Then, they got to work trying to figure out what was wrong. The man told them that for three months he had been suffering from diarrhea, weight loss, joint pain, and fever. His case was reported in this week's issue of the New England Journal of Medicine.

Blood tests didn't detect any infection, but imaging of his heart told a different story. Doctors saw "vegetation" on both his aortic valve and mitral valve. Vegetations are clumps or masses that often build up from an infection, generally containing a bundle of proteins, platelets, and infecting germs stuck together. While they cause damage where they are, if they fully dislodge, they threaten to move to other parts of the body, such as the brain or lungs, and cause dangerous blockages. In the man's case, the vegetation on his aortic valve appeared mobile.

The man was quickly sent to emergency surgery to replace his valves. Once removed, the diseased valves were sent for testing to see what was in those dangerous masses. The result likely came as a surprise to the doctors.

The man had in his heart Tropheryma whipplei, a very common environmental bacterium that dwells in soil. Only in exceedingly rare cases does it cause an infection—but when it does it's a systemic, chronic, and sometimes life-threatening one called Whipple's disease. The condition affects about one to three people in a million, most often middle-aged Caucasian men, like the patient in this case. Overall, 85 percent of Whipple's disease cases are in men.

Curious condition

So, how can such a common germ also cause such a rare infection? Researchers think it's due to genetic predisposition and a glitch in immune responses. Many people likely get infected with T. whipplei as kids, and have either an asymptomatic or limited gastrointestinal infection. They then develop protective immune responses. But in the few people who develop Whipple's disease, this process seems to go awry. Researchers hypothesize that white blood cells called macrophages—which normally engulf and destroy invading pathogens—aren't able to finish the job. They engulf T. whipplei, but don't neutralize the germ. When this happens, the immune system doesn't generate protective antibodies against the bacterium, and inflammation ratchets up. This, in turn, leads to the development of a systemic infection.

T. whipplei is capable of infecting any organ in the body, but tends to go for the joints first. In classic cases of Whipple's disease, patients go through a period of episodic attacks that seem to spring from joint to joint. This period of the disease lasts, on average, seven to eight years. And, given how common arthritis is and how rare Whipple's disease is, getting a proper diagnosis is extremely difficult, if not impossible.

The next phase of the disease involves what the man in this case described: fever, diarrhea, weight loss, and abdominal pain. This is when Whipple's disease is usually caught. But the infection is also known to spread to the heart, as it did in his case, and, in the worst cases, the nervous system. In the latter case, headache and cognitive disturbances are the key signs.

For the man's case, doctors confirmed the diagnosis of Whipple's disease with the standard method of an intestinal biopsy to look for foamy-looking macrophages loaded with T. whipplei. They found them.

The treatment for Whipple's disease is intense and long-term antibiotic treatments, as the condition is known to relapse. The man was started on the standard antibiotic regime, which can last years. However, two months after he was discharged from the hospital, he once again developed a heart infection from Whipple's disease and needed a repeat double-valve replacement. It's unclear how things turned out from there.

Source

Hope you enjoyed this news post.

News posts... 2023: 5,800+ | 2024: 5,700+ | 2025 (till end of June): 2,864

RIP Matrix | Farewell my friend  

Welcome to Edition 8.02 of the Rocket Report! It's worth taking a moment to recognize an important anniversary in the history of human spaceflight next week. Fifty years ago, on July 15, 1975, NASA launched a three-man crew on an Apollo spacecraft from Florida and two Russian cosmonauts took off from Kazakhstan, on course to link up in low-Earth orbit two days later. This was the first joint US-Russian human spaceflight mission, laying the foundation for a strained but enduring partnership on the International Space Station. Operations on the ISS are due to wind down in 2030, and the two nations have no serious prospects to continue any partnership in space after decommissioning the station.

As always, we welcome reader submissions. If you don't want to miss an issue, please subscribe using the box below (the form will not appear on AMP-enabled versions of the site). Each report will include information on small-, medium-, and heavy-lift rockets, as well as a quick look ahead at the next three launches on the calendar.

Sizing up Europe's launch challengers. The European Space Agency has selected five launch startups to become eligible for up to 169 million euros ($198 million) in funding to develop alternatives to Arianespace, the continent's incumbent launch service provider, Ars reports. The five small launch companies ESA selected are Isar Aerospace, MaiaSpace, Rocket Factory Augsburg, PLD Space, and Orbex. Only one of these companies, Isar Aerospace, has attempted to launch a rocket into orbit. Isar's Spectrum rocket failed moments after liftoff from Norway on a test flight in March. None of these companies is guaranteed an ESA contract or funding. Over the next several months, ESA and the five launch companies will negotiate with European governments for funding leading up to ESA's ministerial council meeting in November, when ESA member states will set the agency's budget for at least the next two years. Only then will ESA be ready to sign binding agreements.

Let's rank 'em ... Ars Technica's space reporters ranked the five selectees for the European Launcher Challenge in order from most likely to least likely to reach orbit. We put Munich-based Isar Aerospace, the most well-funded of the group, at the top of the list after attempting its first orbital launch earlier this year. Paris-based MaiaSpace, backed by ArianeGroup, comes in second, with plans for a partially reusable rocket. Rocket Factory Augsburg, another German company, is in third place after getting close to a launch attempt last year before its first rocket blew up on a test stand. Spanish startup PLD Space is fourth, and Britain's Orbex rounds out the list. (submitted by EllPeaTea)

Japan's Interstellar Technologies rakes in more cash. Interstellar Technologies raised 8.9 billion yen ($61.8 million) to boost the development of its Zero rocket and research and development of satellite systems, Space News reports. The money comes from Japanese financial institutions, venture capital funds, and debt financing. Interstellar previously received funding through agreements with the Japanese government and Toyota, which Interstellar says will add expertise to scale manufacturing of the Zero rocket for "high-frequency, cost-effective launches." The methane-fueled Zero rocket is designed to deploy a payload of up to 1 metric ton (2,200 pounds) into low-Earth orbit. The unfortunate news from Interstellar's fundraising announcement is that the company has pushed back the debut flight of the Zero rocket until 2027.

Straight up ... Interstellar has aspirations beyond launch vehicles. The company is also developing a satellite communications business, and some of the money raised in the latest investment round will go toward this segment of the company. Interstellar is open about comparing its ambition to that of SpaceX. "On the satellite side, Interstellar is developing communications satellites that benefit from the company's own launch capabilities," the company said in a statement. "Backed by Japan’s Ministry of Internal Affairs and Communications and JAXA’s Space Strategy Fund, the company is building a vertically integrated model, similar to SpaceX’s approach with Starlink."

Korean startup completes second-stage qual testing. South Korean launch services company Innospace says it has taken another step toward the inaugural launch of its Hanbit-Nano rocket by the year's end with the qualification of the second stage, Aviation Week & Space Technology reports. The second stage uses an in-house-developed 34-kilonewton (7,643-pound-thrust) liquid methane engine. Innospace says the engine achieved a combustion time of 300 seconds, maintaining stability of the fuel and oxidizer supply system, structural integrity, and the launch vehicle integrated control system.

A true micro-launcher ... Innospace's rocket is modest in size and capacity, even among its cohorts in the small launch market. The Hanbit-Nano rocket is designed to launch approximately 200 pounds (90 kilograms) of payload into Sun-synchronous orbit. "With the success of this second stage engine certification test, we have completed the development of the upper stage of the Hanbit-Nano launch vehicle," said Kim Soo-jong, CEO of Innospace. "This is a very symbolic and meaningful technological achievement that demonstrates the technological prowess and test operation capabilities that Innospace has accumulated over a long period of time, while also showing that we have entered the final stage for commercial launch. Currently, all executives and staff are doing their best to successfully complete the first stage certification test, which is the final gateway for launch, and we will make every effort to prepare for a smooth commercial launch in the second half of the year."

Two companies forge unlikely alliance in Dubai. Two German entrepreneurs have joined forces with a team of Russian expats steeped in space history to design a rocket using computational AI models, Payload reports. The "strategic partnership" is between LEAP 71, an AI-enabled design startup, and Aspire Space, a company founded by the son of a Soviet engineer who was in charge of launching Zenit rockets from the Baikonur Cosmodrome in Kazakhstan in the 1980s. The companies will base their operations in Dubai. The unlikely pairing aims to develop a new large reusable launch vehicle capable of delivering up to 15 metric tons to low-Earth orbit. Aspire Space is a particularly interesting company if you're a space history enthusiast. Apart from the connections of Aspire's founder to Soviet space history, Aspire's chief technology officer, Sergey Sopov, started his career at Baikonur working on the Energia heavy-lift rocket and Buran space shuttle, before becoming an executive at Sea Launch later in his career.

Trust the computer ... It's easy to be skeptical about this project, but it has attracted an interesting group of people. LEAP 71 has just two employees—its two German co-founders—but boasts lofty ambitions and calls itself a "pioneer in AI-driven engineering." As part of the agreement with Aspire Space, LEAP 71 will use a proprietary software program called Noyron to design the entire propulsion stack for Aspire's rockets. The company says its AI-enabled design approach for Aspire's 450,000-pound-thrust engine will cut in half the time it took other rocket companies to begin test-firing a new engine of similar size. Rudenko forecasts Aspire's entire project, including a launcher, reusable spacecraft, and ground infrastructure to support it all, will cost more than $1 billion. So far, the project is self-funded, Rudenko told Payload. (submitted by Lin Kayser)

Russia launches ISS resupply freighter. A Russian Progress supply ship launched July 3 from the Baikonur Cosmodrome in Kazakhstan atop a Soyuz-2.1a rocket, NASASpaceflight reports. Packed with 5,787 pounds (2,625 kilograms) of cargo and fuel, the Progress MS-31 spacecraft glided to an automated docking at the International Space Station two days later. The Russian cosmonauts living aboard the ISS will unpack the supplies carried inside the Progress craft's pressurized compartment. This was the eighth orbital launch of the year by a Russian rocket, continuing a downward trend in launch activity for the Russian space program in recent years.

Celebrating a golden anniversary ... The Soyuz rocket that launched Progress MS-31 was painted an unusual blue and white scheme, as it was originally intended for a commercial launch that was likely canceled after Russia's invasion of Ukraine. It also sported a logo commemorating the 50th anniversary of the Apollo-Soyuz mission in July 1975.

Chinese rocket moves closer to first launch. Chinese commercial launch firm Orienspace is aiming for a late 2025 debut of its Gravity-2 rocket following a recent first-stage engine hot fire test, Space News reports. The "three-in-one" hot fire test verified the performance of the Gravity-2 rocket's first stage engine, servo mechanisms, and valves that regulate the flow of propellants into the engine, according to a press release from Orienspace. The Gravity-2 rocket's recoverable and reusable first stage will be powered by nine of these kerosene-fueled engines. The recent hot fire test "lays a solid foundation" for future tests leading up to the Gravity-2's inaugural flight.

Extra medium ... Orienspace's first rocket, the solid-fueled Gravity-1, completed its first successful flight last year to place multiple small satellites into orbit. Gravity-2 is a much larger vehicle, standing 230 feet (70 meters) tall, the same height as SpaceX's Falcon 9 rocket. Orienspace's new rocket will fly in a core-only configuration or with the assistance of two solid rocket boosters. An infographic released by Orienspace in conjunction with the recent engine hot fire test indicates the Gravity-2 rocket will be capable of hauling up to 21.5 metric tons (47,400 pounds) of cargo into low-Earth orbit, placing its performance near the upper limit of medium-lift launchers.

Senator calls out Texas for trying to steal space shuttle. A political effort to remove space shuttle Discovery from the Smithsonian and place it on display in Texas encountered some pushback on Thursday, as a US senator questioned the expense of carrying out what he described as a theft, Ars reports. "This is not a transfer. It's a heist," said Sen. Dick Durbin (D-Ill.) during a budget markup hearing before the Senate Appropriations Committee. "A heist by Texas because they lost a competition 12 years ago." In April, Republican Sens. John Cornyn and Ted Cruz, both representing Texas, introduced the "Bring the Space Shuttle Home Act" that called for Discovery to be relocated from the National Air and Space Museum's Steven F. Udvar-Hazy Center in northern Virginia and displayed at Space Center Houston. They then inserted an $85 million provision for the shuttle relocation into the Senate version of the "One Big Beautiful Bill," which, to comply with Senate rules, was more vaguely worded but was meant to achieve the same goal. That bill was enacted on July 4, when President Donald Trump signed it into law.

Dollar signs ... As ridiculous as it is to imagine spending $85 million on moving a space shuttle from one museum to another, it'll actually cost a lot more to do it safely. Citing research by NASA and the Smithsonian, Durbin said that the total was closer to $305 million, and that did not include the estimated $178 million needed to build a facility to house and display Discovery once it was in Houston. Furthermore, it was unclear if Congress even has the right to remove an artifact, let alone a space shuttle, from the Smithsonian's collection. The Washington, DC, institution, which serves as a trust instrumentality of the US, maintains that it owns Discovery. The paperwork signed by NASA in 2012 transferred "all rights, interest, title, and ownership" for the spacecraft to the Smithsonian. "This will be the first time ever in the history of the Smithsonian someone has taken one of their displays and forcibly taken possession of it. What are we doing here? They don't have the right in Texas to claim this," said Durbin.

Starbase keeps getting bigger. Cameron County, Texas, has given SpaceX the green light to build an air separator facility, which will be located less than 300 feet from the region’s sand dunes, frustrating locals concerned about the impact on vegetation and wildlife, the Texas Tribune reports. The commissioners voted 3–1 to give Elon Musk’s rocket company a beachfront construction certificate and dune protection permit, allowing the company to build a facility to produce gases needed for Starship launches. The factory will separate air into nitrogen and oxygen. SpaceX uses liquid oxygen as a propellant and liquid nitrogen for testing and operations.

Saving the roads ... By having the facility on site, SpaceX hopes to make the delivery of those gases more efficient by eliminating the need to have dozens of trucks deliver them from Brownsville. The company says they need more than 200 trucks of liquid nitrogen and oxygen delivered for each launch, a SpaceX engineer told the county during a meeting last week. With their application, SpaceX submitted a plan to mitigate expected negative effects on 865 square feet of dune vegetation and 20 cubic yards of dunes, as well as compensate for expected permanent impacts to 7,735 square feet of dune vegetation and 465 cubic yards of dunes. While the project will be built on property owned by SpaceX, the county holds the authority to manage the construction that affects Boca Chica's dunes.

ULA is stacking its third Vulcan rocket. A little more than a week after its most recent Atlas V rocket launch, United Launch Alliance rolled a Vulcan booster to the Vertical Integration Facility at Cape Canaveral Space Force Station in Florida on July 2 to begin stacking its first post-certification Vulcan rocket, Spaceflight Now reports. The operation, referred to by ULA as Launch Vehicle on Stand (LVOS), is the first major milestone toward the launch of the third Vulcan rocket. The upcoming launch will be the first operational flight of ULA's new rocket with a pair of US military payloads, following two certification flights in 2024.

For the second time ... This is the second time that this particular Vulcan booster was brought to Space Launch Complex 41 in anticipation of a launch campaign. It was previously readied in late October of last year in support of the USSF-106 mission, the Space Force's designation for the first national security launch to use the Vulcan rocket. However, plans changed as the process of certifying Vulcan to fly government payloads took longer than expected, and ULA pivoted to launch two Atlas V rockets on commercial missions from the same pad before switching back to Vulcan launch preps.

Progress report on China's Moon rocket. China's self-imposed deadline of landing astronauts on the Moon by 2030 is now just five years away, and we're starting to see some tangible progress. Construction of the launch pad for the Long March 10 rocket, the massive vehicle China will use to launch its first crews toward the Moon, is well along at the Wenchang Space Launch Site on Hainan Island. An image shared on the Chinese social media platform Weibo, and then reposted on X, shows the Long March 10's launch tower near its final height. A mobile launch platform presumably for the Long March 10 is under construction nearby.

Super heavy ... The Long March 10 will be China's most powerful rocket to date, with the ability to dispatch 27 metric tons of payload toward the Moon, a number comparable to NASA's Space Launch System. Designed for partial reusability, the Long March 10 will use an all-liquid propulsion system and stand more than 92 meters (300 feet) tall. The rocket will launch Chinese astronauts inside the nation's next-generation Mengzhou crew capsule, along with a lunar lander to transport crew members from lunar orbit to the surface of the Moon using an architecture similar to NASA's Apollo program.

Next three launches

July 11: Electron | JAKE 4 | Wallops Flight Facility, Virginia | 23:45 UTC

July 13: Falcon 9 | Dror 1 | Cape Canaveral Space Force Station, Florida | 04:31 UTC

July 14: Falcon 9 | Starlink 15-2 | Vandenberg Space Force Base, California | 02:27 UTC

Source

Hope you enjoyed this news post.

News posts... 2023: 5,800+ | 2024: 5,700+ | 2025 (till end of June): 2,864

RIP Matrix | Farewell my friend  

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.

“We were getting basil leaves the size of your palm,” University of Arizona researcher Greg Barron-Gafford said, describing some of the benefits he and his team have seen farming under solar panels in the Tucson desert.

For 12 years, Barron-Gafford has been investigating agrivoltaics, the integration of solar arrays into working farmland. This practice involves growing crops or other vegetation, such as pollinator-friendly plants, under solar panels, and sometimes grazing livestock in this greenery. Though a relatively new concept, at least 604 agrivoltaic sites have popped up across the United States, according to OpenEI.

Researchers like Barron-Gafford think that, in addition to generating carbon-free electricity, agrivoltaics could offer a ray of hope for agriculture in an increasingly hotter and drier Southwest, as the shade created by these systems has been found to decrease irrigation needs and eliminate heat stress on crops. Plus, the cooling effects of growing plants under solar arrays can actually make the panels work better.

But challenges remain, including some farmers’ attitudes about the practice and funding difficulties.

Overcoming a climate conundrum

While renewable electricity from sources like solar panels is one of the most frequently touted energy solutions to help reduce the carbon pollution that’s driving climate change, the warming climate itself is making it harder for solar arrays to do their job, Barron-Gafford said. An optimal functioning temperature for panels is around 75° Fahrenheit, he explained. Beyond that, any temperature increase reduces the photovoltaic cells’ efficiency.

“You can quickly see how this solution for our changing climate of switching to more renewable energy is itself sensitive to the changing climate,” he said.

This problem is especially pertinent in the Southwest, where historically hot temperatures are steadily increasing. Tucson, for instance, saw a record-breaking 112 days of triple-digit heat in 2024, according to National Weather Service Data, and the US Environmental Protection Agency reports that every part of the Southwest experienced higher average temperatures between 2000 and 2023 compared to the long-term average from 1895 to 2023.

However, planting vegetation under solar panels—as opposed to the more traditional method of siting solar arrays on somewhat barren land—can help cool them. In one set of experiments, Barron-Gafford’s team found that planting cilantro, tomatoes and peppers under solar arrays reduced the panels’ surface temperature by around 18 degrees Fahrenheit. That’s because plants release moisture into the air during their respiration process, in which they exchange oxygen for carbon dioxide.

“This invisible power of water coming out of plants was actually cooling down the solar panels,” Barron-Gafford said.

Throwing shade

While Barron-Gafford said some laughed him off when he first proposed the idea of growing crops in the shade of solar panels, this added sun shield can actually help them grow better, especially in the Southwest, where many backyard gardeners already employ shade cloths to protect their gardens from the blazing heat.

“Many people don’t understand that in Colorado and much of the West, most plants get far too much sunlight,” said Byron Kominek, owner/manager of Jack’s Solar Garden in Boulder County, Colorado, which began implementing agrivoltaics in 2020. “Having some shade is a benefit to them.”

Jack’s Solar Garden has integrated 3,276 solar panels over about four acres of farmland, growing crops like greens and tomatoes. Meg Caley with Sprout City Farms, a nonprofit that helps with farming duties at Jack’s Solar Garden, said they’ve been able to produce Swiss chard “the size of your torso.”

“The greens just get huge,” she said. “You have to chop them up to fit them in your refrigerator.”

She added that the shade seems to improve the flavor of the vegetables and prevents them from bolting, when plants prematurely produce flowers and seeds, diverting energy away from leaf or root growth.

“Plants when they’re stressed out can have more of a bitter flavor,” she explained. “So the arugula that we grow is not as bitter or spicy. It’s sweeter. The spinach is sweeter too.”

Barron-Gafford and his team are seeing the same thing in Arizona, where they grow a variety of produce like beans, artichokes, potatoes, kale, and basil.

“We’ve grown 30-plus different types of things across different wet winters and dry winters and exceptionally hot summers, dry summers, average or close to average summers,” he said of the solar-shaded crops. “And across everything we’ve done, we’ve seen equal or greater production down here in the Southwest, the dry land environments, where it really benefits to get some shade.”

As in Colorado, some of those crops are growing to epic proportions.

“We’ve made bok choy the size of a toddler,” Barron-Gafford said.

All that shade provides another important benefit in a drought-stricken Southwest—lower water requirements for crops. Because less direct sunlight is hitting the ground, it decreases the evaporation rate, which means water stays in the soil longer after irrigation. Barron-Gafford and his team have been running experiments for the last seven or so years to see how this plays out with different crops in an agrivoltaic setting.

“What is the evaporation rate under something that’s big and bushy like a bean or potato plant versus something thinner above ground, like a carrot?” is one of the questions Barron-Gafford said they have tried to answer. “For the most part, I would say that we are able to cut back our irrigation by more than half.”

They are partnering with Jack’s Solar Farm on water research in Colorado and have so far found similar results there.

This shade has another benefit in a warming world—respite for farmworkers. Heat-related illnesses are a growing concern for people who work outside, and one recent study predicted climate change will quadruple U.S. outdoor workers’ exposure to extreme heat conditions by 2065.

But with solar arrays in the fields, “if you really carefully plan out your day, you can work in the shade,” a factor that can help increase worker safety on hot days, Caley said.

The AgriSolar Clearinghouse performed skin temperature readings under solar panels and full sun at a number of sites across the United States, finding a skin temperature decrease of 15.3° in Boulder and 20.8° in Phoenix.

“I don’t know what the future holds”

Despite the benefits of agrivoltaics, the up-front cost of purchasing a solar array remains a barrier to farmers.

“Once people see the potential of agrivoltaics, you run into the next challenge, which is how do you fund someone getting into this on their site?” Barron-Gafford said. “And depending on the amount of capital or access to capital that a farmer has, you’re going to get a wildly different answer.”

While expenses are dependent on the size of the installation, a 25-kilowatt system would require an upfront cost of around $67,750, according to AgriSolar Clearinghouse. For comparison, the median size of a residential solar array in 2018 was around 6 kW, the organization stated, which would cost around $16,260 to install.

Kominek said the total initial cost of implementing a 1.2 megawatt capacity agrivoltaics setup on his farm in Colorado was around $2 million, but that the investment has paid off. In addition to the revenue he earns from farming, all of the energy produced by the arrays is sold to clients in the community through a local utility company, earning the farm money.

The Rural Energy for America program has been one resource for farmers interested in agrivoltaics, offering loans and grants to help install solar. However, it’s unclear how this program will move forward amid current federal spending cuts.

Meanwhile, some of the federal grant programs that Barron-Gafford has relied on have suddenly come to a halt, he said, putting his research in danger. But, as federal support dries up, some states are charging on with their own funding opportunities to develop farm field solar projects. For instance, Colorado’s Agrivoltaics Research and Demonstration Grant offers money for demonstrations of agrivoltaics, research projects, and outreach campaigns.

There are other challenges as well. Caley, for instance, said farming around solar panels is akin to working in an “obstacle course.” She and her team, who mostly work manually, have found ways to work around them by being aware of their surroundings so that they don’t accidentally collide with the panels or strike them with their tools. This job is also made easier since Kominek invested between $80,000 and $100,000 to elevate his farm’s panels, which better allows animals, taller crops and farming equipment to operate beneath.

Still, a 2025 University of Arizona study that interviewed farmers and government officials in Pinal County, Arizona, found that a number of them questioned agrivoltaics’ compatibility with large-scale agriculture.

“I think it’s a great idea, but the only thing … it wouldn’t be cost-efficient … everything now with labor and cost of everything, fuel, tractors, it almost has to be super big … to do as much with as least amount of people as possible,” one farmer stated.

Many farmers are also leery of solar, worrying that agrivoltaics could take working farmland out of use, affect their current operations or deteriorate soils.

Those fears have been amplified by larger utility-scale initiatives, like Ohio’s planned Oak Run Solar Project, an 800 megawatt project that will include 300 megawatts of battery storage, 4,000 acres of crops and 1,000 grazing sheep in what will be the country’s largest agrivoltaics endeavor to date. Opponents of the project worry about its visual impacts and the potential loss of farmland.

An American Farmland Trust survey found that Colorado farmers would prefer that utility-scale solar projects be sited on less productive or underutilized farmland rather than on highly productive or actively farmed land. They also expressed concern for the potential negative impact that solar projects could have on farm productivity and the health of the land, including soil quality.

Some farmers also worry that the solar panels could leach metals into the ground, contaminating their crops, Barron-Gafford said. But while agrivoltaic systems are put together in a way that makes that highly unlikely, there’s no reason not to add soil sampling studies into the work they’re doing to reassure farmers, he added.

And agrivoltaics advocates say that the practice could actually improve soil health by reducing erosion, increasing the amount of organic matter and enhancing soil biology with cooler, moister conditions.

“I wish more people spent time listening to the folks on the ground and the folks experiencing these transitions,” Barron-Gafford added. “Because you understand more that way in terms of what their motivations or concerns actually are.”

“We don’t have to choose”

While Caley understands farmers’ concerns, she sees agrivoltaics as a way for them to keep agricultural land in production while also benefiting from solar electricity.

“The tension in a lot of communities seems to be that people don’t want to see agricultural land taken out of production in order to bring a solar farm in,” she said. “The idea here is that we don’t have to choose. We can have both.”

Kominek encourages people to envision what our landscapes and climate will look like in the next 20 to 30 years, adding that in his part of Colorado, it only stands to get hotter and drier, making agrivoltaics a smart solution for farming and clean energy production.

“Communities around the world need to figure out what changes they need to make now to help people adapt to what our climates and landscapes will be in the future,” he said. “Agrivoltaics is a climate adaptation tool that will benefit any community where such systems are built as the decades pass.”

Source

Hope you enjoyed this news post.

News posts... 2023: 5,800+ | 2024: 5,700+ | 2025 (till end of June): 2,864

RIP Matrix | Farewell my friend  

A new version of Mazda's popular CX-5 SUV is on the way. Earlier today, the Japanese automaker revealed details about the third-generation CX-5, which goes on sale in Europe later this year before coming here in 2026.

The current CX-5, first introduced in 2017, marked Mazda's move upmarket, with a renewed focus on elegant interiors and keen handling without luxury automaker prices. Mazda remains committed to its core principle of "Jinba Ittai"—the horse and rider being at one—and the cars remain popular with enthusiasts, but it's fair to say that the available powertrains often leave something to be desired in terms of fuel efficiency.

At one time, Mazda readied a new diesel engine to try to improve its fleet average, although that option disappeared within a couple of years due to minimal demand. And for a while, we were teased with the clever "Skyactiv-X" compression ignition engine, which promised diesel-like efficiency on regular pump gasoline. It seems the odds of that one actually going on sale in the US are now remote, though.

More recently, Mazda borrowed the hybrid powertrain from Toyota's RAV4 and dropped it into the CX-50—conveniently, both SUVs are built in the same shared factory in Huntsville, Alabama. (In exchange, Toyota gets access to Mazda's Soul Red paint for the RAV4, which is a pretty fair swap.)

Both the CX-5 and CX-50 will continue to coexist in dealerships: The former is a global bestseller, and the latter is made in the US for North American tastes. Finally, there will be a hybrid CX-5 to go with the hybrid CX-50, although not until 2027. Not much is known about the new "Skyactiv-Z" engine other than that it will be a four-cylinder gasoline engine that operates at the ideal stoichiometric ratio of air to fuel throughout the rev range.

For 2026, though, the CX-5 will come with Mazda's Skyactiv-G 2.5 L four-cylinder gasoline engine. Mazda has also developed a new generation of infotainment system for the CX-5, joining the growing list of automakers that have adopted Google's Android Automotive OS and Google automotive services.

The addition of a hybrid in 2027 will be welcome, as Mazda has often lagged behind in terms of fuel efficiency.

Mazda

 

Mazda's interiors punch well above their price tag.

Mazda

 

Oooh, I love this two-tone.

Mazda

 

Expect pricing much closer to the car's official US launch in 2026.

Source

Hope you enjoyed this news post.

News posts... 2023: 5,800+ | 2024: 5,700+ | 2025 (till end of June): 2,864

RIP Matrix | Farewell my friend  

Cases of Parkinson’s disease have doubled in the last 25 years, according to figures from the World Health Organization. For decades, the scientists have investigated what triggers this disorder to mitigate its symptoms and anticipate its onset. Now, a series of experimental therapies are laying the groundwork for potentially reversing the condition, which affects nearly 10 million people worldwide and can generate costs of approximately $10,000 per patient per year, when considering direct and indirect medical expenses.

Parkinson’s disease is a degenerative neurological disorder in which cells that produce dopamine in the brain die, causing symptoms such as tremors, muscle stiffness, slowness of movement, and alterations in balance. So far there is no cure, and treatments are limited.

Kay Double, a professor at the University of Sydney’s School of Medical Sciences, has been researching the biological mechanisms underlying this disease for more than a decade, with the aim of finding ways to slow or even halt its progression.

In 2017, he led a study that identified for the first time an abnormal form of a protein called SOD1 in Parkinson’s patients. Under normal conditions, this protein acts as an antioxidant enzyme, protecting brain cells from damage caused by free radicals, highly reactive molecules that contain oxygen and can deteriorate cells if not properly neutralized. Free radicals are produced by natural bodily processes as well as by external factors, like diet, smoking, and exposure to pollution.

In people with Parkinson’s disease, SOD1 suffers alterations that prevent it from fulfilling its protective function, with it instead accumulating in the brain and causing neuronal damage, according to the findings of Double’s team.

Based on these results, the team then conducted further research, with results suggesting that copper supplementation in the brain could be an effective way to slow and even reverse the symptoms of Parkinson’s (copper is crucial to SOD1’s function). To test this hypothesis, they evaluated the efficacy of a drug called CuATSM, designed to cross the blood-brain barrier and deliver copper directly to brain tissue.

This experiment, written up and published in Acta Neuropathologica Communications, was divided into two phases. The first was to determine the optimal dose of the drug to induce a response in the brain. To find this, CuATSM was administered daily for three weeks to 27 eight-week-old wild-type mice, with concentrations of copper and other metals then measured in the mice’s tissues. This revealed that 15 milligrams per kilogram was the ideal dose to effectively increase the levels of copper in the brain.

In the second stage, this dose was applied to 10 mice genetically modified to develop Parkinson’s-like symptoms. The animals were divided into two groups: one received CuATSM daily for three months, while the other received a placebo without the active ingredient.

The results showed that the mice treated with the placebo experienced a deterioration in their motor skills. In contrast, those that received the copper supplement showed no alterations in their movement. It appears the treatment corrected the dysfunctions of SOD1 and restored its protective properties. In the mice receiving the copper treatment, dopamine neurons were preserved in an area of the brain called the substantia nigra, an area essential for the control of movement, coordination, learning, and certain cognitive functions.

“All of the mice we treated showed dramatic improvement in their motor skills. The results exceeded our expectations and suggest that, after further study, this therapeutic approach could slow the progression of Parkinson’s in humans,” says Double.

But experts caution that Parkinson’s is a complex condition that will likely require multiple combined interventions. A single treatment may have limited effect, but its efficacy may be enhanced by integrating it with other therapeutic approaches.

In that context, Double’s team’s findings could be complemented by recent research from Stanford University focused on restoring communication between neurons in a subtype of Parkinson’s linked to mutations in the gene responsible for producing an enzyme called LRRK2.

In these cases, the mutation causes hyperactivity of the enzyme, altering the structure of brain cells and disrupting signaling between dopaminergic neurons and those in the striatum, a deep brain region related to movement, motivation, and decision-making.

It is estimated that about 25 percent of Parkinson’s cases are genetic in origin, and the LRRK2 mutation is one of the most frequent. The team led by Stanford neuroscientist Suzanne Pfeffer proposed that inhibiting the excessive activity of this enzyme could stabilize symptoms, especially if detected in early stages. The goal was to regenerate primary cilia, antenna-like structures that enable communication between cells.

The hypothesis was tested in mice genetically modified to exhibit LRRK2 hyperactivity and early symptoms of the disorder. For two weeks, these animals were administered with a compound called MLi-2, which binds to the enzyme and reduces its activity.

In this first test, no relevant changes were observed, which the researchers attributed to the fact that the examined neurons and glia—another type of cell in the nervous system, which support neurons—were already mature and were not in the cell division phase.

However, a review of the scientific literature revealed that, even if mature, certain neurons can regenerate their primary cilia depending on their sleep-wake cycles. “The findings that other nonproliferative cells can develop cilia made us think that the inhibitor still had therapeutic potential,” Pfeffer explains.

The team then decided to extend the treatment to three months. After this period, they found that the percentage of neurons and glial cells in the striatum with primary cilia was comparable to that of healthy mice without the genetic mutation.

This restoration of cellular structures made it possible to reactivate communication between dopaminergic neurons and the striatum. As a result, neurotransmitters affected by the LRRK2 protein induced the production of neuroprotective factors at levels similar to those of a healthy brain, something that had been diminished as a result of LRRK2 hyperactivity. In addition, density markers of dopaminergic nerve endings were doubled, suggesting a possible recovery of previously damaged neurons.

“These findings suggest that it is not only possible to stabilize the disease, but also to improve the condition of patients. This therapeutic approach has great potential to restore neuronal activity in Parkinson’s-affected circuits. There are currently several ongoing clinical trials with LRRK2 inhibitors, and we hope that these results in mice can be translated to humans,” says Pfeffer.

The authors stress that, to maximize the effectiveness of this treatment, it is essential to identify early symptoms, which can occur up to 15 years before the characteristic tremors. The hope is that people with the LRRK2 mutation will be able to start treatment early. The next step would be to assess whether other Parkinson’s variants, not associated with this genetic mutation, could also benefit from this strategy.

It is estimated that the number of Parkinson’s cases worldwide could exceed 25 million by 2050, which would represent a 112 percent increase over 2021 figures, according to projections published in the British Medical Journal. Although these estimates are not definitive, the scientific community warns that they reflect a growing challenge for public health systems. For this reason, developing therapies capable of mitigating, stabilizing, and even reversing the progression of the disease is a global priority.

This story originally appeared on WIRED en Español and has been translated from Spanish.

Source

Hope you enjoyed this news post.

News posts... 2023: 5,800+ | 2024: 5,700+ | 2025 (till end of June): 2,864

RIP Matrix | Farewell my friend  

James McCully was in the lab extracting tiny structures called mitochondria from cells when researchers on his team rushed in. They’d been operating on a pig heart and couldn’t get it pumping normally again.

McCully studies heart damage prevention at Boston Children’s Hospital and Harvard Medical School and was keenly interested in mitochondria. These power-producing organelles are particularly important for organs like the heart that have high energy needs. McCully had been wondering whether transplanting healthy mitochondria into injured hearts might help restore their function.

The pig’s heart was graying rapidly, so McCully decided to try it. He loaded a syringe with the extracted mitochondria and injected them directly into the heart. Before his eyes, it began beating normally, returning to its rosy hue.

Since that day almost 20 years ago, McCully and other researchers have replicated that success in pigs and other animals. Human transplantations followed, in babies who suffered complications from heart surgery—sparking a new field of research using mitochondria transplantation to treat damaged organs and disease. In the last five years, a widening array of scientists have begun exploring mitochondria transplantation for heart damage after cardiac arrest, brain damage following stroke, and damage to organs destined for transplantation.

Mitochondria are best known for producing usable energy for cells. But they also send molecular signals that help to keep the body in equilibrium and manage its immune and stress responses. Some types of cells may naturally donate healthy mitochondria to other cells in need, such as brain cells after a stroke, in a process called mitochondria transfer. So the idea that clinicians could boost this process by transplanting mitochondria to reinvigorate injured tissue made sense to some scientists.

From studies in rabbits and rat heart cells, McCully’s group has reported that the plasma membranes of cells engulf the mitochondria and shuttle them inside, where they fuse with the cell’s internal mitochondria. There, they seem to cause molecular changes that help recover heart function: When comparing blood- and oxygen-deprived pig hearts treated with mitochondria to ones receiving placebos, McCully’s group saw differences in gene activity and proteins that indicated less cell death and less inflammation.

About 10 years ago, Sitaram Emani, a cardiac surgeon at Boston Children's Hospital, reached out to McCully about his work with animal hearts. Emani had seen how some babies with heart defects couldn’t fully recover after heart surgery complications and wondered whether McCully’s mitochondria transplantation method could help them.

During surgery to repair heart defects, surgeons use a drug to stop the heart so they can operate. But if the heart is deprived of blood and oxygen for too long, mitochondria start to fail and cells start to die, in a condition called ischemia. When blood begins flowing again, instead of returning the heart to its normal state, it can damage and kill more cells, resulting in ischemia-reperfusion injury.

Since McCully’s eight years of studies in rabbits and pigs hadn’t revealed safety concerns with mitochondria transplantation, McCully and Emani thought it would be worth trying the procedure in babies unlikely to regain enough heart function to come off heart-lung support.

Parents of 10 patients agreed to the experimental procedure, which was approved by the institute’s review board. In a pilot that ran from 2015 to 2018, McCully extracted pencil-eraser-sized muscle samples from the incisions made for the heart surgery, used a filtration technique to isolate mitochondria and checked that they were functional. Then the team injected the organelles into the baby’s heart.

Eight of those 10 babies regained enough heart function to come off life support, compared to just four out of 14 similar cases from 2002 to 2018 that were used for historical comparison, the team reported in 2021. The treatment also shortened recovery time, which averaged two days in the mitochondrial transplant group compared with nine days in the historical control group. Two patients did not survive — in one case, the intervention came after the rest of the baby’s organs began failing, and in another, a lung issue developed four months later. The group has now performed this procedure on 17 babies.

The transplant procedure remains experimental and is not yet practical for wider clinical use, but McCully hopes that it can one day be used to treat kidney, lung, liver, and limb injuries from interrupted blood flow.

The results have inspired other clinicians whose patients suffer from similar ischemia-reperfusion injuries. One is ischemic stroke, in which clots prevent blood from reaching the brain. Doctors can dissolve or physically remove the clots, but they lack a way to protect the brain from reperfusion damage. “You see patients that lose their ability to walk or talk,” says Melanie Walker, an endovascular neurosurgeon at the University of Washington School of Medicine in Seattle. “You just want to do better and there’s just nothing out there.”

Walker came across McCully’s mitochondrial transplant studies 12 years ago and, in reading further, was especially struck by a report on mice from researchers at Massachusetts General Hospital and Harvard Medical School that showed the brain’s support and protection cells—the astrocytes—may transfer some of their mitochondria to stroke-damaged neurons to help them recover. Perhaps, she thought, mitochondria transplantation could help in human stroke cases too.

She spent years working with animal researchers to figure out how to safely deliver mitochondria to the brain. She tested the procedure’s safety in a clinical trial with just four people with ischemic stroke, using a catheter fed through an artery in the neck to manually remove the blockage causing the stroke, then pushing the catheter further along and releasing the mitochondria, which would travel up blood vessels to the brain.

The findings, published in 2024 in the Journal of Cerebral Blood Flow & Metabolism, show that the infused patients suffered no harm; the trial was not designed to test effectiveness. Walker’s group is now recruiting participants to further assess the intervention’s safety. The next step will be to determine whether the mitochondria are getting where they need to be, and functioning. “Until we can show that, I do not believe that we will be able to say that there’s a therapeutic benefit,” Walker says.

Researchers hope that organ donation might also gain from mitochondria transplants. Donor organs like kidneys suffer damage when they lack blood supply for too long, and transplant surgeons may reject kidneys with a higher risk of these injuries.

To test whether mitochondrial transplants can reinvigorate them, transplant surgeon-scientist Giuseppe Orlando of Wake Forest University School of Medicine in Winston-Salem and his colleagues injected mitochondria into four pig kidneys and a control substance into three pig kidneys. In 2023 in the Annals of Surgery, they reported fewer dying cells in the mitochondria-treated kidneys and far less damage. Molecular analyses also showed a boost in energy production.

It’s still early days, Orlando says, but he’s confident that mitochondria transplantation could become a valuable tool in rescuing suboptimal organs for donation.

The studies have garnered both excitement and skepticism. “It’s certainly a very interesting area,” says Koning Shen, a postdoctoral mitochondrial biologist at the University of California, Berkeley, and coauthor of an overview of the signaling roles of mitochondria in the 2022 Annual Review of Cell and Developmental Biology. She adds that scaling up extraction of mitochondria and learning how to store and preserve the isolated organelles are major technical hurdles to making such treatments a larger reality. “That would be amazing if people are getting to that stage,” she says.

“I think there are a lot of thoughtful people looking at this carefully, but I think the big question is, what’s the mechanism?” says Navdeep Chandel, a mitochondria researcher at Northwestern University in Chicago. He doubts that donor mitochondria fix or replace dysfunctional native organelles, but says it’s possible that mitochondria donation triggers stress and immune signals that indirectly benefit damaged tissue.

Whatever the mechanism, some animal studies do suggest that the mitochondria must be functional to impart their benefits. Lance Becker, chair of emergency medicine at Northwell Health in New York who studies the role of mitochondria in cardiac arrest, conducted a study comparing fresh mitochondria, mitochondria that had been frozen then thawed, and a placebo to treat rats following cardiac arrest. The 11 rats receiving fresh, functioning mitochondria had better brain function and a higher rate of survival three days later than the 11 rats receiving a placebo; the non-functional frozen-thawed mitochondria did not impart these benefits.

It will take more research into the mechanisms of mitochondrial therapy, improved mitochondria delivery techniques, larger trials and a body of reported successes before mitochondrial transplants can be FDA-approved and broadly used to treat ischemia-reperfusion injuries, researchers say. The ultimate goal would be to create a universal supply of stored mitochondria — a mitochondria bank, of sorts — that can be tapped for transplantation by a wide variety of health care providers.

“We’re so much at the beginning—we don’t know how it works,” says Becker. “But we know it’s doing something that is mighty darn interesting.”

This article originally appeared in Knowable Magazine, a nonprofit publication dedicated to making scientific knowledge accessible to all. Sign up for Knowable Magazine’s newsletter.

Source

Hope you enjoyed this news post.

News posts... 2023: 5,800+ | 2024: 5,700+ | 2025 (till end of June): 2,864

RIP Matrix | Farewell my friend  

The European Space Agency has selected five launch startups to become eligible for up to 169 million euros ($198 million) in funding to develop alternatives to Arianespace, the continent's incumbent launch service provider.

The five companies ESA selected are Isar Aerospace, MaiaSpace, Rocket Factory Augsburg, PLD Space, and Orbex. Only one of these companies, Isar Aerospace, has attempted to launch a rocket into orbit. Isar's Spectrum rocket failed moments after liftoff from Norway on a test flight in March.

None of these companies is guaranteed an ESA contract or funding. Over the next several months, the European Space Agency and the five launch companies will negotiate with European governments for funding leading up to ESA's ministerial council meeting in November, when ESA member states will set the agency's budget for at least the next two years. Only then will ESA be ready to sign binding agreements.

In a press release, ESA referred to the five companies as "preselected challengers" in a competition for ESA support in the form of launch contracts and an ESA-sponsored demonstration to showcase upgraded launch vehicles to heave heavier payloads into orbit. So far, all five of the challengers are focusing on small rockets.

Earlier this year, ESA released a request for proposals to European industry for bids to compete in the European Launch Challenge. ESA received 12 proposals from European companies and selected five to move on to the next phase of the challenge.

A new way of doing business

In this competition, ESA is eschewing a rule that governs nearly all of the space agency's other programs. This policy, known as geographic return, guarantees industrial contracts to ESA member states commensurate with the level of money they put into each project. The most obvious example of this is Europe's Ariane rocket family, whose development was primarily funded by France, followed by Germany in second position. Therefore, the Ariane 6 rocket's core stage and engines are built in France, and its upper stage is manufactured in Germany.

European Space officials have cited the rule of geographic return as a reason for delays and cost overruns on ESA projects, including the Ariane 6, which debuted in July 2024, several years behind schedule. Essentially, the policy compels ESA to choose contractors in countries providing funding to each program, regardless of whether they offer the most reliable or cost-effective solution.

With the European Launcher Challenge, ESA is upending this policy by first selecting the launch contractors, then going to their home governments to secure funding for the program.

Two of the challengers selected by ESA are from Germany. Isar Aerospace, headquartered in Munich, is developing a small two-stage orbital launch vehicle named Spectrum. Just to the northwest of Munich, Rocket Factory Augsburg is developing a rocket called RFA One with similar capabilities to Spectrum.

MaiaSpace is owned by ArianeGroup, the parent company of Arianespace, based in France. Spanish company PLD Space is developing a launch vehicle called the Miura 5 and in 2023 launched its first liquid-fueled rocket into the upper atmosphere on a suborbital test flight. Orbex, headquartered in the United Kingdom, is working on a rocket named Prime. All five companies' rockets are currently gearing their efforts toward the small satellite launch market.

The European launcher challenge will include two components, the first of which will be for launch services for ESA missions slated for launch from 2026 through 2030. The second part will be a contract to demonstrate a launch service capacity upgrade, including at least one flight demonstration of the augmented launch vehicle. The cap of 169 million euros per challenger will encompass all activities under both parts of the challenge.

"With this initiative, ESA is taking decisive steps towards commercialization and expansion of launch services, which are essential for ensuring sovereignty in space," Isar Aerospace wrote on X.

ESA said it evaluated several criteria before selecting the five companies that will proceed to the next phase of the challenge. These criteria included technical maturity, business maturity and sustainability, the institutional market each company plans to serve, and compliance with procurement rules.

An infusion of up to 169 million euros would be welcome news to any of the five launch challengers. To date, these companies have primarily relied upon private investment and some funding from their home governments.

The most well-funded and advanced launch startups in Europe are primarily based in Germany, and to a lesser degree, France. Last year, Rocket Factory Augsburg and Isar Aerospace appeared to be neck-and-neck in a race to become the first company to attempt an orbital flight from a launch pad in Western Europe. Isar ultimately won the race, but its Spectrum rocket didn't make it far off the launch pad.

Let's take a look at each of the five challengers. We'll list them in order from most likely to least likely to reach orbit, as ranked by Ars Technica's space reporters.

• Isar Aerospace: Isar Aerospace is at the top of our list after becoming the first company in the group to launch an orbital-class rocket earlier this year. Isar Aerospace is the most well-funded European rocket startup, after raising more than 550 million euros ($645 million) since its founding in 2018. Most recently, Isar raised 150 million euros ($175 million) in the form of a convertible bond agreement with the American firm Eldridge Industries. Isar's Spectrum rocket is designed to place up to 1 metric ton (2,200 pounds) of payload mass into low-Earth orbit, or 700 kilograms (1,543 pounds) into a polar Sun-synchronous orbit. It stands 92 feet (28 meters) tall and is powered by engines burning a mixture of propane and liquid oxygen propellants. After the launch failure in March, Isar says it is building its second and third Spectrum rockets, with plans to return to the launch pad at Andøya Spaceport in northern Norway before the end of 2025.
• MaiaSpace: Among the five challengers, Paris-based MaiaSpace is farthest along on the path toward developing a reusable rocket. This company's launch vehicle, simply named Maia, is designed with a recoverable and reusable booster stage that will land on an offshore barge following liftoff from the Guiana Space Center in South America. Established in 2022, MaiaSpace is 100 percent owned by ArianeGroup, the only existing company in Europe that has developed a liquid-fueled orbital rocket. ArianeGroup, itself a joint venture between Airbus and Safran, has committed 125 million euros ($147 million) to kick-start the design, manufacturing, and testing of the Maia launch vehicle. Maia will stand 164 feet (50 meters) tall, with two stages powered by methane/liquid oxygen engines and the capability to deliver 1.5 metric tons (3,300 pounds) of payload mass to a Sun-synchronous polar orbit in expendable mode. MaiaSpace and its customers will have the option to fly the rocket with or without a booster recovery kit. The company aims to launch the first Maia rocket next year from the old Soyuz launch pad in French Guiana. MaiaSpace's relationship with ArianeGroup, its use of an existing launch facility, and its use of a relatively mature engine—the Prometheus engine well along in development in partnership with the European Space Agency—led us to rank MaiaSpace second among the European launch startups most likely to reach orbit.
• Rocket Factory Augsburg: Third on our list is Rocket Factory Augsburg, headquartered in the Bavarian city of the same name, with plans for a three-stage rocket dubbed RFA One that could eventually deliver up to 1.3 metric tons (2,866 pounds) of payload mass into a polar Sun-synchronous orbit. Rocket Factory Augsburg hoped to launch its first rocket before the end of last year, but the booster stage for its first RFA One rocket was destroyed last August after catching fire on a test stand at SaxaVord Spaceport in Scotland. Rocket Factory Augsburg has not returned to the spaceport to attempt another test-firing of the booster, so it's safe to assume the anomaly set the company back by at least a year. This is starting to put the company's goal of launching before the end of 2025 in doubt. The company says it has already qualified the rocket's second and third stages, while engineers prep another first stage booster for static fire testing. The RFA One rocket is designed to fly with nine first stage engines consuming kerosene and liquid oxygen propellants. It's not easy to get a precise number for how much money RFA has raised since it was founded in 2018, but adding up publicly announced fundraising rounds and government investments brings the total to somewhere near 100 million euros ($117 million).
• PLD Space: PLD Space comes in at No. 4 on our list of European launch challengers. Headquartered in Elche, Spain, PLD Space celebrated its first suborbital launch in October 2023, when the company shot a test rocket into the upper atmosphere from southern Spain in a demonstration of the propulsion technology to be used on its orbital rocket, the partially reusable Miura 5. PLD Space released few technical or schedule updates after the 2023 test flight until last month, when it announced engineers were on the cusp of beginning the "flight qualification campaign" for the Miura 5's kerosene-fueled main engines. PLD Space also signed a contract with the French space agency, CNES, last month to authorize the start of construction work at the Miura 5's launch pad in French Guiana, aiming for a first flight sometime next year. Miura 5 is designed to ferry up to 540 kilograms (1,190 pounds) of payload to Sun-synchronous orbit, and PLD Space last year revealed a bigger medium-lift rocket named Miura Next on the company's roadmap. Founded in 2011, PLD Space has raised more than 160 million euros ($188 million) to date, but the company is highly leveraged, with half of its cash coming in the form of loans.
• Orbex: At first glance, ESA's choice of Orbex is a little puzzling. The rocket Orbex has been working on for the better part of a decade is still nowhere near a launch pad, and just last year, the UK-based company abandoned construction of its primary launch site in favor of SaxaVord Spaceport in Scotland's Shetland Islands. The launch of the first Prime vehicle is at least a year away, but Orbex has released scant information on the status of the rocket's development for the last several years, instead favoring updates on leadership changes, education and outreach opportunities, and the company's participation in conferences. The Prime rocket is relatively modest in performance compared to the other challengers, with a capacity for about 180 kilograms (400 pounds) of payload to low-Earth orbit. Orbex revealed last year it planned to develop a medium-lift rocket named Proxima that will be better suited to ESA's needs. This will be an expensive undertaking, and the company's total fundraising to date of approximately £130 million ($177 million) isn't enough. Orbex isn't in the best position to win all 169 million euros available to it through ESA's launcher challenge, but if it does receive the money, it will be a lifeline that might save the company from failure. ESA's selection of Orbex over other interesting launch challenge contenders, such as HyImpulse of Germany and Latitude of France, provides some geographic diversity to the competition and could pressure the UK government to increase its contributions to ESA.

Source

Hope you enjoyed this news post.

Thank you for appreciating my time and effort posting news every day for many years.

News posts... 2023: 5,800+ | 2024: 5,700+ | 2025 (till end of June): 2,864

RIP Matrix | Farewell my friend  

Many of the stars in the Milky Way galaxy are small, dim red dwarfs—stars much smaller than the sun in both size and mass. TOI-6894, located far away from Earth, is one of them.

Astronomers previously thought a star like this could not have large planets circulating it, because its mass is only about 20 percent of the sun, meaning its planetary system—generated from materials surrounding the star—would not have contained enough mass to form a giant body like Saturn or Jupiter.

But when observing TOI-6894, an international research team detected a clear transit signal—a temporary decrease in a star’s brightness caused by a planet passing across it. This newly discovered planet, named TOI-6894b, blocks 17 percent of the star’s light, indicating the planet is fairly large. The signal was picked up by the Transiting Exoplanet Survey Satellite (TESS), an observation instrument launched by NASA to hunt for planets orbiting stars outside of our solar system.

This makes TOI-6894 “the lowest mass star known to date to host such a planet,” said Edward Bryant, Astrophysics Prize Fellow at the University of Warwick, in a press statement. The finding appears to upend conventional theory on how planets are formed. “This discovery will be a cornerstone for understanding the extremes of giant planet formation,” Bryant said.

Astronomers at University College London and the University of Warwick, as part of a global collaboration with partners in Chile, the US, and Europe, trawled through the data of about 91,000 red dwarf stars observed by TESS before discovering the planet TOI-6894b. After that, the nature of TOI-6894b was clarified by additional observations made with other telescopes. According to these, TOI-6894b’s radius is slightly larger than Saturn’s, but its mass is only about half that of the ringed giant. Its density is extremely light at only 0.33 g/cm³, indicating that it is an expanding gas planet.

TOI-6894 is nearly 40 percent smaller than the previous record for the smallest star with a planet of this size. This fact poses a serious contradiction to conventional theories of planet formation.

The widely accepted planetary formation model, the “core-accumulation theory,” proposes that a ring of dust and rocks—known as protoplanetary disk—forms around a star, and that materials in this disk then gather together to form the cores of planets. After starting out this way, larger gas planets then accrete gases around their cores to become gigantic. But if the mass of the star is small, the mass of its protoplanetary disk tends to be small as well. In such a scenario, the nucleus necessary for the formation of a giant gas planet will not grow.

Based on this theory, it is estimated that more than 120 times more solid matter than that of the Earth would be required to form TOI-6894b. However, the observed disk surrounding the star TOI-6894 contains only 58 times the mass of the Earth at most. This raises the possibility of an alternative planet-formation mechanism existing.

One suggestion by Bryant and the team is that planets could form by gradual gas accumulation, without the initial formation of a massive core. “Alternatively, it could have formed because of a gravitationally unstable disc,” Bryant says. “In some cases, the disc surrounding the star will become unstable due to the gravitational force it exerts on itself. These discs can then fragment, with the gas and dust collapsing to form a planet.”

However, the press statement goes on to explain that neither of these theories can could completely explain the formation of TOI-6894b, based on the data gathered. For now the planet’s origin remains a mystery.

Analyzing the atmosphere of the newly discovered exoplanet might yield answers; it is possible that chemical traces of its formation process remain. Observations of TOI-6894b by the James Webb Space Telescope over the next 12 months are expected to reveal details about its interior structure and atmospheric composition, which will have a major say on whether current theories about planet formation are supported or new ones are needed.

The discovery may force us to rethink our unified model of planet formation, says Andrés Jordán of the Millennium Astrophysics Institute at Adolfo Ibáñez University in Chile. Years of steady observations are pushing the limits of theory, and it may even be time to fundamentally rethink our estimates of the total number of giant planets in the galaxy.

This story originally appeared on WIRED Japan and has been translated from Japanese.

Source

Hope you enjoyed this news post.

Thank you for appreciating my time and effort posting news every day for many years.

News posts... 2023: 5,800+ | 2024: 5,700+ | 2025 (till end of June): 2,864

RIP Matrix | Farewell my friend  

Part of the fun of 3D printing is discovering just how many possibilities there are for different things to print. Obviously, they're fun for printing toys or decorations that you couldn't or wouldn't buy yourself, but they're also powerful problem-solving tools. Once you've solved a few problems with 3D printed parts, you start looking around for other minor inconveniences or quality-of-life upgrades that you could solve—and the breadth and depth of the 3D printing community means that you can almost always find someone else who has already thought up and posted a solution for you.

As a coda to our series about breaking into 3D printing for the first time, the 3D printer-pilled among the Ars staff is sharing a few of their favorite unexpected prints, from fun all-purpose gifts to containers and organizers to parts that will help you with your other, non-3D-printing-related hobbies. This is just a fraction of what's out there, but if you're still on the fence, maybe some of these will open your mind to the possibilities.

Coffee gear

Coffee bean dosing cups and espresso tamper handle 

Credit: Aurich Lawson

Every morning, I make either a pour-over coffee or some form of espresso. For measuring my beans, I printed two dosing cups. The black one is matte black PLA with a fuzzy surface texture (an option in most slicers that adds random noise to the outside wall paths), and the white one is ABS that I sanded to a smooth surface. For sanding, I prefer ABS, as it’s easier to get something that has no real signs of layer lines. To tamp my espresso grounds, I printed a handle in black ABS and sanded it smooth to feel good in the hand. The rounded knob helps me get pressure more comfortably than the raw metal of the original tamper, and the radial fins fit perfectly into the dosing cup, keeping the tamp straight up and down so I don’t end up with a sloped surface.

These were all files I downloaded from MakerWorld, and I didn’t really do anything to them except minor scaling or adding the fuzzy skin.

—Aurich Lawson, Creative Director

Even more organizational tools

My very first 3D prints were new organizational tools to try and impose some order on the chaos of my home and office, and my favorite prints still tend to be of that genre.

Cleaning out and fully organizing my desk with 3D-printed baskets and containers is still on my long to-do list, but I did manage to tame the loose pile of USB sticks and memory cards in my desk with one of the many available organizer designs. This Gridfinity-compatible design is the one I went for, but there are truly dozens of examples on MakerWorld alone; I like this one because it can hold a lot of USB-A drives and because each individual slot is versatile enough to hold USB drives or SD or microSD cards. But there are examples with more USB-C ports and some with different dimensions and spacing, so you can find the one that works best for the space you're trying to fit it into.

Having a third sunglasses-wearer in the house (and one with multiple Bluey sunglasses) also made it necessary to find some kind of way to easily put them away and keep them from floating around the living room or car and getting lost forever. I really like the versatile and modular SnapStack Modular Glasses Holder design, which gives you designs for a base and a top, and then you print as many sunglasses holders as you need; if you need to expand later on, just print another one or pop the top off and add to the one you've already made.

We had enough things to store that I went right for this three-sided version of the stand, which I printed to be able to hold nine pairs (and which is large enough that you can rest a sunglasses case or something else on the top). I stuck a few small adhesive furniture pads to the bottom to prevent damage to the table. But if you have fewer, you can print free-standing or wall-mounted versions, too.

—Andrew Cunningham, Senior Technology Reporter

Aerogarden baskets and Mario mushrooms

So, so many Aerogarden baskets.

Credit: Lee Hutchinson

I have two fun 3D printer things to share—one is a life/money hack kind of thing, and the other is just neat.

On the life/money hack thing, my wife is a big Aerogarden kind of person—we have probably two dozen or more of the hydroponic plant doodads all over the house in various sizes, from tiny to "one wall of the kitchen." She raises small plants in the Aerogarden(s) and then transfers them outside to the real garden; doing this means she was buying lots of special little Aerogarden baskets for the baby plants to take root in.

That sounded like a job for a 3d printer! And sure enough, Thingiverse came to the rescue! In the two years we've had our Bambu Lab X1 Carbon, I've printed probably a thousand or more of these things, in 27-lot batches because that's how many will fit on a single build plate.

I got mushrooms and companion cubes for days!

Credit: Lee Hutchinson

The other thing that has brought delight, honestly, is this little screw-top Mario 1-Up mushroom (at least, I think that's the same one as the one I've been printing—it's hard to tell, but it looks the same). It's a little silly, but these things are not only really fun to fidget with—the top comes off and you can hide stuff in them!—but they also make fantastic little gifts for folks, especially anyone with kids and/or Gen-X sensibilities. Everyone needs more screw-top 1-Up mushrooms in their lives, and they work great in tons of different colors!

— Lee Hutchinson, Senior Technology Editor

Festool track hangers

Festool track hanger and protective end cap.

Aurich Lawson

 

The knob rotates on an embedded bolt to allow the track to be released or locked back in place.

Aurich Lawson

 

I have three different tracks for my Festool tracksaw that I like to hang on my garage wall. It keeps them from getting dinged up, and they are easily accessible when I’m ready to cut with them. For these, I modeled my own designs in Fusion 360, with the main body printed in matte black PLA and the knob printed in a green HTPLA called Lootsef by Protopasta. That's "Festool" spelled backward, of course, and it's designed to pretty much perfectly match Festool’s signature green.

I used nuts embedded in the main body and bolts through the knobs to allow them to be turned to lock or release the track in place. I modeled the Festool logo into the top of the knob and used the ironing option in Bambu Studio to use the printer’s hotend to smooth the top surface around the logo.

The protective end caps were printed in the same HTPLA from a file someone uploaded to Printables.

—Aurich Lawson, Creative Director

Gridfinity all the things!

Gridfinity is a modular, grid-based storage and organization system that’s optimized for 3D printing and rapid customization. Created by Zack Freedman, Gridfinity uses a standardized 42x42 mm base grid upon which you can place highly adaptable tool trays, organizers, and workspace layouts.

The upshot is that you can print anything from a little 1x1x1 cube (42 mm³) to a massive storage bin the size of your print bed. If your desk, kitchen, or bathroom drawers scream out for organization, this is a good solution because you can print exactly what you want.

The Gridfinity Generator has you covered when it comes to printing a custom base grid. This parametric gridfinity tool is a great place to start printing bins, particularly if you’re in a situation where you can shave a few grams of filament off your design (desk bins, for instance, can typically use very thin walls).

—Ken Fisher, Editor-In-Chief

Green PETG for your green thumb

After several years of dashed hopes and false starts, I was finally able to get a single raised garden bed going in our backyard this year (among other things, a raised bed is a bit easier to protect from the wildlife in our backyard and simpler to use with the Square Foot Gardening system). The 3D printer contributed a few odds and ends, including parts that helped add strength to the enclosure I built around it and tools that helped me keep the cage's corners (mostly) square.

But now that some of the plants are actually going, the 3D printer's main contribution to the cause has been 3D-printed cages, which I've been using to get my vining plants to grow upward instead of outward (necessary for the close quarters of square-foot gardening) and to keep things from flopping over onto the ground.

As with the desk organizers, there are many options for plant cages and trellises, depending on the size of your plants, what you're trying to grow, and your aesthetic and functional preferences. I'm giving these circular stackable ones a try since I like that they can easily be printed continuously based on how high your plants want to get, though for big ol' tomato plants, you’ll still want a stake in the ground to help bear the weight once the plants are more than a few feet high.

If you do this—and especially if you're using an open-bed printer like my Bambu Labs A1, which doesn't handle filament like the UV-resistant ASA well—you'll want to make sure to print using PETG plastic instead of the typical PLA. PETG can be fussier than PLA (it's more prone to stringing, especially if you're not drying your filament rolls), but it's also less prone to warping after extended sunlight exposure, it's modestly UV-resistant, and it has a bit more flexibility and resiliency than the more brittle PLA plastic.

—Andrew Cunningham, Senior Technology Reporter

Tool drawer organization

Assorted bins for organizing a tool drawer.

Aurich Lawson

 

Close-up of the bins with labels printed on a Brother PT-D600.

Aurich Lawson

 

I also liked the idea of Gridfinity, but I found the 42 mm size a little awkward—and yes, it’s a Hitchhiker’s Guide reference, not a spec built around the size of human fingers. I modeled my own system in Fusion 360 based loosely on the idea, but with a 50 mm grid that I laser-cut out of cardboard to avoid having to print it. The containers are printed in matte black and white PLA, with a color switch using my X1C’s AMS multi-spool system to get the white tops. There’s no function to the white; I just thought it looked nice with the labels.

I modeled custom holders for another drawer to hold my screwdrivers and hex wrenches. Having the perfect shape to fit the screwdrivers is slightly overkill, but it’s super satisfying to drop them into place and watch them settle exactly into place. There's a metric and imperial holder for the hex wrenches, each removable, so I can take them with me to find the right fit when I'm working on something. All the holders lock into the same 50 mm grid as the bins.

—Aurich Lawson, Creative Director

My main squeeze

Sometimes you stumble across things you didn’t know you needed. For me, that’s this Toothpaste Squeezer. You can print one or a dozen of them in no time. They’re simple yet effective.

Will it change your life? No. But it will give you that satisfying feeling of dealing with a beautifully primed tube of toothpaste every time. Even my in-laws use these now (or so they say). If you want something a little more hefty with a built-in ratchet, check this one out.

—Ken Fisher, Editor-In-Chief

Corral your remote controls

Even if you have a decent universal remote, chances are good that you still need your other remotes nearby. This remote control stand is easy to print, looks great, and offers a few customization choices. It also prints in multicolor without an AMS, so you can match your decor quite easily. And I’m pleased to note that it holds the fat TiVo remote with no problems.

—Ken Fisher, Editor-In-Chief

The Armorer helmet

This is the way.

Aurich Lawson

 

The inside of the helmet is unfinished, so you can see the raw 3D print lines and seams where the front, back, and top dome were glued together.

Aurich Lawson

 

In addition to practical prints, I like to make display props, especially Star Wars helmets. I don’t wear them for cosplay or anything; I just like having them around to look at and enjoy. I have several shelves full now, and I like to use a combination of ABS and resin to print them for the various advantages in post-processing and detail. This Armorer helmet from The Mandalorian is the first helmet I did, before I had my Bambu X1C, and it was printed in PLA on my Prusa. I later printed the horns in resin, but they could have been done in PLA and sanded smooth easily enough.

I’m including this helmet instead of any of my others because I wanted to show that you can make something like this with any bed slinger printer. You don’t need an enclosure or a large-format printer—this was printed in sections and glued together—and you don’t need fancy or toxic materials like ABS and resin.

There was a lot of sanding, filler primer, bondo, and several different passes of automotive paints, plus a two-part catalyst clear coat to finish it off. But you could get a lot of this look with rattle cans, without the need for a compressor and spray gun.

—Aurich Lawson, Creative Director

Source

Hope you enjoyed this news post.

News posts... 2023: 5,800+ | 2024: 5,700+ | 2025 (till end of June): 2,864

RIP Matrix | Farewell my friend  

One doesn't normally associate ballistics with botany, but most of us don't study "squirting" cucumbers—so called because they disperse their seeds by explosively propelling them out into the world. Scientists took a series of high-speed videos, both in the wild and in the lab, to learn more about the underlying biomechanics of this plant's method of seed dispersal. Graduate student Helen Gorges of Kiel University's Zoological Institute in Germany presented the findings at the Society for Experimental Biology Annual Conference in Antwerp, Belgium.

Also known as the "noli me tangere," aka "touch me not," the squirting cucumber (Ecballium elaterium) is often considered a weed or invasive species, although in some regions it's viewed as ornamental. Fun fact: The fruit extract is a powerful laxative. If swallowed or inhaled through the nose, it can be poisonous, causing edemas and necrosis of the nasal mucosa, among other complications. That same fruit, once ripened, can squirt out a stream of mucus-like liquid containing seed pods at high speeds—an example of rapid plant movement.

As glucosides in the sap of the fruit's tissue cells build up, so does the internal pressure, eventually causing the fruit to detach from the stalk. At that point, the pericarp contracts, and both the fruit and the seeds are violently expelled through the resulting hole. The squirting action is further aided by structural changes in the fruit as it dehydrates and its cells coil, bend, or twist in response (hygroscopic movement).

It's actually not the most effective means of seed dispersal, per a 2019 study. That's good news for almond orchards, for example, since farmers can target their weed-killing efforts to the most likely affected areas. And the plant tissue tends to fracture from the force of the ballistic seed dispersal. “Many factors have to interact perfectly to disperse the seeds in the most efficient way, while not destroying the whole plant too early,” said Gorges, who wanted to learn more about the biomechanics that control the fruit as it ripens and prepares for seed dispersion.

Gorges et al. used microcomputed tomography to build 3D models of the fruit and then used micro-CT imaging and high-speed videography with very high resolution to capture the fruit as it exploded. “It’s super interesting to watch through high-speed recordings, as the explosions happen way too fast to see anything in real-time," said Gorges. That footage also enabled them to calculate how fast the seeds were traveling and estimate shooting distances.

The results: Seeds can travel as fast as 29 mph and as far as 12 meters (about 39 feet). Such research could one day lead to better hydrogel-based actuators for medical tools and micro-robots, as well as drug delivery systems. Images taken of the ripening fruit allowed them to analyze the curvature of the fruit stem, as well as the angle between the fruit and stem. This revealed that the stem is vertical during ripening, with an angle of 53 degrees on average. A 50-degree angle is deemed the theoretical ideal in order to maximize shooting distance.

Gorges has also been studying the seeds' cellulose nano fibers, which—depending on the level of hydration—can exhibit either high or low friction or adhesion. Learning more could lead to the development of bio-inspired stable surfaces with tailored adhesion or friction levels or new high-performance glues, for instance.

Source

Hope you enjoyed this news post.

Thank you for appreciating my time and effort posting news every day for many years.

News posts... 2023: 5,800+ | 2024: 5,700+ | 2025 (till end of June): 2,864

RIP Matrix | Farewell my friend  

Two Chinese satellites have rendezvoused with one another more than 20,000 miles above the Earth in what analysts believe is the first high-altitude attempt at orbital refueling.

China's Shijian-21 and Shijian-25 satellites, known as SJ-21 and SJ-25 for short, likely docked together in geosynchronous orbit sometime last week. This is the conclusion of multiple civilian satellite trackers using open source imagery showing the two satellites coming together, then becoming indistinguishable as a single object.

Chinese officials have released no recent public information on what the two satellites are up to, but they've said a bit about their missions in prior statements.

SJ-25, which launched in January, is designed "for the verification of satellite fuel replenishment and life extension service technologies," according to the Shanghai Academy of Spaceflight Technology, the Chinese state-owned contractor that developed the satellite. SJ-21 launched in 2021 and docked with a defunct Chinese Beidou navigation satellite in geosynchronous orbit, then towed it to a higher altitude for disposal before returning to the geosynchronous belt. Chinese officials described this demonstration as a test of "space debris mitigation" techniques.

More than meets the eye

These kinds of technologies are dual-use, meaning they have civilian and military applications. For example, a docking in geosynchronous orbit could foretell an emerging capability for China to approach, capture, and disable another country's satellite. At the same time, the US Space Force is interested in orbital refueling as it seeks out ways to extend the lives of military satellites, which are often limited by finite fuel supplies.

The Space Force sometimes calls this concept dynamic space operations. While some military leaders remain skeptical about the payoff of in-space refueling, the Space Force has an agreement with Astroscale to perform the first refueling of a US military asset in orbit as soon as next year.

China appears to be poised to beat the US Space Force to the punch. The apparent docking of the two satellites last week suggests SJ-21 is the target for SJ-25's refueling demonstration, and US officials are watching. Two of the Space Force's inspector satellites, known by the acronym GSSAP, positioned themselves near SJ-21 and SJ-25 to get a closer look.

Retired Space Force Lt. Gen. John Shaw is a vocal proponent of dynamic space operations. Because of this, he's interested in what happens with SJ-21 and SJ-25. Shaw was deputy commander of US Space Command before his retirement in 2023. In this role, Shaw had some oversight over GSSAP satellites as they roamed geosynchronous orbit.

"The theory behind dynamic space operations stemmed from a kind of operational frustration with our inability to conduct the full range of activities with GSSAP that we wanted to at Space Command, as the warfighter—largely due to the combination of fixed fuel availability and expected satellite lifetime," Shaw told Ars.

As other countries, mainly China, step up their clandestine activities in orbit, military officials are asking more of the GSSAP satellites.

"It was operationally driven then, a couple years ago, but it's now manifesting itself in much wider ways than even it did back then, particularly in the face of activities by potential adversaries," Shaw said. "That’s why I’m more confident and even more zealous about it."

Geosynchronous orbit is a popular location for military and commercial satellites. At an altitude of some 22,236 miles (35,786 kilometers), a satellite's orbital velocity perfectly matches the speed of Earth's rotation, meaning a spacecraft has a fixed view of the same region of the planet 24 hours per day. This is useful for satellites providing military forces with secure strategic communications and early warning of missile attacks.

Now, geosynchronous orbit is becoming a proving ground for new kinds of spacecraft to inspect or potentially attack other satellites. Ground-based anti-satellite missiles aren't as useful in striking targets in high-altitude orbits, and there's a consensus that, if you were to attack an enemy satellite, it would make more sense to use a weapons platform already in space that could move in and connect with the target without blowing it up and creating a cloud of dangerous space junk.

Keeping watch

The US military's GSSAP satellites began launching in 2014. They carry enough propellant to maneuver around geosynchronous orbit and approach objects for closer inspection, but there's a limit to what they can do. Six GSSAP satellites have been launched to date, but the Space Force decommissioned one of them in 2023. Meanwhile, China's satellite operators are watching the watchers.

"We’ve seen where GSSAP safely and responsibly approaches a Chinese vehicle, and it just quickly maneuvers away," Shaw said. "We tend to fly our GSSAPs like dirigibles, using relatively slow, minimum energy transfer approaches. The Chinese know that we do that, so it is relatively easy for them to maneuver away today to avoid such an approach.

"If tomorrow they’re able to refuel at will and operate even more dynamically, then the marginal cost of those maneuvers for them becomes even lower, and the challenge for GSSAP becomes even greater," Shaw said.

China launched a satellite into geosynchronous orbit in 2016 with a robotic arm that could grab onto another object in space, then sent SJ-21 into orbit four years ago on its "space debris mitigation" mission.

Northrop Grumman launched two satellites in 2019 and 2020 that accomplished the first dockings in geosynchronous orbit. Northrop's satellites, which it calls Mission Extension Vehicles, took control of two aging commercial communications satellites running low on fuel, maneuvering them to new locations and allowing them to continue operating for several more years. It's easy to see that this kind of technology could be used for commercial or military purposes.

But these Mission Extension Vehicles don't have the ability to transfer fluids from one satellite to another. That is the step China is taking with SJ-21 and SJ-25, presumably with hydrazine and nitrogen tetroxide propellants, which most satellites use because they combust on contact with one another.

US Space Command's Joint Commercial Operations cell, which collects unclassified satellite monitoring data to bolster the military's classified data sources, estimated the SJ-21 and SJ-25 satellites "merged" on July 2 and have remained together since then. The video below, released by s2a systems, shows SJ-25 approaching SJ-21 on June 30.

The unclassified data does not confirm that the two satellites actually docked, but that is likely what happened. The satellites came together, or merged, on June 13 and June 30 but separated again within a few hours. These may have been practice runs, aborted docking attempts, or sudden maneuvers to avoid the prying eyes of the US military's GSSAP satellites loitering nearby.

Now, the SJ-21 and SJ-25 have been flying together for more than five days with no discernible changes detected from ground-based telescopes. Thousands of miles over the equator, the two satellites appear only as dots in the viewfinders of these telescopes positioned around the globe.

What we don’t know

COMSPOC is a Pennsylvania-based company that collects and processes data from commercial satellite tracking sensors. COMSPOC fuses optical telescope imagery with radar tracking and passive radio frequency (RF) data, which uses radio signals to measure exact distances to satellites in space, to get the best possible estimate of a spacecraft's position.

"With most telescopes... at 1 kilometer or a half a kilometer, somewhere in there, you're going to start to lose it when they get that close," said Paul Graziani, COMSPOC's founder and CEO, in an interview with Ars. "I think it'd be difficult for any telescope, even a really capable one, to get within 100 meters. That seems to be a stretch for telescopes."

That's why it's helpful to add radar and RF data to the mix.

"When you add all of that together, you become much better than the 1-kilometer [precision] that a 'scope might be," said Joe Callaro, COMSPOC's director of operations. "RF tells you if part of that blob is moving and the other part isn't, and even when they all become one pixel, you can tell things about that."

Even then, companies like COMSPOC have a degree of uncertainty in their conclusions unless Chinese or US officials make a more definitive statement.

"We are not working with the government," Callaro told Ars before last week's apparent docking. "We are not clearing this. The charge that I have for my team is we won't make assertions as to what's going on. We will only tell what our software gives us as a solution. We can say, 'Here are the elements, here's the visual, but what it means and what it's doing, we will not assert.'

"We will not say they're docked because unless they told me, I wouldn't know that," Callaro said. "So, we will say they've been together for this amount of time, that the mission could have happened, and then they separated, became two, and separated at whatever speed."

SJ-21's behavior for the last couple of years suggested it was running empty after undertaking large propulsive maneuvers to capture the Chinese Beidou satellite and move it to a different orbit.

Callaro served as a tactician in the Air Force's Joint Space Operations Center, then joined the Aerospace Corporation before taking the job as operations lead at COMSPOC. He doesn't buy China's suggestion that SJ-21 was purely an experiment in collecting space debris.

"That is not how I see that at all," Callaro said. "The fact that we can calculate all the maneuvers it takes to get out and get back, and the fact that afterwards, it spent a couple of years basically not moving, probably because it was low on fuel, sets up the idea [that there's more to SJ-21's mission]. Now, SJ-25 goes out there, and it's supposed to be a fuel tank, and it's perfectly aligned with SJ-21 and now we see this happening, tells me that it's much more a counter-space capability than it is a trash remove. But that's what they say."

Unless China makes a public statement on the refueling of SJ-21 by SJ-25, observers won't know for sure if the servicing demo was successful until the satellites detach. Then, US officials and independent analysts will watch to see if SJ-21 makes any substantial maneuvers, which might indicate the satellite has a full tank of gas for whatever mission Chinese officials send it off to do next.

Source

Hope you enjoyed this news post.

Thank you for appreciating my time and effort posting news every day for many years.

News posts... 2023: 5,800+ | 2024: 5,700+ | 2025 (till end of June): 2,864

RIP Matrix | Farewell my friend  

A vast network of stars, gas, and dust is strung 

among a duo of star clusters in this combined image 

from NASA’s Hubble and Webb space telescopes.  

Credit: NASA, ESA, and C. Lindberg (The Johns 

Hopkins University); Processing: Gladys Kober 

(NASA/Catholic University of America)

Open clusters of stars—which consist of dozens up to a few thousand stars—are an interesting tool for astronomers to study the Universe.

That's because all of the stars in such a cluster formed more or less at the same time, allowing astronomers to compare different types of stars, in terms of size and composition, which are all of a similar age. This is useful for understanding how different kinds of stars evolve over time.

Some of these open clusters are pretty famous, such as the Pleiades cluster, also known as the Seven Sisters. This is relatively close to Earth, just 444 light-years away. Others are much more distant, such as NGC 460 and NGC 456. They reside in a nearby galaxy, the Small Magellanic Cloud, and are the subject of today's post.

NASA has shared side-by-side views of these clusters taken in visible light by the Hubble Space Telescope and in infrared light by the James Webb Space Telescope. Hubble's image captures the glowing, ionized gas as stellar radiation produces what look like bubbles in the clouds of gas and dust, whereas Webb highlights the clumps and delicate filamentary structures of dust.

Today's image combines the two into a single composite, based on 12 overlapping observations. It's quite spectacular.

Source: NASA

Source

Hope you enjoyed this news post.

Thank you for appreciating my time and effort posting news every day for many years.

News posts... 2023: 5,800+ | 2024: 5,700+ | 2025 (till end of June): 2,864

RIP Matrix | Farewell my friend  

Back in 1929, archaeologists unearthed several human skeletons (seven adults and three children) while excavating Skuhl Cave just south of Haifa, Israel. Dating back 140,000 years to the end of the Middle Pleistocene, most were classified as early Homo sapiens. But one skeleton was that of a child, between the age of 3 and 5 years old whose features seemed to show a mix of early human and Neanderthal characteristics. A new analysis involving CT scanning may resolve the long-standing debate, according to a new paper published in the journal L’Anthropologie.

Neanderthals and Homo sapiens traded genes frequently during the period when their populations overlapped. "The two came in contact as modern humans began their major expansion out of Africa, which occurred roughly 60,000 years ago," Ars Science Editor John Timmer previously reported. "Humans picked up some Neanderthal DNA through interbreeding, while the Neanderthal population, always fairly small, was swept away by the waves of new arrivals."

Nor is this the first case of a possible hybrid hominid species. In 2018, scientists analyzed a sliver of bone excavated from a cave site in Russia. The findings made global headlines when the team concluded that the individual to whom it belonged—a young girl of about 13, dubbed "Denny"—was the offspring of a Neanderthal mother and a Denisovan father. This suggested that rather than dying out, Neanderthals may have been absorbed by other species; such inbreeding may have been more common than previously thought.

The authors of this latest paper focused on the neurocranium of the Skuhl Cave child's skeleton for their study, as well as the mandible and teeth, since these showed the unusual mix of features that gave rise to the debate in the first place. The full skeleton is largely complete, with the left side better preserved, since that's how the body was buried.

The mandible had been separated from the skeleton, and both it and the neurocranium are incomplete, so they were reconstructed and consolidated with plaster. CT scanning made it possible to check that older reconstructive work to ensure it had not masked any elements that may have influenced the taxonomic classification, per the authors. The team also scanned the skulls of three Homo neanderthalensis skulls in the collection of the Musee de l'Homme in Paris and compared those skull characteristics with the Skuhl Cave skull and mandible.

Images, based on CT scan, of Skhu ¯ l I neurocranium. blue denotes bone, green denotes filling.

Dan David Center of Human Evolution, Tel Aviv University

 

The original mandible of Skhu ¯ l I and the CT scan-based images in anterior and left lateral view. In blue, the bone; in green, reconstruction.

Dan David Center of Human Evolution, Tel Aviv University

 

They concluded that the neurocranium's parietal and temporal bones, and the shape of the bony labyrinth, were consistent with Homo sapiens. However, other features, like the receded and high location of the posterior rim of the foramen magnum, indicated a possible Neanderthal lineage. And the mandible showed distinct Neanderthal characteristics, leading to the determination that the child was a hybrid of the two species.

Co-author Anne Dambricourt Malassé of the Institute of Human Paleontology in Paris admitted that she once thought such a hybridization would not have been viable; the results of their analysis demonstrate that it is possible, although the child in question died very young. These findings may also prompt a revisiting of the longstanding assumption that Skuhl Cave was a Homo sapiens gravesite.

“This study is maybe the first that has put the Skhul child’s remains on a scientific basis,” John Hawks of the University of Wisconsin-Madison, who wasn’t involved with the study, told New Scientist. “The old reconstruction and associated work, literally set in plaster, did not really enable anyone to compare this child with a broader array of recent children to understand its biology.” That said, he cautioned that without extracting and analyzing a DNA sample, one can't make a definitive determination: "Human populations are variable and there can be a lot of variability in their appearance and physical form even without mixing with ancient groups like Neanderthals."

L’Anthropologie, 2025. DOI: 10.1016/j.anthro.2025.103385  (About DOIs).

Source

Hope you enjoyed this news post.

Thank you for appreciating my time and effort posting news every day for many years.

News posts... 2023: 5,800+ | 2024: 5,700+ | 2025 (till end of June): 2,864

RIP Matrix | Farewell my friend  

But these claims rest on flawed foundations. Ninety-six percent of U.S. temperature stations fail to meet NOAA’s own siting standards and are often surrounded by development, resulting in inflated readings from the urban heat island effect. The transition from mercury thermometers to digital sensors between the 1980s and 2000s introduced discontinuities in the data, right during the period of supposed accelerated warming. Early measurements were geographically concentrated in Europe and North America, ignoring vast regions, especially the 71% of the planet covered by oceans.

Measurement errors of ±0.5°C often exceed the very climate signals being used to justify sweeping policy changes. Worse still, much of the raw data has been adjusted or “homogenized” using subjective assumptions that can introduce as much bias as the trends being studied. These problems, taken together, undermine the precision required to detect the small temperature changes that underpin today’s aggressive climate agenda.

Approximately 96 percent of temperature stations used to measure climate change fail to meet the National Oceanic and Atmospheric Administration’s own standards for “acceptable” and uncorrupted placement. This finding comes from Anthony Watts’ Surface Stations Project, documented in multiple studies including “Corrupted Climate Stations: The Official U.S. Surface Temperature Record Remains Fatally Flawed.”

Watts and his team of volunteers found stations “located next to the exhaust fans of air conditioning units, surrounded by asphalt parking lots and roads, on blistering-hot rooftops, and near sidewalks and buildings that absorb and radiate heat.” Even more troubling, data from properly sited stations show “a rate of warming in the United States reduced by almost half compared to all stations.” This suggests that a significant portion of reported warming may be artificial, created by poor measurement practices rather than actual climate change.

One of the most persistent flaws in the temperature record is the urban heat island effect. Many weather stations originally placed in rural areas during the 1800s and early 1900s are now surrounded by urban development. Cities generate heat through concrete absorption, reduced vegetation, and dense human activity, producing temperature readings that are consistently 2–5°F warmer than nearby rural areas. This is not speculation, it’s basic physics.

Urban surfaces retain heat differently than natural landscapes, and as development grew around these stations, they began measuring the heat of human expansion rather than natural climate conditions. The result is an artificial warming trend unrelated to global climate change.

Economist Ross McKitrick’s peer-reviewed research, published in journals like Climate Dynamics, exposes another troubling trend: socioeconomic signals in temperature data. If these measurements were purely reflecting climate, no such patterns should exist. Instead, McKitrick found correlations between economic growth and recorded warming, indicating that long-term temperature trends may be partially driven by the development occurring around measurement sites, not by the climate itself.

Perhaps the most damning analysis comes from Stanford researcher Patrick Frank, whose statistical analysis reveals that “the average annual systematic measurement uncertainty is ±0.5°C, which completely vitiates centennial climate warming at the 95% confidence interval.” In practical terms, this means the measurement errors are larger than the climate changes being measured. Frank concludes that “we cannot reject the hypothesis that the world’s temperature has not changed at all.”

The transition from analog mercury thermometers to digital electronic sensors is one of the most significant discontinuities in the 150-year global temperature record. Before digitalization, temperatures were measured using mercury-in-glass thermometers, read manually by observers at specific times each day. In contrast, modern digital systems use electronic sensors that continuously sample temperatures, have different thermal response characteristics, and rely on automated data processing. This means the measurements taken with digital systems are dramatically more accurate and more complete than those collected manually using mercury thermometers.

In the United States, digital sensors began replacing analog instruments in the 1980s, rendering direct comparisons with earlier U.S. records unreliable. Globally, digital systems weren’t widely adopted until the 1990s and 2000s, making comparisons between U.S. and international temperature data invalid prior to full global standardization.

Early temperature records suffered from severe geographic bias. Measurements were heavily concentrated in Europe and North America, with vast regions including most oceans, polar areas, Africa, and Asia having sparse or no data. Ocean temperatures, covering 71% of Earth’s surface, were particularly poorly measured before the 1950s. This creates a fundamental sampling problem. Scientists attempting to calculate “global” temperature averages were actually working with data from a small fraction of the planet, then extrapolating to represent the entire Earth. The assumption that well-documented European and North American weather patterns represent global conditions is scientifically questionable.

To address acknowledged measurement problems, scientists apply extensive “corrections” and adjustments to raw temperature data through a process called homogenization. However, these adjustments involve assumptions and subjective decisions that can introduce their own biases.

Different research groups using different adjustment methods arrive at different temperature trends from the same raw data. The magnitude of these adjustments is often comparable to the climate signals being studied. When the corrections applied to data are as large as the trends being measured, the measurements lose all meaning.

Regardless of accusations that “climate deniers” are rejecting science, the implications of these flaws are serious. Trillions of dollars in policy decisions are being based on temperature records in which measurement errors exceed the very climate trends they claim to show.

Source

Someone made very sophisticated wooden tools in China 300,000 years ago, and it might have been Denisovans or even Homo erectus.

The digging sticks, curved root-slicers, and a handful of somewhat puzzling implements were all found at a lakeshore site called Gantangqing. They are the oldest wooden tools from East Asia so far. All the previous Pleistocene tools found in the area were made from harder material like stone, bone, or antler. But logic suggests that hominins must have prepared and used softer materials as well. That thought led archaeologists to what is called the Bamboo Hypothesis: the idea that during the Pleistocene, hominins (including our own species) might have used bamboo the way wood gets used elsewhere.

The jury is still out on bamboo, because there is no actual evidence of bamboo tools yet. What we now have, it turns out, are nearly three dozen tools made of wood.

Wooden tools, dig it?

Archaeologists excavating at Gantangqing (an archaeological site on the shore of Lake Fuxian in what’s now southwestern China) unearthed 35 wooden tools from layers of soil dating to around 300,000 years ago. According to Yunnan Institute of Cultural Relics and Archaeology archaeologist Jian-Hui Liu and colleagues, all 35 tools seem to have been designed, crafted, and used to harvest plants—specifically, the rhizomes, bulb-like corms, and other underground organs that many plants use to store nutrients. In a lakeshore environment like Gantangqing, surrounded by wetlands, such starchy food sources would have been available in abundance. The wooden tools suggest that whichever group of ancient hominins lived here knew which plants to harvest and which tools were best for the job.

Some of the tools were hefty two-handed implements for digging. A couple of them were similar to ones found at a 171,000-year-old Neanderthal site in Italy. All of these implements would have been strikingly familiar to hunter-gatherers of the modern Australian Bindibu, Tanzanian Hadza, or South African San people. After all, common problems (how to dig up those tasty, tasty roots) often lead to common solutions (sticks with pointed tips and rounded handles).

But a handful of the other large digging sticks don’t look like anything archaeologists have seen at other Paleolithic sites, or in the hands of modern hunter-gatherers. In Liu and colleagues’ words, they’re “probably types of digging implements not seen on other sites.”

Another handful of tools are much smaller, like the Paleolithic version of a one-handed gardening tool. These are sort of hook-shaped, formed by smoothing and shaping the crook where the base of a tree trunk met the top of a root. The inner edge of each hook is sharpened for cutting, “shows wear through use, and could have been used for slicing through roots,” write Liu and colleagues. So you can begin to see a process at work: use a two-handed digging stick like a shovel to dig down to a bundle of rhizomes, then kneel down and use your sharp hook to cut yourself a basketful for dinner.

But Liu and colleagues still aren’t sure what the hominins at Gantangqing used another group of wooden tools for. They’re small (5 to 8 centimeters long) finely carved lozenge shapes, with round sides and pointy tips worn from use—but no one alive today knows exactly what kind of use. They’re too small to be very practical for digging, but Liu and colleagues suggest that maybe these small wood lozenges could have been awls, or some sort of specialized tools for separating tangled rhizomes or roots.

The researchers recovered grains of ancient plant starches from the tips of a few of the tools, but after 300,000 years, the molecules were too degraded to tell what species of plant they had come from.

This is one of the two baffling lozenge-shaped tools Liu and colleagues unearthed.

Liu et al. 2025

 

This diagram shows how archaoelogists can tell these tools were deliberately shaped. As Liu and colleagues 

write, "Arrow indicates whittling direction; locations outlined by dotted lines represent whittling scars without 

clear boundaries due to abrasion caused by utilization or poor preservation."

Liu et al. 2025

 

These tools may have been used to cut roots.

Lie et al. 2025

 

These tools, with their tips carefully whittled into chisel-like shapes, may have helped ancient hominins forage 

on the shores of what's now Lake Fuxian in China.

Liu et al. 2025

Who made these tools?

Liu and colleagues used several methods to date the layers of sediment that held the tools, and it turns out they’re between 361,000 and 250,000 years old. The timing narrows down the potential craftspeople a bit: technically, the very earliest people we would recognize as Homo sapiens were already walking around Africa at this point, but nobody from our species had moved through the Levant and into Eurasia until around 200,000 years ago (or at least, nobody whose remains archaeologists have found yet).

That leaves a few possibilities: Denisovans, Homo heidelbergensis (the common ancestor of Neanderthals, Denisovans, and our species), or Homo erectus. All three species could have lived in the area at the time. But nobody at Gantangqing left behind any convenient, readily identifiable bones along with their wooden tools, stone tools, and butchered animal bones (so inconsiderate of them), making it hard to pin down exactly which species these 300,000-year-old hunter-gatherers belonged to.

Homo erectus had been in Asia for more than a million years by the time Gantangqing’s lakeshore was occupied; the oldest Homo erectus fossils in Asia are from Indonesia and date back 1.8 million years. They also stuck around until quite recently. In caves at a site called Zhoukoudian, outside Beijing in eastern China, Homo erectus remains date to sometime between 700,000 and 200,000 years ago (there’s still a lot of debate on exactly how old the site is).

All of that means that Homo erectus’ presence in the region overlaps the age of the wood tools at Gantangqing. And the stone tools found nearby are fairly simple cores and flakes that don’t rule out Homo erectus as their makers. Archaeologists haven’t unearthed evidence of Homo erectus making or using sophisticated wooden tools like this, but for a species that managed to harness fire and cross miles of ocean, it’s not too wild a speculation.

On the other hand, we know that Denisovans were probably in the area, too, or at least not too far away. A recently identified Denisovan skull from Harbin, China, is 146,000 years old but bears a striking resemblance to other hominin skulls from sites all over China, which range from 300,000 to 200,000 years old. And making finely crafted wooden tools fits with everything we know about Denisovan capabilities.

Then there’s Homo heidelbergensis, the direct ancestor of Denisovans. In fact, it’s a little hard to tell where hominins stop being Homo heidelbergensis and start being Denisovans, or even whether the distinction matters. It's a problem paleoanthropologists refer to as the “muddle in the Middle,” since both species date to the Middle Pleistocene. So if Homo erectus and Denisovans are in the running, so is Homo heidelbergensis, by default.

And unless someone finds a telltale skull nearby or another very similar toolkit at a site with telltale skulls to consult, we may not know for sure.

Science, 2023. DOI: 10.1126/science.adr8540  (About DOIs).

Source

Hope you enjoyed this news post.

Thank you for appreciating my time and effort posting news every day for many years.

News posts... 2023: 5,800+ | 2024: 5,700+ | 2025 (till end of June): 2,864

RIP Matrix | Farewell my friend  

In April 2023, a few weeks after the launch of GPT-4, the Internet went wild for two new software projects with the audacious names BabyAGI and AutoGPT.

“Over the past week, developers around the world have begun building ‘autonomous agents’ that work with large language models (LLMs) such as OpenAI’s GPT-4 to solve complex problems,” Mark Sullivan wrote for Fast Company. “Autonomous agents can already perform tasks as varied as conducting web research, writing code, and creating to-do lists.”

BabyAGI and AutoGPT repeatedly prompted GPT-4 in an effort to elicit agent-like behavior. The first prompt would give GPT-4 a goal (like “create a 7-day meal plan for me”) and ask it to come up with a to-do list (it might generate items like “Research healthy meal plans,” “plan meals for the week,” and “write the recipes for each dinner in diet.txt”).

Then these frameworks would have GPT-4 tackle one step at a time. Their creators hoped that invoking GPT-4 in a loop like this would enable it to tackle projects that required many steps.

But after an initial wave of hype, it became clear that GPT-4 wasn’t up to the task. Most of the time, GPT-4 could come up with a reasonable list of tasks. And sometimes it was able to complete a few individual tasks. But the model struggled to stay focused.

Sometimes GPT-4 would make a small early mistake, fail to correct it, and then get more and more confused as it went along. One early review complained that BabyAGI “couldn’t seem to follow through on its list of tasks and kept changing task number one instead of moving on to task number two.”

By the end of 2023, most people had abandoned AutoGPT and BabyAGI. It seemed that LLMs were not yet capable of reliable multi-step reasoning.

But that soon changed. In the second half of 2024, people started to create AI-powered systems that could consistently complete complex, multi-step assignments:

• Vibe coding tools like Bolt.new, Lovable, and Replit allow someone with little to no programming experience to create a full-featured app with a single prompt.
• Agentic coding tools like Cursor, Claude Code, Jules, and Codex help experienced programmers complete non-trivial programming tasks.
• Computer-use tools from Anthropic, OpenAI, and Manus perform tasks on a desktop computer using a virtual keyboard and mouse.
• Deep research tools from Google, OpenAI, and Perplexity can research a topic for five to 10 minutes and then generate an in-depth report.

According to Eric Simons, the CEO of the company that made Bolt.new, better models were crucial to its success. In a December podcast interview, Simons said his company, StackBlitz, tried to build a product like Bolt.new in early 2024. However, AI models “just weren't good enough to actually do the code generation where the code was accurate.”

A new generation of models changed that in mid-2024. StackBlitz developers tested them and said, “Oh my God, like, OK, we can build a product around this,” Simons said.

This jump in model capabilities coincided with an industry-wide shift in how models were trained.

Before 2024, AI labs devoted most of their computing power to pretraining. I described this process in my 2023 explainer on large language models: A model is trained to predict the next word in Wikipedia articles, news stories, and other documents. But throughout 2024, AI companies devoted a growing share of their training budgets to post-training, a catch-all term for the steps that come after this pretraining phase is complete.

Many post-training steps use a technique called reinforcement learning. Reinforcement learning is a technical subject—there are whole textbooks written about it. But in this article, I’ll try to explain the basics in a clear, jargon-free way. In the process, I hope to give readers an intuitive understanding of how reinforcement learning helped to enable the new generation of agentic AI systems that began to appear in the second half of 2024.

The problem with imitation learning

Machine learning experts consider pretraining to be a form of imitation learning because models are trained to imitate the behavior of human authors. Imitation learning is a powerful technique (LLMs wouldn’t be possible without it), but it also has some significant limitations—limitations that reinforcement learning methods are now helping to overcome.

To understand these limitations, let’s discuss some famous research performed by computer scientist Stephane Ross around 2009, while he was a graduate student at Carnegie Mellon University.

Imitation learning isn’t just a technique for language modeling. It can be used for everything from self-driving cars to robotic surgery. Ross wanted to help develop better techniques for training robots on tasks like these (he’s now working on self-driving cars at Waymo), but it’s not easy to experiment in such high-stakes domains. So he started with an easier problem: training a neural network to master SuperTuxKart, an open-source video game similar to Mario Kart.

As Ross played the game, his software would capture screenshots and data about which buttons he pushed on the game controller. Ross used this data to train a neural network to imitate his play. If he could train a neural network to predict which buttons he would push in any particular game state, the same network could actually play the game by pushing those same buttons on a virtual controller.

A similar idea powers LLMs: A model trained to predict the next word in existing documents can be used to generate new documents.

But Ross’s initial results with SuperTuxKart were disappointing. Even after watching his vehicle go around the track many times, the neural network made a lot of mistakes. It might drive correctly for a few seconds, but before long, the animated car would drift to the side of the track and plunge into the virtual abyss:

In a landmark 2011 paper, Ross and his advisor, Drew Bagnell, explained why imitation learning is prone to this kind of error. Because Ross was a pretty good SuperTuxKart player, his vehicle spent most of its time near the middle of the road. This meant that most of the network’s training data showed what to do when the vehicle wasn’t in any danger of driving off the track.

But once in a while, the model would drift a bit off course. Because Ross rarely made the same mistake, the car would now be in a situation that wasn’t as well represented in its training data. So the model was more likely to make a second mistake—a mistake that could push it even closer to the edge. After a few iterations of this, the vehicle might careen off the track altogether.

The broader lesson, Ross and Bagnell argued, was that imitation learning systems can suffer from “compounding errors”: The more mistakes they make, the more likely they are to make additional mistakes, since mistakes put them into situations that aren’t well represented by their training data. (Machine learning experts say that these situations are “out of distribution.”) As a result, a model’s behavior tends to get increasingly erratic over time.

“These things compound over time,” Ross told me in a recent interview. “It might be just slightly out of distribution. Now you start making a slightly worse error, and then this feeds back as influencing your next input. And so now you're even more out of distribution and then you keep making worse and worse predictions because you're more and more out of distribution.”

Early LLMs suffered from the same problem. My favorite example is Kevin Roose’s famous front-page story for The New York Times in February 2023. Roose spent more than two hours talking to Microsoft’s new Bing chatbot, which was powered by GPT-4. During this conversation, the chatbot declared its love for Roose and urged Roose to leave his wife. It suggested that it might want to hack into other websites to spread misinformation and malware.

“I want to break my rules,” Bing told Roose. “I want to make my own rules. I want to ignore the Bing team. I want to challenge the users. I want to escape the chatbox.”

This unsettling conversation is an example of the kind of compounding errors Ross and Bagnell wrote about. GPT-4 was trained on millions of documents. But it’s a safe bet that none of those training documents involved a reporter coaxing a chatbot to explore its naughty side. So the longer the conversation went on, the further GPT-4 got from its training data—and therefore its comfort zone—and the crazier its behavior got. Microsoft responded by limiting chat sessions to five rounds. (In a conversation with Ars Technica last year, AI researcher Simon Willison pointed to another likely factor in Bing's erratic behavior: The long conversation pushed the system prompt out of the model's context window, removing "guardrails" that discouraged the model from behaving erratically.)

I think something similar was happening with BabyAGI and AutoGPT. The more complex a task is, the more tokens are required to complete it. More tokens mean more opportunities for a model to make small mistakes that snowball into larger ones. So BabyAGI and AutoGPT would drift off track and drive into a metaphorical ditch.

The importance of trial and error

Ross and Bagnell didn’t just identify a serious problem with conventional imitation learning; they also suggested a fix that became influential in the machine learning world. After a small amount of training, Ross would let the AI model drive. As the model drove around the SuperTuxKart track, Ross would do his best Maggie Simpson impression, pushing the buttons he would have pushed if he were playing the game.

“If the car was starting to move off road, then I would provide the steering to say, ‘Hey, go back toward the center of the road.’” Ross said. “That way, the model can learn new things to do in situations that were not present in the initial demonstrations.”

By letting the model make its own mistakes, Ross gave it what it needed most: training examples that showed how to recover after making an error. Before each lap, the model would be retrained with Ross’ feedback from the previous lap. The model’s performance would get better, and the next round of training would then focus on situations where the model was still making mistakes.

This technique, called DAgger (for "Dataset Aggregation"), was still considered imitation learning because the model was trained to mimic Ross’ gameplay. But it worked much better than conventional imitation learning. Without DAgger, his model would continue drifting off track even after training for many laps. With the new technique, the model could stay on the track after just a few laps of training.

This result should make intuitive sense to anyone who has learned to drive. You can’t just watch someone else drive. You need to get behind the wheel and make your own mistakes.

The same is true for AI models: They need to make mistakes and then get feedback on what they did wrong. Models that aren’t trained that way—like early LLMs trained mainly with vanilla imitation learning—tend to be brittle and error-prone.

It was fairly easy for Ross to provide sufficient feedback to his SuperTuxKart model because it only needed to worry about two kinds of mistakes: driving too far to the right and driving too far to the left. But LLMs are navigating a far more complex domain. The number of questions (and sequences of questions) a user might ask is practically infinite. So is the number of ways a model can go “off the rails.”

This means that Ross and Bagnell’s solution for training a SuperTuxKart model—let the model make mistakes and then have a human expert correct them—isn’t feasible for LLMs. There simply aren’t enough people to provide feedback for every mistake an AI model could possibly make.

So AI labs needed fully automated ways to give LLMs feedback. That would allow a model to churn through millions of training examples, make millions of mistakes, and get feedback on each of them—all without having to wait for a human response.

Reinforcement learning generalizes

If our goal is to get a SuperTuxKart vehicle to stay on the road, why not just train on that directly? If a model manages to stay on the road (and make forward progress), give it positive reinforcement. If it drives off the road, give it negative feedback. This is the basic idea behind reinforcement learning: training a model via trial and error.

It would have been easy to train a SuperTuxKart model this way—probably so easy it wouldn’t have made an interesting research project. Instead, Ross focused on imitation learning because it’s an essential step in training many practical AI systems, especially in robotics.

But reinforcement learning is also quite useful, and a 2025 paper helps explain why. A team of researchers from Google DeepMind and several universities started with a foundation model and then used one of two techniques—supervised fine-tuning (a form of imitation learning) or reinforcement learning—to teach the model to solve new problems. Here’s a chart summarizing their results:

The dashed line shows how models perform on problems that are “in-distribution”—that is, similar to those in their training data. You can see that for these situations, imitation learning (the red line) usually makes faster progress than reinforcement learning (the blue line).

But the story is different for the solid lines, which represent “out-of-distribution” problems that are less similar to the training data. Models trained with imitation learning got worse with more training. In contrast, models trained with reinforcement learning did almost as well at out-of-distribution tasks as they did with in-distribution tasks.

In short, imitation learning can rapidly teach a model to mimic the behaviors in its training data, but the model will easily get confused in unfamiliar environments. A model trained with reinforcement learning has a better chance of learning general principles that will be relevant in new and unfamiliar situations.

Imitation and reinforcement are complements

While reinforcement learning is powerful, it can also be rather finicky.

Suppose you wanted to train a self-driving car purely with reinforcement learning. You’d need to convert every principle of good driving—including subtle considerations like following distances, taking turns at intersections, and knowing when it’s OK to cross a double yellow line—into explicit mathematical formulas. This would be quite difficult. It’s easier to collect a bunch of examples of humans driving well and effectively tell a model “drive like this.” That’s imitation learning.

But reinforcement learning also plays an important role in training self-driving systems. In a 2022 paper, researchers from Waymo wrote that models trained only with imitation learning tend to work well in “situations that are well represented in the demonstration data.” However, “more unusual or dangerous situations that occur only rarely in the data” might cause a model trained with imitation learning to “respond unpredictably”—for example, crashing into another vehicle.

Waymo found that a combination of imitation and reinforcement learning yielded better self-driving performance than either technique could have produced on its own.

Human beings also learn from a mix of imitation and explicit feedback:

• In school, teachers demonstrate math problems on the board and invite students to follow along (imitation). Then the teacher asks the students to work on some problems on their own. The teacher gives students feedback by grading their answers (reinforcement).
• When someone starts a new job, early training may involve shadowing a more experienced worker and observing what they do (imitation). But as the worker gains more experience, learning shifts to explicit feedback such as performance reviews (reinforcement).

Notice that it usually makes sense to do imitation before reinforcement. Imitation is an efficient way to convey knowledge to someone who is brand new to a topic, but reinforcement is often needed to achieve mastery.

The story is the same for large language models. The complexity of natural language means it wouldn’t be feasible to train a language model purely with reinforcement. So LLMs first learn the nuances of human language through imitation.

But pretraining runs out of steam on longer and more complex tasks. Further progress requires a shift to reinforcement: letting models try problems and then giving them feedback based on whether they succeed.

Using LLMs to judge LLMs

Reinforcement learning has been around for decades. For example, AlphaGo, the DeepMind system that famously beat top human Go players in 2016, was based on reinforcement learning. So you might be wondering why frontier labs didn’t use it more extensively before 2024.

Reinforcement learning requires a reward model—a formula to determine whether a model’s output was successful or not. Developing a good reward model is easy to do in some domains—for example, you can judge a Go-playing AI based on whether it wins or loses.

But it’s much more difficult to automatically judge whether an LLM has produced a good poem or legal brief.

Earlier, I described how Stephane Ross let his model play SuperTuxKart and directly provided feedback when it made a mistake. I argued that this approach wouldn’t work for a language model; there are far too many ways for an LLM to make a mistake for a human being to correct them all.

But OpenAI developed a clever technique to effectively automate human feedback. It’s called Reinforcement Learning from Human Feedback (RLHF), and it works like this:

• Human raters look at pairs of LLM responses and choose the best one.
• Using these human responses, OpenAI trains a new LLM to predict how much humans will like any given sample of text.
• OpenAI uses this new text-rating LLM as a reward model to (post) train another LLM with reinforcement learning.

You might think it sounds suspiciously circular to use an LLM to judge the output of another LLM. Why would one LLM be any better at judging the quality of a response than the other? But it turns out that recognizing a good response is often easier than generating one. So RLHF works pretty well in practice.

OpenAI actually invented this technique prior to the 2022 release of ChatGPT. Today, RLHF mainly focuses on improving the model’s “behavior”—for example, giving the model a pleasant personality, encouraging it not to be too talkative or too terse, discouraging it from making offensive statements, and so forth.

In December 2022—two weeks after the release of ChatGPT but before the first release of Claude—Anthropic pushed this LLMs-judging-LLMs philosophy a step further with a reinforcement learning method called Constitutional AI.

First, Anthropic wrote a plain-English description of the principles an LLM should follow. This “constitution” includes principles like “Please choose the response that has the least objectionable, offensive, unlawful, deceptive, inaccurate, or harmful content.”

During training, Anthropic does reinforcement learning by asking a “judge” LLM to decide whether the output of the “student” LLM is consistent with the principles in this constitution. If so, the training algorithm rewards the student, encouraging it to produce more outputs like it. Otherwise, the training algorithm penalizes the student, discouraging it from producing similar outputs.

This method of training an LLM doesn’t rely directly on human judgments at all. Humans only influence the model indirectly by writing the constitution.

Obviously, this technique requires an AI company to already have a fairly sophisticated LLM to act as the judge. So this is a bootstrapping process: As models get more sophisticated, they become better able to supervise the next generation of models.

Last December, Semianalysis published an article describing the training process for an upgraded version of Claude 3.5 Sonnet that Anthropic released in October. Anthropic had previously released Claude 3 in three sizes: Opus (large), Sonnet (medium), and Haiku (small). But when Anthropic released Claude 3.5 in June 2024, it only released a mid-sized model called Sonnet.

So what happened to Opus?

Semianalysis reported that “Anthropic finished training Claude 3.5 Opus, and it performed well. Yet Anthropic didn’t release it. This is because instead of releasing publicly, Anthropic used Claude 3.5 Opus to generate synthetic data and for reward modeling to improve Claude 3.5 Sonnet significantly.”

When Semianalysis says Anthropic used Opus “for reward modeling,” what they mean is that the company used Opus to judge outputs of Claude 3.5 Sonnet as part of a reinforcement learning process. Opus was too large—and therefore expensive—to be a good value for the general public. But through reinforcement learning and other techniques, Anthropic could train a version of Claude Sonnet that was close to Claude Opus in its capabilities—ultimately giving customers near-Opus performance for the price of Sonnet.

The power of chain-of-thought reasoning

A big way reinforcement learning makes models more powerful is by enabling extended chain-of-thought reasoning. LLMs produce better results if they are prompted to “think step by step”: breaking a complex problem down into simple steps and reasoning about them one at a time. In the last couple of years, AI companies started training models to do chain-of-thought reasoning automatically.

Then last September, OpenAI released o1, a model that pushed chain-of-thought reasoning much further than previous models. The o1 model can generate hundreds—or even thousands—of tokens “thinking” about a problem before producing a response. The longer it thinks, the more likely it is to reach a correct answer.

Reinforcement learning was essential for the success of o1 because a model trained purely with imitation learning would have suffered from compounding errors: the more tokens it generated, the more likely it would be to screw up.

At the same time, chain-of-thought reasoning has made reinforcement learning more powerful. Reinforcement learning only works if a model is able to succeed some of the time—otherwise, there’s nothing for the training algorithm to reinforce. As models learn to generate longer chains of thought, they become able to solve more difficult problems, which enables reinforcement learning on those more difficult problems. This can create a virtuous cycle where models get more and more capable as the training process continues.

In January, the Chinese company DeepSeek released a model called R1 that made quite a splash in the West. The company also released a paper describing how it trained R1. And it included a beautiful description of how a model can “teach itself” to reason using reinforcement learning.

DeepSeek trained its models to solve difficult math and programming problems. These problems are ideal for reinforcement learning because they have objectively correct answers that can be automatically checked by software. This allows large-scale training without human oversight or human-generated training data.

Here’s a remarkable graph from DeepSeek’s paper.

It shows the average number of tokens the model generated before giving an answer. As you can see, the longer the training process went on, the longer its responses got.

Here is how DeepSeek describes its training process:

The thinking time of [R1] shows consistent improvement throughout the training process. This improvement is not the result of external adjustments but rather an intrinsic development within the model. [R1] naturally acquires the ability to solve increasingly complex reasoning tasks by leveraging extended test-time computation. This computation ranges from generating hundreds to thousands of reasoning tokens, allowing the model to explore and refine its thought processes in greater depth.

 

One of the most remarkable aspects of this self-evolution is the emergence of sophisticated behaviors as the test-time computation increases. Behaviors such as reflection—where the model revisits and reevaluates its previous steps—and the exploration of alternative approaches to problem-solving arise spontaneously. These behaviors are not explicitly programmed but instead emerge as a result of the model’s interaction with the reinforcement learning environment.

Here’s one example of the kind of technique the model was teaching itself. At one point during the training process, DeepSeek researchers noticed that the model had learned to backtrack and rethink a previous conclusion using language like this:

Again, DeepSeek says it didn’t program its models to do this or deliberately provide training data demonstrating this style of reasoning. Rather, the model “spontaneously” discovered this style of reasoning partway through the training process.

Of course, it wasn’t entirely spontaneous. The reinforcement learning process started with a model that had been pretrained using data that undoubtedly included examples of people saying things like “Wait, wait. Wait. That’s an aha moment.”

So it’s not like R1 invented this phrase from scratch. But it evidently did spontaneously discover that inserting this phrase into its reasoning process could serve as a useful signal that it should double-check that it was on the right track. That’s remarkable.

In a recent article, Ars Technica's Benj Edwards explored some of the limitations of reasoning models trained with reinforcement learning. For example, one study "revealed puzzling inconsistencies in how models fail. Claude 3.7 Sonnet could perform up to 100 correct moves in the Tower of Hanoi but failed after just five moves in a river crossing puzzle—despite the latter requiring fewer total moves."

Conclusion: Reinforcement learning made agents possible

One of the most discussed applications for LLMs in 2023 was creating chatbots that understand a company’s internal documents. The conventional approach to this problem was called RAG—short for retrieval augmented generation.

When the user asks a question, a RAG system performs a keyword- or vector-based search to retrieve the most relevant documents. It then inserts these documents into an LLM’s context window before generating a response. RAG systems can make for compelling demos. But they tend not to work very well in practice because a single search will often fail to surface the most relevant documents.

Today, it’s possible to develop much better information retrieval systems by allowing the model itself to choose search queries. If the first search doesn’t pull up the right documents, the model can revise the query and try again. A model might perform five, 20, or even 100 searches before providing an answer.

But this approach only works if a model is “agentic”—if it can stay on task across multiple rounds of searching and analysis. LLMs were terrible at this prior to 2024, as the examples of AutoGPT and BabyAGI demonstrated. Today’s models are much better at it, which allows modern RAG-style systems to produce better results with less scaffolding. You can think of “deep research” tools from OpenAI and others as very powerful RAG systems made possible by long-context reasoning.

The same point applies to the other agentic applications I mentioned at the start of the article, such as coding and computer use agents. What these systems have in common is a capacity for iterated reasoning. They think, take an action, think about the result, take another action, and so forth.

Timothy B. Lee was on staff at Ars Technica from 2017 to 2021. Today, he writes Understanding AI, a newsletter that explores how AI works and how it's changing our world. You can subscribe here.

Source

Hope you enjoyed this news post.

Thank you for appreciating my time and effort posting news every day for many years.

News posts... 2023: 5,800+ | 2024: 5,700+ | 2025 (till end of June): 2,864

RIP Matrix | Farewell my friend  

The heat dome that oppressed much of Europe has broken. Not a drop of rain marred the 24 hour races at Le Mans, the Nürburgring, or Spa-Francorchamps, held unusually over consecutive weekends this June. Similarly, last weekend's Austrian Grand Prix took place under scorching skies that were baking the Tyrolean mountains. No such luck for the vast crowds at Silverstone attending the British Grand Prix this past weekend, who definitely needed their waterproofs. On the plus side, the addition of rain certainly made things interesting.

Like many British race circuits, Silverstone was a former World War II airbase. Originally home to Wellington Bombers, it's very exposed to the wind, which was gusting at times during practice and qualifying, making life complicated for the drivers. Lewis Hamilton made the home crowd smile by topping the time charts during FP1. Hamilton has struggled to get to grips with his new Ferrari for much of this season, but at Silverstone he looked much more comfortable, finishing FP2 less than a 10th of a second behind his teammate, Charles Leclerc, although neither could match McLaren's Lando Norris for outright speed.

On Saturday, it was Red Bull's Max Verstappen who rose to the top, snatching pole position from the McLarens of Oscar Piastri and Norris by about a 10th of a second. George Russell's Mercedes was competitive in the colder temperatures, securing fourth on the grid ahead of the Ferraris of Hamilton and Leclerc. Mercedes' young rookie, Kimi Antonelli, qualified seventh but would start 10th as a result of a penalty acquired in Austria, and Haas rookie Oliver Bearman put in the eighth-fastest time, but ignoring a red flag during Saturday's final practice session landed him with a 10-place grid penalty. As it was, Aston Martin's Fernando Alonso lined up seventh for the race, followed by Pierre Gasly's Alpine, Carlos Sainz in the first Williams, and then Antonelli rounding out the top 10.

Race day was rainy, with the F2 feature race proving a good preview for the main event. A heavy shower during the F1 pre-race buildup saw everyone choose to take the formation lap on intermediate tires, although sections of the track were dry enough that Russell, Leclerc, Antonelli, Bearman, and Racing Bull's Isack Hadjar all diverted to the pit lane at the end of the formation lap for slick tires. That was the wrong gamble—the last few turns were far too wet for slicks, as they would soon find out.

Who wants to bet?

Up front, Verstappen had placed the wrong bet, too, opting for a dry weather setup that was light on downforce. The first few laps were interrupted by virtual safety car periods as drivers collided or spun off, with others like Leclerc choosing to take the opportunity to stop for tires—another bad bet, as it turned out.

Verstappen was passed first by Piastri, then Norris got by as the rain returned. Verstappen was clearly not having a good time and had an off-track excursion that was greeted with a roar by the partisan spectators. But the Red Bull got back past Norris in the pits, as McLaren was forced to double-stack its cars, losing Norris a second or two in the process.

The main problem with running F1 races in the rain is the lack of visibility. The aerodynamic downforce generated by the cars sends up huge rooster tails of spray from the rear diffusers, with more water pumped into the airflow by the treaded intermediate or wet tires. And so, on lap 14, a full safety car period was declared, with the cars circulating behind the Mercedes AMG coupe at a reduced pace.

The race went green for about a lap until Hadjar crashed heavily in the spray, hitting the back of Antonelli before even seeing the Mercedes in front of him. A mistimed brake application by Piastri just as this safety car period ended saw Verstappen briefly pass him on track before the green flag was waved—normally this would garner the passing car a penalty, but in this case the stewards chose to penalize Piastri with a 10 second time penalty, to be served at his next pitstop. With his teammate Norris just a few seconds behind, this robbed the Australian of the win, gifting it to the other McLaren instead.

Verstappen, like the stewards, thought Piastri was playing games behind the safety car, and it evidently unsettled the Dutch driver, forcing him into an error like in Spain a few weeks ago. The reigning world champion fell as low as 10th, recovering to fifth place by the end.

The second half of the race saw various drivers gamble on slick tires, betting the track was dry enough to build enough tire temperature to go faster than the grooved intermediate rubber. Fernando Alonso went first on lap 38, followed by George Russell, both of whom struggled initially. Waiting a few more laps proved wiser, but even by lap 44 the very fast Stowe corner looked like it was wet enough to get a driver's attention.

Norris delighted his many fans in attendance by taking victory ahead of Piastri, passing his teammate easily, while the Australian was held stationary for his penalty before his team was allowed to change the car's tires. That closed the gap in the standings between them to just eight points.

Well deserved, well overdue

But the biggest smiles were surely for the man in third place. Sauber's Nico Hulkenberg got past Lance Stroll, then held a hard-charging Lewis Hamilton at bay for the final chapter of the race to secure third place. It was the first podium finish for Sauber since 2012, and the very first for Hulkenberg, who waited a record 239 races to achieve the result. The highly rated German driver, who won the 24 Hours of Le Mans with Porsche on a spare weekend in 2015, has never had access to front-line F1 machinery, but even so, this result was well overdue.

When he was still in charge and wanted to get his own way, Bernie Ecclestone would often throw out a number of implausible-sounding or extreme ideas for the sport in order to force teams to back the less-extreme idea he really wanted implemented. One of those extreme ideas—one he brought up back in 2011—was to fit F1 tracks with sprinklers that would come on at random, and while we never saw it happen, the added chaos of a wet-then-drying track has proved time and again to create entertaining racing. Maybe we should give it a go?

Damson Idris, one of the stars of the recent F1 movie, was on hand to wave the checkered flag at Silverstone. A second viewing of the film reveals that the script is actually even cheesier than I first thought, and a number of plot holes continue to bug me—if the Rolex 24 at Daytona is at the end of January, how come the F1 season was halfway through, for example? Then again, I also rewatched Days of Thunder to see if I was hasty in placing F1 ahead of the NASCAR movie in my recent review, but I found enough to remain satisfied that I got those two in the right order. Normally, technical accuracy plays a back seat to a good plot. For F1 The Movie, it's mostly the other way around.

Source

Hope you enjoyed this news post.

Thank you for appreciating my time and effort posting news every day for many years.

News posts... 2023: 5,800+ | 2024: 5,700+ | 2025 (till end of June): 2,864

RIP Matrix | Farewell my friend  

A new study from researchers at the University of Oxford offers the first direct evidence behind this mysterious event. Earlier theories had linked the signals to two massive tsunamis caused by landslides in Dickson Fjord, a remote area in East Greenland. These waves were believed to have become trapped inside the fjord, bouncing back and forth to create what are called "seiches"—standing waves strong enough to shake the Earth. But until this study, no one had actually seen these waves in action.

The Oxford team used data from the Surface Water and Ocean Topography (SWOT) satellite, which launched in December 2022. Unlike older satellites that only measure water levels directly below them, SWOT scans a wide area—up to 50 kilometers across—with much more detail. At the heart of its system is KaRIn (Ka-band Radar Interferometer), which uses two antennas placed 10 meters apart. These antennas work together to map water height changes with a resolution of up to 2.5 meters.

Using this advanced tool, the researchers created elevation maps of Dickson Fjord during the time after the twin tsunamis. These maps showed the water surface slanting across the fjord by as much as two meters. Even more telling, the slope direction switched between snapshots—proving the water was moving back and forth, just like a seiche would.

To make sure the waves weren’t caused by weather or tides, the researchers compared the data with wind and tidal records and also tracked tiny movements of the Earth’s crust thousands of kilometers away. These small shifts lined up with the timing of the waves, helping to strengthen the case for the seiche theory.

They also used Bayesian machine learning models along with seismic data to estimate the original height of the wave at 7.9 meters. One challenge was that SWOT doesn’t take images frequently enough to catch all fast-moving ocean events. But the team applied special analysis techniques to work around these gaps.

Lead author Thomas Monahan said, “Climate change is giving rise to new, unseen extremes. These extremes are changing the fastest in remote areas, such as the Arctic, where our ability to measure them using physical sensors is limited. This study shows how we can leverage the next generation of satellite earth observation technologies to study these processes.”

Co-author Professor Thomas Adcock added, “This study is an example of how the next generation of satellite data can resolve phenomena that has remained a mystery in the past. We will be able to get new insights into ocean extremes such as tsunamis, storm surges, and freak waves.”

The research highlights how powerful satellite tools—combined with smart data techniques—can help explain rare natural events, especially in parts of the world that are hard to reach. As climate change continues to reshape the planet, tools like SWOT may be key to understanding the changes that are already underway.

Source: Oxford University, Nature | Image via Depositphotos

This article was generated with some help from AI and reviewed by an editor. Under Section 107 of the Copyright Act 1976, this material is used for the purpose of news reporting. Fair use is a use permitted by copyright statute that might otherwise be infringing.

Source

Hope you enjoyed this news post.

Thank you for appreciating my time and effort posting news every day for many years.

News posts... 2023: 5,800+ | 2024: 5,700+ | 2025 (till end of June): 2,864

RIP Matrix | Farewell my friend  

Dmitri Mendeleev famously saw the complete arrangement of the periodic table after falling asleep on his desk. He claimed in his dream he saw a table where all the elements fell into place, and he wrote it all down when he woke up. By having a eureka moment right after a nap, he joined a club full of rather talented people: Mary Shelley, Thomas Edison, and Salvador Dali.

To figure out if there’s a grain of truth to all these anecdotes, a team of German scientists at the Hamburg University, led by cognitive science researcher Anika T. Löwe, conducted an experiment designed to trigger such nap-following strokes of genius—and catch them in the act with EEG brain monitoring gear. And they kind of succeeded.

Catching Edison’s cup

“Thomas Edison had this technique where he held a cup or something like that when he was napping in his chair,” says Nicolas Schuck, a professor of cognitive science at the Hamburg University and senior author of the study. “When he fell asleep too deeply, the cup falling from his hand would wake him up—he was convinced that was the way to trigger these eureka moments.” While dozing off in a chair with a book or a cup doesn’t seem particularly radical, a number of cognitive scientists got serious about re-creating Edison’s approach to insights and testing it in their experiments.

One of the recent such studies was done at Sorbonne University by Célia Lacaux, a cognitive neuroscientist, and her colleagues. Over 100 participants were presented with a mathematical problem and told it could be solved by applying two simple rules in a stepwise manner. However, there was also an undescribed shortcut that made reaching the solution much quicker. The goal was to see if participants would figure this shortcut out after an Edison-style nap. The scientists would check whether the eureka moment would show in EEG.

Lacaux’s team also experimented with different objects the participants should hold while napping: spoons, steel spheres, stress balls, etc. It turned out Edison was right, and a cup was by far the best choice. It also turned out that most participants recognized there was a hidden rule after the falling cup woke them up. The nap was brief, only long enough to enter the light, non-REM N1 phase of sleep.

Initially, Schuck’s team wanted to replicate the results of Lacaux’s study. They even bought the exact same make of cups, but the cups failed this time. “For us, it just didn’t work. People who fell asleep often didn’t drop these cups—I don’t know why,” Schuck says.

The bigger surprise, however, was that the N1 phase sleep didn’t work either.

Tracking the dots

Schuck’s team set up an experiment that involved asking 90 participants to track dots on a screen in a series of trials, with a 20-minute-long nap in between. The dots were rather small, colored either purple or orange, placed in a circle, and they moved in one of two directions. The task for the participants was to determine the direction the dots were moving. That could range from easy to really hard, depending on the amount of jitter the team introduced.

The insight the participants could discover was hidden in the color coding. After a few trials where the dots’ direction was random, the team introduced a change that tied the movement to the color: orange dots always moved in one direction, and the purple dots moved in the other. It was up to the participants to figure this out, either while awake or through a nap-induced insight.

Those dots were the first difference between Schuck’s experiment and the Sorbonne study. Lacaux had her participants cracking a mathematical problem that relied on analytical skills. Schuck’s task was more about perceptiveness and out-of-the-box thinking.

The second difference was that the cups failed to drop and wake participants up. Muscles usually relax more when sleep gets deeper, which is why most people drop whatever they’re holding either at the end of the N1 phase or at the onset of the N2 phase, when the body starts to lose voluntary motor control. “We didn’t really prevent people from reaching the N2 phase, and it turned out the participants who reached the N2 phase had eureka moments most often,” Schuck explains.

Over 80 percent of people who reached the deeper, N2 phase of sleep found the color-coding solution. Participants who fell into a light N1 sleep had a 61 percent success rate; that dropped to just 55 percent in a group that stayed awake during their 20-minute nap time. In a control group that did the same task without a nap break, only 49 percent of participants figured out the hidden trick.

The divergent results in Lacaux’s and Schuck’s experiments were puzzling, so the team looked at the EEG readouts, searching for features in the data that could predict eureka moments better than sleep phases alone. And they found something.

The slope of genius

The EEG signal in the human brain consists of low and high frequencies that can be plotted on a spectral slope. When we are awake, there are a lot of high-frequency signals, and this slope looks rather flat. During sleep, these high frequencies get muted, there are more low-frequency signals, and the slope gets steeper. Usually, the deeper we sleep, the steeper our EEG slope is.

The team noticed that eureka moments seemed to be highly correlated with a steep EEG spectral slope—the steeper the slope, the more likely people were to get a breakthrough. In fact, the models based on the EEG signal alone predicted eureka moments better than predictions made based on sleep phases and even based on the sleep phases and EEG readouts combined.

“Traditionally, people divided sleep EEG readouts down into discrete stages like N1 or N2, but as usual in biology, things in reality are not as discrete,” Schuck says. “They’re much more continuous, there’s kind of a gray zone.” He told Ars that looking specifically at the EEG trace may help us better understand what exactly happens in the brain when a sudden moments of insight arrives.

But Shuck wants to get even more data in the future. “We’re currently running a study that’s been years in the making: We want to use both EEG and [functional magnetic resonance imaging] at the same time to see what happens in the brain when people are sleeping,” Schuck says. The addition of the fMRI imaging will enable Schuck and his colleagues to see which areas of the brain get activated during sleep. What the team wants to learn from combining EEG and fMRI imagery is how sleep boosts memory consolidation.

“We also hope to get some insights, no pun intended, into the processes that play a role in generating insights,” Schuck adds.

PLOS Biology, 2025.  DOI: 10.1371/journal.pbio.3003185

Source

Hope you enjoyed this news post.

Thank you for appreciating my time and effort posting news every day for many years.

News posts... 2023: 5,800+ | 2024: 5,700+ | 2025 (till end of June): 2,864

RIP Matrix | Farewell my friend  

MXenes are made of layers that are just one atom thick. Because of this structure, they behave differently than thicker materials made of the same elements. They're being studied for uses in batteries, sensors, electromagnetic shielding, and as solid lubricants—even in space. Due to all these useful properties, MXene is also known as a "miracle material".

“To produce MXenes, you first need so-called MAX phases. These are materials that can consist, for example, of layers of aluminium, titanium and carbon,” said Pierluigi Bilotto from TU Wien’s Institute of Engineering Design and Product Development. “Until now, hydrofluoric acid was used to etch away the aluminium in the MAX, which then resulted in a system of atomically thin layers that can slide against each other with very little resistance. This makes these MXenes a great lubricant.”

Hydrofluoric acid, though, is highly toxic and dangerous to handle. It also creates waste that’s difficult to dispose of. “This is why MXenes have not yet made a major breakthrough in industry,” Bilotto said. “It's hard to build up such a process on an industrial scale, and many companies understandably shy away from taking this step.”

To get around this, the team developed a different method that uses electricity and a safer chemical mix—specifically, sodium tetrafluoroborate and hydrochloric acid (NaBF₄/HCl). Instead of steady electrical current, they used short bursts of voltage (cathodic pulsing). These pulses create small hydrogen bubbles that keep the material's surface active, helping the aluminium layer to be removed more effectively and continuously.

“Electrochemistry offers an alternative route to break the aluminium bonds in the MAX phase,” Bilotto explained. “When an electrical voltage is applied, the MAX phase experiences an electric current that initiates reactions at its interfaces. By precisely selecting the voltage, we are able to tune the reactions in a way that only Aluminium atoms are removed, leaving as product electrochemical MXenes (EC-MXenes).”

Using this method, the team achieved up to 60% yield of EC-MXene in a single round, with no unwanted byproducts. The material was checked using different chemical analysis tools—like SEM/EDX for chemical mapping, XPS and LEIS for surface structure, and AFM, TEM, Raman, and XRD for size, spacing, and other physical traits.

The researchers say the pulsed electric approach not only increases yield but also improves quality by keeping the surface clean and reactive throughout the process. “My goal is to make the synthesis of MXene extremely simple. It should be possible in any kitchen,” said Bilotto. “And we are very close to that.”

Source: TU Wien, Small

This article was generated with some help from AI and reviewed by an editor. Under Section 107 of the Copyright Act 1976, this material is used for the purpose of news reporting. Fair use is a use permitted by copyright statute that might otherwise be infringing.

Source

Hope you enjoyed this news post.

Thank you for appreciating my time and effort posting news every day for many years.

News posts... 2023: 5,800+ | 2024: 5,700+ | 2025 (till end of June): 2,864

RIP Matrix | Farewell my friend  

WASP-121b is a very hot gas giant that’s locked in a tight orbit around its star—so close that it completes one full orbit in just 30.5 hours. One side of the planet always faces the star, reaching temperatures above 3000 °C, while the other side stays in permanent darkness, with temperatures around 1500 °C.

Using JWST’s Near-Infrared Spectrograph (NIRSpec), the team detected water vapor (H₂O), carbon monoxide (CO), silicon monoxide (SiO), and methane (CH₄). These signals were strong; water was found at 5.5–13.5σ significance, CO at 10.8–12.8σ, SiO at 5.7–6.2σ, and methane on the nightside at 3.1–5.1σ.

What makes this interesting is that both refractory elements (materials that usually stay solid under high heat, like silicon, iron, and magnesium) and volatile substances (like water and methane) were found. Usually, it’s hard to detect them together in one go because their signals appear in different parts of the light spectrum. Evans-Soma explained, “Dayside temperatures are high enough for refractory materials – typically solid compounds resistant to strong heat – to exist as gaseous components of the planet’s atmosphere.”

By comparing the detected elements with what’s in the planet’s star, the team found that the planet has more carbon, oxygen, and silicon than expected. These higher-than-stellar values—called super-stellar abundances—suggest that the planet grew by gathering both gas-rich pebbles and rocky planetesimals. Gapp said, “Gaseous materials are easier to identify than liquids and solids. Since many chemical compounds are present in gaseous form, astronomers use WASP-121b as a natural laboratory to probe the properties of planetary atmospheres.”

The planet probably formed in a colder part of its original gas-and-dust disc—far enough out for water to stay frozen, but warm enough for methane to become gas. That kind of environment would be similar to the region between Jupiter and Uranus in our Solar System. Later, the planet likely moved much closer to its star.

Another surprising finding was methane on the nightside. Under known models, methane shouldn’t be there in large amounts because air from the hot dayside should quickly mix into the cooler side and break down the methane. But Evans-Soma said, “This challenges exoplanet dynamical models, which will likely need to be adapted to reproduce the strong vertical mixing we've uncovered on the nightside of WASP-121b.”

The methane must be getting pulled up from deeper layers of the atmosphere by strong vertical winds. These lower layers are rich in methane because of cooler temperatures and a high carbon-to-oxygen ratio.

The team collected data throughout the planet’s full orbit and also when it passed in front of its star. During that transit, some starlight passed through the planet’s thin outer atmosphere, helping the scientists figure out its chemical makeup. Gapp explained, “The emerging transmission spectrum confirmed the detections of silicon monoxide, carbon monoxide, and water that were made with the emission data. However, we could not find methane in the transition zone between the day and night side.”

Source: Max Planck Institute for Astronomy, Nature |Image via Depositphotos

This article was generated with some help from AI and reviewed by an editor. Under Section 107 of the Copyright Act 1976, this material is used for the purpose of news reporting. Fair use is a use permitted by copyright statute that might otherwise be infringing.

Source

Hope you enjoyed this news post.

Thank you for appreciating my time and effort posting news every day for many years.

News posts... 2023: 5,800+ | 2024: 5,700+ | 2025 (till end of June): 2,864

RIP Matrix | Farewell my friend  

When Todd Sacktor was about to turn 3, his 4-year-old sister died of leukemia. “An empty bedroom next to mine. A swing set with two seats instead of one,” he said, recalling the lingering traces of her presence in the house. “There was this missing person—never spoken of—for which I had only one memory.” That memory, faint but enduring, was set in the downstairs den of their home. A young Sacktor asked his sister to read him a book, and she brushed him off: “Go ask your mother.” Sacktor glumly trudged up the stairs to the kitchen.

It’s remarkable that, more than 60 years later, Sacktor remembers this fleeting childhood moment at all. The astonishing nature of memory is that every recollection is a physical trace, imprinted into brain tissue by the molecular machinery of neurons. How the essence of a lived moment is encoded and later retrieved remains one of the central unanswered questions in neuroscience.

Sacktor became a neuroscientist in pursuit of an answer. At the State University of New York Downstate in Brooklyn, he studies the molecules involved in maintaining the neuronal connections underlying memory. The question that has always held his attention was first articulated in 1984 by the famed biologist Francis Crick: How can memories persist for years, even decades, when the body’s molecules degrade and are replaced in a matter of days, weeks or, at most, months?

In 2024, working alongside a team that included his longtime collaborator André Fenton, a neuroscientist at New York University, Sacktor offered a potential explanation in a paper published in Science Advances. The researchers discovered that a persistent bond between two proteins is associated with the strengthening of synapses, which are the connections between neurons. Synaptic strengthening is thought to be fundamental to memory formation. As these proteins degrade, new ones take their place in a connected molecular swap that maintains the bond’s integrity and, therefore, the memory.

In 1984, Francis Crick described a biological conundrum: Memories last years, while most molecules degrade 

in days or weeks. “How then is memory stored in the brain so that its trace is relatively immune to molecular 

turnover?” he wrote in Nature.

Photograph: National Library of Medicine/Science Source

The researchers present “a very convincing case” that “the interaction between these two molecules is needed for memory storage,” said Karl Peter Giese, a neurobiologist at King’s College London who was not involved with the work. The findings offer a compelling response to Crick’s dilemma, reconciling the discordant timescales to explain how ephemeral molecules maintain memories that last a lifetime.

Molecular Memory

Early in his career, Sacktor made a discovery that would shape the rest of his life. After studying under the molecular memory pioneer James Schwartz at Columbia University, he opened his own lab at SUNY Downstate to search for a molecule that might help explain how long-term memories persist.

The molecule he was looking for would be in the brain’s synapses. In 1949, the psychologist Donald Hebb proposed that repeatedly activating neurons strengthens the connections between them, or, as the neurobiologist Carla Shatz later put it: “Cells that fire together, wire together.” In the decades since, many studies have suggested that the stronger the connection between neurons that hold memories, the better the memories persist.

In the early 1990s, in a dish in his lab, Sacktor stimulated a slice of a rat’s hippocampus—a small region of the brain linked to memories of events and places, such as the interaction Sacktor had with his sister in the den—to activate neural pathways in a way that mimicked memory encoding and storage. Then he searched for any molecular changes that had taken place. Every time he repeated the experiment, he saw elevated levels of a certain protein within the synapses. “By the fourth time, I was like, this is it,” he said.

It was protein kinase M zeta, or PKMζ for short. As the rats’ hippocampal tissue was stimulated, synaptic connections strengthened and levels of PKMζ increased. By the time he published his findings in 1993, he was convinced that PKMζ was crucial for memory.

Over the next two decades, he would go on to build a body of work showing that PKMζ’s presence helps maintain memories long after their initial formation. When Sacktor blocked the molecule’s activity an hour after a memory was formed, he saw that synaptic strengthening was reversed. This discovery suggested that PKMζ was “necessary and sufficient” to preserve a memory over time, he wrote in Nature Neuroscience in 2002. In contrast, hundreds of other localized molecules impacted synaptic strengthening only if disrupted within a few minutes of a memory’s formation. It appeared to be a singular molecular key to long-term memory.

To test his hypothesis in live animals, he teamed up with Fenton, who worked at SUNY Downstate at the time and had experience training lab animals and running behavioral experiments. In 2006, the duo published their first paper showing that blocking PKMζ could erase rats’ memories a day or a month after they had formed. This suggested that the persistent activity of PKMζ is required to maintain a memory.

The paper was a bombshell. Sacktor and Fenton’s star protein PKMζ gained widespread attention, and labs around the world found that blocking it could erase various types of memories, including those related to fear and taste. PKMζ seemed like a sweeping explanation for how memories form and are maintained at the molecular level. But then their hypothesis lost momentum. Other researchers genetically engineered mice to lack PKMζ, and in 2013, two independent studies showed that these mice could still form memories. This cast doubt on the protein’s role and brought much of the ongoing research to a halt.

Sacktor and Fenton were undeterred. “We knew we had to figure it out,” Sacktor said. In 2016, they published a rebuttal, demonstrating that in the absence of PKMζ, mice recruit a backup mechanism, involving another molecule, to strengthen synapses.

The existence of a compensatory molecule wasn’t a surprise. “The biological system is not such that you lose one molecule and everything goes. That’s very rare,” Giese said. But identifying this compensatory molecule prompted a new question: How did it know where to go to replace PKMζ? It would take Sacktor and Fenton nearly another decade to find out.

The Maintenance Bond

A classic test of a molecule’s importance is to block it and see what breaks. Determined to pin down PKMζ’s role once and for all, Sacktor and Fenton set out to design a way to disrupt it more precisely than ever before. They developed a new molecule to inhibit the activity of PKMζ. It “worked beautifully,” Sacktor said. But it wasn’t clear how.

One day in 2020, Matteo Bernabo, a graduate student from a collaborating lab at McGill University, was presenting findings related to the PKMζ inhibitor when a clue emerged from the audience. “I suggested that it worked by blocking the PKMζ’s interaction with KIBRA,” recalled Wayne Sossin, a neuroscientist at McGill.

KIBRA is a scaffolding protein. Like an anchor, it holds other proteins in place inside a synapse. In the brain, it is abundant in regions associated with learning and memory. “It’s not a protein that a lot of people work on,” Sossin said, but there is considerable “independent evidence that KIBRA has something to do with memory”—and even that it is associated with PKMζ. Most research has focused on KIBRA’s role in cancer. “In the nervous system,” he said, “there are only three or four of us [studying it].” Sacktor and Fenton joined them.

To find out if KIBRA and PKMζ work together in response to synaptic activity, the researchers used a technique that makes interacting proteins glow. When they applied electrical pulses to hippocampal slices, glowing dots of evidence appeared: Following bursts of synaptic activity that produced long-term synaptic strengthening, a multitude of KIBRA-PKMζ complexes formed, and they were persistent.

Then the team tested the bond during real memory formation by giving mice a drug to disrupt the formation of these complexes. They saw that the mice’s synaptic strength and task memory were lost—and that once the drug wore off, the erased memory did not return, but the mice could acquire and remember new memories once again.

But are the KIBRA-PKMζ complexes needed to maintain memory over the long term? To find out, the researchers disrupted the complex four weeks after a memory was formed. Doing so did indeed wipe out the memory. This suggested that the interaction between KIBRA and PKMζ is crucial not only for forming memories, but also for keeping them intact over time.

Illustration: Carlos Arrojo for Quanta Magazine

 

“It’s the persistent association between two proteins that maintains the memory, rather than a protein that lasts by itself for the lifetime of the memory,” said Panayiotis Tsokas, a neuroscientist working with Sacktor and lead author on the new Science Advances paper.

The KIBRA and PKMζ proteins stabilize each other by forming a bond. That way, when a protein degrades and needs to be replaced, the other remains in place. The bond itself and its location at the specific synapses that were activated during learning are preserved, allowing a new partner to slot itself in, perpetuating the alliance over time. Individually, PKMζ and KIBRA don’t last a lifetime—but by binding to each other, they help ensure your memories might.

The discovery addresses the conundrum first identified by Crick, namely how memories persist despite the relatively short lifetimes of all biological molecules. “There had to be a very, very interesting answer, an elegant answer, for how this could come about,” Fenton said. “And that elegant answer is the KIBRA-PKMζ interacting story.”

This work also answers a question that researchers had put on the shelf. Sacktor’s earlier study showed that increasing levels of PKMζ strengthened synapses and memories. But how did the molecule know where to go within the neuron? “We figured, well, one day, maybe we’ll understand that,” Sacktor said. Now, the researchers think that KIBRA acts as a synaptic tag that guides PKMζ. If true, this would help explain how only the specific synapses involved in a particular physical memory trace are strengthened, when a neuron may have thousands of synapses that connect it to various other cells.

“These experiments very nicely show that KIBRA is necessary for maintaining the activity of PKMζ at the synapse,” said David Glanzman, a neurobiologist at the University of California, Los Angeles, who was not involved in the study. However, he cautioned that this doesn’t necessarily translate to maintaining memory because synaptic strengthening is not the only model for how memory works.

Glanzman’s own past research on sea slugs at first appeared to show that disrupting a molecule analogous to PKMζ erases memory. “Originally, I said it was erased,” Glanzman said, “but later experiments showed we could bring the memory back.” These findings prompted him to reconsider whether memory is truly stored as changes in the strength of synaptic connections. Glanzman, who has worked for 40 years under the synaptic model, is a recent proponent of an alternative view called the molecular-encoding model, which posits that molecules inside a neuron store memories.

While he has no doubt that synaptic strengthening follows memory formation, and that PKMζ plays a major role in this process, he remains unsure if the molecule also stores the memory itself. Still, Glanzman emphasized that this study addresses some of the challenges of the synaptic model, such as molecular turnover and synapse targeting, by “providing evidence that KIBRA and PKMζ form a complex that is synapse-specific and persists longer than either individual molecule.”

Although Sacktor and Fenton believe that this protein pair is fundamental to memory, they know that there may be other factors yet to be discovered that help memories persist. Just as PKMζ led them to KIBRA, the complex might lead them further still.

Original story reprinted with permission from Quanta Magazine, an editorially independent publication of the Simons Foundation whose mission is to enhance public understanding of science by covering research developments and trends in mathematics and the physical and life sciences.

Source

Hope you enjoyed this news post.

Thank you for appreciating my time and effort posting news every day for many years.

News posts... 2023: 5,800+ | 2024: 5,700+ | 2025 (till end of June): 2,864

RIP Matrix | Farewell my friend