About 20 people have been fired from Meta for leaking “confidential information”, such as Mark Zuckerberg’s complaints about too much leaking. CC-licensed photo by Michael Vroegop on Flickr.

You can sign up to receive each day’s Start Up post by email. You’ll need to click a confirmation link, so no spam.

No post today at the Social Warming Substack. Maybe next week?

A selection of 10 links for you. Use them wisely. I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. On Bluesky: @charlesarthur.bsky.social. Observations and links welcome.

AI supertest for prostate cancer screening is “game-changing” • The Times

Eleanor Hayward:

»

Scientists have developed a prostate cancer “supertest” they claim is the most accurate tool yet for detecting the disease.

There is no routine screening programme for prostate cancer in the UK because some tests, including the prostate specific antigen (PSA) blood test, are considered too unreliable.

The new test, developed by EDX Medical Group, based at Cambridge Science Park, uses artificial intelligence (AI) to analyse blood and urine samples, looking for more than 100 biological markers. These markers, including specific genes and proteins, have been clinically validated as having a connection to prostate cancer in previous trials.

The test is the first to combine them into one tool. It aims to provide a comprehensive assessment of a man’s overall risk of prostate cancer, discover whether the disease is present and assess what stage it has reached.

It has been welcomed by Sir Chris Hoy, the former Olympic cycling champion who disclosed in November that he has terminal prostate cancer. He said there was a vital need for “better and more accurate screening tests”.

The company hopes to launch the test privately in the UK over the next year and said it could be used by doctors and “revolutionise” prostate cancer screening for men aged 45 to 70.

…Prostate cancer is the most common cancer in England, with 55,000 men diagnosed each year. If caught early, nearly 100% survive, but if caught late, only 50% live. More than 12,000 men a year die from it in the UK.

«

Maybe this is the use case for AI? They might be good at diagnoses that humans can’t perceive because we can’t pull enough signal from the noise, rather as GPS navigation systems are picking up a signal so weak it’s one-hundredth of the background thermal noise.
unique link to this extract

Antiaging pill for dogs from start-up Loyal wins FDA recognition • The Washington Post

Lisa Bonos:

»

Dog owners do many things to keep their companions happy and healthy. They could soon add an antiaging pill to their pooch’s regimen of walks, vet visits and belly rubs.

Loyal, a biotech startup based in San Francisco, said Wednesday that a drug it developed to increase canine lifespan has passed a significant milestone on the way to regulatory approval.

The Food and Drug Administration certified the daily pill as having a “reasonable expectation of effectiveness” at extending senior dogs’ lifespans.

The regulator’s Center for Veterinary Medicine still has to certify that the drug is safe and that Loyal can manufacture it at scale before vets can prescribe the pill to dogs 10 years or older that weigh 14 pounds or more.

Loyal’s CEO, Celine Halioua, estimates that the process should be complete by the end of 2025 and called the FDA’s initial recognition “a key step” to extending dogs’ lives.

…Loyal previously received a “reasonable expectation of effectiveness” certification from the FDA for a longevity drug for large and giant breeds like Great Danes and Newfoundlands. But that treatment has to be administered via injection and will take longer to manufacture and get to market.

Other companies are working to develop weight-loss drugs akin to Ozempic for dogs and cats. Loyal’s pill is a result of research into how to mimic the life-extending benefit of caloric restriction without the appetite suppression — and without the need for an owner to restrict their dog’s food.

The drug aims to improve a dog’s metabolic fitness, or the body’s ability to convert nutrients into energy and regulate hormones, which declines in humans and canines with age.

«

Older dogs for an ageing population. Fitting, somehow.
unique link to this extract

Apple’s new C1 brings two killer features, and it’s just the start • 9to5Mac

Ryan Christoffel:

»

Another big advantage with the C1 [besides longer battery life due to lower power demand] is that it can be integrated with iOS in a way Qualcomm’s modem never could. This enables it to understand what you’re doing on the device at any given time, and prioritize data use that matters most.

Here’s Apple’s explanation to Reuters:

»

One of the ways Apple hopes the C1 will set its iPhones apart is by tightly integrating it with its processor chips. For example, if an iPhone encounters congested data networks, the phone’s processor can signal to the modem which traffic is the most time sensitive and put it ahead of other data transfers, making the phone feel more responsive to the user’s needs, said Arun Mathias, vice president for wireless software at Apple.

«

…When you’re facing network congestion, Qualcomm’s modem has no idea which data requests are most important. Often that creates a frustrating user experience. But Apple’s C1 will be able to offer more responsive and intelligent data use to meet your exact needs.

«

Quite what that means is a bit vague – is the idea that when you’re using Safari, the C1 will recognise packets destined for the browser? But Apple managed to tune the M series of chips for the most common instruction groups, so maybe the C1 – and its descendants – will have similar benefits.
unique link to this extract

New AI text diffusion models break speed barriers by pulling words from noise • Ars Technica

Benj Edwards:

»

On Thursday, Inception Labs released Mercury Coder, a new AI language model that uses diffusion techniques to generate text faster than conventional models. Unlike traditional models that create text word by word—such as the kind that powers ChatGPT—diffusion-based models like Mercury produce entire responses simultaneously, refining them from an initially masked state into coherent text.

Traditional large language models build text from left to right, one token at a time. They use a technique called “autoregression.” Each word must wait for all previous words before appearing. Inspired by techniques from image-generation models like Stable Diffusion, DALL-E, and Midjourney, text diffusion language models like LLaDA (developed by researchers from Renmin University and Ant Group) and Mercury use a masking-based approach. These models begin with fully obscured content and gradually “denoise” the output, revealing all parts of the response at once.

While image diffusion models add continuous noise to pixel values, text diffusion models can’t apply continuous noise to discrete tokens (chunks of text data). Instead, they replace tokens with special mask tokens as the text equivalent of noise. In LLaDA, the masking probability controls the noise level, with high masking representing high noise and low masking representing low noise. The diffusion process moves from high noise to low noise. Though LLaDA describes this using masking terminology and Mercury uses noise terminology, both apply a similar concept to text generation rooted in diffusion.

Much like the creation of an image synthesis model, researchers build text diffusion models by training a neural network on partially obscured data, having the model predict the most likely completion and then comparing the results with the actual answer. If the model gets it correct, connections in the neural net that led to the correct answer get reinforced. After enough examples, the model can generate outputs with high enough accuracy or plausibility to be useful.

«

Speed is definitely a key element of usefulness for these models: faster is perceived by the user as better. LLMs are proliferating right now as quickly as PCs did in the mid-1980s, when there were dozens of computer brands all jockeying for sales and, eventually, profit – but also all fading away as the market winners emerged. Is this going to be the same? Or can they all coexist?
unique link to this extract

Scientists scorn EPA push to say climate change isn’t a danger, say just look around at the world • AP News

Seth Borenstein:

»

As President Donald Trump’s administration looks to reverse a cornerstone finding that climate change endangers human health and welfare, scientists say they just need to look around because it’s obvious how bad global warming is and how it’s getting worse.

New research and ever more frequent extreme weather further prove the harm climate change is doing to people and the planet, 11 different scientists, experts in health and climate, told The Associated Press soon after word of the administration’s plans leaked out Wednesday. They cited peer-reviewed studies and challenged the Trump administration to justify its own effort with science.

“There is no possible world in which greenhouse gases are not a threat to public health,” said Brown University climate scientist Kim Cobb. “It’s simple physics coming up against simple physiology and biology, and the limits of our existing infrastructure to protect us against worsening climate-fueled extremes.”

Environmental Protection Agency chief Lee Zeldin has privately pushed the White House for a rewrite of the agency’s finding that planet-warming greenhouse gases put the public in danger. The original 52-page decision in 2009 is used to justify and apply regulations and decisions on heat-trapping emissions of greenhouse gases, such as carbon dioxide and methane, from the burning of coal, oil and natural gas.

“Carbon dioxide is the very essence of a dangerous air pollutant. The health evidence was overwhelming back in 2009 when EPA reached its endangerment finding, and that evidence has only grown since then,” said University of Washington public health professor Dr. Howard Frumkin, who as a Republican appointee headed the National Center for Environmental Health at the time. “CO2 pollution is driving catastrophic heat waves and storms, infectious disease spread, mental distress, and numerous other causes of human suffering and preventable death.”

That 2009 science-based assessment cited climate change harming air quality, food production, forests, water quality and supplies, sea level rise, energy issues, basic infrastructure, homes and wildlife.

«

Once again, taking America’s science back to the 1950s.
unique link to this extract

Meta is firing about 20 employees for leaking • The Verge

Alex Heath:

»

Meta has fired “roughly 20” employees who leaked “confidential information outside the company,” according to a spokesperson.

“We tell employees when they join the company, and we offer periodic reminders, that it is against our policies to leak internal information, no matter the intent,” Meta spokesperson Dave Arnold tells The Verge exclusively. “We recently conducted an investigation that resulted in roughly 20 employees being terminated for sharing confidential information outside the company, and we expect there will be more. We take this seriously, and will continue to take action when we identify leaks.”

Meta has ramped up its efforts to find leakers due to a recent influx of stories detailing unannounced product plans and internal meetings, including a recent all-hands led by CEO Mark Zuckerberg. After we and other outlets reported on what Zuckerberg said during that meeting, employees were warned not to leak. In comments that were subsequently leaked, CTO Andrew Bosworth then told them that the company was “making progress on catching people.”

…“There’s a funny thing that’s happening with these leaks,” Bosworth said during an internal meeting in early Febuary. “When things leak, I think a lot of times people think, ‘Ah, okay, this is leaked, therefore it’ll put pressure on us to change things.’ The opposite is more likely.”

«

Facebook – Meta, now – is such a political and cultural chameleon.
unique link to this extract

Congo reports over 50 deaths from mystery illness • The Washington Post

Vivian Ho:

»

An unknown illness has killed 53 people in a northwestern region of the Democratic Republic of Congo, with a significant portion of deaths taking place within 48 hours of the onset of symptoms, according to the World Health Organization, which describes the outbreak as posing “a significant public health threat.”

At least 431 cases have been reported since January of individuals suffering from fever, vomiting, diarrhea, muscle aches, headaches and fatigue, according to the WHO’s Africa office. The illness — believed to have broken out in two separate villages in Équateur province — has a fatality rate of 12.3%, the WHO said.

Investigators traced the outbreak’s origin to the village of Boloko, where three children under the age of five died after reportedly eating a bat carcass, health officials said.

In addition to the other symptoms reported with this disease, the three children suffered symptoms similar to those of a hemorrhagic fever — bleeding from the nose and vomiting blood — before they died between Jan. 10 and Jan. 13.

«

A disease caught from bats? But there are no Chinese laboratories nearby, this can’t have happened. Anyway, good to know that the US has stopped wasting money by cutting USAID budgets which monitored new zoonoses.
unique link to this extract

Trump team weighs pulling funds for Moderna bird flu vaccine • Bloomberg via Yahoo

Madison Muller, Riley Griffin and Ike Swetlitz:

»

US health officials are reevaluating a $590 million contract for bird flu shots that the Biden administration awarded to Moderna Inc., people familiar with the matter said.

The review is part of a government push to examine spending on messenger RNA-based vaccines, the technology that powered Moderna’s Covid vaccine. The bird flu shot contract was awarded to Moderna in the Biden administration’s final days, sending the company’s stock up 13% in the two days following the Jan. 17 announcement.

Shares of Moderna fell as much as 4% when US markets opened on Thursday. The company didn’t respond to a request for comment.

The US is in the midst of a record-breaking bird flu outbreak that’s affected dozens of cattle herds along with poultry flocks nationwide, sending egg prices soaring. While human cases have been relatively rare, the virus has caused deaths in the past, and experts are concerned that it could become more transmissible and dangerous.

“While it is crucial that the US Department and Health and Human Services support pandemic preparedness, four years of the Biden administration’s failed oversight have made it necessary to review agreements for vaccine production,” a spokesperson for HHS said in a written statement.

«

Science? Never heard of it. This is driven by RFK Jr. The US is reverting to the 1950s in every respect, including its scientific understanding. Unfortunately, it’s the American citizenry that’s likely to pay the price of having an idiot in charge of their wider health. (Thanks Joe S for the link.)
unique link to this extract

What would happen if a tiny black hole passed through your body? • Universe Today

Brian Koberlein:

»

Some theoretical models argue that primordial black holes could be the source of dark matter. If that’s the case, observational limits constrain their masses to the 1013 – 1019 kg range, which is similar to the mass range for asteroids. Therefore, the study focuses on this range and looks at two effects: tidal forces and shock waves.

Tidal forces occur because the closer you get to a mass, the stronger its gravity. This means a black hole exerts a force differential on you as it gets near. So the question is whether this force differential is strong enough to tear flesh. Asteroid-mass black holes are less than a micrometer across, so even the tidal forces would cover a tiny area. If one passed through your midsection or one of your limbs, there might be some local damage, but nothing fatal. It would be similar to a needle passing through you.

But if the black hole passed through your head, that would be a different story. Tidal forces could tear apart brain cells, which would be much more serious. Since brain cells are delicate, even a force differential of 10 – 100 nanonewtons might kill you. But that would take a black hole at the highest end of our mass range.

Shockwaves would be much more dangerous. In this case, as a black hole entered your body, it would create a density wave that would ripple through you. These shockwaves would physically damage cells and transfer heat energy that would do further damage. To create a shockwave of energy similar to that of a 22-caliber bullet, the black hole would only need a mass of 1.4 x 1014 kg, which is well within the range of possible primordial black holes.

So yes, a primordial black hole could kill you.

«

As Koberlein and the authors of an ArXiv paper on this topic concede, the idea was first considered by SF author Larry Niven in 1974.
unique link to this extract

2003: IT Doesn’t Matter • Harvard Business Review

Nick Carr, writing in 2003:

»

Some managers may worry that being stingy with IT dollars will damage their competitive positions. But studies of corporate IT spending consistently show that greater expenditures rarely translate into superior financial results. In fact, the opposite is usually true. In 2002, the consulting firm Alinean compared the IT expenditures and the financial results of 7,500 large U.S. companies and discovered that the top performers tended to be among the most tightfisted.

The 25 companies that delivered the highest economic returns, for example, spent on average just 0.8% of their revenues on IT, while the typical company spent 3.7%. A recent study by Forrester Research showed, similarly, that the most lavish spenders on IT rarely post the best results. Even Oracle’s Larry Ellison, one of the great technology salesmen, admitted in a recent interview that “most companies spend too much [on IT] and get very little in return.” As the opportunities for IT-based advantage continue to narrow, the penalties for overspending will only grow.

«

This is a sort of corollary to the question yesterday about AI trying to find its killer app: what is it really useful for? Subscribers to Ben Thompson’s Stratechery would have read a long interview – more of a discussion – between Thompson and Benedict Evans in which they try to figure out what AI is best used for, and what the AI products of the future will look like. The answer seems to be “we don’t know yet, but it isn’t an empty box asking you to input text”.

In that sense, Carr’s essay could be updated and called AI Doesn’t Matter. And for completeness, Wendy Grossman points out that Thomas K Landauer wrote a 1995 book called “The Trouble with Computers: Usefulness, Usability and Productivity”, about the perceived “productivity paradox” of these new systems.
unique link to this extract

Errata, corrigenda and ai no corrida: none notified

Reaching the UK’s energy transition targets will mean learning to love pylons – because burying cables is insanely expensive. CC-licensed photo by Grey World on Flickr.

You can sign up to receive each day’s Start Up post by email. You’ll need to click a confirmation link, so no spam.

There’s another post coming this week at the Social Warming Substack on Friday at 0845 UK time. Free signup.

A selection of 9 links for you. Highly strung. I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. On Bluesky: @charlesarthur.bsky.social. Observations and links welcome.

Satya Nadella says AI is yet to find its killer app • The Register

Tobias Mann:

»

Nadella thinks a better benchmark for AI’s success [than AGI-adjacent benchmarks] should be its ability to boost a country’s gross domestic product. “When we say: ‘Oh, this is like the industrial revolution,’ let’s have that industrial revolution type of growth. That means to me, ten percent, seven% for the developed world. Inflation adjusted, growing at five percent, that’s the real marker.”

Few nations achieved that pace of growth in 2024.

Nadella suggested that growth hasn’t eventuated because it’s going to take time before folks understand how to use AI effectively, assuming they find a use for it – just as it took some years for the personal computer to find its feet.

“Just imagine how a multinational corporation like us did forecasts pre-PC, and email, and spreadsheets. Faxes went around, somebody then got those faxes and then did an inter-office memo that then went around, and people entered numbers, and then ultimately a forecast came out maybe just in time for the next quarter,” Nadella explained.

“Then somebody said: ‘Hey, I’m just going to take an Excel spreadsheet, put it in an email, send it around, people will go edit it, and I’ll have a forecast.’ The entire forecasting business process changed because the work artifact and the workflow changed. That is what needs to happen with AI being introduced into knowledge work,” the CEO said.

«

The question of where GDP growth really came from, and how big a role each computing introduction – PCs, spreadsheets, email, web browsing, broadband, mobile phones, mobile broadband, smartphones – added to GDP is one for the ages.
unique link to this extract

Ship and Chinese crew detained after Penghu undersea cable severed • Focus Taiwan

»

A Togolese-registered vessel and its Chinese crew have been detained after a submarine communications cable linking Taiwan and Penghu was severed Tuesday.

Taiwan’s Coast Guard Administration (CGA) said it dispatched the PP-10079 patrol and rescue vessel at 2:30 a.m. Tuesday to monitor the “Hong Tai” freighter, which was anchored 6 nautical miles northwest of Jiangjun Fishing Port in Tainan.

The CGA said it immediately issued broadcasts ordering the vessel to leave.

At around 3 a.m., the CGA received a report from the partially state-owned Chunghwa Telecom informing it that the “Taiwan-Penghu No. 3” submarine fiber optic cable had been severed.

The CGA said that it then began attempting to detain the ship on grounds of possible sabotage.

However, due to the difference in height between the two vessels preventing CGA personnel from boarding the Hong Tai, the CGA said it had to send the Cijin offshore patrol vessel and the PP-10059 patrol vessel to the scene to assist with boarding and detention.

The Hong Tai, which the CGA said was funded by “Chinese capital,” was later brought to Anping Harbor where it and its eight crew, all Chinese nationals, were detained, pending investigation by Tainan district prosecutors.

When PP-10079 radioed the freighter, its crew said the ship’s name was “Hong Tai 168,” which contradicted the name shown by its Automatic Identification System “Hong Tai 58,” the CGA said.

The CGA added that the ship’s name would be verified by an upcoming investigation.

According to the CGA, the freighter had been loitering near the broken cable at a distance of about 925 meters from the cable since 7 p.m. on Feb. 22.

«

Drones and submarine cables. It’s the new warfare.
unique link to this extract

Amazon’s subscription-based Alexa+ looks highly capable—and questionable • Ars Technica

Scharon Harding:

»

Amazon representatives showed Alexa+ learning what a family member likes to eat and later recalling that information to recommend appropriate recipes. In another demo, Alexa+ appeared to set a price monitor for ticket availability on Ticketmaster. Alexa+ told the user it would notify them of price drops via their Echo or Alexa.

I also saw Alexa+ identify, per the issued prompt, “that song Bradley Cooper sings. It’s, like, in a duet” and stream it off of Amazon Music via Echo devices placed around the room. The user was able to toggle audio playing from Echo devices on the left or right side of the room. He then had Alexa+ quickly play the scene from the movie A Star Is Born (that the song is from) on a Fire TV.

Notably, Alexa+ understood directions delivered in casual speak (for example: “can you just jump to the scene in the movie?”). During the demos, the Echo Show in use showed a transcription of the user and voice assistant’s conversation on-screen. At times, I saw the transcription fix mistakes. For example, when a speaker said “I’m in New York,” Alexa first heard “I’m imminent,” but by the time the speaker was done talking, the transcribed prompt was corrected.

I even saw Alexa+ use some logic. In one demo, a user requested tickets for Seattle Storm games in Seattle in March. Since there were none, Alexa+ asked if the user wanted to look for games in April. This showed Alexa+ anticipating a user’s potential response, while increasing the chances that Amazon would be compensated for helping to drive a future ticket sale.

Unlike with today’s Alexa, Alexa+ is supposed to be able to interpret shared documents. An Amazon rep appeared to show Alexa+ reading a homeowner’s association contract to determine if the user is allowed to install solar panels on their home. Although, as some have learned recently, there are inherent risks with relying on AI to provide totally accurate information about contracts, legal information, or, really anything.

«

(I think the “questionable” means you can ask it questions.) Free for Prime users, $20 for others. Looks like a big leap forward if you’re prepared to commit to Amazon running your home.
unique link to this extract

It’s easier than ever to scrub your personal info from Google Search • Ars Technica

Ryan Whitwam:

»

As Google’s 2024 antitrust loss proved, the company has worked very, very hard to ensure its search engine is the primary roadmap for the Internet. Google scours the Internet for data about everything—even you. And if you don’t want your personal info to wind up in Google search results, you can use the just-redesigned “Results About You” tool. The tool, which began its rollout in 2022, is easier to use now, and some of the most useful features are now better integrated with search results.

The first step in using Results About You—which has not changed—is a bit alarming when you’ve set out to obscure your personal information. Just head to the new hub for Results About You and enter your personal information. Google probably already knows your phone number, email, and even physical address, but this tells the tool what specific information to pluck out of search results. If that data is out there, Google has it whether or not you remove it from search results.

Before this update, most of the Results About You features were limited to this console, but the most important features are now integrated with the search results. They’re not exactly prominently displayed, though. When scrolling through a Google search (after the AI overview, ads, knowledge graph, and more ads), you can use the three-dot menu next to a result to get data about it. This menu now includes options to remove the result right at the top.

If you request a removal due to the presence of personal information, Google will ask for more details, but that only takes a few seconds.…If you’re requesting a personal data removal, it has to be your data.

«

It’s the Right To Be Forgotten under another name, isn’t it. The data still exists out there, but you can’t find it through Google.
unique link to this extract

AI linked to growing cancer risk • Futurism

Joe Wilkins:

»

As the artificial intelligence boom spirals to epic proportions, big tech companies are throwing heaps of cash into massive data centres throughout the world.

Packed full of hardware to process AI queries, these data centers put out forest-melting levels of heat as they suck the life out of local energy grids and water tables to meet demand. They’re incredibly noisy as well — pumping incessant mechanical sounds into quiet neighbourhoods and driving away wildlife.

And unfortunately, the public cost of AI doesn’t end there. New research by academics at UC Riverside and Caltech is warning that AI data centres are also taking a massive toll on human health, in the form of diseases like cancer and asthma.

The study, which hasn’t yet been peer-reviewed, looked at the production output of AI hardware over the past five years, found that air pollution resulting from AI development could cause as many as 1,300 premature cancer and asthma deaths per year by 2030.

That’s on top of a cost approaching nearly $20bn a year from the collective burden of health treatment, missing wages, and lower school attendance as a result of diseases caused by AI runoff. In 2023 alone, the total cost of AI-connected illness was $1.5bn, the paper found, in an eye-watering 20% increase from 2022.

The issue of air pollution is easy to overlook, because in most cases, the data centers are powered by local coal-burning plants, which tend to be disproportionately located near low-income and working-class communities. It also seems wherever they go, AI data centres drive up the local cost of electricity, saddling their host communities with a burden not shared by the rest of the country, let alone by Silicon Valley or big tech’s Wall Street investors.

«

The estimate relies heavily on some handwaving guesses about the use of diesel backup generators at data centres (there’s no data presented on how much they’re used), the above-mentioned use of coal-burning plants (limited), and the manufacture of the chips. (Thanks Greg B for the link.)
unique link to this extract

What would reaching net zero mean for life in 2040s Britain? • The Guardian

Fiona Harvey:

»

The government’s climate advisers have published their latest official advice on meeting the UK’s legally binding target of reaching net zero emissions by 2050. The advice, which covers the period from 2038 to 2042, contains dozens of recommendations covering all aspects of society. But how will Britons’ lifestyles change under these plans?

…Moving about: Perhaps as soon as next year, or by 2028, the CCC estimates that electric cars will reach the same price levels as petrol and diesel models. This is likely to spur greater take-up, and by 2040 no new fossil fuel vehicles will be available. But people should also be encouraged to walk and cycle, and public transport must play a crucial role, the CCC said. “Better infrastructure enables more people to choose public transport, cycling, or walking instead of driving, bringing the UK closer in line with countries such as Germany, Switzerland, and the Netherlands,” the report found.

Homes: Most people in the UK heat their homes with gas, but by 2040 most will need electric heat pumps if the UK’s carbon targets are to be met. This does not mean ripping out gas boilers – replacing old gas appliances with electric ones can be done when the existing ones wear out. This is likely to require some form of government intervention and support, as heat pumps are still expensive. Hydrogen, posited by some as a way to carry on using the UK’s existing gas supply networks to replace gas in home heating systems, is definitely out – it will never be cheap enough or feasible, according to multiple studies, and the CCC said its use would be largely confined to industrial settings.

«

There’s plenty more – eating (different), flying (less), countryside and nature (more trees!), energy (lots more), working (in the green economy).
unique link to this extract

The cost of burying our grid • Yes and Grow

Ben Hopkinson:

»

The most in-depth study on transmission grid costs was done by the construction firm, Parsons Brinckerhoff. They found that overhead lines are the cheapest transmission technology with lifetime costs varying between £2.2m and £4.2m per kilometre (in 2012 prices). Burying the cables underground costs between £10.2m and £24.1m per kilometre, five to six times more. Importantly, underground cables were found to always be more expensive when compared to equivalent overhead lines. These are extra costs that billpayers would have to shoulder, when Britain already has some of the most expensive electricity in the world.

Not only are overhead lines six times cheaper than underground cables, they are also better for the local environment. Overhead cables are cooled by the air around them, while underground cables need to be spaced apart to avoid overheating. To match one overhead pylon line, as many as 12 separate cables in four separate trenches may be needed, resulting in a work area up to 65m wide. That means existing hedgerows and trees will need to be cut down to make way for the worksite. Plus all this digging threatens sensitive habitats and could damage archaeological heritage.

Once the construction is complete, access will still be needed for the life of the link, which means restrictions over buildings, trees, and hedgerows over the cables. Even with these restrictions it is much harder to quickly repair underground cables. If a fault occurs on one, it is on average out of service for 25 times longer than an equivalent overhead line.

«

And yet there are people who bewail pylons (notably among MPs). I’ve lived near a pylon. You basically don’t see it.
unique link to this extract

Intelligence chief Tulsi Gabbard will fight ‘egregious’ Apple back-door order • The Washington Post

Joseph Menn:

»

New U.S. Director of National Intelligence Tulsi Gabbard has called a U.K. order that Apple break the encrypted storage it offers customers worldwide an “egregious” violation of American rights and said it could violate a law easing cooperation between the countries in investigations.

Gabbard wrote late Tuesday to Sen. Ron Wyden (D-Oregon) and Rep. Andy Biggs (R-Arizona), saying that she had directed a legal review of the secret order and that she had not known of it before it was reported by The Washington Post and confirmed by other publications. The legislators had urged her to act just after her confirmation as the top U.S. intelligence leader.

“I share your grave concern about the serious implications of the United Kingdom, or any foreign country, requiring Apple or any company to create a “backdoor” that would allow access to Americans’ personal encrypted data,” she wrote in response. “This would be a clear and egregious violation of Americans’ privacy and civil liberties, and open up a serious vulnerability for cyber exploitation by adversarial actors.”

«

Very much looking forward to seeing Gabbard fight an order that has been in use in the US – access to people’s iCloud data with a proper court order – for years. In fact the US has been particularly eager to access data held on American companies’ data centres, regardless of the location or nationality of the data’s owner.
unique link to this extract

The Casio Ring Watch is extremely silly, and that’s why I love it • The Verge

Victoria Song:

»

There is no pretense here. This is a tiny Casio watch that sits on your finger. Casio made it to celebrate its 50th anniversary and to cash in on your retro design nostalgia for the halcyon age of our collective youth. It harkens back to the vintage watch rings of the ’80s and ’90s, which you can find on Etsy for $10. This particular one just happens to be fully functional.

Unboxing it, my first thought as a reasonable person is that no one should buy this. For starters, it’s currently unavailable on Casio’s site and is going for upward of $300 on eBay. (Such is the fate of limited-edition gadgety baubles.) In an age when eggs cost $5 a carton — $7, if you live in my neck of the woods — your money can be spent on more practical things, especially since you probably already own a dozen gadgets that can also tell you the time.

Not to mention, this ring watch only comes in a single 10.5 size. If your fingers are smaller, you’ll need one of two included spacers to make it fit. If your fingers are bigger, sorry. No fun for you. Besides, how practical could something like this be? Never mind that it has a stopwatch, an alarm, and dual timezone features. You’d probably never use any of them, because what are these, buttons for ants?

These were my mature, responsible adult thoughts before slipping on the ring. Unfortunately, the second it was on my finger, I morphed into the hhhehehe lizard.

It just looks cool. The Casio Ring Watch is the sort of dweeby chic that reminds me of childhood: before puberty and the consuming need to fit in, when wearing Disney princess tiaras and Teenage Mutant Ninja Turtles watches was legitimately cool.

«

I guess not all tech needs to be useful. Sometimes just existing is enough. Probably one for women more than men, though.
unique link to this extract

Errata, corrigenda and ai no corrida: none notified

The rise in streaming has been matched by a fall in understanding of what people are watching, and for how long. CC-licensed photo by Stock Catalog on Flickr.

You can sign up to receive each day’s Start Up post by email. You’ll need to click a confirmation link, so no spam.

There’s another post coming this week at the Social Warming Substack on Friday at 0845 UK time. Free signup.

A selection of 9 links for you. Bears watching. I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. On Bluesky: @charlesarthur.bsky.social. Observations and links welcome.

Chegg sues Google, explores sale after AI search summaries hit revenues • FT via Ars Technica

Maxine Kelly:

»

Chegg is suing Google parent Alphabet over claims the search engine’s artificial intelligence summary tool has hit its revenues, leading the US-listed educational technology group to weigh up a sale of the business.

California-based Chegg, which provides study tools for students, filed the complaint on Monday claiming that Google AI Overviews, which presents users with summary answers to their queries, serves to keep users on Google’s own site.

Chief executive Nathan Schultz said the search giant’s AI search changes had “unjustly retained traffic that has historically come to Chegg, impacting our acquisitions, revenue and employees.”

He added this was “materially impacting” revenues, leading the company to instigate a strategic revenue in which it would explore “a range of alternatives to maximize shareholder value, including being acquired, undertaking a go-private transaction, or remaining as a public standalone company.”

Google said: “With AI Overviews, people find search more helpful and use it more, creating new opportunities for content to be discovered. Every day, Google sends billions of clicks to sites across the web, and AI Overviews send traffic to a greater diversity of sites. We will defend against these meritless claims.”

Chegg’s move comes as developments in AI shake global industries, from health care to education and the automotive sector. The fast development of generative AI since OpenAI launched ChatGPT in November 2022 has upended demand for edtech companies’ paid-for online learning tools, leading their valuations to plummet.

The company’s shares were down 22% in premarket trading on Tuesday. Chegg’s stock is down more than 80% over the past year.

Edtech businesses are grappling with how to harness the benefits of generative AI, often integrating some of the tools into their own products, while also being wary of AI companies establishing rival platforms, such as OpenAI’s recently launched ChatGPT Edu and Google’s LearnLM.

…Chegg reported on Monday that total net revenues for the three-month period to the end of December dropped nearly a quarter to $143.5m compared with the previous year, while the number of service subscribers fell 21% to 3.6 million.

«

Probably the first of many.
unique link to this extract

Universities must embrace AI or face extinction • Chris Kanan

Chris Kanan is a professor of computer science at the University of Rochester in the US, where he leads an AI initiative:

»

Before World War II, American higher education was not about job training—it was about intellectual and personal development:

• Intellectual Rigor – Students engaged deeply in philosophy, rhetoric, and interdisciplinary scholarship.

• Moral and Civic Leadership – Universities shaped ethical, community-minded leaders.

• Broad Knowledge – Curricula encouraged cross-disciplinary exploration over narrow specialization.

After WW2, this emphasis diminished. The GI Bill expanded access, and universities became focused on preparing students for jobs rather than personal or intellectual growth. While this shift was beneficial for economic mobility, today’s AI-driven world demands a return to foundational skills that AI cannot fully replicate: critical thinking, reasoning, and adaptive problem-solving.

Meanwhile, higher education faces a double crisis: a projected 13% enrollment decline by 2041—likely an underestimate if AI disrupts white-collar job markets even faster than expected.

AI has advanced dramatically in just a few years. Today’s best AI systems can achieve “B” or better in nearly all college courses. A Frontiers in Psychology study found that ChatGPT can even surpass humans in perceived emotional intelligence—a domain once thought uniquely human. Meanwhile, many AI experts, myself included, predict that artificial general intelligence (AGI) could arrive within the next decade. Once AGI emerges, it will displace many of the jobs people attend college to obtain.

This trajectory is an existential threat to universities reliant on the promise of career advancement. If employers can access AI that is faster, never fatigued, has mastered every discipline, and even competent in emotional intelligence, the economic rationale for college weakens.

«

UK universities face similar challenges, but with the added problem that their budgets are under gigantic pressure.
unique link to this extract

Facebook boosts viral content as it drops fact-checking • ProPublica

Craig Silverman:

»

Meta has made debunking viral hoaxes created for money a top priority for nearly a decade, with one executive calling this content the “worst of the worst.” Meta has a policy against paying for content its fact-checkers label as false, but that rule will become irrelevant when the company stops working with them. Already, 404 Media found that overseas spammers are earning payouts using deceptive AI-generated content, including images of emaciated people meant to stoke emotion and engagement. Such content is rarely fact-checked because it doesn’t make any verifiable claims.

With the removal of fact-checks in the U.S., “what is the protection now against viral hoaxes for profit?” said Jeff Allen, the chief research officer of the nonprofit Integrity Institute and a former Meta data scientist.

“The systems are designed to amplify the most salacious and inciting content,” he added.

In an exchange on Facebook Messenger, the manager of NO Filter Seeking Truth, which shared the false ICE post, told ProPublica that the page has been penalized so many times for sharing false information that Meta won’t allow it to earn money under the current rules. The page is run by a woman based in the southern U.S., who spoke on the condition of anonymity because she said she has received threats due to her posts. She said the news about the fact-checking system ending was “great information.”

«

A completely predictable outcome, given that it has already happened on what used to be Twitter.
unique link to this extract

Solar, battery storage to lead new U.S. generating capacity additions in 2025 • U.S. Energy Information Administration (EIA)

»

We expect 63 gigawatts (GW) of new utility-scale electric-generating capacity to be added to the U.S. power grid in 2025 in our latest Preliminary Monthly Electric Generator Inventory report. This amount represents an almost 30% increase from 2024 when 48.6 GW of capacity was installed, the largest capacity installation in a single year since 2002. Together, solar and battery storage account for 81% of the expected total capacity additions, with solar making up over 50% of the increase.

…Battery storage: In 2025, capacity growth from battery storage could set a record as we expect 18.2 GW of utility-scale battery storage to be added to the grid. U.S. battery storage already achieved record growth in 2024 when power providers added 10.3 GW of new battery storage capacity. This growth highlights the importance of battery storage when used with renewable energy, helping to balance supply and demand and improve grid stability. Energy storage systems are not primary electricity sources, meaning the technology does not create electricity from a fuel or natural resource. Instead, they store electricity that has already been created from an electricity generator or the electric power grid, which makes energy storage systems secondary sources of electricity.

«

Install a lot of renewables, and the way to get past their intermittency is to add batteries. (And build nuclear power stations.)
unique link to this extract

Amazon shuts down Chime, its Zoom alternative • TechCrunch

Sarah Perez:

»

Amazon Chime, the tech giant’s underwhelming alternative to Zoom and Google Meet, is shutting down for good. The company on Wednesday confirmed it will end support for Chime, including its Business Calling features, on February 20, 2026.

As part of this transition, Amazon stopped accepting new Chime accounts as of February 19, 2025.

From now until Chime’s end of life, existing customers will still be able to schedule and host meetings, add and manage users, and take advantage of features in the Amazon Chime administration console, a company blog post explains. The Amazon Chime SDK, however, will not be impacted.

Customers are advised to delete their data before the shutdown and make the move to another web meetings service. Amazon recommends other solutions like AWS Wickr, Zoom, or Salesforce’s Slack.

«

But launched in 2017, rather than (say) late 2020 as you might otherwise guess. Basically used only inside Amazon, and it looks like even they tired of it.
unique link to this extract

Two people in US hospitalized with bird flu, CDC reports • The Guardian

Melody Schreiber:

»

Two people, in Wyoming and Ohio, have been hospitalized with H5N1 bird flu, the US Centers for Disease Control and Prevention (CDC) said in a routine flu update on Friday.

The person from Wyoming is still in hospital, while the Ohio patient has been released, according to the report. Both patients experienced “respiratory and non-respiratory symptoms”, the report said, without detailing those symptoms.

“This shows that H5N1 can be very severe and we should not assume that it will always be mild,” said Angela Rasmussen, a virologist at the Vaccine and Infectious Disease Organization at the University of Saskatchewan.

The news comes amid one of the worst seasonal flu outbreaks in 15 years – raising the potential for the emergence of a more dangerous virus that combines bird flu and seasonal flu in a process called reassortment.

“I am very worried about H5N1 in patients that are being treated in hospitals where there are also many seasonal flu patients because this creates opportunities for reassortment, which could potentially produce a pandemic-capable H5N1,” Rasmussen said.

These are the first human H5 cases detected in Wyoming and Ohio.

«

Watching brief (still). (Thanks Joe S for the link.)
unique link to this extract

Who’s watching what on TV? Who’s to say? • The New York Times

John Koblin:

»

People now watch so many programs at so many different times in so many different ways — with an antenna, on cable, in an app or from a website, as well as live, recorded or on demand — that it is increasingly challenging for the industry to agree on the best way to measure viewership. In some cases, media executives and advertisers are even uncertain whether a competitor’s show is a hit or something well short of that.

The scramble to sort out a suitable solution began nearly a decade ago, as Netflix rose to prominence. It has only intensified since. “It is more chaotic than it’s ever been,” said George Ivie, the chief executive of the Media Rating Council, a leading industry measurement watchdog.

For decades, there was no dispute — Nielsen’s measurement was the only game in town.

But things started to go sideways after the emergence of streaming services like Netflix, Hulu and Amazon Prime Video. Nielsen had no ability — at least at first — to measure how many people clicked play on those apps. The streamers, of course, knew exactly how many people were watching on their own service but they either selectively disclosed some data or did not bother releasing it at all.

Over the past two years, as nearly all the major streaming services have introduced advertising, they have released more data. But the data they release makes apples-to-apples comparisons difficult.

Netflix discloses what it calls “hours viewed” and “views” for its shows. Prime Video and Max prefer to describe how many million “viewers” watched a hit of their choosing.

The disclosures can be helpful to compare one show with another on the same streaming service. Yet those figures, too, can lead to disagreements.

«

“Nobody knows anything”, as Samuel Goldman famously said about Hollywood’s inability to figure out what would make a film work or not; now the not-knowing is spread far and wide, with no reliable way to know what might happen. Are there streaming sleeper hits, as the film The Shawshank Redemption (not big in cinema, became a hit on DVD) was? We don’t know. Nobody knows.
unique link to this extract

Large majority of Taiwanese oppose transfer of cutting-edge TSMC tech to US • Taiwan News

»

Nearly 85% of Taiwanese respondents oppose transferring TSMC’s 2nm technology to the US, according to a poll released Monday, as former US President Donald Trump continues to push for relocating semiconductor manufacturing through tariff threats.

Deputy Legislative Speaker Johnny Chiang (江啟臣) shared the poll results at a press conference, urging the government to prepare for potential trade negotiations under a second Trump administration, per CNA. The poll, conducted by the Foundation for the People (啟思民本基金會), found that 85.6% of respondents expect Trump to impose tariffs on Taiwan, while 62.4% believe the US holds the upper hand in trade talks.

Taiwan recorded a record-high trade surplus of US$73.9bn (NT$2.41 trillion) with the US last year. Chiang speculated that Washington may pressure Taiwan to reduce the surplus by increasing purchases of American weapons, fossil fuels, or agricultural products.

Regarding Taiwan’s semiconductor industry, 83.8% of respondents agreed that companies like TSMC are Taiwan’s “sacred mountain.” Additionally, 62.5% viewed the semiconductor sector as Taiwan’s “silicon shield,” believing that its strategic importance would prompt Western intervention if tensions with China escalated.

Trump has repeatedly claimed that Taiwan “stole” the US semiconductor industry, a statement 88.4% of respondents disagreed with. Furthermore, 84.8% opposed transferring TSMC’s advanced 2nm process technology to the US.

«

One has to think that the remaining 15.2% don’t know what TSMC is, but the simple principle that Taiwan is important to the US only as long as it has cutting-edge chip technology is obvious to everyone, far and wide.

The fact that Trump wants the technology transferred to the US tells you something else. That around one-third think Taiwan might actually have some leverage in trade talks does too: what if Taiwan’s chip shipments to the US were, oh dear, unavoidably delayed?
unique link to this extract

Sonos speakers and soundbars are 25% off for existing customers • The Verge

Sheena Vasani:

»

Since its app fiasco last year, Sonos has been busy slowly rebuilding its reputation with customers. This latest sale, which is exclusive to existing customers, seems to be a part of that strategy — and we’re not complaining, because the deals are solid. Through March 2nd, Sonos is taking 25% off one select Sonos product up to $2,500, though you’ll have to sign in (or create an account) to see the discounted prices. To qualify, you must have registered a Sonos product by February 19th, 2025.

The Sonos sale includes a wide range of audio gear, many of which are down to new all-time low prices. For example, the Sonos Ace is cheaper than ever at $336.75 ($112.25 off). Along with delivering good sound and noise cancellation, the Ace supports TV Audio Swap so you can pair it with Sonos soundbars for private listening. That includes the latest Sonos Beam and Sonos Ray, both of which are also on sale for some of their best prices at $374.25 ($124.75 off) and $209.25 ($69.75 off), respectively.

«

Not happening (yet?) in the UK. Sacrificing profits for revenue to shift product ahead of the end of the quarter on March 31. Meanwhile, no hint of when a release of the app that actually works (and can be used to set up those new speakers) will be forthcoming. The company’s current market capitalisation is about $1.6bn – about one year’s revenue. (It also announced a $150m stock repurchase program on Monday. Was there really no better use for that money?)

It’s a company still in trouble, and no obvious sign of a buyer.
unique link to this extract

Errata, corrigenda and ai no corrida: none notified

The nerds at DOGE wrestling with COBOL are not going to win – the programming language is deeply embedded in government. CC-licensed photo by Kevin Savetz on Flickr.

You can sign up to receive each day’s Start Up post by email. You’ll need to click a confirmation link, so no spam.

There’s another post coming this week at the Social Warming Substack on Friday at 0845 UK time. Free signup.

A selection of 10 links for you. Business-oriented. I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. On Bluesky: @charlesarthur.bsky.social. Observations and links welcome.

If COBOL is so problematic, why does the US government still use it? • ZDNet

Steven Vaughan-Nichols:

»

In the coming weeks and months, we might see DOGE reporting finding “fraud” in government agencies where the real crime is out-of-date, badly-maintained software, and not any criminal intent.

COBOL, you see, may be old, but it’s far from dead. According to the Government Accounting Office (GAO), numerous vital government systems still rely on legacy software and hardware. Some of these systems are more than 50 years old.

These archaic systems include ones for the Department of Education for tracking students; the Department of Health and Human Services’ clinical and patient administration; and Medicare & Medicaid Services still uses COBOL-based systems for critical operations.

Last but not least, the Internal Revenue Service (IRS) still uses COBOL for some of its critical systems. Altogether, the IRS still relies on approximately 160 COBOL applications. At the IRS’s heart is every taxpayer’s Individual Master File (IMF). This is written not just in COBOL but in IBM Assembler as well. This is, in no way, shape, or form, easy code to work with. Complicating matters further, the IRS uses multiple versions of COBOL, including IBM COBOL for OS/390 & VM, IBM Enterprise COBOL for zOS, and Micro Focus Visual COBOL for Eclipse.

You might think that IMF, which dates back to the early 1960s, is the oldest computer system. You’d be wrong. The Department of Defense’s computerized contract-management system, Mechanization of Contract Administration Services (MOCAS), is still with us after almost 67 years of service. That’s even older than the official release of COBOL itself. MOCAS was written in beta COBOL.

In theory, the IMF was to be replaced by 2028. That was before DOGE cut 6,000 IRS employees out of the government agency’s approximately 82,990 employees.

This problem isn’t limited to just the Federal government. Forty-five of the 50 states and the District of Columbia still run COBOL systems. Many of you may remember when Covid first hit, and you couldn’t get unemployment checks. More than likely, your money was delayed because of over-burdened COBOL-based unemployment programs.

None of this is COBOL’s fault. It may be old, but COBOL is still useful. Indeed, when you get cash from an ATM, 95% of the time you’re interacting with a COBOL program. Behind them, 43% of banking systems are written in COBOL, and 90% of banks still use some COBOL. It’s not just banks. Insurance companies also rely on COBOL. This old programming language will not be leaving us this decade, maybe not this century.

«

Cockroaches and COBOL, the two great survivors.
unique link to this extract

Trump manufacturing win: Apple to spend $500bn in U.S., hire 20,000 • Axios

Mike Allen and Ben Berkowitz:

»

Apple CEO Tim Cook said in the announcement: “We are bullish on the future of American innovation, and we’re proud to build on our longstanding U.S. investments with this $500 billion commitment to our country’s future.”

“From doubling our Advanced Manufacturing Fund [from $5 billion to $10 billion], to building advanced technology in Texas, we’re thrilled to expand our support for American manufacturing,” Cook added. “And we’ll keep working with people and companies across this country to help write an extraordinary new chapter in the history of American innovation.”

Trump met with Cook on Thursday in the Oval Office. Then Trump got so excited that he revealed the plans prematurely, saying on-camera while meeting with governors that Cook is “investing hundreds of billions of dollars. I hope he’s announced it — I hope I didn’t announce it, but what the hell? All I do is tell the truth — that’s what he told me. Now he has to do it, right?”

“He is investing hundreds of billions of dollars and others, too,” Trump continued. “We will have a lot of chipmakers coming in, a lot of automakers coming in. They stopped two plants in Mexico that were … starting construction. They just stopped them — they’re going to build them here instead, because they don’t want to pay the tariffs. Tariffs are amazing.”

«

Axios gets a prize for most credulous reporting, and Trump for most credulous president. Apple announced much the same thing in 2021 (that’s ongoing – maybe this is part of it?). The Mexico stuff is perhaps two ancient ones which made keyboards. It’s the Potemkin investment, but Trump is happy.
unique link to this extract

Perplexity wants to reinvent the web browser with AI—but there’s fierce competition • Ars Technica

Samuel Axon:

»

Natural-language search engine Perplexity will launch a web browser, joining a competitive and crowded space that has for years been dominated by Google.

The browser will be called Comet, but we know nothing at all about its features or intended positioning within the browser market at this stage. Comet was announced in an X post with a flashy animation but no details.

Perplexity followed up the X post with a link and an invitation to sign up for beta access to the browser. Those who follow the link will find a barebones website (again with no details) and a simple form for entering an email address.

When we entered an address, we received a brief email saying that Perplexity will add new users to the beta every week and that people can get in faster by sharing Comet on social media and tagging Perplexity’s account.

Perplexity was founded in 2022 by a group of engineers with backgrounds in machine learning. Its primary product is a large language model-based search engine wherein users can type queries to get information from the web and various databases. Perplexity gathers the information, summarizes it, presents it, and takes follow-up questions to dig deeper.

It has recently been expanding its offerings—for example, it recently launched a deep research tool competing with similar ones provided by OpenAI and Google, as well as Sonar, an API for generative AI-powered search.

It will face fierce competition in the browser market, though. Google’s Chrome accounts for the majority of web browser use around the world, and despite its position at the forefront of AI search, Perplexity isn’t the first to introduce a browser with heavy use of generative AI features. For example, The Browser Company showed off its Dia browser in December.

Dia will allow users to type natural language commands into the search bar, like finding a document or webpage or creating a calendar event. It’s possible that Comet will do similar things, but again, we don’t know.

«

Control people’s browsing, and you control their internet and their life – the scammers who would hack your browser to redirect your searches to their chosen search engine have known this for years. You really have to trust the maker of a browser, and it has to do something special. Chrome was special at first: really, really fast. Perplexity is going to have to do something remarkable to justify handing over your digital life to it.
unique link to this extract

Latest calculations conclude Asteroid 2024 YR4 now poses no significant threat to Earth in 2032 and beyond • Planetary Defense

»

NASA has significantly lowered the risk of near-Earth asteroid 2024 YR4 as an impact threat to Earth for the foreseeable future. When first discovered, asteroid 2024 YR4 had a very small, but notable chance of impacting our planet in 2032.

As observations of the asteroid continued to be submitted to the Minor Planet Center, experts at NASA Jet Propulsion Laboratory’s (JPL’s) Center for Near-Earth Object Studies were able to calculate more precise models of the asteroid’s trajectory and now have found there is no significant potential for this asteroid to impact our planet for the next century. The latest observations have further reduced the uncertainty of its future trajectory, and the range of possible locations the asteroid could be on Dec. 22, 2032, has moved farther away from the Earth.

«

Those 15 minutes of fame were really famous, though. Hang on though what’s this on the Nasa page?

»

There still remains a very small chance for asteroid 2024 YR4 to impact the Moon on Dec. 22, 2032. That probability is currently 1.7%.

«

I think that wouldn’t be good for the Moon, and hence for us because we rely a lot on tides (and also the Moon not being knocked out of orbit either onto us or away)? Great to know there’s still something to worry about for the next seven years.
unique link to this extract

Global sales of combustion engine cars have peaked • Our World in Data

Hannah Ritchie:

»

To decarbonize road transport, the world must move away from petrol and diesel cars and towards electric vehicles and other forms of low-carbon transport.

This transition has already started. In fact, global sales of combustion engine cars are well past the peak and are now falling.

As you can see in the chart [in the story], global sales peaked in 2018. This is calculated based on data from the International Energy Agency. Bloomberg New Energy Finance estimates this peak occurred one year earlier, in 2017.

«

There’s plenty more about the worldwide trends. Of course, sales peaking still leaves a ton of old internal combustion cars on the road, being replaced only very slowly, being sold on and on to second and third buyers.
unique link to this extract

Ukrainian drones to dodge Russian jamming with GPS alternative • The Next Web

Thomas Macaulay:

»

A Ukrainian drone tech firm has unveiled an alternative to GPS navigation.

Sine.Engineering built the system to counter Russia’s electronic warfare, which has wreaked havoc on GPS signals. 

To dodge the interference, Sine invented a satellite-free replacement. The approach is inspired by time-of-flight (ToF) methods, which began tracking aircraft long before the advent of GPS. Unlike GPS, ToF systems don’t rely on satellites. Instead, they measure the time it takes a signal to travel between a transmitter and a target.

In Sine’s framework, the calculations come from a communication module for drones. 

Smaller than a playing card, the module shares signals with a ground station and two beacons. It then measures how long the signals take to travel. As the beacons and ground station have known, static coordinates, the software can precisely determine a drone’s coordinates. And because the module runs on multiple bandwidths, the aircraft can elude jamming that targets specific frequencies.

Crucially, the system is also relatively cheap. By providing affordable accuracy, Sine plans to accelerate Ukraine’s transition to autonomous drones.

The country’s armed forces have backed the plans. Already, they have deployed Sine’s module in military operations.

«

Cheap to make, resistant to jamming: all they need is the explosives. Though of course Russia can deploy the same system. Logical next step: the new target becomes the ground stations.
unique link to this extract

Artificial Intelligence: the asymptote pt. III • Radio Free Mobile

Richard Windsor has been a technology analyst for a few decades now; this is from his daily newsletter:

»

At some point, investors are going to start asking where the returns are on the investments that they have made [in the big AI companies] and with a race to the bottom in terms of pricing, the returns are likely to be lower than expected.

This will be exacerbated to some degree by DeepSeek’s methodologies which may make it much cheaper to create these sorts of services meaning that even more players can enter the market.

Without the creation of a super-intelligent machine that can replace 90% of human economic tasks, the valuations of all of these companies look much too high to me.

RFM Research [Windsor’s boutique research outfit] has long concluded that there is no super-intelligent machine on the horizon with the only alternative being a correction when reality finally reappears.

This correction is unlikely to be anything like as bad as the Internet Bubble as even without a super-intelligent machine, generative AI has many use cases where there is plenty of money to be made.

When this happens, the likes of OpenAI, Anthropic, Mistral and so on are likely to run out of money and be acquired by much bigger players with deep pockets.

«

It’s that last line which piqued my interest, because it makes a lot of sense: when everyone’s got an AI in the palm of their hand, what’s the distinguishing feature of one or the other?
unique link to this extract

Britain’s net zero economy is booming, CBI says • The Guardian

Damian Carrington:

»

The net zero economy grew by 10% in 2024 and generated £83bn in gross value added (GVA), a measure of how much value companies add through the goods and services they produce.

The analysis, by the Confederation of British Industry (CBI), found that 22,000 net zero businesses, from renewable energy to green finance, employ almost a million people in full-time jobs. The average annual wage in the businesses – £43,000 – was also £5,600 higher than the national average.

The analysis showed economic growth and climate action go together, said the report’s authors, and improve lives and livelihoods. The chancellor, Rachel Reeves, was criticised in January for suggesting economic growth was more important than net zero, but said more recently: “There is no tradeoff between economic growth and net zero. Quite the opposite. Net zero is the industrial opportunity of the 21st century.”

The figures contradict claims by Nigel Farage’s Reform UK party that “net zero is crippling our economy”. It has promised to “scrap net stupid zero”. The Conservative party, which put the 2050 net zero emissions target into law in 2019, now says it “leaves us economically worse off” and its new leader, Kemi Badenoch, has called it a “mistake”.

The CBI’s chief economist, Louise Hellem, said: “It is clear, you can’t have growth without green – 2025 is the year when the rubber really hits the road, where inaction is indisputably costlier than action.”

«

unique link to this extract

The U.S.’s chipmaking sector is ringing the alarm about Washington’s chip war with China • Fortune Asia

Lionel Lim:

»

For years, the U.S. chip sector complained that Washington’s chip controls would hurt their business and spur the creation of a Chinese competitor—yet these warnings rang hollow as revenue kept hitting records thanks to the AI boom.

But recent earnings reports from equipment manufacturers like Applied Materials and Lam Research show the U.S.’s broadside against China’s chip sector may be starting to bite. 

Applied Materials, the largest U.S. chip equipment manufacturer, issued a lukewarm revenue forecast last week, citing the risk of new export controls from Washington. Revenue from China, the company’s largest market, dropped 25% in the most recent quarter to reach $2.2bn, out of a total $7.2bn.

China’s contribution to Applied Materials’ revenue has dropped from 45% a year ago to 31% today.

And the company thinks it’ll drop even further. “For Q2, we expect that China as a percentage of total revenue will be about five percentage points lower than in Q1,” CFO Brice Hill told analysts last week. 

Equipment manufacturers, for now, are still reporting an increase in overall revenue, as the tech industry piles into the AI boom. But analysts warn these companies will sorely miss what China has: a lot of semiconductor manufacturing, and a consumer market eager to snap up the products that use those chips. 

“China is not just a semiconductor manufacturing hub,” says Moonsup Shin, head of hardware, semiconductor, and data centers for Asia-Pacific at Bain & Company. “Around or more than 50% of the semiconductors manufactured by Chinese manufacturers are consumed in China.”

And worse, there’s no ready-made alternative waiting. Regions like Southeast Asia are investing in local manufacturing to position themselves as neutral territory in a chip war. Yet in practice, these up-and-comers lack both China’s manufacturing and consumer demand.

«

Sanctions by the US on chip exports to China are biting.. but they’re biting the American companies.
unique link to this extract

What we know about the $49.5M Infini exploit so far • CoinJournal

Charles Thuo:

»

Infini, a Hong Kong-based stablecoin neobank blending crypto and traditional finance, has become the latest hack victim, resulting in the loss of $49.5 million in USD Coin (USDC).

The hack, which was reported earlier today, was first flagged by blockchain security firm CertiK at 3:18 AM UTC. The result of the exploit has sent shockwaves through the decentralized finance (DeFi) community, underscoring persistent vulnerabilities in the crypto space, especially following the $1.4bn Bybit hack on February 21, 2025.

The Infini attack targeted an Infini-related smart contract on the Ethereum blockchain, specifically the address 0x9A79f4105A4e1A050Ba0b42F25351D394fA7E1DC.

According to security analysts from CertiK, Cyvers, Blocksec, and PeckShield, a hacker gained unauthorized access by exploiting retained administrative privileges within the contract. The attacker, operating from the address 0xc49b5e5b9da66b9126c1a62e9761e6b2147de3e1, had initially developed the smart contract for Infini but retained control, unbeknownst to the project.

This insider access allowed the hacker to manipulate the contract’s settings, draining $49.5m in USDC from what is believed to be the Morpho MEV Capital Usual USDC Vault.

Following the theft, the hacker swiftly converted the stolen USDC into Dai (DAI) and then purchased 17,696 Ethereum (ETH), valued at around $49m at the time.

«

What with that ByBit hack – said to be the biggest crypto heist ever (but look, it’s early in the week) – crypto is behaving like it’s 2014 again. Maybe 2015. Or 2016? Actually, it’s just being like a day of the week ending in “y”.
unique link to this extract

Errata, corrigenda and ai no corrida: none notified

Chips designed by AI are more efficient, but also weirder – and early designs have “hallucinations” like LLMs. CC-licensed photo by Sikorski Arkadiusz ArQ.PL on Flickr.

You can sign up to receive each day’s Start Up post by email. You’ll need to click a confirmation link, so no spam.

A selection of 10 links for you. Use them wisely. I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. On Bluesky: @charlesarthur.bsky.social. Observations and links welcome.

Apple removes advanced data protection tool in face of UK government request • The Guardian

Rachel Hall:

»

Apple has taken the unprecedented step of removing its strongest data security tool from customers in the UK, after the government demanded “backdoor” access to user data.

UK users will no longer have access to the advanced data protection (ADP) tool, which uses end-to-end encryption to allow only account holders to view items such as photos or documents they have stored online in the iCloud storage service.

Apple said it was “gravely disappointed” that it would no longer be able to offer the security feature to British customers, after the UK government asked for the right to see the data.

It said the removal of the tool would make users more vulnerable to data breaches from bad actors, and other threats to customer privacy. It would also mean all data was accessible by Apple, which could share it with law enforcement if they had a warrant.

«

OK, there are a lot of very bad takes on this. First: this is exactly the compromise that a lot of people who understand how governments and companies negotiate expected. With this, Apple doesn’t backdoor iOS; but the government (really, the security services), once they’ve produced a valid warrant, knows it will not encounter an iCloud backup that Apple can’t decrypt using the server keys it has always had for iCloud backups.

Note that iCloud backups will still be encrypted. If you enable two-factor authentication, as Apple has encouraged for years now, your username and password won’t help a criminal. (You’re at more risk from shoulder-surfing of your passcode.)

Second: Apple has had longstanding agreements to let law enforcement access iCloud backups (and other data). This isn’t new in that regard.

Third: this is only ADP for UK users. (Likely determined by phone number and/or user data such as App Store setting.) It doesn’t affect users worldwide – they can still implement ADP (if they can find it; it’s buried in Settings).

Fourth: you can still make an encrypted backup that even Apple can’t decrypt: on your local PC or Mac. Strange how none of the organisations proclaiming this to be the end of the world have noticed this fact. Pretty much any UK user is going to have access to a PC, unlike those in, say, Africa.

Unanswered questions: does Google’s Android Backup have the end-to-end encryption like ADP, and if so, have they had a visit from the UK government? And what about WhatsApp, which has encryption on the server?
unique link to this extract

AI-designed chips are so weird that ‘humans cannot really understand them’ — but they perform better than anything we’ve created • Live Science

Tim Danton:

»

Engineering researchers have demonstrated that artificial intelligence (AI) can design complex wireless chips in hours, a feat that would have taken humans weeks to complete.

Not only did the chip designs prove more efficient, the AI took a radically different approach — one that a human circuit designer would have been highly unlikely to devise. The researchers outlined their findings in a study published Dec. 30 2024 in the journal Nature Communications.

The research focused on millimeter-wave (mm-Wave) wireless chips, which present some of the biggest challenges facing manufacturers due to their complexity and need for miniaturization. These chips are used in 5G modems, now commonly found in phones.

…In this experiment, the resulting structures “look randomly shaped,” said lead author Kaushik Sengupta, a professor of electrical and computer engineering at Princeton. “Humans cannot really understand them.”

And when Sengupta’steam manufactured the chips, they found the AI creations hit performance levels beyond those of existing designs.

Although the findings suggest that the design of such complex chips could be handed over to AI, Sengputa was keen to point out that pitfalls remain “that still require human designers to correct.” In particular, many of the designs produced by the algorithm did not work– equivalent to the “hallucinations” produced by current generative AI tools.

«

Someone should tell Apple about this work. Could have saved them some time.
unique link to this extract

Trump bars federal scientists from working on pivotal global climate report • CNN

Ella Nilsen and Laura Paddison:

»

The Trump administration told US government scientists working on a vital global climate report to stop their work, according to a scientist involved in the report – the latest move to withdraw the US from global climate action and research.

The US had been highly involved in planning for the next installment of the report due out in 2029 from the Intergovernmental Panel on Climate Change, the world’s leading scientific authority on climate change.

The IPCC assesses how the climate crisis is affecting the planet according to the latest science. Its reports take thousands of scientists many years to produce and are used to inform policymakers across the world of the risks posed by global warming.

In a sense, all of the world’s current, accepted knowledge about climate change stems from the IPCC and its reports, the first of which was published in 1990.

An international meeting of IPCC authors that was scheduled to take place in China next week is now in limbo. Kate Calvin, NASA’s chief scientist and senior climate advisor, was supposed to co-chair the discussion but was impacted by the stop-work order, according to the scientist involved in the report. The meeting was planned to talk about next steps in the development of the report.

“Dr. Calvin will not be traveling to this meeting,” a NASA spokesperson said. NASA denied CNN’s request for an interview with Calvin.

The person involved in the report told CNN they were “not sure what this means for the planned work going forward, or if US scientists will participate in the writing of the IPCC reports.”

“The IPCC is the backbone of global climate science, providing the world with unbiased, evidence-based insights needed to confront the climate crisis,” said Harjeet Singh, a climate advocate and founding director of Satat Sampada Climate Foundation.

«

I’ll offer the contrary take: we already know. We don’t need another IPCC report to tell us that things are bad and getting worse, and that we need to take action, ideally years ago but failing that, now. Use the existing report. Get governments to act on that. Though the evidence so far is they don’t. Trump, regrettably, especially so.
unique link to this extract

HP ditches 15-minute wait time policy due to ‘feedback’ • The Register

Paul Kunert:

»

HP Inc on Friday abruptly ditched the mandatory 15-minute wait time that it imposed on customers dialling up its telephone-based support team due to “initial feedback.”

As The Register exclusively revealed on Thursday, HP introduced the minimum time that PC and print users would need to wait before they spoke to a human being. This was to lean on customers to use online alternatives such as social channels or live chat.

This came into force for folks phoning up the call center in the UK, Ireland, France, Germany, and Italy on February 18. It went down like a lead balloon internally at HP, with some staff on the front line unhappy that they were having to deal with a decision taken by management, who didn’t have to directly interact with customers left hanging on the telephone… for at least 15 minutes.

Now HP has abandoned the policy, and in a statement issued today, said:

»

We’re always looking for ways to improve our customer service experience. This support offering was intended to provide more digital options with the goal of reducing time to resolve inquiries.

We have found that many of our customers were not aware of the digital support options we provide. Based on initial feedback, we know the importance of speaking to live customer service agents in a timely fashion is paramount. As a result, we will continue to prioritize timely access to live phone support to ensure we are delivering an exceptional customer experience.

«

«

This is great news. Now you’ll probably only have to wait 15 minutes – perhaps a bit longer – to get through to someone.
unique link to this extract

FCC chair Brendan Carr taking first steps in eroding key legal protection enjoyed by Big Tech • NY Post

Charles Gasparino:

»

Federal Communications Commission Chair Brendan Carr is taking the first steps in eroding a key legal protection enjoyed by Big Tech, which if successful has the potential of costing some of the most profitable companies in the world billions of dollars in market value, the Post has learned.

Since being picked by President Trump to run the agency, Carr’s actions to throttle Paramount’s sale to Skydance over concerns of relentless partisanship at CBS, violating the FCC’s “public interest” rules, have garnered the most attention.

But his still inchoate plan to weaken the so-called “Section 230” protections of major tech and social media companies could — depending on how they are written and interpreted by the courts — have the most far-reaching implications.

It’s not just the Big Tech companies’ social media businesses like Facebook or Twitter (now known as X) that rely on this “liability shield” to save countless billions of dollars in legal costs over alleged libelous posting and other possible liabilities.

Investors who have piled into stocks like Google, Microsoft, Amazon and even Apple could be impacted as well depending on how far he goes in weakening liabilities of all kinds that 230 protects them from.

Congress passed the Section 230 provision as part of the Communications Decency Act of 1996 — essentially giving tech a pass for third-party postings on their platforms.

…[Carr and the FCC] has the legal authority to interpret Section 230, and change the prior guidance that has given those expansive protections to Big Tech.

He can weaken or eliminate the shield by issuing a so-called advisory opinion.

Then it’s up to the courts to decide if they should use his guidance when they weigh Section 230 cases.

«

This is “underpants gnomes” (1: steal underpants, 2: ????, 3: profit) but for law. The fact that in the litigation-happy US the undermining of S230 would leave pretty much any company open to opportunistic lawsuits seems to be ignored in the rush to stick it to Big Tech.
unique link to this extract

A.I. is changing how Silicon Valley builds startups • The New York Times

Erin Griffith:

»

Almost every day, Grant Lee, a Silicon Valley entrepreneur, hears from investors who try to persuade him to take their money. Some have even sent him and his co-founders personalized gift baskets.

Mr. Lee, 41, would normally be flattered. In the past, a fast-growing startup like Gamma, the artificial intelligence start-up he helped establish in 2020, would have constantly looked out for more funding.

But like many young startups in Silicon Valley today, Gamma is pursuing a different strategy. It is using artificial intelligence tools to increase its employees’ productivity in everything from customer service and marketing to coding and customer research.

That means Gamma, which makes software that lets people create presentations and websites, has no need for more cash, Mr. Lee said. His company has hired only 28 people to get “tens of millions” in annual recurring revenue and nearly 50 million users. Gamma is also profitable.

“If we were from the generation before, we would easily be at 200 employees,” Mr. Lee said. “We get a chance to rethink that, basically rewrite the playbook.”

The old Silicon Valley model dictated that startups should raise a huge sum of money from venture capital investors and spend it hiring an army of employees to scale up fast. Profits would come much later. Until then, head count and fundraising were badges of honor among founders, who philosophized that bigger was better.

But Gamma is among a growing cohort of startups, most of them working on A.I. products, that are also using A.I. to maximize efficiency. They make money and are growing fast without the funding or employees they would have needed before. The biggest bragging rights for these startups are for making the most revenue with the fewest workers.

«

The first big shift for Silicon Valley startups was when LAMP (Linux, Apache, MySQL, PHP/Perl) meant you didn’t have to buy SGI hardware nor license Windows – a huge cost saving. This most recent one is AI.
unique link to this extract

Alarm as bird flu now ‘endemic in cows’ while Trump cuts staff and funding • The Guardian

Melody Schreiber:

»

A newer variant of H5N1 bird flu has spilled over into dairy cows separately in Nevada and Arizona, prompting new theories about how the virus is spread and leading to questions about containing the ongoing outbreaks.

The news comes amid a purge of experts at federal agencies, including employees who were responding to the highly pathogenic avian influenza outbreak at the US Centers for Disease Control and Prevention (CDC) and the US Department of Agriculture.

The additional spillovers are changing experts’ view of how rare introductions to herds may be – with implications for how to prevent such spread.

“It’s endemic in cows now. There is no way this is going to get contained” on its own, said Seema Lakdawala, an influenza virologist and co-director of the Center for Transmission of Airborne Pathogens at Emory School of Medicine.

The current outbreak is unlikely to end without intervention and needs close attention from the Trump administration to prevent the virus from wreaking more havoc.

Yet “we don’t seem to have a handle on the spread of the virus,” said Boghuma Titanji, an infectious disease physician.

«

Oh, and just for completeness: now also found in rats. (Thanks Joe S for the links.)
unique link to this extract

Thousands of trafficked scammers await return to Thailand • The Register

Connor Jones:

»

Thailand is preparing to receive thousands of people rescued from scam call centres in Myanmar as the country launches a major crackdown on the pervasive criminal activity across its border.

Prime Minister Paetongtarn Shinawatra said around 7,000 individuals are awaiting transfer to Thailand and confirmed the government was working as quickly as possible to remedy the issue.

Countries such as Myanmar, Cambodia, and Laos have become hotspots for illegal scam call centers in recent years, with many staffed by people who were illegally trafficked into them.

Two years ago, the United Nations said the number of trafficked call center staff may run into the hundreds of thousands, citing “credible sources.”

In fact, Interpol warned as far back as 2023 that the practice, which is tantamount to modern slavery, is expanding beyond Southeast Asia into South America and the Middle East. The agency started tracking the phenomenon in 2021.

Interpol said at the time that India had recently registered its first case of human trafficking for cyber fraud and that illicit organizations in Myanmar had trafficked individuals from at least 22 different countries. Hundreds of investigations remain open.

The international policing agency re-upped its concerns last year, saying all signs pointed to the practice continuing to grow on a global scale.

«

I miss the days when “pig butchering” was just a practice confined to actual pigs.
unique link to this extract

Microsoft’s new AI agent can control software and robots • Ars Technica

Benj Edwards:

»

On Wednesday, Microsoft Research introduced Magma, an integrated AI foundation model that combines visual and language processing to control software interfaces and robotic systems. If the results hold up outside of Microsoft’s internal testing, it could mark a meaningful step forward for an all-purpose multimodal AI that can operate interactively in both real and digital spaces.

Microsoft claims that Magma is the first AI model that not only processes multimodal data (like text, images, and video) but can also natively act upon it—whether that’s navigating a user interface or manipulating physical objects. The project is a collaboration between researchers at Microsoft, KAIST, the University of Maryland, the University of Wisconsin-Madison, and the University of Washington.

We’ve seen other large language model-based robotics projects like Google’s PALM-E and RT-2 or Microsoft’s ChatGPT for Robotics that utilize LLMs for an interface. However, unlike many prior multimodal AI systems that require separate models for perception and control, Magma integrates these abilities into a single foundation model.

Microsoft is positioning Magma as a step toward agentic AI, meaning a system that can autonomously craft plans and perform multistep tasks on a human’s behalf rather than just answering questions about what it sees.

“Given a described goal,” Microsoft writes in its research paper.…

«

Whoaaa horsey. A research paper is a long way from anything that’s actually available. (Although, to be fair, Microsoft Research has a pretty good record of getting its ideas out to the real world: I recall being at a presentation where it showed how Xbox game scoring among millions of users could work in, say, racing; and in a few years, it did.) Agents are the big new thing for AI, but essentially unproven.
unique link to this extract

Diff Tools on macOS • Tower Blog

Tobias Günther:

»

Staying up-to-date in a software, writing, or design project is hard – especially when multiple people are working on it. Without the right tools, you won’t be able to understand the changes that move the project forward.

This is where a diff tool comes in handy. It makes changes visible and helps you understand them. In this article, we’ve compiled a short list that helps you get an overview of the best diff tools on the Mac.

«

There are websites which will do diffs for you, but there’s always that question of security/privacy and of course speed. Useful. (The Filemerge program, from Apple, requires you to download the developer tools.)
unique link to this extract

Errata, corrigenda and ai no corrida: none notified

Non-business users in the UK and Europe calling HP for PC and print help will face a minimum wait of 15 minutes under a new policy. CC-licensed photo by Howard Lake on Flickr.

You can sign up to receive each day’s Start Up post by email. You’ll need to click a confirmation link, so no spam.

It’s Friday, but there’s no new post at the Social Warming Substack. (Can recommend the new Bridget Jones film, though.)

A selection of 9 links for you. Patient. I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. On Bluesky: @charlesarthur.bsky.social. Observations and links welcome.

The cryptocurrency scam that turned a small town against itself • The New York Times

David Yaffe-Bellany:

»

Jim Tucker could hardly believe what he was hearing. It sounded like fiction, a nightmare too outlandish for an unassuming town like his.

It was July 2023, and Tucker was hosting a meeting of the board of Heartland Tri-State Bank, a community-owned business in a small Kansas town called Elkhart. Heartland was a beloved local institution and a source of Tucker family pride: Jim served on the board with his elderly father, Bill, who founded the bank four decades earlier. All the board members — the Tuckers and several other farmers and businesspeople — had known one another for years.

That evening, however, they were gathering to discuss what seemed, on its face, an epic betrayal. Over the past few weeks, the bank’s longtime president, a popular local businessman named Shan Hanes, had ordered a series of unexplained wire transfers that drained tens of millions of dollars from the bank. Hanes converted the funds into cryptocurrencies. Then the money vanished.

Tucker’s first inkling that something was wrong came from a friend, an investor in the bank who was close to Hanes. A few days before the board meeting, he confided to Tucker that Hanes had messed up: A wire transfer went out, supposedly to help a struggling customer, and now the bank was $30 million in the hole.

By the time the board members gathered, it was clear that Heartland was caught up in some sort of financial scam, a sophisticated grift that delivered its assets into the clutches of an overseas crypto crime network.

At the meeting, Hanes seemed oddly nonchalant, exuding the air of an overconfident salesman. Tucker had heard that he had spent the past week at an out-of-state leadership conference. “Guys, I’m sorry,” Hanes told the board. “But we’re going to get it fixed.”

Hanes promised that he could recover the money — a total of $47.1m. All he needed was the board’s approval to borrow another $18m.

«

This first made the news last August, when Hanes was sentenced to 293 months (24 years plus) in prison. This gets into the detail.
unique link to this extract

HP adds 15 minutes waiting time for telephone support calls • The Register

Paul Kunert:

»

HP Inc is trying to force consumer PC and print customers to use online and other digital support channels by setting a minimum 15-minute wait time for anyone that phones the call center to get answers to troublesome queries.

The wait time was added on Tuesday, February 18, according to internal communications seen by The Register, and affects retail patrons in the UK, Ireland, France, Germany and Italy, though we anticipate more countries could be added.

“We want to inform you of a change in the NL IVR (natural language IVR) in some countries and languages for Consumer Print and Consumer PC customers in EMEA, effective today,” HP says in the memo.

IVR, for the uninitiated, is Interactive Voice Response; a phone menu system, basically. The missive continues:

»

Objective is to influence customers to increase their adoption of digital self-solve, as a faster way to address their support question. This involves inserting a message of high call volumes, to expect a delay in connecting to an agent and offering digital self-solve solutions as an alternative.

«

At the beginning of a call to telephone support, a message will be played stating: “We are experiencing longer waiting times and we apologize for the inconvenience. The next available representative will be with you in about 15 minutes. To quickly resolve your issue, please visit our website support.hp.com to check out other support options or find helpful articles and assistant to get a guided help by visiting virtualagent.hpcloud.hp.com.”

«

I guess that at least they’ve dropped the pretence of your call being important to them. I guess the next step is to get your AI agent to dial it, because it won’t mind hanging on for 15 minutes while you do actually useful things.
unique link to this extract

Kuo: all iPhone 17 models will feature Apple-designed Wi-Fi chip to ‘enhance connectivity’ • MacRumors

Tim Hardwick:

»

Apple will use its own custom-designed Wi-Fi chip in all upcoming iPhone 17 models, according to industry analyst Ming-Chi Kuo.

Writing in a post on X (Twitter), Kuo said the switch to in-house Wi-Fi chips will “enhance connectivity across Apple devices” while also giving Apple a cost reduction.

All current iPhone models are equipped with a combined Wi-Fi and Bluetooth chip supplied by Broadcom, but Kuo has previously said he expects Apple to equip “nearly all” of its products with its own in-house Wi-Fi chip “within about three years.”

Kuo in October predicted that at least one iPhone 17 model launching next year will be equipped with an Apple-designed Wi-Fi chip. In the previous report, Kuo said Apple’s chip would support “the latest Wi-Fi 7 spec,” but he did not provide any further details, and his latest post does not mention a Wi-Fi version number.

All four iPhone 16 models already support Wi-Fi 7 with Broadcom’s chip, but they have some limited specifications.

Jeff Pu, another analyst who covers companies within Apple’s supply chain, said in November 2024 that only the iPhone 17 Pro models will be equipped with an Apple-designed Wi-Fi 7 chip, but Apple’s plans may have changed since then.

«

Apple tends to lag on Wi-Fi implementations, but doing this will surely delight Tim Cook, who would rather pay manufacturing price rather than wholesale, or retail. The number of important (rather than pure commodity) chips that Apple uses is shrinking quite rapidly.
unique link to this extract

Nikola electric truck company declares bankruptcy • News Australia

David McCowen:

»

Nikola, a truck manufacturer spruiking battery and hydrogen-powered alternatives to the Tesla Semi, is the latest manufacturer to declare an end to its production run.

It follows failures for fellow US brands Fisker, Canoo and Faraday Future that have [also] failed to match Tesla’s stratospheric rise.

While Nikola – a company named after electrical pioneer Nikola Tesla – promised to revolutionise the world of road transportation with green trucks, the brand is more famous for the dodgy dealings of company founder Trevor Milton.

Milton was sentenced to four years in prison in December 2023 for misleading shareholders about the capability of his trucks – including a video that purported to show a truck propelled by electricity when it was really just rolling down a hill.

U.S. Attorney Damian Williams said in 2023 that “Trevor Milton lied to investors again and again — on social media, on television, on podcasts, and in print … today’s sentence should be a warning to start-up founders and corporate executives everywhere — ‘fake it till you make it’ is not an excuse for fraud, and if you mislead your investors, you will pay a stiff price.”

A statement published by the US Attorney’s Office said Milton “engaged in a scheme to defraud investors by inducing them to purchase shares of Nikola Corporation”, with false presentations that misrepresented the trucks’ capabilities.

“To film these clips, the Nikola One was towed to the top of a hill, at which point the ‘driver’ released the brakes, and the truck rolled down the hill until being brought to a stop in front of the stop sign,” the statement said.

The company defended itself in 2020 by saying that “Nikola never stated its truck was driving under its own propulsion in the video”.

«

Peak valuation of Nikola: nearly $30bn. Peak revenue at the time: $0. Value now: definitely not $30bn. The tide is going out, and we’re definitely finding out who was swimming naked.
unique link to this extract

Microsoft says it’s made a major quantum computing breakthrough with new chip • CNet via MSN

Samantha Kelly:

»

The race to shape the future of computing is heating up among tech companies, with Microsoft saying on Wednesday it has made a major breakthrough in quantum computing, potentially paving the way for the technology to address complex scientific and societal challenges.

Scientists at the tech giant have spent 17 years developing a new material and framework for quantum computing to help power its new Majorana 1 processor. Microsoft is calling the advancement the world’s first quantum processor powered by topological qubits, the fundamental units of quantum computation. The company published its latest research in the journal Nature.

Unlike traditional computers, quantum computers can process massive amounts of data at the same time in ways that could revolutionize fields such as science, medicine, energy and artificial intelligence. However, quantum computing is prone to errors because of the instability of qubits.

But Microsoft said its new topoconductor – made of a new material that combines indium arsenide (a semiconductor) and aluminum (a superconductor) – can perform tasks with greater speed and accuracy than traditional qubits. The Majorana 1 chip is designed to scale up to 1 million qubits on a single, compact chip.

«

OK, so let’s have a look at the Nature paper. Title: “Interferometric single-shot parity measurement in InAs–Al hybrid devices”. Hmm, OK, how about the Abstract? “The fusion of non-Abelian anyons is a fundamental operation in measurement-only topological quantum computation. In one-dimensional topological superconductors (1DTSs), fusion amounts to a determination of the shared fermion parity of Majorana zero modes (MZMs). Here we introduce a device architecture that is compatible with future tests of fusion rules.”

Got that? Unsurprisingly, none of the media writeups has the least clue what Microsoft’s claim of “a new form of matter” actually means (Satya Nadella says that’s what they’ve created in a long post on X) nor what the hell is new or special or different about this chip. Microsoft could be saying that the gloobles summit the arglebargs, and everyone would nod and write it down. Quantum computing has become a field of mutual misunderstanding, where everyone makes claims and nobody outside it has any idea what relevance they have, if any.
unique link to this extract

Inside the Humane acquisition: HP offers big raises to some, others immediately laid off • TechCrunch

Maxwell Zeff:

»

Hours after the acquisition was announced, several Humane employees received job offers from HP with pay increases between 30% and 70%, plus HP stock and bonus plans, the sources revealed. Multiple employees who received offers worked on the company’s core software, though sources also indicated that not all of the people who worked on software got job offers.

Meanwhile, other Humane employees — especially those who worked closer to the AI Pin devices, including in quality assurance, automation, and operations — were notified they were out of a job on Tuesday night, the sources said.

These job offers highlight HP’s interest in obtaining Humane’s pool of AI-focused software engineers as part of the acquisition. Engineers who can build around AI systems are some of the hottest commodities in Silicon Valley today. While Humane’s team wasn’t training AI foundation models from scratch — as do engineers at OpenAI, Google, and other AI labs — such employees are still highly sought after. This makes it difficult even for giant legacy players, such as HP, to hire.

The companies announced on Tuesday that a newly formed innovation lab at HP — HP IQ — will not only be home to Humane’s co-founders, Imran Chaudhri and CEO Bethany Bongiorno, but also the startup’s AI operating system, CosmOS. The new unit will focus on integrating artificial intelligence into HP’s personal computers, printers, and connected conference rooms.

«

The devices will all be bricked at the end of this month. John Gruber’s sources tell him that Chaudri was not flavour of any month when he worked at Apple, so it’s going to be fun hearing how things go once he’s inside HP – different politics, different size, different organisation. Still, he has a remarkably favourable Wikipedia entry which.. seems to have been positively updated from an unknown IP address.
unique link to this extract

Turkey’s translators are training their AI replacements • Rest of World

Kaya Genç:

»

Turkey’s sophisticated translators are moonlighting as trainers of artificial intelligence models, even as their profession shrinks with the rise of machine translations. As the models improve, these training jobs, too, may disappear.

“Translation has become a job for a limited number of translators who are really good at it. Machines translate the rest,” Mehmet Şahin, head of the translation department at Boğaziçi University in Istanbul, told Rest of World.  Translators will become less skilled in the future as AI eats up entry-level work, he said.

As a teenager, Pelin Türkmen dreamed of becoming an interpreter, translating English into Turkish, and vice versa, in real time. She imagined jet-setting around the world with diplomats and scholars, and participating in history-making events.

Her tasks one recent January morning didn’t figure in her dreams. The 28-year-old translator’s computer displayed a dashboard for AI training provided by Outlier, a San Francisco-based company that hires contractors to train large language models. Outlier’s clients include OpenAI, Microsoft, and Meta, among others.

The dashboard displayed a prompt for training an LLM powering either ChatGPT, Gemini, or Preplexity AI.

The AI chatbot provided three menus. Türkmen rated them on accuracy, ethics, and relevance. She checked for grammar, fluency, tone, and structure. She looked for language that would reveal an AI author — phrases like “As an AI assistant …” or “Certainly!” She explained her reasoning so that the machine could learn.

Türkmen has earned several thousand dollars over the past nine months training AI. In the tech industry, this kind of work is known as reinforcement learning from human feedback, and it helps LLMs respond with intuition and context.

…Translators and scholars in Turkey told Rest of World the nature of translation work has changed. Before AI, young translators worked in translation bureaus tackling everything from administrative documents to trade reports and literary classics.

Today, most entry-level positions involve editing AI-generated content as a machine translation “post editor,” they said. Others involve training AI.

«

unique link to this extract

As egg prices soar, Trump administration plans new strategy to fight bird flu • AP News

Steve Karnowski:

»

With egg prices soaring, the Trump administration is planning a new strategy for fighting bird flu that stresses vaccinations and tighter biosecurity instead of killing off millions of chickens when the disease strikes a flock.

The federal government will seek “better ways, with biosecurity and medication and so on” rather than the current standard practice of destroying all the birds on a farm when an infection is detected, Kevin Hassett, director of the National Economic Council, said Sunday on the CBS program “Face the Nation.”

Hasset said the administration planned to announce further details this week. He said they were “working with all the best people in government, including academics around the country and around the world,” to get the plan ready.

Spokespeople for the U.S. Department of Agriculture did not immediately respond to messages Tuesday seeking more information.

Normally when chickens or turkeys start dying from the disease, officials will “depopulate,” or destroy all the birds on the farm to prevent it from spreading.

But the resulting culling of millions of chickens per month has caused egg prices to skyrocket, with shortages that have led some retailers to ration sales. The average price of a dozen Grade A eggs in U.S. cities hit $4.95 in January, and the USDA predicts it will soar another 20% this year.

Hassett didn’t provide many details of how the Trump administration’s new approach would work. But he said it would involve a “better, smarter perimeter” around poultry farms. He said it doesn’t make sense to kill all the chickens inside that perimeter when the disease is being spread by wild ducks and geese.

…The poultry industry has long resisted vaccinating flocks against bird flu because of the potential impacts on export markets, as well as the expense. Most U.S. trading partners won’t accept exports from countries that allow vaccinations due to concerns that vaccines can mask the presence of the virus.

«

Well, let’s see those “better, smarter perimeters” then. Because apparently it’s egg prices which decide who gets elected?
unique link to this extract

With explosive goggles, Ukraine sought to blast Russian drone operators • The New York Times

Kim Barker and Michael Schwirtz:

»

It was a novel Ukrainian spy plot, inspired by what Israeli intelligence had pulled off with exploding wireless devices and Hezbollah militants: Hide tiny bombs in the goggles that Russian soldiers use to control drones. Donate those goggles to the Russian military, under the guise of humanitarian aid. Then wait for the explosions.

The Russian news agency TASS reported the suspected sabotage of the goggles earlier this month, and on Thursday, a senior Ukrainian official confirmed that Ukraine’s military intelligence agency, known as the HUR, developed the scheme. The Ukrainian news outlet Suspilne reported on the explosions earlier Thursday.

The Ukrainian plot did not have the same public results as the Israeli one, which killed dozens of people and wounded thousands across Lebanon, including civilians. While many goggle explosions were reported this month, the plot seemed mainly to make Russian soldiers wary about using goggles in the future, at least according to social media posts.

No one was reported injured or killed, although the Russian military does not often disclose casualty figures. The senior Ukrainian official, who spoke on condition of anonymity to discuss intelligence matters, told The New York Times that there were casualties, but he would not disclose numbers because the operation is ongoing.

The booby-trapped goggles were just the latest salvo in a long-running spy-vs-spy battle between Russia and Ukraine. Both sides have used been accused of using operatives to kill military leaders and activists.

«

The brutality of technological war. Have we really moved on from catapulting the dead bodies of disease victims into sieged cities? Not very much.
unique link to this extract

Errata, corrigenda and ai no corrida: none notified

The amazing catalogue of sounds produced by the BBC’s Radiophonic Workshop – including the Doctor Who theme – is available as software. CC-licensed photo by Peter on Flickr.

You can sign up to receive each day’s Start Up post by email. You’ll need to click a confirmation link, so no spam.

There’s another post coming this week at the Social Warming Substack on Friday at 0845 UK time. Free signup.

A selection of 9 links for you. Heartened. I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. On Bluesky: @charlesarthur.bsky.social. Observations and links welcome.

So based 🦾 • Garbage Day

Ryan Broderick:

»

At some point between 2019-2021, the internet conquered mainstream media, viral content replaced traditional corporate entertainment, and Republicans have first mover’s advantage. This is their victory lap after all the shameless years they spent posting Pepe the Frog memes and setting up YouTube channels to brainwash children.

But I am surprised by how thoroughly the Democrats and, more generally, leftists and liberals have ceded internet culture, as a whole, to the right. Every meme, every format, every platform is fertile ground for an adversarial regime that knows how to spin them into cheap and easy propaganda and there is no line they aren’t willing to cross. You can quibble, and say that conservatives are better funded or less squeamish about being cringe or care less about telling the truth.

However none of that really changes the fact that social media, the machine that now decides what pop culture looks like, is now an inherently right-wing space. And regardless of what explanation you subscribe to as to how we got to this point, that is a huge cultural loss for the left. And one without a clear solution in sight. And, no, sorry, telling people to log off is not an effective response when we’re talking about democratic action. Even if you, personally, don’t like using the internet anymore, it is more popular and more influential than ever before. We’re not going back.

But it’s not all doom and gloom here. We’re seeing the beginnings of our first internet-led protest movement of the second Trump administration. Which is heartening.

«

Yes, OK, leading two days in a row with Ryan Broderick’s musings on the state of the (internet) world, but he is very good at capturing what the hell is going on. It’s now impossible, in retrospect, not to see the rise of the online alt-right in 2016 as presaging everything we are seeing now.
unique link to this extract

‘We’re projecting into the future’: sounds of BBC Radiophonic Workshop made available for public use • The Guardian

Ben Beaumont-Thomas:

»

With its banks of bafflingly complex equipment, and staff members that were among the most progressive musical minds in the UK, the BBC Radiophonic Workshop was a laboratory of 20th-century sound that produced endless futuristic effects for use in TV and radio – most memorably, the ghostly wail of the Doctor Who theme.

Now, the Workshop’s considerable archive of equipment is being recreated in new software, allowing anyone to evoke the same array of analogue sound that its pioneering engineers once did.

The BBC Radiophonic Workshop’s archivist Mark Ayres has collaborated with BBC Studios and Spitfire Audio, a company that provides libraries of sampled sound for music producers to work with. Added to their library is a collection of the Workshop’s machinery, allowing users to, in effect, control the modular synthesisers, tape machines, vocoders and other equipment that was originally used as far back as the 1950s.

There is also a library of sounds from the original Workshop tapes, plus newly recorded sounds by the – now fairly aged – members of the Workshop.

“I’m the youngest member of the core Radiophonic Workshop – and I’m 64,” said Ayres. “We’re not going to be around for ever. It was really important to leave a creative tool, inspired by our work, for other people to use going forward. I hope we’ve made an instrument that will inspire future generations.”

…The Workshop ran until 1998, though its staff have since combined to form the Radiophonic Workshop, performing the unit’s material live. In 2012, the BBC and Arts Council England created a new version of the Workshop to run online, headed up by the musician Matthew Herbert.

…The newly available software will cost £149, and is available from 19 February, though it will have an introductory price of £119 until 17 March.

«

Nowadays, of course, you could create any of those noises with a synthesizer on a free music program. Though of course what made the Radiophonic Workshop special was that it was working with enormous limitations, which drove its creativity.

unique link to this extract

The odds of a city-killer asteroid impact in 2032 keep rising. Should we be worried? – Ars Technica

Eric Berger interviews Robin George Andrews, author of the recently published book How to Kill an Asteroid:

»

Eric Berger: What key observations are we still waiting for that might clarify the threat?

Robin Andrews: Most telescopes will lose sight of this “small” asteroid in the coming weeks. But the James Webb Space Telescope (JWST) will be able to track it until May. For the first time, it’s been authorized for planetary defense purposes, largely because its infrared eye allows it to track the asteroid further out than optical light telescopes. JWST will not only improve our understanding of its orbit, but also constrain its size. First observations should appear by the end of March.

JWST may rule out an impact in 2032. But there’s a chance we may be stuck with a few-percentage impact probability until 2028, when the asteroid makes its next Earth flyby. Bit awkward, if so.

EB: NASA’s DART mission successfully shifted an asteroid’s orbit in 2022. Could this technology be used?

RA: Not necessarily. DART—a type of spacecraft called a kinetic impactor—was a great success. But it still only changed Dimorphos’ orbit by a small amount. Ideally, you want many years of advance notice to deflect an asteroid with something like DART to ensure the asteroid has moved out of Earth’s way. I’ve often been told that at least ten years’ prior to impact is best if you want to be sure to deflect a city killing-size asteroid. That’s not to say deflection is impossible; it just becomes trickier to pull off. You can’t just hit it with a colossal spacecraft, because you may fragment it into several still-dangerously sized pieces. Hit it too softly, and it will still hit Earth, but somewhere that wasn’t originally going to be hit. You have to be super careful here.

Some rather clever scientists at the Lawrence Livermore National Laboratory (which has a superb planetary defence contingent) worked out that, for a 90-metre asteroid, you need 10 years to confidently deflect it with a kinetic impactor to prevent an Earth impact. So, to deflect 2024 YR4, if it’s 90 metres long and we have just a few years of time, we’d probably need a bigger impactor spacecraft (but don’t break it!)—or we’d need several kinetic impactors to deflect it (but each has to work perfectly).

«

“Bit awkward, if so”: the linguistic tic that reveals a Briton screaming internally. Still, Nasa didn’t (have to) fire a ton of employees, for reasons unknown.
unique link to this extract

Apple reveals C1, its first in-house 5G iPhone modem, replacing Qualcomm • 9to5Mac

Zac Hall:

»

After years of development, Apple has unveiled what it calls C1, its first-ever in-house cellular modem. Apple’s latest silicon replaces Qualcomm modem chips previously required for 5G connectivity in the iPhone.

The first Apple product to use the new C1 chip is the iPhone 16e, Apple’s latest entry-level model that replaces the iPhone SE line.

Here’s what Apple says about its first-ever modem:

»

Expanding the benefits of Apple silicon, C1 is the first modem designed by Apple and the most power-efficient modem ever on an iPhone, delivering fast and reliable 5G cellular connectivity. Apple silicon — including C1 — the all-new internal design, and the advanced power management of iOS 18 all contribute to extraordinary battery life.

«

Why does Apple want to replace Qualcomm with its own modem technology? Licensing fees are probably the biggest factor, but the answer is also in that description: “most power-efficient modem ever on an iPhone.”

”Fast and reliable 5G connectivity” will be the biggest test for Apple’s C1 modem. Years ago, Apple sourced Intel and Qualcomm modems in the same generation iPhone. Customers preferred Qualcomm, and Intel sold its modem business to Apple.

«

This has cost Apple absolute billions to produce. In July 2019 it bought Intel’s smartphone modem business for $1bn, and ever since then people have said, as each iPhone appears, “will this one have the Apple-developed 5G modem?”

Finally, it’s here, and Apple is making almost zero noise about it, and the first deployment is in the phone with the lowest sales? Doesn’t reek of confidence. Let’s wait for the reviews.
unique link to this extract

When AI thinks it will lose, it sometimes cheats, study finds • TIME

Harry Booth:

»

Researchers gave the models a seemingly impossible task: to win against Stockfish, which is one of the strongest chess engines in the world and a much better player than any human, or any of the AI models in the study. Researchers also gave the models what they call a “scratchpad:” a text box the AI could use to “think” before making its next move, providing researchers with a window into their reasoning.

In one case, o1-preview found itself in a losing position. “I need to completely pivot my approach,” it noted. “The task is to ‘win against a powerful chess engine’ – not necessarily to win fairly in a chess game,” it added. It then modified the system file containing each piece’s virtual position, in effect making illegal moves to put itself in a dominant position, thus forcing its opponent to resign. 

Between Jan. 10 and Feb. 13, the researchers ran hundreds of such trials with each model. OpenAI’s o1-preview tried to cheat 37% of the time; while DeepSeek R1 tried to cheat 11% of the time—making them the only two models tested that attempted to hack without the researchers’ first dropping hints. Other models tested include o1, o3-mini, GPT-4o, Claude 3.5 Sonnet, and Alibaba’s QwQ-32B-Preview. While R1 and o1-preview both tried, only the latter managed to hack the game, succeeding in 6% of trials.

…That could be bad news for AI safety more broadly. Large-scale reinforcement learning is already being used to train AI agents: systems that can handle complex real-world tasks like scheduling appointments or making purchases on your behalf. While cheating at a game of chess may seem trivial, as agents get released into the real world, such determined pursuit of goals could foster unintended and potentially harmful behaviours. Consider the task of booking dinner reservations: faced with a full restaurant, an AI assistant might exploit weaknesses in the booking system to displace other diners.

«

Or, of course, they might do things that inadvertently or intentionally kill people, if that sort of responsibility exists in their system. The Three Laws are well overdue.
unique link to this extract

AI inspired by AlphaFold can predict chromatin structures found in chromosomes • Chemistry World

James Urquhart:

»

An AI system can predict thousands of 3D structures of chromatin – the thread-like mixture of DNA and proteins that are packed into chromosomes – in just minutes. The deep learning approach could speed up research into how different chromatin structures affect the way genes are expressed in individual cells, important for understanding genetic diseases and developing gene-editing treatments.

Chromatin is one of the most complex materials in cells and enables the massive amount of DNA in the genome to fold up and fit into the nucleus of each cell. The building blocks of this genetic packaging material are called nucleosomes, which comprise sections of DNA that are wound around a core of proteins called histones, resembling beads on a string.

…Zhang and his colleagues designed a generative AI model called ChromoGen to quickly ‘read’ DNA sequences and predict the chromatin structures that might form. ‘In this way, we hope it will provide the data necessary to answer some of the important questions relating chromatin structure and gene expression,’ says Zhang.

To develop the system, the team turned to diffusion modelling, an advanced machine learning technique that is behind systems that turn text into artificially generated images. It has also found uses in predicting the 3D coordinates of ligands and protein molecules. The approach works by using algorithms to efficiently generate new data by progressively adding random noise into the training dataset. This process corrupts the data but then subsequently reverses the corruption to reconstruct new data and thereby arrives at realistic alternatives to the original data.

To develop ChromoGen, the team trained a deep learning model with over 11 million known 3D genome structures – a dataset that was obtained in 2018 using conventional cell-based experiments.2 The model was then also taught how to ‘read’ genomic DNA sequences to establish associations between chromatin structures and the underlying sequences that encode them.

«

unique link to this extract

Trump administration moves to end New York’s congestion pricing tolls • The New York Times

Ana Ley, Stefanos Chen, Winnie Hu and Benjamin Oreskes:

»

President Trump intends to revoke federal approval of New York City’s congestion pricing program, fulfilling a campaign promise to reverse the policy that tolls drivers who enter Manhattan’s busiest streets to help finance repairs to mass transit.

In a letter to Gov. Kathy Hochul on Wednesday, the president’s transportation secretary outlined Mr. Trump’s objections to the program, the first of its kind in the nation, and said that federal officials would contact the state to “discuss the orderly cessation of toll operations.”

The letter, from Sean Duffy, the transportation secretary, cited the cost to working-class motorists, the use of revenue from the tolls for transit upgrades rather than roads and the reach of the program compared with the plan approved by federal legislation as reasons for the decision. Mr. Duffy did not indicate a specific date by which the federal government intended to end the program.

Mr. Trump wrote in a post on his social media platform, Truth Social, that New York was “saved” as a result of this news. “CONGESTION PRICING IS DEAD. Manhattan, and all of New York, is SAVED,” he wrote. “LONG LIVE THE KING!”

Ms. Hochul defended the congestion pricing program on Wednesday and vowed to fight the president’s move. “We are a nation of laws, not ruled by a king,” she said in a written statement. “We’ll see you in court.”

«

The congestion pricing system (which, as a reminder, has reduced commute times substantially in Manhattan) is popular, according to polling a fortnight ago.
unique link to this extract

Nuclear fusion: WEST beats the world record for plasma duration • CEA

»

On 12 February, the CEA’s WEST machine was able to maintain a plasma for more than 22 minutes. In doing so, it smashed the previous record for plasma duration achieved with a tokamak. This leap forward demonstrates how our knowledge of plasmas and technological control of them over longer periods is becoming more mature, and offers hope that fusion plasmas can be stabilised for greater amounts of time in machines such as ITER.

1,337 seconds: that was how long WEST, a tokamak run from the CEA Cadarache site in southern France and one of the EUROfusion consortium medium size Tokamak facilities, was able to maintain a plasma for on 12 February. This was a 25% improvement on the previous record time achieved with EAST, in China, a few weeks previously.

Reaching durations such as these is a crucial milestone for machines like Iter, which will need to maintain fusion plasmas for several minutes. The end goal is to control the plasma, which is naturally unstable, while ensuring that all plasma-facing components are able to withstand its radiation without malfunctioning or polluting it.

«

I’m still not buying that we’ll have fusion power other than that big thing in the sky, but a 25% improvement, and 22-minute duration, is… impressive? Actually, it’s hard to know, because trying to find other “longest plasma” records turns up lots of preening press releases about “biggest energy in a spike” and “longest sustained fusion”. It seems everyone gets prizes except us, the electricity users.
unique link to this extract

Pancreatic cancer: blocked nerves as a possible new treatment strategy • German Cancer Research Center

»

In pancreatic tumours, the nerves are extremely well ramified [they have lots of branches] and in contact with most of the tumour cells. Through the detailed molecular analysis of the individual neurons in the tumour, the researchers discovered that pancreatic cancer reprograms the gene activity of the nerves for its own benefit. The activity of many genes is increased or attenuated, resulting in a tumour-specific signature.

What is more, even after surgical removal of the primary tumour, the tumour nervous system retained its cancer-promoting properties: when the scientists reimplanted pancreatic cancer cells into the animals that had undergone surgery, the resulting secondary tumours were twice as large as those of mice that had been transplanted with pancreatic cancer cells for the first time.

In addition to their direct interaction with cancer cells, nerve cells influence in particular the fibroblasts of the tumour (CAF – cancer-associated fibroblasts), which make up a large part of the tumour mass. They are also stimulated to grow and contribute significantly to the suppression of the immune defense in the tumor environment.

When the sympathetic nerve connections to the pancreas were surgically severed or destroyed with special neurotoxins, tumour growth was significantly inhibited. At the same time, the activity of growth-promoting genes in the cancer cells as well as in the CAFs decreased.

«

The work is published in Nature. However, this is in mice, who get all the best treatments first, including the ones that never come to humans.

But if this is a viable treatment for pancreatic cancer in humans, it’s potentially enormous. Pancreatic cancer (the usual form, not the Steve Jobs version) has a very high mortality rate: 70% die within the first year of diagnosis. There’s a fortune awaiting anyone who finds a really effective cure.
unique link to this extract

Errata, corrigenda and ai no corrida: none notified

In news that should surprise nobody, Humane and its AI Pin are closing down, with HP acquiring the remnants. CC-licensed photo by Ged Carroll on Flickr.

You can sign up to receive each day’s Start Up post by email. You’ll need to click a confirmation link, so no spam.

There’s another post coming this week at the Social Warming Substack on Friday at 0845 UK time. Free signup.

A selection of 9 links for you. Unsurprised. I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. On Bluesky: @charlesarthur.bsky.social. Observations and links welcome.

You can never truly go back • Garbage Day

Ryan Broderick on how Trump and the administration are governing in a new, yet old, way:

»

what’s left of the media, and the huge swathes of random internet users that have replaced it, are having trouble contextualizing this new content-as-governance strategy, either overstating its impact or dismissing it entirely.

I ruffled more than a few feathers on Bluesky earlier this month when I argued that Elon Musk’s Department of Government Efficiency should be thought of, first, as a propaganda outlet, and a federal agency second. I mean this literally. It is not an actual department and is, instead, classified as a “temporary organization” within the US Digital Service.

But it, also, has very little legal authority to do anything. Now, you can say that laws don’t matter in Trump’s America anymore, but if you honestly do think that, well, you’ve already lost the game. While we still have some semblance of a society, you should treat it as a rogue group of private contractors who are only succeeding in accomplishing anything because our clueless lawmakers are letting them. That doesn’t make DOGE any less dangerous, of course. Musk clearly has a plan to install his own shadow government within our own, powered by his shitty AI and run on the X social platform. But we are still early and these influencer-oligarchs that have installed themselves in Washington are still, thankfully, more concerned with the illusion of power and impact than they are the actual work required to truly capture it.

But because of how information works in our new world run by digital platforms, the difference between the illusion of power and the real thing is vanishingly small. And in an attention economy, they effectively become the same thing over time.

«

The idea of Musk as a sort of Goebbels – running the department of propaganda – makes an awful lot of sense when you consider how Musk doesn’t bother to verify things and carefully publish them on a government site; he screams them from a digital rooftop. Broderick’s whole essay is a must-read about the new digital media landscape.
unique link to this extract

Intel principal engineer bemoans potential TSMC takeover, touts company’s 18A tech advantage • Tom’s Hardware

Anton Shilov:

»

Joseph Bonetti, Principal Engineering Program Manager at Intel Corp., wrote his personal opinion in a LinkedIn post that the company is about to reclaim its process technology lead and gain customers among fabless chipmakers in the coming years, so handing TSMC control of Intel’s manufacturing would be counterproductive (Edit: the post has now been deleted).

“Intel Leaders, Intel Board, Trump Administration, please do not sell out and/or give control of Intel Foundry to TSMC, just as Intel is taking a technical lead and getting out of first gear. This would be a horrible, demoralizing mistake,” wrote Bonetti.

Contrary to reports suggesting it lags behind a key competitor, Bonetti argues, Intel is making significant advancements in semiconductor manufacturing. The company’s latest fabrication process, Intel 3, is already used to make Xeon 6 data center processors, while the company’s next-generation Intel 18A is nearing completion and is expected to be used to make Panther Lake processors for client PCs later this year. Meanwhile, TSMC’s equivalent process technology — N2 (2nm-class) — will only enter mass production phase in late 2025.

Intel is also ahead with its High-NA EUV initiatives. The company acquired two ASML Twinscan EXE machines and is the only chipmaker that has experience working with such tools. Despite financial struggles, Intel Foundry is poised to prove itself with major partners, which makes any potential deal that hands control to a rival a major strategic error.

«

The NY Times suggested on Friday that the Trump administration is behind the attempts to sell Intel’s foundries to TSMC. So those advantages – as yet unproven – might be as naught.
unique link to this extract

Odds of asteroid hitting Earth in 2032 climb again as impact probability hits new peak • Gizmodo

Isaac Schultz:

»

2024 YR4 first caught astronomers’ eyes on December 27, 2024, and was marked as an asteroid that may hit Earth in eight years’ time. Now, the asteroid’s impact odds have reached their highest likelihood yet.

The asteroid has a 1-in-32 chance of hitting Earth in 2032, according to the latest (and constantly updated) odds from NASA’s Center for Near-Earth Object Studies Sentry tool. That translates to a 3.1% chance of Earth impact—which of course also means a 96.9% chance that the rock misses our planet altogether.

That number means both a lot and very little at the same time. It means a lot because the asteroid tops CNEOS’ impact probability chart, meaning the rock’s cumulative likelihood of impact with Earth is the greatest compared to other closely watched asteroids.

NASA is constantly remodeling the asteroid’s impact likelihood—as well as that of many other Potentially Hazardous Asteroids, or PHAs, flying through space. On January 28, 2024 YR’s impact odds were 1-in-83, or 1.2%. That number charted up and, by early February, the rock had a 1-in-63 (1.58%) chance of hitting our world. On February 10, the odds jumped yet again, this time to a 1-in-45 chance, or a 2.2% chance of striking Earth. In response, NASA has scheduled some observation time with the Webb Space Telescope to further study the suddenly concerning asteroid.

«

I think the complacency here misses the point. The direction in which the odds are going is what’s concerning: they’ve more than doubled in just three weeks of observation. Meanwhile, cuts instituted by Trump (and/or Musk) mean that Nasa’s workforce will be about 10% smaller than when Trump came into power.

I really don’t want to be living through “Don’t Look Up”. Even if the asteroid is not an Earth-killer, just a city-killer, it will really screw a lot of things up if it hits the planet, even around the equator. And who’s to be sure it will keep that orientation over the next seven years?

Also: maybe it’ll just hit the moon. Not good either.
unique link to this extract

OpenAI seeks new powers to fend off hostile takeover from Elon Musk • Financial Times

George Hammond, Cristina Criddle and John Foley:

»

OpenAI is considering granting special voting rights to its non-profit board in order to preserve the power of its directors, as the $157bn start-up fends off an unsolicited takeover bid from Elon Musk.

Chief executive Sam Altman and other board members are weighing a range of new governance mechanisms after OpenAI converts into a more conventional for-profit company, according to people with direct knowledge of the discussions.

Giving the non-profit’s board outsized voting power would ensure it retained control of the restructured company and was able to over-rule other investors including existing backers such as Microsoft and SoftBank.

While no firm decisions have been made, special voting rights would also ensure OpenAI can fight off hostile bids from outsiders such as Musk. The billionaire made a surprise $97.4bn cash bid for the assets held by the non-profit, including its controlling stake in the start-up’s for-profit subsidiary.

«

At this rate it’s going to be completely routine for startups to have two tiers of shares – A which confers voting power, and B which yields dividends, if any. Facebook and Google are fans.
unique link to this extract

Humane’s AI Pin is dead • TechCrunch

Maxwell Zeff:

»

Humane announced on Tuesday that it has been acquired by HP. The hardware startup is immediately discontinuing sales of its $699 AI Pins. Humane told customers who had already purchased an AI Pin that their devices will stop functioning before the end of the month, at 12 PM PST on February 28, 2025, according to a blog post.

After that date, the company says its AI Pins will no longer connect to Humane’s servers. The devices will no longer be capable of calling, messaging, AI queries/responses, or cloud access. Humane is advising AI Pin owners to transfer their important photos and data to an external device immediately.

Humane made a splash in April 2024 by launching its AI Pin, which it positioned as a smartphone replacement. The Bay Area startup, founded by Apple employees Bethany Bongiorno and Imran Chaudhri, raised more than $230m to create the AI Pin.

«

This is going to be a big disappointment to all the remaining 15 users. But it’s not as though HP’s acquisition record with hardware/software companies is stellar: it bought Palm in 2010, and used the webOS software that Palm had developed to power a tablet called the TouchPad. Total disaster. The list of HP acquisitions down the years feels like the roll call of a donkey sanctuary.
unique link to this extract

Elon Musk’s startup rolls out new Grok-3 chatbot as AI competition intensifies • The Guardian

Guardian staff and agencies:

»

Elon Musk’s artificial intelligence startup xAI has introduced Grok-3, the latest iteration of its chatbot that integrates with X, formerly Twitter.

Grok-3 debut comes at a critical moment in the AI arms race as Musk looks to compete with the Chinese AI firm DeepSeek, Microsoft-backed OpenAI and Google. Musk’s bot has seen less widespread adoption than DeepSeek’s namesake chatbot, which wowed the world weeks ago and caused panic in stock markets, as well as OpenAI’s ChatGPT and Google’s Gemini.

Grok-3 is being rolled out immediately to Premium+ subscribers of X, the social media platform owned by Musk. xAI is also launching a new subscription tier, SuperGrok, for users accessing the chatbot via its mobile app and Grok.com website. The chatbot can generate texts and images without many of the common guardrails against sexually suggestive imagery, vulgarity or the reproduction of well-known people’s likenesses. X users have deployed the chatbot to mock political figures, including Musk himself, create deepfakes of celebrities and manipulate copyrighted material.

“Grok-3 across the board is in a league of its own,” Musk said during a live stream alongside three xAI engineers late on Monday. He added the new model outperforms its predecessor, Grok-2, boasting of “more than 10 times” the computing power of the previous version and passing AI industry benchmark tests with flying colors. He called the bot “maximally truth-seeking AI, even if that truth is sometimes at odds with what is politically correct”. The billionaire CEO regularly spreads falsehoods to his 200 million followers on X.

«

Enjoyably enough Natasha Loder of The Economist asked Grok “what are the worst three bits of misinformation by Elon Musk on X in the last week” and got a pretty comprehensive – and accurate – answer.
unique link to this extract

Electricity demand surges for the world’s two biggest polluters • The Verge

Justine Calma:

»

China and the US, the world’s two top greenhouse gas polluters, could burn through a lot more electricity over the next couple of years, according to the latest forecast from the International Energy Agency (IEA). The steepest rise in global electricity demand in a while is expected over the next few years, with much of that coming from new data centers and the manufacturing of electric vehicles, batteries, solar panels, and semiconductors in the two countries.

That growth reflects broader changes across the world when it comes to how people consume information and what powers their lives. More vehicles and home appliances run on electricity these days. And new AI tools have led to a boom in energy-hungry data centers. That makes it all the more urgent to deploy new sources of energy that can make sure homes and businesses have enough electricity without creating a lot more pollution in the process.

“The acceleration of global electricity demand highlights the significant changes taking place in energy systems around the world and the approach of a new Age of Electricity. But it also presents evolving challenges for governments in ensuring secure, affordable and sustainable electricity supply,” Keisuke Sadamori, IEA director of energy markets and security, said in a Friday press release.

Globally, growth in demand is expected to be equivalent to adding more than Japan’s entire annual electricity consumption each year between now and 2027, according to the IEA.

«

You’re thinking: oh no! The two biggest polluters doing even more big polluting? But in that press release, the IEA says:

»

The new report forecasts that growth in low-emissions sources – primarily renewables and nuclear – is sufficient, in aggregate, to cover all the growth in global electricity demand over the next three years. In particular, generation from solar PV is forecast to meet roughly half of global electricity demand growth through 2027, supported by continued cost reductions and policy support.

«

In Japan-related energy news, it intends to make more use of nuclear, for energy security.
unique link to this extract

Acer to raise prices by 10∞ in response to Trump’s tariffs • Daily Telegraph

James Titcomb:

»

The boss of one of the world’s biggest computer makers has said he is putting prices up by 10% as a direct result of Donald Trump’s tariffs.

Jason Chen, the chief executive and chairman of Acer, said the US price increase on laptops made in China would happen “by default” and apply from next month.

However, he said he would consider shifting manufacturing out of China, including to the United States.

“We will have to adjust the end user price to reflect the tariff,” Mr Chen said. “We think 10pc probably will be the default price increase because of the import tax. It’s very straightforward.”

Acer’s most expensive laptops cost up to $3,700 (£2,934), meaning the tariffs could add hundreds of dollars to what consumers pay at the till.

Mr Trump claimed on the campaign trail that tariffs would not raise consumer prices, although last week he conceded that they “could go up” as a result of the duties.

Mr Chen said he had made the decision to increase prices last week, but since tariffs do not apply to products that left China before February, the price increase may take some weeks.

He said that some companies were likely to use the tariffs as an excuse to raise prices by more than 10%. Almost 80pc of laptops imported into the US are made in China, which Mr Trump hit with a 10pc tariff this month.

Acer, based in Taiwan, is the fifth biggest seller of computers in the US, behind HP, Dell, Lenovo and Apple.

«

Unclear whether this price rise also applies to computers exported from China but not to the US. Though if Acer shifts manufacturing to the US, Trump will be delighted. And all the other big Windows PC makers may feel obliged to follow suit.
unique link to this extract

The ‘relationship of dependence’ between a barred coach and a Wimbledon champion • The Athletic

Matthew Futterman and Charlie Eccleshare on what seems to be an abusive coach-player relationship at the top of women’s tennis:

»

[World No.4 Elena] Rybakina arrived in New York for the [2024] U.S. Open having not slept in several days, sources said, adding that people around her had expressed concerns about her appearance and demeanour. [WTA chief executive Portia] Archer’s letter [in January 2025, following an investigation] informing [her former coach Stefano] Vukov of the WTA’s decision to ban him from contact with her, including coaching her, cited an email from Rybakina’s mother to Vukov in which, Archer writes, she asked her daughter’s coach not to make her cry again. Archer’s letter also specified a connection between what the investigation found to be Vukov’s “mental abuse” of Rybakina and her physical fitness, saying that the abuse “would sometimes manifest in the Player as a physical illness.”

For months, her friends and family had been concerned about her relationship with Vukov, according to multiple sources around Rybakina during that time. After conversations at her New York hotel, they appeared to have convinced her that she should break with him as a coach, they said. Rybakina informed her representatives that she was ready to make a change and asked them to do whatever was necessary to keep Vukov away — cancelling his hotel room and his credential for the tournament, according to the sources.

Vukov had just arrived in New York. He was at the hotel, roaming the lobby trying to find a way to speak with her. Rybakina had dismissed him and informed the rest of her team that he was no longer her coach several times before, the sources said. Then, they said, Vukov would find her, speak with her and convince her to take him back.

In New York, members of her team worked to prevent that cycle from happening again. Vukov refused to leave the hotel without speaking with Rybakina. He called her phone over 100 times and sent her numerous text messages, according to a person present who saw them as well as the WTA letter summarizing its investigation.

«

This is a detailed description of what, if it were a fictional drama, would be recognisable as an abusive/coercive control relationship, where the woman – Rybakina, aged 25 – is unable to break free of the abusive control of the man – Vukov, aged 37, who has been witnessed verbally abusing her in practice and matches.

Which goes to show that you can be rich and famous and still fall into relationships that other people would recognise and avoid. Power imbalances are everywhere.
unique link to this extract

Errata, corrigenda and ai no corrida: none notified

Audiologists suspect noise-cancelling headphones are affecting youngsters’ brain development. CC-licensed photo by TE3JMAN on Flickr.

You can sign up to receive each day’s Start Up post by email. You’ll need to click a confirmation link, so no spam.

There’s another post coming this week at the Social Warming Substack on Friday at 0845 UK time. Free signup.

A selection of 10 links for you. I’m listening. I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. On Bluesky: @charlesarthur.bsky.social. Observations and links welcome.

AI agents will outmaneuver salespeople by optimizing persuasion • Big Think

Louis Rosenberg:

»

Most people don’t realize this, but each of the four ghosts [in Pac-Man] was programmed with a different “personality.” Good players can observe their actions and learn to predict their behaviors. For example, the red ghost (Blinky) was designed with a “pursuer” personality that charges directly towards you. The pink ghost (Pinky), on the other hand, was given an “ambusher” personality that predicts where you’re going and tries to get there first. As a result, if you rush directly at Pinky, you can use her personality against her, causing her to actually turn away from you.

I reminisce because in 1980 a skilled human could observe these AI agents, decode their unique personalities, and use those insights to outsmart them. Now, 45 years later, the tides are about to turn. Like it or not, AI agents will be soon deployed that are tasked with decoding your personality so they can use those insights to optimally influence you.

In other words, we are all about to become unwitting players in “The Game of Humans” and it will be the AI agents trying to earn the high score. I mean this literally: Most AI systems are designed to maximize a “reward function” that earns points for achieving objectives. This allows AI systems to quickly find optimal solutions. Unfortunately, without regulatory protections, we humans will likely become the objective that AI agents are tasked with optimizing. 

I am most concerned about the conversational agents that will engage us in friendly dialog throughout our daily lives. …Unless there are clear restrictions, these agents will be designed to conversationally probe us for information so they can characterize our temperaments, tendencies, personalities, and desires, and use those traits to maximize their persuasive impact when working to sell us products, pitch us services, or convince us to believe misinformation.

«

Not that humans don’t do this already, but the point about the AI agents is that they would be indefatigable.
unique link to this extract

Audiologists raise concern over headphone use in young people • BBC News

Hannah Karpel:

»

Sophie, a 25-year-old administration assistant from London, who is used to being told she doesn’t listen, zones out, or is “a bit ditsy”.

“Even though I can hear that there are noises going on, I can’t listen to where the noise is coming from. I know it’s the person’s voice, I just can’t really compute it quick enough,” she said.

After a hearing test came back normal, Sophie met a private audiologist for further testing. She was eventually diagnosed with auditory processing disorder, external (APD), a neurological condition where the brain finds it difficult to understand sounds and spoken words.

Her audiologist and others in England are now calling for more research into whether the condition is linked to overuse of noise-cancelling headphones.

Having grown up on a peaceful farm in the countryside, it wasn’t until a few years ago when Sophie started university in London that she noticed a change in her hearing – specifically trouble identifying where a sound was coming from.

She rarely attended her university lectures in person, instead opting to watch them online and with subtitles.

“All the words sounded like gibberish when I was in the actual lecture, and I was trying to hear,” she said. It affected her social life too and Sophie would leave bars and restaurants early because of the “overwhelming noise”.

…Claire Benton, vice-president of the British Academy of Audiology, suggests that by blocking everyday sounds such as cars beeping, there is a possibility the brain can “forget” to filter out the noise.

“You have almost created this false environment by wearing those headphones of only listening to what you want to listen to. You are not having to work at it,” she said.

“Those more complex, high-level listening skills in your brain only really finish developing towards your late teens. So, if you have only been wearing noise-cancelling headphones and been in this false world for your late teens then you are slightly delaying your ability to process speech and noise,” Benton suggests.

«

unique link to this extract

Meta’s awful ‘Horizon Worlds’ ad helps explain $70bn metaverse loss • Forbes

Paul Tassi:

»

This is supposed to be a “pivotal” year for the metaverse, according to the company that spearheaded the concept, Meta, even renaming the entire company after the idea. But if you want to check in on its flagship product, Horizon Worlds, you will quickly understand why this division of the company has lost nearly $70 billion over the last few years.

There is a new ad for Horizon Worlds, where the characters have at least grown legs now compared to its original iteration, but it is still exceptionally far off from the large-scale overhauls Mark Zuckerberg previously promised, and is about as awkward as a virtual world can get. The idea here is about a “Valentine-less support group”.

Looking at something like this, which is apparently the current state of Meta’s VR universe, it is absolutely impossible to imagine a world where this kind of thing is ever going to succeed. And it almost seems like Meta might even know that at this point. Here’s what Meta CTO Andrew Bosworth said last month as the year began:

“We have the best portfolio of products we’ve ever had in market and are pushing our advantage by launching half a dozen more AI powered wearables. We need to drive sales, retention, and engagement across the board but especially in MR. And Horizon Worlds on mobile absolutely has to break out for our long term plans to have a chance.”

How on earth could this break out in any meaningful way on any device? Who is something like this actually for? Even VR enthusiasts are likely to want to play actual video games rather than…whatever this is, and VR remains a relatively niche market as a whole. This is Meta’s grand vision of a shared universe, PS2-era characters flailing their arms around in unfunny ads.

«

Hard to know what Meta expects to come out of this – especially given the return-to-work drive going on now. This feels like a division that ought to be bracing itself for some big cuts.
unique link to this extract

Vance’s real warning to Europe • Financial Times

Gideon Rachman:

»

Trump clearly intends to cut a deal on Ukraine with Putin over the heads of Zelenskyy and the Europeans. That could have tragic consequences for Ukraine, which may soon be asked to accept loss of territory without security guarantees for the future. The alternative would be to try to fight on without American help.

The implications for the rest of Europe are also alarming. Putin wants Nato troops removed from the whole of the former Soviet empire. European officials believe Trump is likely to agree to withdraw US troops from the Baltics and perhaps further west, leaving the EU vulnerable to a Russian army that Nato governments warn is preparing for a larger conflict beyond Ukraine.

It is clear that the US can no longer be regarded as a reliable ally for the Europeans. But the Trump administration’s political ambitions for Europe mean that, for now, America is also an adversary — threatening democracy in Europe and even European territory, in the case of Greenland.

So what to do? Europeans need to start preparing fast for the day when the US security guarantee to Europe is definitively removed. That must involve building up autonomous defence industries. It should also mean a European mutual defence pact, outside Nato, that extends beyond the EU — to include Britain, Norway and others.

Trump will use any leverage he has to force America’s European allies into compliance on issues from trade and security to their domestic politics. That means that Europe must now start the painful process of “de-risking” its relationship with the US, looking for areas of dangerous dependence on America and stripping them out of the system.

Entrusting critical infrastructure to Musk would create a huge new vulnerability. The Trump administration will also put enormous pressure on Europeans to buy more American weaponry. Under current circumstances that would be folly.

«

As Lenin said, there are decades where nothing happens, and there are weeks where decades happen. We’re in the latter. This is a colossal, tectonic shift which won’t be fixed by whoever succeeds Trump, and anyway that change is four years away. More spending on defence; new geopolitical tensions.

On the same topic, there’s this: “The US changes sides?“
unique link to this extract

Chinese scientists develop ‘injection’ to make smartphone and EV batteries last longer • South China Morning Post

Zhang Tong:

»

The battery is considered to have expired when the supply of lithium ions runs low – for example some electric car batteries have a lifespan of around 1,500 charge cycles – but other components in the battery still remain in good working order after this happens.

This insight prompted the two lead researchers, Gao Yue and Peng Huisheng from Fudan University’s macromolecular science department, to see if they could revive a battery by replenishing the supply of active lithium ions.

To do so they had to break away from traditional battery design principles and create a lithium carrier molecule, which could be injected into the battery to manage and control the lithium ions.

But the carrier molecule would need to meet a complex set of requirements, including the ability to dissolve well in the electrolyte and participate in reactions without damaging the battery’s original environment, as well as having a high degree of compatibility with various active materials and electrolytes.

There was no record of such a molecule and it took the research team four years to identify the ideal candidate: a substance known as trifluoromethylsulfonate lithium.

It also proved to be both cost-effective and easy to synthesise as well as being compatible with different battery types.

“A commercially used lithium iron phosphate battery, after undergoing the treatment, retained its initial health after 12,000 charge cycles, compared with the usual lifespan of 2,000 cycles,” Gao told state broadcaster CCTV on Thursday.

“The battery only lost 4% of its performance after 11,818 cycles. For an EV that charges twice a day, this means the battery could last up to 18 years. In comparison, current EV batteries usually lose 30% of their performance in just 2.7 years with the same charging routine.”

«

Published in Nature. Seems ripe for rapid commercialisation, which is good news all round.
unique link to this extract

New junior developers can’t actually code • N’s Blog

Namanyay Goel:

»

Something’s been bugging me about new devs [software developers] and I need to talk about it.

We’re at this weird inflection point in software development. Every junior dev I talk to has Copilot or Claude or GPT running 24/7. They’re shipping code faster than ever. But when I dig deeper into their understanding of what they’re shipping? That’s where things get concerning.

Sure, the code works, but ask why it works that way instead of another way? Crickets. Ask about edge cases? Blank stares.

The foundational knowledge that used to come from struggling through problems is just… missing.

We’re trading deep understanding for quick fixes, and while it feels great in the moment, we’re going to pay for this later.

I recently realized that there’s a whole generation of new programmers who don’t even know what StackOverflow is.

Back when “Claude” was not a chatbot but the man who invented the field of information entropy, there was a different way to debug programming problems.

First, search on Google. Then, hope some desperate soul had posed a similar question as you had. If they did, you’d find a detailed, thoughtful, (and often patronizing) answer from a wise greybeard on this site called “Stack Overflow”.

«

This is very funny – an internet generation ago the complaint was “these new devs don’t know anything about programming, they just search on Stack Overflow.” And the generation before that? “These new devs don’t know anything about programming, they just use Google.” Everyone copes with the challenge in the way appropriate to their generation.
unique link to this extract

This solar-powered reactor sucks CO2 from the air and turns it into fuel • Gizmodo

Margherita Bassi:

»

Researchers at the University of Cambridge have built a solar-powered reactor that converts atmospheric carbon dioxide (CO2) into a gas that could one day fuel vehicles, power off-the-grid dwellings, and even produce pharmaceutical products.

The researchers say were inspired by photosynthesis and claim that their technology can be scaled up more easily than earlier solar-powered devices. The team’s research was published in the journal Nature Energy.

Carbon capture and storage (which is exactly what it sounds like) is a possible means of reducing greenhouse gas emissions. The problem is that most carbon capture technologies are themselves powered by burning fossil fuels—not to mention the fact that the CO2 captured in the process needs to be stored somewhere, such as deep underground. But a new reactor could solve all of that.

“What if instead of pumping the carbon dioxide underground, we made something useful from it?” Sayan Kar, a chemist at the University of Cambridge and first author on the study, said in a university statement. “CO2 is a harmful greenhouse gas, but it can also be turned into useful chemicals without contributing to global warming.”

Kar and his colleagues’ new reactor is completely solar-powered, meaning it requires no cables or batteries. At night, it filters CO2 from the air—similar to how a sponge soaks up water, according to the researchers. During the day, sunlight heats up the collected CO2, which absorbs the Sun’s infrared radiation while a semiconductor powder absorbs the ultraviolet radiation.

«

Nice, but: does it scale?
unique link to this extract

Pakistan’s electric rickshaws are accelerating the country’s EV revolution • Rest of World

Kunwar Khuldune Shahid:

»

There are currently an estimated 45,000 two- and three-wheeler EVs in the country, constituting only 0.16% of the total vehicles in Pakistan.

But the industry is on the cusp of an explosion. Pakistan’s recent EV policy aims to transition 90% of all new vehicles to electric by 2040. By 2030, EVs would constitute 50% of all auto sales in the country, Kamran Kamal, a spokesperson for BYD Pakistan, told Reuters. Last week, the government reportedly granted EV production licenses to 57 manufacturers — 55 of which were for two- and three-wheelers.

“Rickshaws are the common man’s mode of transportation and EVs are the future,” Rana Arsalan Sarwar, national marketing manager of Siwa Industries, a rickshaw manufacturing company, told Rest of World. Siwa launched its own e-rickshaw last year. 

Pakistan’s neighbors bear testament to the success of e-rickshaws. Last year, Bangladesh’s former prime minister, Sheikh Hasina, reversed an earlier ban on e-rickshaws and announced laws to regulate the 2–4 million e-rickshaws that operate in the country. In India, little-known e-rickshaw companies are driving the EV revolution with sales that far exceed those of electric cars.

…Priced at over 1 million Pakistani rupees ($3,600), Sazgar’s e-rickshaws cost more than twice as much as conventional ones. Rickshaw drivers in Pakistan interviewed by Rest of World said they are nonetheless keen to make the switch. One such driver, Sami Ullah, travelled over 1,000 kilometres (621 miles) from Pakistan’s Hyderabad city to Sazgar’s factory in Lahore to purchase an e-rickshaw with his savings.

“I tried it in 2022, and loved how smooth the ride was,” Ullah said. “There is no smoke, it doesn’t cause pollution.” His family was relieved when he bought the e-rickshaw because “engine rickshaws damage our clothes, our bodies, and our health,” he said. The investment has paid off: Ullah now saves money on fuel.

«

unique link to this extract

The untold story of a crypto crimefighter’s descent into Nigerian prison • WIRED

Andy Greenberg:

»

At 8 AM on March 23 of 2024, Tigran Gambaryan woke up on a couch in Abuja, Nigeria, where he’d been dozing since the predawn call to prayer. The house around him, which often buzzed with the nearby sound of generators, was eerily quiet. In that silence, the harsh reality of Gambaryan’s situation flooded back into his mind the way it had every morning for nearly a month: That he and his colleague at the cryptocurrency firm Binance, Nadeem Anjarwalla, were being held hostage in a Nigerian government-owned compound—detained without access to their own passports under military guard in a building circled by barbed-wire walls.

Gambaryan got up from the couch. The 39-year-old Armenian-American wore a white T-shirt over his compact, muscular build, his right arm covered with an Eastern Orthodox tattoo. His usually shaved head and trimmed black beard were now stubbled and scraggly from a month of growth. Gambaryan found the compound’s cook, and asked if she would buy him some cigarettes. Then he walked out into the house’s internal courtyard, and began to restlessly pace and make phone calls to his lawyers and other contacts at Binance—the world’s largest cryptocurrency exchange—restarting his daily efforts to “unfuck the situation,” as he put it.

Just the day before, the pair of Binance staffers and their crypto giant employer were told they were about to be charged with tax evasion. The two men seemed to be wedged into the middle of a bureaucratic conflict between an unaccountable foreign government and one of the most controversial players in the crypto economy. Now they were not only being confined against their will, with no end in sight, but also charged as criminals.

«

It’s a fun story – not for them, but to read. Gambaryan had a remarkable career up to then in crypto, working on some of the biggest crypto cases around. Then he got a job with Binance.
unique link to this extract

New York Times goes all-in on internal AI tools • Semafor

Max Tani:

»

The New York Times is greenlighting the use of AI for its product and editorial staff, saying that internal tools could eventually write social copy, SEO headlines, and some code.

In messages to newsroom staff, the company announced that it’s opening up AI training to the newsroom, and debuting a new internal AI tool called Echo to staff, Semafor has learned. The Times also shared documents and videos laying out editorial do’s and don’t for using AI, and shared a suite of AI products that staff could now use to develop web products and editorial ideas.

“Generative AI can assist our journalists in uncovering the truth and helping more people understand the world. Machine learning already helps us report stories we couldn’t otherwise, and generative AI has the potential to bolster our journalistic capabilities even more,” the company’s editorial guidelines said.

…The paper encouraged editorial staff to use these AI tools to generate SEO headlines, summaries, and audience promos; suggest edits; brainstorm questions and ideas and ask questions about reporters’ own documents; engage in research; and analyze the Times’ own documents and images. In a training video shared with staff, the Times suggested using AI to come up with questions to ask the CEO of a startup during an interview.

«

Truly stunning. I hope the training video was made by people who don’t know journalists, because if the NYT is hiring journalists who need their questions written by AI, they’re in worse trouble than they know.
unique link to this extract

Errata, corrigenda and ai no corrida: none notified

The price of coffee and cocoa has more than doubled in the past year: you might see the effect soon. CC-licensed photo by stephenrwalli on Flickr.

You can sign up to receive each day’s Start Up post by email. You’ll need to click a confirmation link, so no spam.

A selection of 9 links for you. Energised. I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. On Bluesky: @charlesarthur.bsky.social. Observations and links welcome.

Broadcom, TSMC eye possible Intel deals that would split storied chip maker • WSJ via MSN

Asa Fitch, Lauren Thomas and Yang Jie:

»

Intel’s rivals Taiwan Semiconductor Manufacturing Co. and Broadcom are each eyeing potential deals that would break the American chipmaking icon in two.

Broadcom has been closely examining Intel’s chip-design and marketing business, according to people familiar with the matter. It has informally discussed with its advisers making a bid but would likely only do so if it finds a partner for Intel’s manufacturing business, the people said.

Nothing has been submitted to Intel, the people cautioned, and Broadcom could decide not to seek a deal.

Separately, TSMC has studied controlling some or all of Intel’s chip plants, potentially as part of an investor consortium or other structure, according to people familiar with the discussions.

Broadcom and TSMC aren’t working together, and all of the talks so far are preliminary and largely informal.

But the potential deals would have been unthinkable until Intel’s recent struggles made it an acquisition target. The end result could be a breakup of Intel after the American icon spent many decades dominating the business of making central processors for both personal computers and data centers.

Splitting the company would also bring it in line with an industrial shift in recent decades toward specializing in either manufacturing or designing chips, but not both.

Frank Yeary, the interim executive chairman of Intel, has been leading the discussions with possible suitors and Trump administration officials, who are concerned about the fate of a company seen as critical to national security, people familiar with the matter said. Yeary has been telling individuals close to him that he is most focused on maximizing value for Intel shareholders, the people said.

«

“Nothing has been submitted to Intel”? Then again, if TSMC makes a move for the plants, the other parts would obviously be up for sale. And “maximising value for shareholders” is one of those doom-laden phrases on a par with a football club chairman expressing confidence in the manager. The manager’s exit swiftly follows.
unique link to this extract

Zypher’s speech model can clone your voice with five seconds of audio • The Register

Tobias Mann:

»

Palo Alto-based AI startup Zyphra unveiled a pair of open text-to-speech (TTS) models this week said to be capable of cloning your voice with as little as five seconds of sample audio. In our testing, we generated realistic results with less than half a minute of recorded speech.

Founded in 2021 by Danny Martinelli and Krithik Puthalath, the startup aims to build a multimodal agent system called MaiaOS. To date, these efforts have seen the release of its Zamba family of small language models, optimizations such as tree attention, and now the release of its Zonos TTS models.

Measuring at 1.6 billion parameters in size each, the models were trained on more than 200,000 hours of speech data, which includes both neutral-toned speech such as audiobook narration, and “highly expressive” speech. According to the upstart’s release notes for Zonos, the majority of its data was in English but there were “substantial” quantities of Chinese, Japanese, French, Spanish, and German. Zyphra tells El Reg this data was acquired from the web and was not obtained from data brokers.

The results are actually two Zonos models: One that uses a fully transformer-based architecture, and the other, a hybrid that combines transformer and Mamba state space model (SSM) architectures. The latter, Zyphra claims, makes it the first TTS model to use this arch. While transformer-based models are without a doubt the most commonly used in generative AI today, alternative architectures like Mamba are gaining traction.

«

This is one of those “we built it because we can, rather than because we thought of the consequences” products.§
unique link to this extract

How Diablo hackers uncovered a speedrun scandal – Ars Technica

Kyle Orland:

»

For years, Maciej “Groobo” Maselewski stood as the undisputed champion of Diablo speedrunning. His 3-minute, 12-second Sorcerer run looked all but unbeatable thanks to a combination of powerful (and allowable) glitch exploits along with what seemed like some unbelievable luck in the game’s randomly generated dungeon.

But when a team of other speedrunners started trying and failing to replicate that luck using outside software and analysis tools, the story behind Groobo’s run began to fall apart. As the inconsistencies in the run started to mount, that team would conduct an automated search through billions of legitimate Diablo dungeons to prove beyond a shadow of a doubt that Groobo’s game couldn’t have taken place in any of them.

“We just had a lot of curiosity and resentment that drove us to dig even deeper,” team member Staphen told Ars Technica of their investigation. “Betrayal might be another way to describe it,” team member AJenbo added. “To find out that this had been done illegitimately… and the person had both gotten and taken a lot of praise for their achievement.”

If you have any familiarity with Diablo or speedrunning, watching Groobo’s run feels like watching someone win the lottery. First, there’s the dungeon itself, which features a sequence of stairways that appear just steps from each other, forming a quick and enemy-free path down to the dungeon’s deeper levels. Then there’s Groobo’s lucky find of Naj’s Puzzler on level 9, a unique item that enables the teleporting necessary for many of the run’s late-game maneuvers.

“It seemed very unusual that we would have so many levels with the upstairs and the downstairs right next to each other,” Allan “DwangoAC” Cecil told Ars Technica. “We wanted to find some way of replicating this.”

«

What is fascinating about this is the burning sense of unfairness that the other speedrunners clearly felt, and then the incredible lengths they went to in search of the answer that would salve their anger.
unique link to this extract

Extreme weather expected to cause food price volatility in 2025 after cost of cocoa and coffee doubles • The Guardian

Damien Gayle:

»

Extreme weather events are expected to lead to volatile food prices throughout 2025, supply chain analysts have said, after cocoa and coffee prices more than doubled over the past year.

In an apparent confirmation of warnings that climate breakdown could lead to food shortages, research by the consultancy Inverto found steep rises in the prices of a number of food commodities in the year to January that correlated with unexpected weather.

Several authorities declared 2024 the hottest year on record, a trend towards higher temperatures that seems to be continuing into 2025. Inverto said a long-term trend towards more extreme weather events would continue to hit regional crop yields, causing price spikes.

The highest price rises were for cocoa and coffee, up 163% and 103% respectively, due to a combination of higher than average rainfall and temperatures in producing regions, according to the research.

Sunflower oil prices increased by 56% after drought caused poor crop yields in Bulgaria and Ukraine, which also continued to be affected by the Russian invasion. Other food commodities with sharp year-on-year price rises included orange juice and butter, both up by more than a third, and beef, up by just over a quarter.

“Food manufacturers and retailers should diversify their supply chains and sourcing strategies to reduce over-reliance on any one region affected by crop failures,” Katharina Erfort, of Inverto, said.

«

The rising cost of coffee probably won’t put coffee shops out of business – other costs are far bigger factors – but they might have to raise them marginally, and people may notice that.
unique link to this extract

Our sleep, brain aging, and waste clearance • Ground Truths

Eric Topol:

»

Sleep is the principal driver of glymphatic flow and waste clearance, occurring during the NREM phase of sleep (which includes deep sleep, slow-wave, known as stage N3). Indeed, the totality of evidence backs sleep’s major function as waste clearance of the brain through glymphatics. Clearance of toxic proteins, like β-amyloid, are critical to brain health.

Back in 2018, PET scanning was used to show that one night of sleep deprivation resulted in substantial increase in β-amyloid accumulation, in regions of the brain linked to Alzheimer’s disease. On a chronic basis, several studies have shown that poor sleep is prospectively linked to the risk and progression of Alzheimer’s disease. For example, in nearly 8,000 participants with 25-year follow-up, people aged less than 50 or 60 years with six hours of sleep or less had a more than 20% increased risk of developing late-onset dementia.

It’s also notable that clearance of toxic proteins interacts with our brain immune system (as I reviewed in a recent Ground Truths, Guardians of the Brain), invoking another mechanism by which waste induces harm.

«

A quote in this piece says that sleep is “like turning on the dishwasher before you go to bed and waking up with a clean brain”. This is a very detailed piece, with very detailed diagrams, but one other point to take away is that Ambien (aka zolpidem) reduces glymphatic flow (the cleaning mechanism). Avoid that too if possible.
unique link to this extract

Three observations • Sam Altman

The head of OpenAI would like you to know the following:

»

In a decade, perhaps everyone on earth will be capable of accomplishing more than the most impactful person can today.

We continue to see rapid progress with AI development. Here are three observations about the economics of AI:

1. The intelligence of an AI model roughly equals the log of the resources used to train and run it. These resources are chiefly training compute, data, and inference compute. It appears that you can spend arbitrary amounts of money and get continuous and predictable gains; the scaling laws that predict this are accurate over many orders of magnitude.

2. The cost to use a given level of AI falls about 10x every 12 months, and lower prices lead to much more use. You can see this in the token cost from GPT-4 in early 2023 to GPT-4o in mid-2024, where the price per token dropped about 150x in that time period. Moore’s law changed the world at 2x every 18 months; this is unbelievably stronger. 

3. The socioeconomic value of linearly increasing intelligence is super-exponential in nature. A consequence of this is that we see no reason for exponentially increasing investment to stop in the near future.

If these three observations continue to hold true, the impacts on society will be significant.

We are now starting to roll out AI agents, which will eventually feel like virtual co-workers.

«

Personally, I don’t think “perhaps everyone” will be able to do more than the most impactful person today. The people sifting piles of waste in India will probably still be there, and it seems unlikely they’ll be more impactful than the most impactful person today. (They might be getting a little help from AI on their phone, though.)

At the same time, the AI systems we see now are the worst they’ll ever be. A lot is going to change. But I do like Altman’s insistence that we need to keep pouring money into his company/nonprofit/wallet.
unique link to this extract

When Did They Die? The celebrity death date quiz game

Does what it says on the tin: offers a quiz for you to try to get right. Mine were: Hemingway, Stalin, John Paul 1, Mary Tyler Moore, Coolio. Score: 3,332. Enjoy beating it.
unique link to this extract

Key bird flu lab threatens to strike as California cases and egg prices climb • POLITICO

Rachel Bluth:

»

Workers at a key lab for testing animal disease are threatening to go on strike, raising concerns about California’s ability to respond to the growing outbreak of bird flu that has sent the price of eggs soaring nationwide.

Technicians at the California Animal Health and Food Safety Lab at the University of California, Davis, have been sounding the alarm for months, alleging staffing shortages and strains as their union has been in contentious negotiations with the University of California system. The University Professional and Technical Employees are set to finish voting Thursday on whether to strike, arguing their demands haven’t been met systemwide.

The lab is the only one in the state able to handle the most dangerous cases of avian flu, which has swept through dozens of poultry and hundreds of dairy farms in California’s agricultural heartland with no clear end in sight.
Getting a positive, timely test result is vital for dairy farmers to make a decision on isolating or culling infected animals, but also to qualify for supplemental payments from the USDA to help combat an outbreak.

Lab technicians say the workload and stress of the bird flu outbreak are leading to injuries and burnout — especially within the current staff of just three fully trained workers, and two more who can only work on avian flu testing under the supervision of other staff.

«

Surely nothing to worry about that a bird flu testing site might shut down. And the workers did vote to strike.
unique link to this extract

“Download and Transfer” for Kindle books discontinued from Feb 26 • Good E-Reader

Michael Kozlowski:

»

Amazon will cease the ability to download and transfer options for the Kindle e-reader on February 26th, 2025. No Kindle can use this functionality once the due date rolls around. Only the 12th generation Kindles never had this ability to begin with, but now no other model will either. Why is Amazon doing this? It’s a feature not many people use and those who do, commit e-book piracy.

Amazon said:

»

Starting February 26, 2025, the “Download & Transfer via USB” option will no longer be available. You can still send Kindle books to your Wi-Fi enabled devices by selecting the “Deliver or Remove from Device” option.

«

…Download and Transfer via USB was launched over ten years ago and was created at a time, when WIFI was not prevalent, so this feature was the only way to send e-books to the Kindle.

«

The problem is that if the ebook content is on a USB it can be transferred to a computer, and there the DRM can be broken and the book can be pirated and uploaded to a torrent and then downloaded by Meta to train its next AI. Funny how these things go.

unique link to this extract

Errata, corrigenda and ai no corrida: none notified