Gaming PC company Razer has paused direct laptop sales in the US. Guess why. CC-licensed photo by Jon Russell 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. Clean cut. I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. On Bluesky: @charlesarthur.bsky.social. Observations and links welcome.
UK creating “murder prediction” tool to identify people most likely to kill • The Guardian
Vikram Dodd:
»
The UK government is developing a “murder prediction” programme which it hopes can use personal data of those known to the authorities to identify the people most likely to become killers.
Researchers are alleged to be using algorithms to analyse the information of thousands of people, including victims of crime, as they try to identify those at greatest risk of committing serious violent offences.
The scheme was originally called the “homicide prediction project”, but its name has been changed to “sharing data to improve risk assessment”. The Ministry of Justice hopes the project will help boost public safety but campaigners have called it “chilling and dystopian”.
The existence of the project was discovered by the pressure group Statewatch, and some of its workings uncovered through documents obtained by Freedom of Information requests.
Statewatch says data from people not convicted of any criminal offence will be used as part of the project, including personal information about self-harm and details relating to domestic abuse. Officials strongly deny this, insisting only data about people with at least one criminal conviction has been used.
The government says the project is at this stage for research only, but campaigners claim the data used would build bias into the predictions against minority-ethnic and poor people.
The MoJ says the scheme will “review offender characteristics that increase the risk of committing homicide” and “explore alternative and innovative data science techniques to risk assessment of homicide”.
«
Ooh, I can help on this. Strong predictors: 1) male 2) any history of domestic abuse. It is a strange how people try to recreate Minority Report again and again, when there’s so much past evidence. If they’re hoping to predict deadly knife crime – that’s highly unlikely to work. Won’t stop some people poring over datasets though.
unique link to this extract
Nvidia’s AI chip empire is under threat from Trump’s tariffs • Rest of World
Viola Zhou talks to Stephen Witt, whose new book is about the rise of Nvidia:
»
Viola Zhou: TSMC’s founder Morris Chang said globalization was “almost dead” when his company had to build a factory in Arizona. Now, we have these tariffs. How do you think Jensen Huang will navigate Trump’s policy?
Stephen Witt: Jensen doesn’t talk politics, not with me, not with anybody. He is probably the most powerful tech figure in America to not attend Donald Trump’s inauguration. Now, Jensen must deal with politics. The disruption to his business is too severe.
Nvidia had long benefited from no U.S. manufacturing. It’s completely outsourced, mostly to Taiwan. Personally, I’m certain he thinks these tariffs are a terrible idea. Whether he will ever say that out loud, I don’t know.
For Nvidia, one option is to just pay the tariffs. Nvidia can afford, almost uniquely, to pay these tariffs, because the margins are high. But profits will go down, and stock prices will go down.
Second is to attempt to rebuild or onshore manufacturing into the United States. I think they are reluctant to do so because it seems likely that any politician, even another Republican, would undo or reverse these tariffs. The third option would be to just raise prices.
VZ: Nvidia has also been opposing the U.S. government’s attempts to restrict chip exports to China. After the most powerful chips got banned, it has been selling billions of dollars worth of slightly inferior chips to Chinese tech companies. Why is Nvidia doing this?
SW: Jensen has made the point that China will simply just innovate out of this problem. He was saying that a couple of years ago, and then DeepSeek came along and proved that they could do so.
This is probably Jensen’s point of view: If we ban this stuff, [China] will just build it better and cheaper. It would actually be in our long-term strategic and competitive advantage to keep selling them because then there’s not going to be a domestic competitor.
«
Multiple companies, all struggling with how to navigate these tariffs. Nvidia with perhaps the most to lose – but American companies also with the most to lose.
unique link to this extract
Why we distrust technology • Quillette
Marian Tupy:
»
For most of our history, human survival depended less on technological ingenuity and more on cooperation and social cohesion. Our ancestors did not invent their way out of problems; they solved them through alliances, negotiations, and collective rulemaking. Food shortages, for instance, were addressed not by developing advanced agricultural techniques—those came much later—but by rationing resources, redistributing wealth within the tribe, and reinforcing norms against hoarding.
This survival strategy shaped our psychology. Over generations, humans became attuned to social fixes as the primary way to navigate crises. We evolved to seek consensus, enforce norms, and reward conformity—traits that helped small groups function efficiently in an unpredictable environment. As a result, when confronted with modern challenges, we instinctively default to social regulation over technological adaptation.
Today, this bias manifests in the way we talk about, for example, climate change. The dominant discourse does not emphasise nuclear fusion, carbon capture, or geoengineering, despite their potential to dramatically cut emissions. Instead, we hear calls for people to consume less, fly less, drive less, eat differently—as though the best way to tackle a global problem is through personal sacrifice. This isn’t a rational economic approach; it’s a deeply ingrained cognitive reflex.
Beyond evolutionary psychology, several well-documented cognitive biases reinforce our scepticism toward technological solutions. One of the most powerful is negativity bias, the tendency to focus more on potential downsides than on possible benefits. Innovations—especially large-scale ones like nuclear power or geoengineering—are often accompanied by uncertainties. A nuclear plant meltdown is a vivid disaster; the slow, cumulative benefits of abundant clean energy are far less emotionally gripping.
«
Good points. It’s just that the technological solutions are making only the merest dent on the problem, and the chemistry of carbon dioxide militates against any removal solution that doesn’t have unlimited free energy.
unique link to this extract
The AI therapist can see you now • NPR
Katia Riddle:
»
New research suggests that given the right kind of training, AI bots can deliver mental health therapy with as much efficacy as — or more than — human clinicians.
The recent study, published in the New England Journal of Medicine, shows results from the first randomized clinical trial for AI therapy.
Researchers from Dartmouth College built the bot as a way of taking a new approach to a longstanding problem: The U.S. continues to grapple with an acute shortage of mental health providers. “I think one of the things that doesn’t scale well is humans,” says Nick Jacobson, a clinical psychologist who was part of this research team. For every 340 people in the U.S., there is just one mental health clinician, according to some estimates.
While many AI bots already on the market claim to offer mental health care, some have dubious results or have even led people to self-harm.
More than five years ago, Jacobson and his colleagues began training their AI bot in clinical best practices. The project, says Jacobson, involved much trial and error before it led to quality outcomes.
“The effects that we see strongly mirror what you would see in the best evidence-based trials of psychotherapy,” says Jacobson. He says these results were comparable to “studies with folks given a gold standard dose of the best treatment we have available.”
«
The NEJM is somewhat second-division in science publications these days (it’s printed a lot of stuff that has been shown to be wrong, but hasn’t retracted). But anyway, this feels like the final vindication of Eliza, the original chatbot, just under 60 years later.
unique link to this extract
Razer pauses direct laptop sales in the US as new tariffs loom • The Verge
Antonio G. Di Benedetto:
»
Razer’s upcoming Blade 16 and other laptops are no longer available for preorder or purchase on its US site. The configurator for preordering its new Blade 16 laptop was available as recently as April 1st, according to the Internet Archive — one day before the Trump administration announced sweeping US tariffs on China, Taiwan, and others that make laptop components. When asked recently if tariffs might affect Razer’s prices or availability, its public relations manager Andy Johnston told The Verge, “We do not have a comment at this stage regarding tariffs.”
Razer may not be openly talking about the impact of tariffs, but Framework halted sales of its entry-level Laptop 13 in the US on April 7th, and Micron reportedly confirmed surcharges for its memory chips will apply once the tariffs take effect after midnight tonight.
The direct link to the Blade 16 configurator now takes you to a 404 error page, and its product page only has a “notify me” button instead of anywhere to submit your preorder.
While shopping for other laptops on Razer’s site, only skins and accessories are available for purchase.
«
Razer in an American-Singaporean-Hong Kong company, but it’s not clear where its PCs are manufactured.
unique link to this extract
Microsoft backs out of $1bn central Ohio data center plan • Colombus Dispatch
Max Filby:
»
A major tech company has backed away from its plans to build data centers in central Ohio, at least for the time being.
On Monday, Microsoft said it would no longer move forward with its plans to build data centers in Licking County. The company had planned to invest $1 billion initially toward three data center campuses in New Albany, Heath and Hebron.
“We will continue to evaluate these sites in line with our investment strategy,” a Microsoft spokesperson told The Dispatch. “We sincerely appreciate the leadership and partnership of Ohio government officials and the support of Licking County residents.”
The spokesperson later on Monday said the company will continue to own the land and intends to proceed with the project at some point, although a specific timeframe was not given.
Microsoft plans to ensure the land at two of the three sites will be able to be used for farming and will still carry out development agreements to fund roadway and utility project upgrades, according to the company.
Each of the three data center campus were supposed to have one building at first with the potential for several buildings on each site, The Dispatch reported in 2024.
Microsoft’s initial investment was to consist of $700m in building costs and $300m in machinery expenses. The project was to create 20 jobs to start with.
…Microsoft walked back its plans for Licking County data centers as the company has ended leases for sizeable data center capacity in the U.S., Reuters reported. The move suggests there’s a potential oversupply at Microsoft as it builds out artificial intelligence infrastructure to meet a looming surge in demand.
«
History suggests a tariff-induced downturn may cause a reckoning in the venture capital industry • Fortune
Jessica Mathews:
»
First you had a venture boom caused by more than a decade of low interest rates. Then a 2022 bear market as the post-pandemic recovery upended many business plans, followed by unprecedented amounts of money being plowed into AI companies. But the AI boom was missing a standard element of the private market ecosystem: IPOs and M&A. As a result, limited partners haven’t been getting many distributions for three years.
No surprise then that VC fundraising has been freefalling ever since 2022—and nearly all the capital that is available has flowed to a small group of funds. Last year, 75% of all the capital raised by VCs went to only 30 venture capital firms, according to PitchBook (see data here). Just nine firms raised half of all that capital. Nearly eight in 10 limited partners say they declined to re-up investments into at least one of the VCs in their portfolio this past year, according to Coller Capital’s annual survey.
The expectations going into this year were that the VC sector was due to get its groove back. Many Silicon Valley elites have been hopeful that Trump’s anti-regulation approach will revive M&A activity. And the IPO pipeline was starting to fill up again. CoreWeave’s public market debut wasn’t the blockbuster some might have wished for—the AI datacenter company ended up slashing the amount of money it raised and its stock has been whipsawed since it started trading on the Nasdaq—but there was hope that other IPO candidates, with cleaner balance sheets, might fare better.
As the Trump tariffs take effect, however, the equation is changing. Investors and startup founders must now consider the very real possibility of a sustained bear market or a recession. Companies like Klarna and StubHub have already decided to put their IPO plans on hold.
«
Question: which AI company will be the first to IPO? I’m pretty sure that the first search company to IPO was not Google (it didn’t go until 2004). Is the runway of their funds long enough to get them to takeoff?
unique link to this extract
40% of world’s power clean but emissions rising – report • BBC News
Jonah Fisher:
»
More than 40% of the world’s electricity was generated without burning fossil fuels in 2024, according to a new report from the think-tank Ember..
But carbon dioxide emissions, which warm the planet, have risen to an all time high, the report says, with hot weather pushing up the overall demand for power.
That meant an increase in the use of fossil fuel burning power stations.
Solar power continues to be the fastest-growing energy source, with the amount of electricity it generates doubling in the last three years.
“Solar power has become the engine of the global energy transition,” said Phil Macdonald, the managing director of Ember. “Amid the noise, it’s essential to focus on the real signal. Hotter weather drove the fossil generation increase in 2024, but we’re very unlikely to see a similar jump in 2025.”
In a separate report, the European Copernicus climate service said March 2025 was the second hottest on record, extending a spell of record or near record breaking temperatures.
…Cheap and relatively easy to install, for the twentieth year in a row solar is the fastest growing electricity source. According to Ember, the amount generated by solar panels has doubled every three years since 2012.
…In the last five years, fast-growing Asian economies, in particular India and China, have continued expanding their use of fossil fuels to meet rapidly rising demand for electricity.
«
So it’s one of those good news, bad news things. At least it isn’t only bad news.
unique link to this extract
Google DeepMind’s weapon in the AI talent war: aggressive noncompetes • Business Insider
Hugh Langley and Hasan Chowdhury:
»
The battle for AI talent is so hot that Google would rather give some employees a paid one-year vacation than let them work for a competitor.
Some Google DeepMind staff in the UK are subject to noncompete agreements that prevent them from working for a competitor for up to 12 months after they finish work at Google, according to four former employees with direct knowledge of the matter who asked to remain anonymous because they were not permitted to share these details with the press.
Aggressive noncompetes are one tool tech companies wield to retain a competitive edge in the AI wars, which show no sign of slowing down as companies launch new bleeding-edge models and products at a rapid clip. When an employee signs one, they agree not to work for a competing company for a certain period of time.
Google DeepMind has put some employees with a noncompete on extended garden leave. These employees are still paid by DeepMind but no longer work for it for the duration of the noncompete agreement.
Several factors, including a DeepMind employee’s seniority and how critical their work is to the company, determine the length of noncompete clauses, those people said. Two of the former staffers said six-month noncompetes are common among DeepMind employees, including for individual contributors working on Google’s Gemini AI models. There have been cases where more senior researchers have received yearlong stipulations, they said.
“Our employment contracts are in line with market standards,” a Google spokesperson told Business Insider in a statement.
«
I guess it’s a question of definition, but to me the one-year vacation isn’t strictly a “non-compete”. It’s a “you’re still an employee, but we don’t want you to do anything for us.” A non-compete would be “goodbye, but you can’t go to companies X or Y” (to which X and Y would agree – something which got Apple and Google into very expensive hot water in California a decade ago).
unique link to this extract
| • Why do social networks drive us a little mad? • Why does angry content seem to dominate what we see? • How much of a role do algorithms play in affecting what we see and do online? • What can we do about it? • Did Facebook have any inkling of what was coming in Myanmar in 2016? Read Social Warming, my latest book, and find answers – and more. |
Errata, corrigenda and ai no corrida: none notified
The Overspill: when there's more that I want to say 