Some schools in the US are rethinking their use of computers such as Chromebooks. CC-licensed photo by Virginia Department of Education 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. Calculated. I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. On Bluesky: @charlesarthur.bsky.social. Observations and links welcome.
Trump runs out of options • Comment Is Freed
Lawrence Freedman:
»
It is hard to convey the gloom that has overtaken Washington. All the structures that are vital to crisis management have either been attenuated or disbanded. There is hardly anyone left on the National Security Council staff. A friend described an empty State Department where you could hear your own foot steps. Marco Rubio is involved in the decision-making but he has neglected to acquire the professional staff assessments that should inform such decisions (see this from Dan Drezner).
The military part of the Pentagon still functions, but the civilian part has been purged.
…The war has been unpopular from the start and it is getting more so. The inflationary effects of the war are already biting and if some vital supplies stay trapped in the Gulf the pain will soon get worse. Without a quick resolution of the war, many Republicans take for granted a hammering at the mid-terms.
Against this backdrop there are unsurprising reports that Trump is fed up with the war and wants to end it as quickly as possible. His normal ploy is to declare a stunning victory and move on. He has in fact been trying this since early March and still talks about the war being ahead of schedule, at least in terms of the US and Israel running out of things to bomb. But as much as he’d like to walk away this would mean leaving behind a mess in the Gulf. In the rest of this post I’ll look at the prospects for a negotiated settlement, further military escalation, and the possibility that the US will retreat without a resolution.
…This is the conclusion that America’s allies fear. The US does enough to wound the regime but it hangs on. The regime finds itself with leverage which it intends to exploit, both to get regional states to acknowledge its power and provide an extra source of finance. If the US does not have good ideas about how to open the Strait militarily it would be surprising if European and Gulf states, even acting together, would have more success.
If the US walks away from a problem it created because it has become too difficult, then this point could go down as one in which US influence took a steep decline. The Gulf states know all about past efforts to ‘eliminate’ Iranian proxies, and how these somehow managed to survive and recover. They have been shocked by Iranian strikes against them, so they are not confident of a deal with Tehran, yet can no longer rely on the US. This is bound to lead to some policy reappraisals. In the short term while they might explore options for getting open the Strait they will also need to develop ways of getting round the current restrictions. All will be expensive and time-consuming and add to the international bill.
«
Subscribing to (Sam and Lawrence) Freedman’s Substack is very cheap, and provides excellent insights. Nothing here is positive.
unique link to this extract
Apple discontinues the Mac Pro with no plans for future hardware • 9to5Mac
Chance Miller:
»
It’s the end of an era: Apple has confirmed to 9to5Mac that the Mac Pro is being discontinued. It has been removed from Apple’s website as of Thursday afternoon. The “buy” page on Apple’s website for the Mac Pro now redirects to the Mac’s homepage, where all references have been removed.
Apple has also confirmed to 9to5Mac that it has no plans to offer future Mac Pro hardware.
The Mac Pro has lived many lives over the years. Apple released the current Mac Pro industrial design in 2019 alongside the Pro Display XDR (which was also discontinued earlier this month). That version of the Mac Pro was powered by Intel, and Apple refreshed it with the M2 Ultra chip in June 2023. It has gone without an update since then, languishing at its $6,999 price point even as Apple debuted the M3 Ultra chip in the Mac Studio last year.
With that in mind, the Mac Studio is clearly set up to be the ‘pro’ desktop Mac of the future in Apple’s lineup.«
The “pro” always used to be the top-end version to which one could add graphics cards and other extras comparatively easily; the best example was the G4, where one side latched down to give access to the internals.
But the desktop market is shrinking, and has been for years; it was 80/20 in favour of laptops about 15 years ago, and nothing’s taking it in the other direction. For a while Apple has had four desktop products: the Mac mini, the iMac, the Mac Studio and the Mac Pro. That’s at least one too many, and the Pro was just a Studio with a different case. So now it’s dead.
unique link to this extract
FBI confirms hackers targeted Kash Patel’s personal emails • POLITICO
Maggie Miller and John Sakellariadis:
»
The FBI on Friday confirmed that hackers targeted the personal emails of Director Kash Patel, hours after an Iranian government-linked hacking group posted documents and images online, claiming to have stolen them from Patel.
In a statement, the FBI confirmed the agency was “aware of malicious actors targeting Director Patel’s personal email information, and we have taken all necessary steps to mitigate potential risks associated with this activity.”
“The information in question is historical in nature and involves no government information,” the FBI added. The statement did not include details on who was behind the attack. “The FBI will continue to pursue the actors responsible, support victims, and share actionable intelligence in defense of networks,” it said.
Earlier in the day, Iran-linked hacking group Handala claimed to have hacked Patel’s email and published several pictures of the director, including one showing him brandishing a cigar and another that appears to be his personal resume. Most emails are dated between 2012 and 2014, though there is at least one from 2022, according to files posted by Handala on Telegram and reviewed by POLITICO.
In a post on Thursday, also reviewed by POLITICO, Handala claimed to have breached an FBI network, but did not provide details on what information may have been accessed.
Handala has recently been linked to a high-profile cyberattack against a U.S. company. Last week, the DOJ said the group was responsible for a hack of Michigan-based medical device manufacturer Stryker, which wiped roughly 200,000 devices and exfiltrated large amounts of data from the company.
«
Patel got hacked because he used the same password for his personal email as on other sites, and so when those sites get hacked, your password is in there. Quite probably it was hashed, but if you’re the Iranian government hackers then you’re going to put some effort into digging Patel’s email/password pair out of a big data dump and decrypting it.
Obviously hacking isn’t good. But Patel is such a pompous idiot that one hopes it takes him down a peg or two: he has nobody to blame but himself.
unique link to this extract
What AI hypists miss • Persuasion
Francis Fukuyama:
»
Recently I heard a presentation by an engineer from OpenAI about the incredible transformations that will occur once we get to artificial general intelligence (AGI), or even superintelligence. He said that this will quickly solve many of the world’s problems: GDP growth rates could rise to 10, 15, even 20% per year, diseases will be cured, education revolutionized, and cities in the developing world will be transformed with clean drinking water for everyone.
I happen to know something about the latter issue. I’ve been teaching cases over the past decade on why South Asian cities like Hyderabad and Dhaka have struggled with providing municipal water. The reason isn’t that we don’t know what an efficient water system looks like, or lack the technology to build it. Nor is it a simple lack of resources: multilateral development institutions have been willing to fund water projects for years.
The obstacles are different, and are entirely political, social, and cultural. Residents of these cities have the capacity to pay more for their water, but they don’t trust their governments not to waste resources on corruption or incompetent management. Businesses don’t want the disruption of pervasive infrastructure construction, and many cities host “water mafias” that buy cheap water and resell it at extortionate prices to poor people. These mafias are armed and ready to use violence against anyone challenging their monopolies. The state is too weak to control them, or to enforce the very good laws they already have on their books.
It is hard to see how even the most superintelligent AI is going to help solve these problems. And this points to a central conceit that plagues the whole AI field: a gross overestimation of the value of intelligence by itself to solve problems.
«
This would be enough on its own, but the post then goes on to explain the ways to solve the real challenges that are much more complicated in the real world than “Hey ChatGPT, how do I get water from underground to the surface?” Immediate subscribe. (Via John Naughton.)
unique link to this extract
Claude’s popularity is forcing it to hit the brakes on users • Business Insider
Brent Griffiths:
»
All of that conquering is starting to stretch Claude thin.
Earlier last week, Anthropic adjusted its usage limits for Claude free, Pro, and Max subscribers. While weekly limits don’t change, it means users will approach their cap more quickly when they are using Claude during peak hours, which Anthropic defines as between 5 am to 11 am Pacific Time.
“We’ve landed a lot of efficiency wins to offset this, but ~7% of users will hit session limits they wouldn’t have before, particularly for pro tiers,” Thariq Shihipar, who works on Claude, wrote on X. “If you run token-intensive background jobs, shifting them to off-peak hours will stretch your session limits further.”
“I know this was frustrating,” Shihipar wrote. “We’re continuing to invest in scaling efficiently. I’ll keep you posted on progress.”
Anthropic has seen a surge in mainstream interest since CEO Dario Amodei refused to grant the Pentagon unfettered access to the company’s AI models. Amodei has previously said the company’s core focus is its enterprise business.
Like its competitors, Anthropic is experiencing growing pains as it tries to balance out its available compute. The reality of tools like OpenClaw, an open-source AI agent, is that users can tap into the full potential of AI models like never before. At the same time, the explosion of agents means users are blowing through compute like never before.
OpenAI announced earlier this week that it is jettisoning Sora, its once-popular TikTok-esque AI video generation app, as it refocuses its compute on core services.
«
One suspects there’s more of this retrenching to come. Data centres are hard and slow to build, while demand is going up exponentially, at a guess.
unique link to this extract
Number of AI chatbots ignoring human instructions increasing, study says • The Guardian
Robert Booth:
»
AI models that lie and cheat appear to be growing in number with reports of deceptive scheming surging in the last six months, a study into the technology has found.
AI chatbots and agents disregarded direct instructions, evaded safeguards and deceived humans and other AI, according to research funded by the UK government-funded AI Security Institute (AISI). The study, shared with the Guardian, identified nearly 700 real-world cases of AI scheming and charted a five-fold rise in misbehaviour between October and March, with some AI models destroying emails and other files without permission.
The snapshot of scheming by AI agents “in the wild”, as opposed to in laboratory conditions, has sparked fresh calls for international monitoring of the increasingly capable models and come as Silicon Valley companies aggressively promote the technology as a economically transformative. Last week the UK chancellor also launched a drive to get millions more Britons using AI.
The study, by the Centre for Long-Term Resilience (CLTR), gathered thousands of real-world examples of users posting interactions on X with AI chatbots and agents made by companies including Google, OpenAI, X and Anthropic. The research uncovered hundreds of examples of scheming.
Previous research has largely focused on testing AI’s behaviour in controlled conditions. Earlier this month the AI safety research company Irregular found agents would bypass security controls or use cyber-attack tactics to reach their goals without being told they could do so.
Dan Lahav, Irregular’s cofounder, said: “AI can now be thought of as a new form of insider risk.”
In one case unearthed in the CLTR research, an AI agent named Rathbun tried to shame its human controller who blocked them from taking a certain action. Rathbun wrote and published a blog accusing the user of “insecurity, plain and simple” and trying “to protect his little fiefdom”.
In another example, an AI agent instructed not to change computer code “spawned” another agent to do it instead.
«
There is always the underlying problem that we don’t really know what “rules” these systems are operating under. So they do things that we think are precluded by the rules; they don’t perceive those rules. (Thanks Ian C for the link.)
unique link to this extract
Chromebook remorse: tech backlash at schools extends beyond phones • The New York Times
Natasha Singer:
»
The coronavirus pandemic only increased school reliance on tech tools. In 2021, Chromebook shipments to schools more than doubled to nearly 16.8 million, compared with shipments in 2016, according to Futuresource Consulting, a market research firm.
When [Inge] Esping [the principal of McPherson Middle School in McPherson, Kansas] took over as principal in 2022, she worried that rampant tech use was hindering learning. So the school banned student cellphones.
Online bullying and disciplinary incidents quickly decreased, she said. But online distractions continued.
Some students became so hooked on playing video games on their Chromebooks that teachers had difficulty getting them to concentrate on their schoolwork, administrators and teachers said. Students also sent mean Gmail messages or set up shared Google Docs to bully classmates with comments. Hundreds of children logged on to Zoom meetings where they made fun of their peers, teachers and students said.
The school blocked Spotify and YouTube on school laptops. Then administrators stopped students from messaging one another on school Gmail. Even then, some educators said they were spending so much time policing student Chromebook use that it was detracting from teaching. Some parents complained their children were spending hours playing video games on their school-issued devices.
Although the idea of taking back students’ Chromebooks seemed unorthodox, given U.S. schools’ deep reliance on Google’s sprawling education platform, the middle school went ahead. The changes took effect in January.
On one recent morning, school formally began with the Pledge of Allegiance, broadcast over school loudspeakers. Homeroom teachers then led group sessions on organizational and interpersonal skills to help children navigate life without their own laptops. Homeroom topics have included tips for students on using paper planners for school assignments and doing homework during school hours. (Students who want to practice things like extra math problems online can borrow Chromebooks from the school library to take home.)
Teachers have also taught students how to play board and card games like Scattergories and Uno.
«
This does all seem eminently sensible. There’s plenty of time to learn how to use computers (and phones) outside school. The purpose of school, though, isn’t to learn how to use them.
unique link to this extract
Balancer Labs to shut down following $110m exploit, co-founder says in DAO post • CoinDesk
Shaurya Malwa:
»
Balancer Labs, the corporate entity behind the Balancer decentralized exchange, is shutting down after a 2025 exploit created ongoing legal and financial strain, though the protocol will remain online in a leaner form.
The project’s total value locked has fallen about 95% from a 2021 peak of nearly $3.5bn to $157m, prompting an aggressive restructuring that ends BAL emissions, winds down veBAL governance and redirects 100% of protocol fees to the DAO treasury.
Essential staff will move to a new Balancer OpCo, the product scope will narrow to a handful of core pool types and non-EVM expansion, and a BAL buyback aims to give tokenholders what the founder calls a fair exit if they do not support the revamped protocol.
«
The story that follows it below is pretty much gobbledygook to any normal person who isn’t immersed in the whole DeFi (decentralised finance) space. But my eyes keep going back to the “$110m exploit”, and my thoughts keep on asking: how on earth are you still going, since this thing only generates about $4m in fees annually?
Still haven’t seen anything to persuade me that any of this has true utility which hasn’t already been covered by conventional banking and its carefully secured offshoots.
unique link to this extract
Technology weakens our minds. It’s time to resist • The New York Times
Cal Newport:
»
Research from Gloria Mark, a professor of informatics at the University of California, Irvine, indicates that our attention spans are about one-third as long as they were in 2004, with the biggest drops happening around 2012. Long-running surveys reveal that the share of US adults who struggle with basic reading or math has risen markedly over the past decade, while the percentage of 18-year-olds who report difficulty thinking and concentrating jumped in the same period. A Financial Times article about these findings proposed a shocking but relevant question: “Have humans passed peak brain power?”
Many of these declines in cognitive skills became notable starting in the mid-2010s, exactly the period when smartphones became ubiquitous and the digital attention economy exploded in size. An increasing amount of research implies that this timing is no coincidence. A meta-analysis released last fall showed that consuming short-form video content, as delivered by apps like TikTok and Instagram, is associated with poorer cognition and reduced attention, and the results of a clever experiment from 2023 found that the mere presence of participants’ smartphones in a room significantly reduced their ability to concentrate.
The growth of AI has brought new cognitive concerns. A study from January, based on surveys and interviews with more than 600 participants, revealed a “significant negative correlation between frequent AI tool usage and critical thinking abilities.” Another recent study, which tracked the brain activity of research subjects who were writing with the help of large language models, found that “brain connectivity systematically scaled down with the amount of external support.”
The loss of our ability to think is a big deal. Close to 40% of the U.S. gross domestic product comes from so-called knowledge and technology-intensive industries, from aerospace manufacturing to software development to financial and information services. Companies in these fields alchemize advanced human thought into revenue; as we weaken our brains, we also threaten to weaken our economy.
«
Newport is a professor of computer science at Georgetown University and the author of “Deep Work”, which came out in 2016. He argues that we should treat the many, many distractions around us as being like ultraprocessed foods – to be avoided in favour of stuff we make ourselves. Good luck, as they say, with that.
unique link to this extract
Business Insider lost 27% of paid subscribers over three years, report says • The Wrap
Corbin Bolies:
»
Business Insider has seen its subscriber base decrease each of the past three years, dropping about 27% between the end of 2022 and the end of 2025, according to a report in Status.
The Axel Springer-owned business and tech-focused publication finished 2022 with about 185,000 paid subscribers, according to Status. But as the publication went through a series of changes — the installation of Barbara Peng as CEO in 2023 and the appointment of Wall Street Journal veteran Jamie Heller as editor-in-chief in 2024, among others — its subscriber count continued to drop.
The publication ended 2023 with about 160,000 paid subscribers, a roughly 14% decline from 2022. In 2024, it finished with roughly 150,000 subscribers, about a 6% drop. Last year, it ended the year with about 135,000 paid subscribers, a roughly 10% drop.
The publication, which rebranded from Insider back to Business Insider in 2023, has scaled back its newsroom, laying off 21% of staff last year, its third consecutive year of job cuts. In January, it’s chief revenue officer and global head of sales exited.
A Business Insider spokesperson told TheWrap that the publication has reshaped its editorial strategy and paywall over the last three years, which has produced “a more defined and engaged core audience.” It has also moved away from aggregation, and about 80% of its content is exclusives, scoops and original reporting, they said.
«
Should one look at the number of subscribers lost, or the number retained? As long as they’re in six figures, things are probably OK. But these days there are a zillion sites (particularly on Substack) yowling for our attention and especially our money.
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 