What is it that railways really produce? It’s not what you might expect, even though you’ve used it. CC-licensed photo by Hugh Llewelyn 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. Ride on time. 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 perils of letting AI plan your next trip • BBC Travel

Lynn Brown:

»

Miguel Angel Gongora Meza, founder and director of Evolution Treks Peru, was in a rural Peruvian town preparing for a trek through the Andes when he overheard a curious conversation. Two unaccompanied tourists were chatting amicably about their plans to hike alone in the mountains to the “Sacred Canyon of Humantay”.  

“They [showed] me the screenshot, confidently written and full of vivid adjectives, [but] it was not true. There is no Sacred Canyon of Humantay!” said Gongora Meza. “The name is a combination of two places that have no relation to the description. The tourist paid nearly $160 (£118) in order to get to a rural road in the environs of Mollepata without a guide or [a destination].”

What’s more, Gongora Meza insisted that this seemingly innocent mistake could have cost these travellers their lives. “This sort of misinformation is perilous in Peru,” he explained. “The elevation, the climatic changes and accessibility [of the] paths have to be planned. When you [use] a program [like ChatGPT], which combines pictures and names to create a fantasy, then you can find yourself at an altitude of 4,000m without oxygen and [phone] signal.”

In just a few years, artificial intelligence (AI) tools like ChatGPT, Microsoft Copilot and Google Gemini have gone from a mere novelty to an integral part of trip planning for millions of people. According to one survey, 30% of international travellers are now using generative AI tools and dedicated travel AI sites such as Wonderplan and Layla to help organise their trips.

While these programs can offer valuable travel tips when they’re working properly, they can also lead people into some frustrating or even dangerous situations when they’re not. This is a lesson some travellers are learning when they arrive at their would-be destination, only to find they’ve been fed incorrect information or steered to a place that only exists in the hard-wired imagination of a robot.

«

Who could have imagined.
unique link to this extract

---

How to try Chrome’s hidden AI model • Pete Warden’s blog

Pete Warden:

»

There’s an LLM hiding in Chrome. Buried in the browser’s basement, behind a door with a “Beware of Leopard” sign.

But I’ll show you how to find it. In a couple minutes, you’ll have a private, free chatbot running on your machine.

We’re going to enable some developer flags in desktop Chrome so you can get full access to the AI model. We have to do this because the functionality is only being slowly rolled out by Google, and by turning on these developer options we can skip to the front of the line. There’s also a screencast version of these instructions if you’d like to follow along on YouTube.

You’ll need access to Chrome’s internal debugging pages to try out the model, so enter chrome://chrome-urls/ into the URL bar, scroll down, and click on “Enable internal debugging pages”.

[Instructions in post.]

…Why does this matter?

It’s free: These models work with the PC you have and require no subscriptions. Your usage is only limited by the speed of the model.

It’s 100% privacy-safe: None of your questions or answers leave your PC. Go ahead, turn off your WiFi and start prompting – everything works perfectly.

It works offline: The first time I used a local model to help with a coding task while flying on an airplane without WiFi, it felt like magic. There’s something crazy about the amount of knowledge these models condense into a handful of gigabytes.

It’s educational: This is the main reason you should bother with local LLMs right now. Just trying out this model demystifies the field, and should be an antidote to the constant hype the AI industry fosters. By getting your hands just slightly dirty, you’ll start to understand the real-world trajectory of these things.

It’s the future: Local models are only getting better and faster, while cloud-based chatbots like Claude and ChatGPT plateau. The market is inevitably going to shift to free models like this that are integrated into platforms and operating systems.

«

Pete has been exploring machine learning systems for more than a decade. And if he says it matters, it really matters. It’s also very neat. As much as anything, it might teach people the limitations of chatbots if they’re using one in their browser all the time.
unique link to this extract

---

Apologies: you have reached the end of your free-trial period of America! • The Atlantic

Alexandra Petri:

»

We were so excited to reach 340 million free users. But now it’s time to streamline our product so that it appeals more to paid subscribers, and that means some changes for everyone else. We are adding a lot of features no one asked for that will make your experience worse and also cost a lot of money! Freedom isn’t free! Nor is it, exactly, the freedom you’ve been used to! Yes, that is the National Guard in your city. We know that you didn’t request it; it’s just a new feature we’re rolling out, possibly for 30 days, possibly for even longer!

You were pretty vocal about what attracted you to America in the first place: personal liberty, economic opportunity, something called the American dream, and, of course, the perennial threat of gun violence. (That last feature developed over time, but it seems that our users are pretty attached? We offered you many opportunities to opt out.) But we knew what was really keeping you here: inertia, and the challenge of finding an alternative that sells decent breakfast burritos. We are banking on that going forward.

…Our new CEO does hate a large portion of our current user base, but he’s not totally ignorant of the culture here. He is very excited to bring back some things that past users described as “great,” such as Depression, Recession, and White-Shark Attacks. It was also his brilliant idea to add the features of autocracy—State Control of Business, General Encouragement of Groveling, Masked Men Who Yank Your Neighbors Into an Unmarked Van to Whisk Them Off to a Gulag—to our core democratic product.

You heard it right: The government you knew for Weather Data and Medical Research is going all in on Despot Whims. This costs money, so bedrock features such as Separation of Powers, No Troop Quartering, and Due Process are being phased out, even for premium subscribers. We are also getting rid of most of our Health and Science. But you can have a career in ICE.

We are retaining some features for premium users. Want rule of law? That’s premium. The right to run your company without government interference? That’s a paid feature now. An explanation from the Supreme Court as to why it just ruled against you? Maybe!

«

Petri used to write her brilliant humour for the Washington Post. But you can see why it just wouldn’t fit there now.
unique link to this extract

---

Elon Musk fans are still losing ridiculous amounts of money to crypto scams • Gizmodo

Matt Novak:

»

Gizmodo filed a Freedom of Information Act (FOIA) request with the FTC for complaints from people who say they were tricked by scammers posing as Musk or who used online ads with the billionaire’s likeness. Gizmodo has filed such FOIA requests with the FTC before, and it’s instructive to learn what new tricks and tools scammers are using to lure their victims.

Some scammers appear to convince victims they’re actually Elon Musk, even bringing in supporting characters like Elon’s mother, Maye Musk, to vouch for a given “investment.” Other times, scammers will use AI-generated photos or videos to make it appear Elon is promoting a given cryptocurrency or online platform where someone can supposedly get rich.

Some of the stories are heartbreaking, like the woman in Florida who filed a complaint with the FTC because her elderly husband has cancer and isn’t thinking clearly as he engages with a person he’s convinced is the real Elon. She writes that her husband, who’s in his 70s, is “heavily medicated and does not believe he is being scammed.” He lost at least $10,000.

Scammers frequently set up YouTube livestreams during major SpaceX or Tesla events, making them appear as official as possible, but imploring people to send cryptocurrency that will be “matched” by Musk or his companies. A Florida man in his 60s wrote to the FTC about getting scammed out of $225,000 that way, sending crypto to Coinbase earlier this year.

The scams are all from the past year, meaning that there are sometimes cameos by President Donald Trump. Like the complaint from Michigan, where a person in their 70s who lost $10,000 was convinced it had to be Musk because, “His profile picture was of him on Airforce 1 [sic] with Trump. He presented detailed information that Musk would know.”

«

It is so dismaying how easily people are taken in by this sort of stuff. There must be a particular mindset that makes people believe Musk would get in touch. Then again, he does discuss things with completely random people on X, so perhaps that part of the scam isn’t so unbelievable.
unique link to this extract

---

The product of the railways is the timetable • Benedict’s Substack

Benedict Springbett:

»

This post is about a simple, yet crucial insight: the product of the railways is the timetable. Not the tracks, not the trains, but the timetable.

First ask yourself: what is the purpose of a railway? The job of the railways is to move people around. A person gets a train because they want to travel from A to B.

A small group of enthusiasts (me included) will actively choose to get the train if at all possible, but the vast majority of people are not particularly loyal to how their user need is met. They might drive, they might fly, they might cycle, they might decide not to make the journey and just do a video call. All of these things are the competitors of the railways.

We could therefore say that the product of the railways (and roads, and airports) is travel. But we can be more specific than that. A railway is not like a road. A road is built, and then it is open for anybody to use it at any time. There is no need to plan out precisely when cars move along the road. The movements of trains, by contrast, have to be planned out months in advance. It would neither be possible nor sensible to run trains ad hoc. They are not taxis, free to roam the roads whenever they like. Railway tracks are a network; everything depends on everything else. The service from Cambridge to Norwich affects the service from Norwich to London Liverpool Street, which in turn affects the service from Liverpool Street to Southend. To optimise the use of the tracks, train movements have to be planned out well in advance with precision.

This planning is what we call the timetable, the mapping between space and time that determines which train occupies which track at which time. The railways offer travel to the public via the timetable: a traveller buys the (supposed) fact that the 12.32 from Reigate gets into London Victoria at 13.19. The product of the railways is the timetable.

«

This is rather like the insight that people are buying holes in the wall when they purchase a drill: it produces a different way of seeing what people want. If the timetable is too confusing (including having puzzling connections or weird ticket pricing schemes), fewer people will take the train. That’s not because the trains aren’t running at the right times.
unique link to this extract

---

BBC apologises for airing fake claim about Digital ID scheme • Politics Home

Zoe Crowther:

»

The BBC has apologised for airing a false claim about Tony Blair’s son’s company being awarded a government contract to produce the new mandatory digital ID scheme.

The episode of comedy quiz show Have I Got News For You was taken down from BBC iPlayer on Saturday morning, before being put up again with the false claim edited out.

Last week, the government announced a new digital ID scheme, which it said would help combat illegal working and make it easier for the public to use vital government services. It will be mandatory for Right to Work checks by the end of this Parliament.

On Friday evening, the BBC’s Have I Got News For You host Victoria Coren Mitchell incorrectly said that Multiverse – owned by Euan Blair, former prime minister Tony Blair’s son – was producing the digital ID scheme, which she described as a “happy coincidence”.

This claim is untrue and was fact-checked by the independent organisation Full Fact earlier this week. Blair’s company, Multiverse, does not develop its own software, but offers apprenticeship programmes and training on AI and tech.

Both Multiverse and the Department for Science, Innovation and Technology have confirmed there is “no truth” to the claims that the company is involved in any way. PoliticsHome understands that no decisions have been made regarding which third-party companies will be awarded contracts associated with the scheme, which is still in the early stages of development and will be subject to a public consultation later this year.

«

This was a pretty egregious error, from a rumour which originated on social media and which – as the story notes – had been factchecked already. It would be surprising, though, if Digital ID survives the consultation.

The story also has this at the bottom: Correction: Victoria Coren’s name was misspelled in an earlier version of this post. Oh, irony, thou are ubiquitous. By the way, all the mistakes in this newsletter are intentional.
unique link to this extract

---

Finland unable to prosecute alleged Baltic Sea cable-cutters, court rules • Financial Times

Richard Milne:

»

Finland lacks the jurisdiction to prosecute sailors accused of cutting underwater electricity and data cables in one of the most high-profile recent sabotage events in the Baltic Sea area, a court in Helsinki ruled.

The case against the captain of the Eagle S and two pilots was dismissed on Friday by Helsinki district court as it found the severed cables were outside Finland’s territorial waters, highlighting the difficulties of holding those accused of sabotage and hybrid attacks accountable.

Finland received plaudits after armed border guards abseiled on to the deck of the Eagle S on Boxing Day and seized control of the tanker after it had dragged its anchor along the Baltic Sea for almost 100km.

It became the first Nato country to bring charges against the crew of a “shadow fleet” vessel, used by Russia to circumvent western sanctions on its oil trade, and prosecutors wanted two-and-a-half years’ jail time for the trio.

But lawyers for captain Davit Vadatchkor and pilots Robert Egizaryan and Santosh Kumar Chaurasia argued that Finland had no jurisdiction as the alleged cable-cutting took place in international waters and the Eagle S was only stopped once it turned into Finnish territory.

Europe has been hit by a wave of sabotage and hybrid attacks, some attributed to Russia and some merely suspected as being ordered by them.

«

unique link to this extract

---

Why Apple should steal the Fairphone 6 Moments switch for the iPhone • Stuff

Craig Grannell:

»

It felt quite magical when a tame Fairphone rep flicked a switch on the side of the [Android] device and instantly transformed the home screen. The grid of icons vanished, replaced by a boring list of app names. And, to be clear, boring in this context is good. It means a phone without distractions. A device that lets you be present, rather than tempting you every waking moment with yet more shiny icons.

Of course, there’s nothing complicated happening here. Fairphone just mapped a switch to a launcher. But the implementation is everything. The physicality – the deliberate act of making your phone less distracting – made something in my brain go ding. It reminded me of Bear Focus Timer, a Pomodoro app that only works when your phone is face down.

The ritual matters more than the mechanism – it’s a psychological thing. You flip the phone over to put it “out of reach”. You commit to focus. It’s far more effective than tapping a virtual button. And so it felt with the Fairphone 6, even if, alas, that device lacks scowling cartoon bears should you abandon focus mode.

It also reminded me of something else: my iPhone 16 Pro has a perfectly serviceable Action button that I barely use. Cue: epiphany! Why not shamelessly steal Fairphone’s great idea and bodge it into iOS?

I duly set up my standard iOS home screen to feature a single Dumb Phone widget with a few apps I use, and turned all the others off. Then I created a new one called Badness, with a dozen home screens of icons in all their appy glory.

«

He created a Shortcut (a short script) to switch between modes – icons, or app name list – and linked it to the Action button to create a quick switching method. If you find your phone too distracting, it might work for you too. (You might be able to create a Focus mode that does the same.)
unique link to this extract

---

X-ray scans reveal the hidden risks of cheap batteries • The Verge

Andrew Liszewski:

»

Lumafield has released the results of a new study of lithium-ion batteries that “reveals an enormous gap in quality between brand-name batteries and low-cost cells” that are readily available through online stores including Amazon and Temu. The company used its computed tomography (CT) scanners, capable of peering inside objects in 3D using X-rays, to analyze over 1,000 lithium-ion batteries. It found dangerous manufacturing defects in low-cost and counterfeit batteries that could potentially lead to fires and explosions.

The study tested 18650 lithium-ion battery cells, which are used in various products including electric toothbrushes, power tools, e-bikes, power banks, and even electric cars. The batteries were purchased from ten different brands: three OEMs, including Samsung and Panasonic, sourced from “highly reviewed, specialized suppliers,” three vendors selling rewraps (typically OEM batteries with their plastic outer wrapping replaced) sourced from “specialized battery sites” or the brand’s own web stores, and four companies selling low-cost or counterfeit batteries sourced from “large, general online retailers such as Temu.”

Lumafield scanned 1,054 batteries – around 100 from each brand – and found 33 of them had a serious manufacturing defect known as negative anode overhang. The defect “significantly increases the risk of internal short-circuiting and battery fires” and can reduce the overall life of the battery,” according to Lumafield. All 33 of the batteries with the defects came from the 424 sold by low-cost brands or brands selling counterfeits. One of the counterfeits was identified by its pink wrapper designed to match the one Samsung uses on its 30Q cells.

«

Cheap can be expensive, particularly if it burns down your house.
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