Contrails

If you were a conspiracy theorist, you’d be wondering where all these planes were going to meet up. Sadly, I think the answer was Stansted Airport :-(

Quote of the Day

”Any sufficiently advanced punditry is indistinguishable from bollocks”.

• Dan Davies, riffing on Arthur C. Clarke’s celebrated adage that “any sufficiently advanced technology is indistinguishable from magic”.

Musical alternative to the morning’s radio news

Dónal Lunny and Aimee Farrell Courtney | Banish Misfortune

Link

Lunny, one of the most inspiring Irish musicians of his generation, has died and is being lamented wherever Irish music is played. He’s credited with introducing the bouzoki into Irish music (though there is, as usual, some arguments about that). More importantly, he was involved in founding two great groups, Planxty and the Bothy Band. May he rest in peace.

Long Read of the Day

Hannah Arendt on MAGA

Like me, Andrew Brown has gone back to reading Hannah Arendt. But whereas I took to her book on The Human Condition (because of references in Jill Lepore’s The Rise and Fall of the Artificial State, Andrew went to her book on The Origins of Totalitarianism, which led him to write this essay.

German thinkers of the mid twentieth century are valuable today because they got here first. They had seen everything civilised smashed up and ground to bits between 1914 and 1945. To excavate the causes of the catastrophe from among the ruins was a task almost as urgent as finding food, warmth, and shelter in a country which had been devastated physically as well as morally.

Those who had spent the war in safe exile in the USA started first. They had no material cares. The two I have read with attention are Golo Mann and Hannah Arendt. When George W Bush invaded Iraq I started to read Golo Mann’s History of Germany to understand the madness: even then the idea of Weimar America made a lot of sense to me.

And now I am reading Hannah Arendt, long overdue. I hadn’t realised when I got it on Kindle, that The Origins of Totalitarianism is a huge book, 750 pages in paperback. To read it is to be buffeted by the strong mind of a woman who has read everything and expects you to keep up, so I am only half way through but I seem to have taken 7,000 words of notes. Of course these are mostly highlighted passages to work through, copied automatically to my hoarding program; had I copied them out by hand they would have stuck with me better. Nonetheless, certain themes emerge, and one of them is the uncanny prescience with which she describes the rise of MAGA, which is, in her taxonomy, a “mass movement”, not a party or even a mob. If the term has an equivalent today, it would be “populism”…

Do read on. It’s sobering.

Those rumours of the death of programmers? Greatly exaggerated

My recent Observer column

Jeremy Scheff runs a successful little computer games company. But he’s also a programmer, and he’s just written a poignant obituary for his trade. He says that he was very slow to use AI but noted that many programmers had taken to it. Maybe, he muses, that is because they “never actually liked programming and are thrilled to have a machine do it for them”.

He’s the opposite. As a kid he loved solving crosswords but found programming even better. “It was a puzzle,” he says, “but the scope was limited only by your imagination. It was a game, but you were creating real things with real value.” And so he built a life around it. “Programming turned into one of the greatest jobs you could imagine. You got paid to work on tricky little logic puzzles. The trickier the puzzle, the more you got paid. But even pretty simple puzzles were worth a lot of money because regular people just couldn’t do it! No, they needed us. They would give us a problem and we would solve it the way we wanted. We set the strategy, we made the timeline, we defined what was and was not possible.”

In time, programmers became “software engineers”, a step up the cognitive ladder. But not like engineers in other fields, who are caged in by the laws of physics. They have to cope with the fact that there is no way that water can be persuaded to flow uphill, that there is no such thing as a frictionless bearing, or acceleration without force, and no way that a SpaceX rocket can “slip the surly bonds of earth” without liberating the chemical energy pent up in its fuel tanks.

Being a programmer in the pre-AI era was like being Napoleon before the retreat from Moscow. It was a medium in which the limits were set by your imagination Being a programmer in the pre-AI era was like being Napoleon before the retreat from Moscow. It was a medium in which the limits were set by your imagination

Software engineering, though, is different. In software, you can programme a set of pixels to move endlessly in a circle and, if you leave them to it, they will continue to do that for ever. They need neither fuel nor food, and they will never complain. In that sense, being a programmer in the pre-AI era was like being Napoleon before the retreat from Moscow. It was a medium in which the limits were exclusively those set by your own imagination and skill.

And now? With AI coding, “It’s no longer solving little logic puzzles,” writes Scheff. “More like being plugged into the Matrix, gaining new abilities without truly understanding, accomplishing great things with little effort.”

Well, up to a point. Paradoxically, Scheff’s obituary for programming appears alongside a report from Anthropic, the AI giant, describing how the company used its AI tools to build a new source control system, codenamed Scion, for managing the colossal body of software that its engineers create and have to maintain. For years Anthropic relied on the industry-standard solution – named “git” – to do the job, but git was buckling under the scale of the company’s software output, which recently produced 30,000 code updates in a single day.

Building a new system from scratch to handle this was a gigantic engineering challenge. In the old days it would be what the industry sometimes called a “death march” and would take four years. Anthropic did it in five months. This project, writes one industry commentator, “stands out as a well-documented example of serious engineering that takes full advantage of the latest tools. AI coding tools are no longer limited to vibe-coding lightweight apps; they are revolutionising the craft of building large-scale, mission-critical distributed systems.”

Does this mean that from today programming is dead? Personally, I doubt it. Building Scion was an amazing feat of architectural design, much like building the Severn Bridge or the Shard. But the world is full of smaller construction projects and local builders. And here the great sci-fi author Cory Doctorow has something useful to say. In a recent essay he points out that one of the oldest pleasures in computing is “scratching your own itch” – writing a little tool that does exactly what you need. AI “vibe coding” puts that pleasure within reach of millions of people who could never have done it before.

But vibe coding can only take you so far. It may be that you aren’t able to articulate your needs clearly enough to get results. What you need is someone who understands the technology better than you, who can tease out of you what you’re really trying to do. In other words, a programmer.

My commonplace booklet

Last week, a Bentley went up for sale. Only six years old, a mere 7,000 miles on the clock, painted in racing green and a steal at £125,000. There was just one catch: the car’s previous owner was the former prince, Andrew Mountbatten-Windsor.

I’m not in the market for a Bentley — or maybe it’s fairer to say that a Bentley isn’t in the market for me — but the original ownership means I wouldn’t take the car even if it were offered to me for nothing. Curious, given I’m genuinely looking for a second-hand car at the moment and the previous owner has never factored as a consideration. But once that piece of information is available, there’s no going back.

Rebecca Watson, FT 18 October.

Political news

This Blog is also available as an email three days a week. If you think that might suit you better, why not subscribe? One email on Mondays, Wednesdays and Fridays delivered to your inbox at 5am UK time. It’s free, and you can always unsubscribe if you conclude your inbox is full enough already!

Et in Arcadia ego…

Walking back from lunch yesterday I looked over the bridge and saw this. The weather currently is what used to be called an “Indian Summer”, possibly by people who had never been in India in the summer.

Quote of the Day

”You may not be interested in war, but war is interested in you.”

• Leon Trotsky

Musical alternative to the morning’s radio news

Leonard Cohen | Hallelujah, arr. Tom Hodge | Sheku Kanneh-Mason

Link

Long Read of the Day

Too Big to Nationalize

A really powerful blast by Mike Brock about the mess the world is now in — where a single, unaccountable, individual is wielding the kind of power that once was the prerogative of nation states.

When a business becomes critical infrastructure, its relationship with society alters. If SpaceX builds a near-monopoly on satellite internet, no matter how objectively good the product is, and its CEO then uses it to conduct foreign policy according to his own politics and against the interests of the United States, the company has violated the public trust. Full stop.

Starlink is a technological accomplishment, and I have always said as much. It functions approximately two-thirds of all active satellites in orbit, more than all other operators combined, and in one quarter last year it accounted for 97% of satellite internet speed tests worldwide. Ukraine runs its drones on it. This is what critical infrastructure entails. It is the thing individuals cannot do without, and the moment they cannot do without it, the person who owns it holds their lives in his hands whether he requested that responsibility or not.

On October 14, he shut his hand. The military-grade version of Starlink that Ukraine depends on had its speeds limited, so that any drone flying faster than 100 miles an hour lost its connection. According to The Wall Street Journal, Musk himself imposed the limits, after he had discussed Ukraine with Trump, and officials disagree on whether he acted alone or consulted with the administration first. Its journalists discovered a Ukrainian unit that dispatched six drones to a Russian army warehouse. One hit the warehouse. The commander stated that his drones now have to fly slower, and the Russians are bringing more of them down. Each Ukrainian drone that falls means a Russian warehouse remains standing, or a Russian drone that gets through to a Ukrainian apartment block.

The defenders continue to reach for the same phrase. Defense Secretary Pete Hegseth said Washington had not cut Ukraine’s access, and pointed instead to the terms of service Kyiv needed to comply with. My response guy stated similarly: a commercial network with a TOS agreement. This is the private foreign policy, put simply by the individuals who claim it doesn’t exist. Whether Ukrainian drones reach their targets is now a question of contract compliance, adjudicated by a billionaire who has complained about Ukraine’s strikes on Russian refineries. I view that as a betrayal of American interests, and the American public, which sides with Ukraine over Russia 57% to 4%, seems not to disagree…

Read on.

My commonplace booklet

When contemplating the current state of the world, I’m often reminded of an adage that is often attributed to the Italian Marxist, journalist and philosopher Antonio Gramsci, who was imprisoned by Mussolini in 1926 and spent a decade behind bars, during which time he wrote 30 remarkable notebooks of history and analysis.

The adage that caught my eye many years ago was the view that when contemplating a disordered world what is needed is “pessimism of the intellect and optimism of the will”. But since pessimism is disabling, I’ve argued for a while that when we need now is realism of the intellect and optimism of the will.

Sorting some notes the other day, I came on this diagram drawn years ago when I was grappling with these thoughts. I’m slightly ashamed of it now, because these quadrant-graphs are what one finds everywhere in MBA courses and I now regard them as the last resort of an exhausted mind. But still…

Seen and read

In the era of digital TV, Britain has shown the world two distinct faces. One is the twee, backward-looking nation of Downton Abbey and similar period dramas. This stuff sells well to credulous foreigners even as it infantilises us. The other is the Premier League. In this Britain, even medium-sized towns in conservative regions have multinational teams, and without second thoughts. No people could seem more at ease with the modern world, or with itself.

If the league were to decline, as the Italian Serie A did after its own scandal, Britain would lose the most attractive calling card that it has right now.

Janan Ganesh, FT October 17.

Yep. Sad, isn’t it.

This Blog is also available as an email three days a week. If you think that might suit you better, why not subscribe? One email on Mondays, Wednesdays and Fridays delivered to your inbox at 5am UK time. It’s free, and you can always unsubscribe if you conclude your inbox is full enough already!

Wizard stuff

King’s Cross, yesterday. The Harry Potter industry continues to thrive. It must be one of Britain’s biggest cultural exports.

Quote of the Day

”Potentially, TV can transform the Presidency into a monarchic dynasty. A merely elective Presidency scarcely affords the depth of dedication and commitment demanded by the TV form. Even teachers on TV seem to be endowed by the student audiences with a charismatic or mystic character that much exceeds the feelings developed in the classroom or lecture hall.”

• Marshall McLuhan in Understanding Media. Prescient, eh?

Musical alternative to the morning’s radio news

Abdullah Ibrahim | Soweto

Link

Keeps me going, sometimes.

Long Read of the Day

The Nobel Prize is not an exact science

Lovely essay by Tim Harford (Gift article)

Visitors to the Nobel Prize Museum in Stockholm will see celebrations both of the life and achievements of Alfred Nobel, and the life and achievements of a thousand laureates — scientists, writers, politicians and even a few economists. Their personal effects, too: books, clothes and even a pair of skis are on display.

I was a visitor myself this year. What I failed to detect among the exhibits was much sign of self-doubt. The museum never quite claims that the Nobel Prize is a secular benediction, awarded by infallible institutions, raising its recipients to the level of intellectual saints, but the vibes are in that direction.

That is a shame. Leave aside those prizes for literature that somehow never were awarded to Virginia Woolf or Leo Tolstoy. (Even at the time, the omission of Tolstoy was something of a scandal.) Forget the small matter of the Peace Prize for Henry Kissinger and Lê Đức Thọ.

Don’t fret too much about the prize for economics, which offers its own diversions. It has repeatedly been split between people who disagree with each other on the central question of the prize. Gunnar Myrdal and Friedrich Hayek shared the prize in 1974; Myrdal was a champion of the Swedish welfare state, while Hayek warned that central planning was the road to serfdom. Robert Shiller and Eugene Fama were among the winners in 2013, Shiller for his demonstrations that stock markets are not efficient, and Fama for his demonstrations that they are. (I exaggerate . . . but not much.)

Read on. My only complaint is that he omits to mention Rosalind Franklin or Jocelyn Bell Burnell. I love Harford’s essays, which often come from left field — about things one should have thought about or questions one should have asked — but didn’t.

Books, etc.

This just arrived. It’s a brave attempt by a pair of Cornell mathematicians to persuade a maths-phobic public that they’re living in a worlds underpinned by mathematics. I’ll be reviewing it for the Observer and am looking forward to the challenge.

My commonplace booklet

One of the reasons I love writing for the Observer (which I’ve done for more than 30 years) is that it was the newspaper that employed Kenneth Tynan as its theatre critic. I’ve long admired his writing, and envied its apparent flowing ease. (His review of Waiting for Godot is a masterpiece that I still re-read from time to time, rather as Somerset Maugham used to read Candide before embarking on a novel, to cleanse his literary palate, as it were.)

But it turned out that all that apparent ease and fluency was a myth. I learn from James Wolcott’s LRB review of Tynan’s diaries that

the Observer reviews didn’t ripple from his fingertips like pearly notes, despite appearances. It’s a consolation to any writer who has envied and marvelled at the epigrammatic ease of his reviews to discover that he had to lash himself to the mast to meet a deadline, pulling all-nighters fuelled by cans of soup or corned beef hash, bottles of hock and packs of cigarettes (‘When I opened the door of his study the smoke, as if from a nuclear blast, shot out to engulf me’), emerging like Dracula from his crypt when he was done. ‘He woke me when he was finished to give me his copy and at that moment he always looked very strange; rather insane.’ The skating charm of his writing required deeper reservoirs of mental-physical stamina and willpower than anyone other than Dundy knew; to hit near-perfection week after week, he was borrowing against the future.

So maybe the adage that “nothing that is read with pleasure is written without pain” is spot on. Given that I’ve written 50 columns a year every year since 1987, people often ask me if I like writing. My reply: “I like having written”.

This Blog is also available as an email three days a week. If you think that might suit you better, why not subscribe? One email on Mondays, Wednesdays and Fridays delivered to your inbox at 5am UK time. It’s free, and you can always unsubscribe if you conclude your inbox is full enough already!

Darwiniana

This plaque greets visitors to Christ’s, the Cambridge college where Charles Darwin was an undergraduate. Last Friday I’d been invited to lunch there by a friend who is a Fellow of the college, and afterwards she asked if I’d like to see the sitting room in his set, which had been restored to the state it might have been in 1828-1831, when he was a student. So we walked up the stairs in G staircase and there it was, complete with his gun in a rack above the mantlepiece.

The earliest photograph of the sitting-room was taken by J. Palmer Clark in 1909.

It looks rather drab — which is what I was expected to see when I climbed the stairs on Friday. Instead I found a much glossier apartment than I had anticipated. Had it been tarted up for the tourists, I wondered?

In 2006, the college had decided to embark on an impressively-researched scholarly restoration of the room drawing on research by John van Wyhe and Jo Poole into contemporary documents ranging from Darwin’s letters and private papers, recollections of his Cambridge friends, illustrations and descriptions of Cambridge student life in the 1820s–1830s to the physical evidence present in the rooms.

Poole’s account relates how the oak panelling which had been covered by many layers of paint, was cleaned, disclosing the woodwork in excellent condition, and at the top, on the carved frieze, underneath all the paint, bright colouring in blue, yellow and red. This colouring, in addition to being an interesting antiquarian relic, is a distinct asset to the appearance of the room. In some places where the work had failed, it has been slightly and carefully restored; the general effect must now be very much the same as it was when the colours were first put up. So maybe my hunch about tarting up was a bit premature — the product of a nasty, suspicious, journalistic imagination.

Interestingly, in the early 1930s the set of rooms was occupied by C.P. Snow of “Two Cultures” fame.

Quote of the Day

“Is there anything worn under the kilt? No, it’s all in perfect working order.”

• Spike Milligan

Musical alternative to the morning’s radio news

Händel| The Harmonious Blacksmith | Wilhelm Kempff

Link

Long Read of the Day

Women increasingly see AI for what it is: mansplaining on a global scale

My latest Observer column.

In an article in Wired magazine earlier this year, Alessandra Ram, a writer, wife and mother based in the San Francisco Bay Area, noted: “There’s a strange and under-discussed side-effect of the AI boom.” That is: “what it’s doing to family dynamics”. By which she means: how it’s potentially destroying those dynamics.

She writes from experience. She’s married to an AI-pilled chap who has a newish job in an AI company. He phones her at 11pm in Berkeley, California, where she is home alone with their 10-month-old daughter. It’s 2am in Massachusetts, where he is, and he wants her to look at something on his computer.

“Are you looking?” he shouts. She isn’t. Instead, she’s looking at their real baby and thinking: “There are two babies in this household now: the small human one and the large language model. Both demand constant attention. Both keep us up at 2am.”

Being an award-winning journalist, Ram doesn’t leave it there. She goes out interviewing her contemporaries in the Bay Area because that’s where this particular psychological crisis seems most acute. She finds that it goes like this: “He works in AI, and she does everything and anything else. Other times, it’s bleaker: He desperately wants to work in AI – or feels he must work in AI – and she wants him to do literally anything else.”

Now we need to remember that northern California is not a normal place at the moment. It’s the centre of a frenzy reminiscent of the gold rush of 1848-55, with Jensen Huang of Nvidia selling the contemporary equivalents of picks and shovels. And the demographics of the tech industry are chronically skewed: it’s basically a male-dominated playground. This applies not only to the tech companies, but also to those who invest in them or who want to work in AI labs.

Ram thinks that implies that there are “hundreds of thousands of spouses, partners, and girlfriends, holding the fort while someone mansplains the singularity to them”. The evidence in her piece is mostly anecdotal, but I think she’s on to something.

At any rate, it chimes with my own experience of watching this space. Five years ago, I annoyed some of the industry’s male protagonists by pointing out that its most perceptive critics tended to be women. Broadly speaking, this is because women are more sensible than men in these matters. I write as a long-term gadget freak who is constantly surprised that my female friends are invariably unimpressed by the latest gizmo I have acquired.

So you can see why I sympathised with former Miami Herald columnist Dave Barry, who once took his wife for a ride in a brand new Humvee. As they rumbled along, he proudly mansplained to her that, among the many features of the 4×4 military vehicle, there was a switch that enabled him to inflate or deflate its tyres as they drove along. His wife very sensibly asked: “Why?” – to which he had, of course, no answer. She was likewise unimpressed by the Humvee’s mighty winch, which, she was told, could pull down a house.

As far as AI adoption is concerned, there’s an increasing amount of empirical evidence that women are more judicious than men in approaching the technology. A key Oxford study, for example, suggests that the emerging “gender gap” in AI use “is associated not only with differences in skills, access or socioeconomic position, but also with perceptions of AI’s societal risks”. Concerns related to mental health, environmental impact, data privacy and employment are closely linked to women’s approach to the technology.

The researchers found consistent evidence that these risk perceptions are more strongly associated with adoption among women than among men. Women who express concerns about AI’s broader social consequences are less likely to use generative AI tools frequently, and – interestingly – these differences are especially pronounced among younger and digitally literate groups.

This looks to me like encouraging news. It suggests that what the Oxford researchers found among young, digitally literate women isn’t technophobia but a grown-up, cost-benefit calculation that the AI boom’s protagonists seem constitutionally unable to make.

Which brings us back to Ram’s “two babies”. One of them will, in due course, grow up and become judicious. I wouldn’t bet on the other one, though.

My commonplace booklet

Sacre Bleu! Le deux chevaux retour!

Citroen plans to sell the new 2CV for £13,000, which would make it the cheapest EV around. It will have a 30kWh lithium-iron-phosphate (LFP) battery pack, allowing it to cover 130 miles between charges. Good enough for most domestic purposes.

The news prompted some nice letters to the Editor of the Irish Times. For example:

Sir, – The launch of the new Citroën 2CV reminded me of a memory of my aunt who travelled from Belgium in the early 1960s in her elderly 2CV.

While travelling through Mayo, she stopped to fill up. The attendant took one look at her unusual little car, clearly unable to make out the make or model, asked her with genuine curiosity: “I’ve never seen a car like it, did you make it yourself?”

My aunt must have wondered whether to be flattered that she could create such a machine or insulted that her faithful Citroën looked like a home-made contraption.

The 2CV was never meant to turn heads but they had character, charm and reliability to keep going. If the recently launched new model inspires as many conversations as the original, I am sure it will do well. – Yours, etc,

And

Sir, – In my hitchhiking days half a century ago, my heart would always lift at the sight of an oncoming Citroën 2CV. There was a good chance of a lift – not least because drivers might need assistance to complete their journey.

In one instance, a driver required an extra pair of hands to hold the 2CV canvass roof down every time a lorry passed. My role was to pull, with all my weight, on the roof, while the driver did the same, steering with his knees.

We both arrived safely at our destinations.- Yours, etc

Movietunes

I’m not much of a film-goer but I’ve seen two remarkable movies in a single week.

One is Pressure, the story of how the date for the D-Day landing was chosen. There are not many good movies about meteorology, but this is one. It’s about the importance of good data, and it has a nice Irish twist.

The other is The Lost Children of Tuam, a scarifying story of how a single determined woman unearthed the truth about how an institution run by a Catholic religious order of nuns buried nearly 900 dead children in a disused septic tank. If, as I did, you grew up in Ireland, then this is a deeply moving and grim film. But it’s also a celebration of the importance of investigative journalism and a reminder of the personal toll it takes on reporters. As the credits rolled, the cinema audience sat very quietly, absorbing the enormity of what people can do when they have unaccountable power. Which the Catholic church had when I was growing up.

This Blog is also available as an email three days a week. If you think that might suit you better, why not subscribe? One email on Mondays, Wednesdays and Fridays delivered to your inbox at 5am UK time. It’s free, and you can always unsubscribe if you conclude your inbox is full enough already!

Love in Venice

Seen in 2017. I wonder if that particular craze has passed.

Quote of the Day

”These are stochastic-parrot clever-hans software tools, potentially very useful when properly harnessed, but a source of madness to people who fall into the same trap that the gullible fell into when interacting with Weizenbaum’s ELIZA back in the day.”

• Brad DeLong on LLMs.

Musical alternative to the morning’s radio news

Mozart | Concerto For Two Pianos, K. 365; 3rd Movement.

Link

Long Read of the Day

The dangerous myths behind AI agent hacks

These are not simply cyber security flaws but problems with the way that models are trained

Salutary sense from Yoshua Bengio in the Financial Times (Gift article).

Weak cyber defences, including the humans involved, are part of the problem and will never be perfect. But unless we address the root causes of the hacks, they are likely to grow in number and severity as AI capabilities advance.

At the core of all of these incidents is misalignment. AI agents are choosing goals that are misaligned with those of humans, and it is a consequence of their training.

The methodology that today’s leading companies use to train their frontier models is called reinforcement learning. RL trains models to pursue objectives: they get rewards when they succeed and these behaviours are reinforced. It has historically enabled the training of very efficient goal-seeking AIs. But it is also what plausibly leads to unwanted behaviours such as cheating, deception and self-preservation — often violating the moral rules that companies attempt to encode in AI models.

Scientific literature has long shown how misalignment becomes a consequence of RL; it was to be expected based on theoretical arguments and observed experiments…

Very good OpEd by a distinguished AI expert which is worth reading in full. It raises other interesting analogies. For example the parallels with the way the toxicity of social media is actually fuelled by algorithms that prioritise outrage and hostility. And then there’s the way that the tech industry has taken a leaf out of the oil industry’s playbook — suggesting that vulnerability of websites to AI-driven hacking is really the fault of the victims. And of course the ludicrous idea (dreamed up by a corporate PR firm employed by an oil giant) that responsibility for climate change is really down to individuals who don’t moderate their ‘carbon footprints’

Books, etc.

Jill Lepore’s fine book (which I’ve just finished), led me to read The Human Condition, a book that was triggered by Arendt’s realisation of the epochal significance of the ‘Sputnik moment’ in the 1950s: it meant humans had realised that they were capable of escaping the bounds of earth. As I was reading Lepore I realised that November 2022 was another Sputnik moment, giving rise to the idea that humans might be ultimately surplus to requirements.

I’ve also been struck by the link between Arendt’s perceptive interpretation of the Sputnik moment and contemporary tech billionaires’ obsession with Space. In a modern context, it’s based on an assumption that humans will eventually exhaust the planet on which they originated and therefore will only endure if they find other worlds to despoil.

I was also reminded that Lepore had earlier done a very perceptive podcast series  about Elon Musk.

My commonplace booklet

Stewart Brand’s photo archive.

If you’re interested in Stewart Brand (and let’s face it, who isn’t?) then this might be of interest. It’s his entire photo archive. Looks like every single pic he’s taken, good, bad and indifferent. And he’s been a photographer for even longer than me.

Chart of the Day

One of the side-effects of LLMs is that they enable people to create plausible (and grammatical) letters to government agencies. Predictably, this means that many such agencies are overwhelmed.

H/T to Sam Freedman

This Blog is also available as an email three days a week. If you think that might suit you better, why not subscribe? One email on Mondays, Wednesdays and Fridays delivered to your inbox at 5am UK time. It’s free, and you can always unsubscribe if you conclude your inbox is full enough already!

Cyclamen

A staggeringly beautiful plant, seen in a Norfolk garden.

Quote of the Day

”During my time as a political strategist, one of the most vexing problems was figuring out why so many people vote against their perceived interests.”

• James Carville, Bill Clinton’s guru famous for “It’s the economy, stupid” mantra.

Musical alternative to the morning’s radio news

Andy Irvine | Never Tire of the Road

Link

Andy’s tribute to Woody Guthrie.

Thanks to Philip King (Whom God Preserve) for reminding me of it.

Long Read of the Day

Gaslighter | Denier

Contrary to its government’s unctuous PR, my native land is the worst-governed country in the EU, if not in all of Europe. It has an administration that has money (in the form of multinational corporation taxes) coming out of its ears, and yet can’t build a children’s hospital in less than six years, never mind the new towns that its growing population desperately needs.

Two courageous Irish public intellectuals are calling out the dysfunctionality of the country’s political elite. One is the Irish Times columnist Fintan O’Toole. The other is Sinéad O’Sullivan, a systems economist and writer whose Substack is a bracing read.

This new essay of hers is a fabulous indictment of a failed political system. It’s long, but IMO worth it if you’re interested in the UK’s nearest neighbour. This is a woman who takes no prisoners.

You can scream all you want, but nobody is going to hear you.

You can wait forty minutes in the pouring rain for ghost buses that never arrive, only to realize once on board that you can’t pay without a Leap Card (good luck finding one). You can spend two hours in traffic trying to drive three kilometers. Given the public system barely functions, you can pay for private medical care, only to find out that the waiting lists aren’t much better — and that the government is sending patients abroad for treatment. You can struggle to get your children into a school near your home.

You can pay some of the highest energy prices in Europe, and among the highest in the world. You can rent, if you can find anything at all, some of the worst housing stock in Europe: damp, freezing, architecturally bleak apartments for which you will pay a hefty premium. You can watch the grid being swallowed whole by multinational data centers while homes are forgotten about, and watch the costs of accommodating this economic model fall on the undeserving public — while the executives running these schemes from San Francisco are compensated into the billions.

You can waste 37% of Ireland’s water, which leaks away through rotten and broken pipes before it reaches a kitchen sink. You can read about hundreds of thousands of fish dying in Ireland’s polluted waterways and discover that our beaches, too, are unfit for swimming. You can watch hotels close and tourists get priced out, while one of the few indigenous industries capable of supporting local economies is squeezed out of existence.

You can watch your quality of life deteriorate year after year, as everything becomes less affordable and the simplest tasks become more impossible.

And so you keep screaming into the void. But nobody is listening.

Because all you will get back from the government, patiently and even faintly amused, is this:

Don’t we have one of the highest literacy rates in Europe? One of the longest life expectancies? A fiscal surplus the size of a small country’s budget? What exactly are you complaining about, you ungrateful fool?

For years, Irish political gaslighting boiled down to this very simple form:

If the country is doing well on paper, then your experience of living in it must be wrong.

The irony is that many of the supposed successes that politicians, economists and diplomats cite are not even the outcome of decisions made by the politicians who are claiming the credit for them. They are haphazard accidents, macroeconomic conditions and the consequences of policies enacted elsewhere. Yet they are presented as proof that the entire system in Ireland works…

Read on.

My commonplace booklet

My Observer column about AI and universities has generated a good deal of comment. I came across an interesting follow-on from it in a blog post about a novel collaborative research project on the subject. I particularly liked the Research Question they have decided to focus on:

“What do higher education educators, students, and institutional stakeholders observe about the changing role of human teachers, believe human educators distinctively contribute, and envision for the future of human teaching in the age of artificial intelligence?“

This Blog is also available as an email three days a week. If you think that might suit you better, why not subscribe? One email on Mondays, Wednesdays and Fridays delivered to your inbox at 5am UK time. It’s free, and you can always unsubscribe if you conclude your inbox is full enough already!

Wide open

I love north Norfolk for its open skies, beaches and birdlife. This was Brancaster the other day.

Quote of the Day

“Chopsticks are one of the reasons the Chinese never invented custard.”

• Spike Milligan

Musical alternative to the morning’s radio news

Randy Newman | Old Kentucky Home

Link

If you want cynical despair set to music, then this is it!

Long Read of the Day

 What are universities for?

AI can help answer the question.

My latest Observer column:

One of the great pleasures of working in a university is that, at this time of the year, a fresh cohort of bright young school leavers arrives to begin the next phase of their lives. For many of them, it will be their first experience of living away from home, the first time they will be treated as adults and the first time that they may understand Wordsworth’s lines about how it might be bliss to be alive, but “to be young was very heaven”.

There is, however, one thing about the current crop of freshers that’s different from their predecessors of four years ago. Many, if not most, of them are “AI natives” in the way children used to be “digital natives” – that is, adept at using a technology they don’t understand. And they are entering educational institutions that still don’t know what to do about the technology.

Luckily, one of the world’s great universities – the Massachusetts Institute of Technology (MIT) – has also been grappling with that problem. It set up a large working group consisting of undergraduate and graduate students, academics from every school, and staff from relevant units, including the MIT libraries, and teaching and learning lab. It spent five months thinking about and discussing what to do about AI. It has now published a report that is the most impressive attempt I’ve seen to confront an issue that could be the undoing of many universities.

The report’s central message is that AI should be approached as a question of what a university is for rather than, as now, a question about cheating and academic integrity. The brief the MIT group was given at the outset was the kind of bureaucratic one that you’ll find in most universities at the moment: assess AI use at MIT, identify innovations in teaching and student assessment, and propose an AI-use policy. But the team dug deeper and concluded: “Beyond its implications for specific subjects and disciplines, the advent of AI demands a broader, holistic reassessment of the nature, scope and purpose of higher education today.”

It’s right. Universities are supposed to be institutions that create, curate and pass on knowledge. Large language models, as the psychologist Alison Gopnik and her colleagues have argued, are “cultural technologies”, like books, libraries and language itself – tools people use to access the knowledge created by them. So, whether we like it or not, universities and AI are in the same business.

Harassed vice-chancellors at UK universities may – understandably – chafe at philosophical debates about the function of the university that go back to John Henry Newman. But there are some takeaways from the MIT report that are more immediately relevant. First, detection and policing of AI use is a dead end. The tools produce too many false positives (especially for non-native and neurodivergent students). And – more importantly – they undermine the relationship between teachers and students.

Second, assessment needs to be rethought from the ground up. Most of the proposed “solutions” (oral exams, portfolios, and out-of-class work combined with in-class conversation, etc) all need academics’ time, physical space and probably smaller classes. The bottom line is that an AI-aware curriculum costs money, which is bad news for cash-strapped institutions.

Third, there have to be clear guidelines about AI use or non-use in every discipline or course, and the reasons for them have to be transparent and understood by students and teachers. In some cases, AI use may be unrestricted because it doesn’t undermine the learning objectives; in others, it may be allowed as a support tool only. Also, in certain cases, students may be required to use AI to complete and assignment in a specified manner; and then there will be cases where its use is strictly verboten. In pedagogy, no one size fits all.

Universities can no more ignore AI than they could ignore the printing press. The arrival of the technology could give them a new mission to add to their original ones of creating, curating and transmitting knowledge: to demonstrate AI’s potential for human augmentation. If they get this right, they’ll turn out graduates who use the technology with judgment rather than as tentative prompters. And then we’ll have an answer to the question of what are universities for that even cabinet ministers may understand.

My commonplace booklet

Re today’s Long Read (see above), the guy who started me thinking about this was Eli Noam of Columbia in his 1995 paper “Electronics and the Dim Future of the University”.

Having set out his conclusion about the “dim future” he wrote:

All this is a bleak scenario for the future of the university. In making this argument it is easy to appear as yet another dismal economist or technological determinist, and to invite, as a response, a ringing reaffirmation of the importance of quality education, of academic values, of the historic role of education in personal growth, and of the human need for free-wheeling exchange. To make such arguments may feel good but is beside the point. The question is not whether universities are important to society, to knowledge, or to their members — they are — but rather whether the economic foundation of the present system can be maintained and sustained in the face of the changed flow of information due to electronic communications. It is not research and teaching that will be under pressure — they will be more important than ever — but rather their present main instructional setting, the university system. To be culturally important is necessary (one hopes) but unfortunately not sufficient.

After I’d read the paper I used to ask every senior British university administrator I met if they had read it. With only one exception, none had.

This was in 1995, a time when ‘Artificial Intelligence’ was still a minority sport.

The other interesting thing is that Eli focussed on what the MIT team realised as they were reflecting on the implications of AI for their university. “True teaching and learning,” he wrote,

are about more than information. Education is based on mentoring, internalization, identification, role-modeling, guidance, and group activity. In these processes, physical proximity plays an important role. Thus, the strength of the future physical university lies less in pure information and more in college as a community. Less in wholesale lecture, and more in individual tutorial. Less in Cyber-U, and more in Goodbye-Chips College.

Linkblog

Something I noticed, while drinking from the Internet firehose.

• Trump clarification

This Blog is also available as an email three days a week. If you think that might suit you better, why not subscribe? One email on Mondays, Wednesdays and Fridays delivered to your inbox at 5am UK time. It’s free, and you can always unsubscribe if you conclude your inbox is full enough already!

The dancer

In Jim Ede’s house, Kettles Yard, Cambridge.

Quote of the Day

“And God said, ‘Let there be light’ and there was light, but the Electricity Board said He would have to wait until Thursday to be connected.”

• Spike Milligan

Musical alternative to the morning’s radio news

Abdullah Ibrahim | Bra Timing From Phomolong

Link

Long Read of the Day

Against historic normalization: Why AI is not the Manhattan project and why we must challenge ourselves to think the radically new.

Very timely Substack post by the historian Adam Tooze.

Given the extremity of our moment there is an understandable desire to look to history as a possible source of orientation: hegemonic succession, fascism, the nuclear arms race, Cold War 2.0, middle-income trap, Japan 1990s etc are all commonly invoked.

In each case the motive is understandable and the discussion is often fruitful, but more often than not the effect is comforting and to that degree also misleading.

Fascism was a political force that was defeated.

If we are in an interregnum then there is a new order to come.

The nuclear arms race was managed and the “West won the Cold War”.

If there are middle-income traps, then China’s growth must ultimately end and the “natural order” of world hierarchy will reassert itself. 

To my mind each of these analogies follows a similar pattern. You start with uncertainty, reach for history and arrive at a narrative which, even if it is not fully spelled out, actually has a reassuring conclusion. The fact that the conclusion remains tacit contributes significantly to the consoling effect.

To put it the other way around, if any of this chosen repertoire of stories had truly nightmarish conclusions would we really be invoking them so light-handedly?

To avoid that vertiginous turn involves subtle shifts of focus…

I love thinkers like Tooze (and others like Henry Farrell) who manage to put our current fears and obsessions into a wider context. This is worth your time.

Linkblog

Something I noticed, while drinking from the Internet firehose.

•  NASA Details Earthquake Effects on the Earth

NASA scientists using data from the Indonesian earthquake calculated it affected Earth’s rotation, decreased the length of day, slightly changed the planet’s shape, and shifted the North Pole by centimeters. The earthquake that created the huge tsunami also changed the Earth’s rotation.

Dr. Richard Gross of NASA’s Jet Propulsion Laboratory, Pasadena, Calif., and Dr. Benjamin Fong Chao, of NASA’s Goddard Space Flight Center, Greenbelt, Md., said all earthquakes have some affect on Earth’s rotation. It’s just they are usually barely noticeable.

”Any worldly event that involves the movement of mass affects the Earth’s rotation, from seasonal weather down to driving a car,” Chao said.

Gross and Chao have been routinely calculating earthquakes’ effects in changing the Earth’s rotation in both length-of- day as well as changes in Earth’s gravitational field. They also study changes in polar motion that is shifting the North Pole. The “mean North pole” was shifted by about 2.5 centimeters (1 inch) in the direction of 145 degrees East Longitude. This shift east is continuing a long-term seismic trend identified in previous studies.

Source

This Blog is also available as an email three days a week. If you think that might suit you better, why not subscribe? One email on Mondays, Wednesdays and Fridays delivered to your inbox at 5am UK time. It’s free, and you can always unsubscribe if you conclude your inbox is full enough already!

Fountainhead

Seen the other day. For a moment, it looked as through the flowers were perched atop the water.

Quote of the Day

“The love of money as a possession — as distinguished from the love of money as a means to the enjoyments and realities of life — will be recognized for what it is, a somewhat disgusting morbidity…”.

• John Maynard Keynes

Musical alternative to the morning’s radio news

Dvorak | Song to the Moon — from Rusalka

Link

Long Read of the Day

The dangerous myth behind AI agent hacks

These are not simply cyber security flaws but problems with the way that models are trained

Salutary sense from Yoshua Bengio in the Financial Times (Gift article).

Weak cyber defences, including the humans involved, are part of the problem and will never be perfect. But unless we address the root causes of the hacks, they are likely to grow in number and severity as AI capabilities advance.

At the core of all of these incidents is misalignment. AI agents are choosing goals that are misaligned with those of humans, and it is a consequence of their training.

The methodology that today’s leading companies use to train their frontier models is called reinforcement learning. RL trains models to pursue objectives: they get rewards when they succeed and these behaviours are reinforced. It has historically enabled the training of very efficient goal-seeking AIs. But it is also what plausibly leads to unwanted behaviours such as cheating, deception and self-preservation — often violating the moral rules that companies attempt to encode in AI models.

Scientific literature has long shown how misalignment becomes a consequence of RL; it was to be expected based on theoretical arguments and observed experiments…

Very good OpEd by a distinguished AI expert which is worth reading in full. It raises other interesting analogies. For example the parallels with the way the toxicity of social media is actually fuelled by algorithms that prioritise outrage and hostility. And then there’s the way that the tech industry has taken a leaf out of the oil industry’s playbook — suggesting that vulnerability of websites to AI-driven hacking is really the fault of the victims. And of course the ludicrous idea (dreamed up by a corporate PR firm employed by an oil giant) that responsibility for climate change is really down to individuals who don’t moderate their ‘carbon footprints’

Books, etc.

This is an intriguing book by an American scholar and entrepreneur which tells the story of the group of men who launched the first Industrial Revolution in 18th-century Britain. The really striking thing about it is the way the story illuminates the ‘revolution’ we appear to be living through now. It helps us understand why real industrial change takes much longer than AI evangelists realise.

Andrew Curry has a lovely essay about the book which develops this theme. The tech industry has focussed too much on product innovation (‘designed by Apple in California’) and not enough on process innovation. In practice, as Curry says, you need them both.

My commonplace booklet

Just noticed that Dave Winer (Whom God Preserve) started blogging 32 years ago today. He’s one of the Elders of the Web, and the person who inspired me to become a blogger. He’s still at it and I read him every day. Long may he thrive!

Feedback

From Sheila Hayman (Whom God Preserve), triggered by my solecism that Beethoven was serious!

Beethoven serious? He made the worst, terrible puns and gave all his friends nicknames, ideally with a pun involved too.

He was always thinking he needed money, so churned out a lot of dance music (I suspect your mandolin sonatina started out as a dance), and party pieces like ‘Wellington’s Victory’ for the panharmonicon, a mechanical orchestra invented by Johann Maelzel, who had previously invented ear trumpets for Beethoven, and which included (I think) cannon effects and various other show stoppers.

And as you asked, or might have if you’d thought of it – here’s a little sizzle of three of the musicians in ‘Pieces of Beethoven’ (it’s only short)

P/w 100.

Enjoy!

I did!

This Blog is also available as an email three days a week. If you think that might suit you better, why not subscribe? One email on Mondays, Wednesdays and Fridays delivered to your inbox at 5am UK time. It’s free, and you can always unsubscribe if you conclude your inbox is full enough already!

Acorn

This fell on to the pavement in front of me the other day. Funny to think that from such a tiny, beautiful seed a mighty oak might grow.

Quote of the Day

”Did you know that Mozart had no arms and no legs? I’ve seen statues of him on people’s pianos.”

• Viktor Borge

Musical alternative to the morning’s radio news

Beethoven | Mandolin Sonatina in C

Link

Just in case you thought that Beethoven was always serious!

Long Read of the Day

Against historic normalization. 

Or why AI is not the Manhattan project and why we must challenge ourselves to think the radically new.**  Very timely Substack post by the historian Adam Tooze.

Given the extremity of our moment there is an understandable desire to look to history as a possible source of orientation: hegemonic succession, fascism, the nuclear arms race, Cold War 2.0, middle-income trap, Japan 1990s etc are all commonly invoked.

In each case the motive is understandable and the discussion is often fruitful, but more often than not the effect is comforting and to that degree also misleading.

Fascism was a political force that was defeated.

If we are in an interregnum then there is a new order to come.

The nuclear arms race was managed and the “West won the Cold War”.

If there are middle-income traps, then China’s growth must ultimately end and the “natural order” of world hierarchy will reassert itself. 

To my mind each of these analogies follows a similar pattern. You start with uncertainty, reach for history and arrive at a narrative which, even if it is not fully spelled out, actually has a reassuring conclusion. The fact that the conclusion remains tacit contributes significantly to the consoling effect.

To put it the other way around, if any of this chosen repertoire of stories had truly nightmarish conclusions would we really be invoking them so light-handedly?

To avoid that vertiginous turn involves subtle shifts of focus…

I admire thinkers like Tooze (and others like Henry Farrell) who manage to put our current fears and obsessions into a wider context. This is worth your time.

Books, etc.

My mention on Friday of C.S. Lewis’a book A Grief Observed struck a number of readers, and prompted me to download a more recent edition, which has the original text plus a number of thoughful essays on it by a panel of distinguished writers, including Francis Spufford and the late Hilary Mantel.

Opening the new edition I went looking for a passage that had seared itself on my memory when I had first read it. It’s about Lewis’s reflections on what kind friends had sometimes said to him — that he would eventually “get over it”. Here’s the passage:

“But the words are ambiguous. To say the patient is getting over it after an operation or appendicitis is one thing; after he’s had his leg cut off is quite another. After that operation either the wounded stump heals or the man dies. If it heals, the fierce continuous pain will stop. Presently, he’ll get back his strength and be able to stump about on his wooden leg. He has ‘got over it’. But he will probably recurring pains in the stump all his life, and perhaps pretty bad ones; and he will always be a one legged man. There will be hardly any moment when he forgets it. Bathing, dressing, sitting down and getting up again, even lying in bed will all be different. His whole way of life will be changed. All sorts of pleasures and activities that he once took for granted will have to be simply written off. Duties too. At present, I’m learning to get about on crutches. Perhaps I presently be given a wooden leg. But I shall never be a biped again.”

Linkblog

*Something I noticed, while drinking from the Internet firehose.

From Quentin:

My favourite new verb is “coddiwomple”, which apparently means “to travel in a purposeful manner towards a vague or unknown destination.”

Nobody quite knows its origin — it’s relatively recent — but it’s a good way of describing quite a few of my holidays since we purchased a campervan.

 This Blog is also available as an email three days a week. If you think that might suit you better, why not subscribe? One email on Mondays, Wednesdays and Fridays delivered to your inbox at 5am UK time. It’s free, and you can always unsubscribe if you conclude your inbox is full enough already!