One of Labour’s key policy goals was to “kick-start economic growth”, a goal that would save a lot of trouble if they could achieve it, but one that was always going to be easier to say than to do.

Perhaps that is why the chancellor has looked for a quick crowd-pleaser. Her recent announcement of “Great British Summer Savings” (really?) promises to fleetingly cut VAT on children’s cinema tickets, children’s restaurant meals (a definitional headache) and visits to the zoo, along with free bus rides for under-16s and tariff suspensions aimed at giving us cheaper chocolate biscuits. (Families are more likely to need help when the energy crisis bites in the winter months, but it’s possible that the chancellor thinks that anything that happens after September will be somebody else’s problem.)

It would be giving this policy too much credit to call it a sugar hit; it’s more like a dusting of saccharine. A small, temporary, rather scattershot tax break is just moving small change around the economy. If you are prosperous enough to take your children to restaurants and amusement parks, you’ll save a few quid at best, and can look forward to the bill in due course. While you wait, you can ponder whether the chancellor really has nothing better to spend her scarce resources on.

Much of the money, in any case, is likely to be snaffled by the companies in question, which are under no obligation to pass on the tax cut. Those in the spotlight may decide that it would be prudent to make a show of doing so. Yet many theatres, theme parks and museums have no direct local competitors and will surely be tempted to pocket this brief summer bonus.

And if they do pass on the full benefit to customers? Merlin Entertainments, which runs Legoland, Alton Towers and Thorpe Park, has promised to do just that. But have you been to Alton Towers in a school holiday? Whatever families gain in lower ticket prices, they are likely to relinquish in longer waits for rides. The only real solution is to increase supply — an intriguing thought, no?

Still, it is nice to see rollercoasters on the front page, because for an insight into what ails the UK economy beyond the nearsighted unseriousness of successive governments, theme parks are not the worst place to start. They may have little practical significance but they make a fine symbol of the country’s ills: the UK’s rollercoasters are underwhelming for much the same reason that the UK economy in general is underwhelming.

I speak with some authority, as the proud parent of a teenage rollercoaster addict. Harford Junior has sampled much of what the UK’s rollercoaster scene has to offer, and now has his eyes firmly fixed on a theme park in Stockholm, another in Belgium, a third near Paris and a fourth near Krakow. (He’ll be lucky.)

This wanderlust speaks eloquently of the power of rollercoaster influencers on YouTube, but also of the fact that the UK just doesn’t have much to offer the connoisseur.

Thorpe Park has the extravagantly tall Hyperia and Alton Towers has The Smiler, in all its labyrinthine dystopian glory. Both are memorable coasters, but both have equally memorable queues.

Speaking as a man who has been dragged to more rollercoaster parks than you might imagine, my experience has been that the European parks are simply more pleasant experiences — shorter queues, better landscapes and themes, nicer food, longer opening hours. 

Compare Alton Towers with Europa Park in south-west Germany. Alton Towers is cheaper, which means it is overcrowded and the queues are absurd. A fast-track pass to skip them is available — for three times the price of admission, at busy times. Europa Park doesn’t sell a queue-jumping pass, and it has enough capacity that in my experience you don’t really need one.

These are subjective impressions, admittedly. But according to the Themed Entertainment Association, none of the world’s 25 most popular theme parks is in the UK. (There are four in Europe.) Only one UK theme park is in Europe’s top 10 — Alton Towers, in 10th place.

Why doesn’t Alton Towers just expand its capacity? A good question. It may be something to do with the limited competition: the UK’s three most popular theme parks are all run by the same operator. There are plans to build a Universal Studios theme park near Bedford, but major theme park openings in the UK are not a frequent occurrence.

It may be a response to planning restrictions — Alton Towers rides are obliged to stay below the tree line, lest they annoy the locals. Alton’s population is not much more than a thousand people, while the annual attendance at Alton Towers is well over two million. As so often in the UK, we let the planning-permission tail wag the economic dog.

The UK economy has suffered decades of under-investment from private and public sector alike, and you can see that reflected in our rollercoaster game. At least the fast-track passes serve as a shining example of how to profit in a low-growth economy: when life hands you long lines, make lemonade by charging people for the privilege of standing at the front of them.

So there we are. The rollercoasters don’t really matter, but what they symbolise does: a supply-constrained economy, hemmed in by rules and habits that don’t just make it hard to build something new and better, but make it hard even to imagine what “new and better” might look like. A brief, poorly targeted and poorly timed tax cut is no more likely to help than handing out fast-track passes is likely to reduce the average queue times at Alton Towers.

What might work? Treating the tax system strategically as a source of revenue and an encouragement to productivity would help — rather than using it as a fun-house source of distraction. Taxes are too complex and don’t raise enough revenue to meet any recent government’s spending plans, so cutting some randomly chosen tax rates for a few weeks is hardly addressing that problem.

But the real challenge lies in the nation’s supply side. Encouraging a few more theme parks would do wonders to improve the experience for coaster fans everywhere. And on the off-chance that we actually get serious? More power lines, houses, vehicle charging stations, roads, railway lines and power generation would be nice. I’ll cope without the discount on my chocolate biscuits.

Written for and first published in the Financial Times on 3 June 2026.

The children of North End, Boston, play in the shadow of an enormous steel tank of molasses. The thick, sticky sugar syrup is being used to make munitions for the First World War. When a worker notices dark molasses seeping from the tank, hewarns the company that there could be a leak. But the man in charge, Arthur Jell, has more important things to worry about: schedules to meet and profits to make. Besides, it’s only sugar. How dangerous could it be? 

This episode was released last year exclusively to members of the Cautionary Club. Along with monthly exclusive Cautionary Tales episodes, club members also receive ad-free listening, monthly video episodes and a monthly newsletter.

[Apple] [Spotify] [Stitcher]

Further Reading

The key source for this episode was Dark Tide: The Great Boston Molasses Flood of 1919 (2019) by Stephen Puleo. This episode also drew on various contemporary news reports published between 1914 and 1931 in The Boston Globe, The Philadelphia Inquirer and The New York Daily News.

Other sources included:

“Brittle Fracture: The Great Molasses Flood”, Epic Engineering Failures and the Lessons They Teach, by Stephen Ressler and The Great Courses Plus

The North End: A Brief History of Boston’s Oldest Neighborhood (2009) by Alex R. Goldfeld

Massachusetts Disasters: True Stories of Tragedy and Survival (2018) by Larry Pletcher, with additional stories by David J. Krajicek

“A sticky problem — a new light shone on Boston’s great molasses spillage” by Tony Fishwick, Loss Prevention Bulletin 264, Institute of Chemical Engineers,December 2018.

“Solving a mystery behind a deadly ‘Tsunami of Molasses’ of 1919” by Erin McCann, The New York Times, 26 November 2016.

“Move Fast And Break Things: Pros and Cons of the Concept“, MasterClass, 21 June 2022.

Six weeks later, someone asked the man to take off his shirt. He had a genetic condition, not a delusion. “It was in fact true: he was growing breasts.”

It was a lesson to learn: even experts — no, especially experts — can become sidetracked by elaborate ideas, overlooking the simple and direct approach. No surprise, then, that Dawes would become fascinated by the research of Ted Sarbin and Paul Meehl, psychologists who had studied the surprising power of simple statistical predictions in areas such as clinical diagnoses or academic performance.

Sarbin, for example, used a linear regression — almost the simplest statistical rule imaginable — to predict the college grades (GPA) of graduating students based on their high-school class rank and their score on the entrance test. That method was more accurate than the opinion of clinical psychologists armed with the same data and vastly more besides. Meehl found many more examples of cases where simple statistical rules beat the diagnosis or forecasts of the experts.

But just how simple could those rules be? A standard linear regression predicts an output based on a combination of various inputs. For example, the probability that an offender would be rearrested might be a function of their age, sex, number of previous convictions and severity of previous convictions. How much weight each factor gets is determined by a mathematical formula to most closely fit the historical data.

Instead, Dawes suggested what he called “improper” linear regression, where the weights weren’t optimised, but chosen arbitrarily — perhaps equally weighted, or even chosen at random.

To pick an example suitable for the Financial Times, think of the choice of an optimal investment portfolio. The output is the portfolio return; with the right allocation across different assets, we could maximise the predicted return for any given level of risk. Harry Markowitz, who shared a Nobel memorial prize in 1990, showed in the 1950s how to choose the weights in such a perfect portfolio. The Robyn Dawes school of thought says not to bother. Instead, follow a rule such as “Just invest your money equally in the 50 largest public companies” — or maybe even “half in stocks, half in bonds”.

That can’t possibly work, can it? Well, in the first job Markowitz took after publishing his theory, he had to decide how to allocate his pension contributions. He plumped for half in stocks, half in bonds. That’s an improper weight for you. But it’s not clear that Markowitz was wrong, even if he was self-contradictory: a 2009 paper by Victor DeMiguel, Lorenzo Garlappi and Raman Uppal found that the simple strategy of investing equally across a bunch of assets is surprisingly effective.

Over drinks after an academic conference discussion, a fellow panellist challenged Dawes: “Could you . . . use one of your improper linear models to predict how well my wife and I get along together?”

Dawes thought he could. He had colleagues who had been gathering data on sex and relationships, and he proposed the following improperly weighted predictor: that couples would probably describe their relationship as “happy” if they had sex more often than they had fights, and “unhappy” if the frequency of fights exceeded the frequency of sex.

Two variables, equally weighted — surely there’s a more accurate model than that? Yet the absurdly simple theory fit the evidence. A colleague had data on 12 unhappy couples; all of them fought more often than they had sex. Of 30 happy couples, 28 had sex more often than they had arguments. Subsequent small studies reached the same conclusion.

(An important caveat: ask the couples about the quality of the relationship first, and the quantity of sex and arguments afterwards, otherwise the totting-up may provoke a self-fulfilling crisis. One woman counted the sex, and the fights, and decided it was time to file for divorce.)

“The conclusion is that if we love more than we hate, we are happy; if we hate more than we love, we are miserable,” wrote Dawes in a 1979 article, “The Robust Beauty of Improper Linear Models in Decision Making”, adding: “This conclusion is not very profound, psychologically or statistically. The point is that this very crude improper linear model predicts a very important variable.”

Why do these almost laughably simple models work? One answer is that while the weights are arbitrary, there is already some expertise smuggled into the choice of variables to throw into the mix. Dawes might have claimed that marital happiness was a function of average monthly rainfall in Nigeria; another simple model but not a very good one.

Another answer is that complicated-seeming outcomes often reflect fairly straightforward combinations of variables. It’s almost always a bad sign if an offender has a string of previous convictions, no matter what else might be true. And regardless of the psychodramas surrounding a couple’s relationship, it’s probably a good sign if they’re having plenty of sex.

But a third factor is that the typical dataset only captures a slice of what’s really going on. Marital happiness is hard to measure precisely. Risk is hard to measure precisely. Even the frequency of sex is harder to measure than it might seem — who’s counting, and what do they think counts? Alongside all this noise is the fact that everything changes.

As a result, the optimal-seeming estimate may prove overconfident as time goes by and more data arrives. A simpler, cruder method may be a bit more robust. Victor DeMiguel and his colleagues reckoned that in order for the allegedly optimal estimates to reliably outperform the simple equal-shares rule for a 50-asset portfolio, the analyst would need a dataset five centuries long.

The point is not that simple — “improper” — analysis is always best, just that it’s a surprisingly strong baseline. It doesn’t mess about or claim too much. It can be performed on a napkin in a bar, or jotted on a doctor’s notepad. Before unrolling a great analytical edifice, sometimes it’s worth asking to check under the shirt.

Written for and first published in the Financial Times on 27 May 2026.

Loyal readers might enjoy How To Make The World Add Up.

“Nobody makes the statistics of everyday life more fascinating and enjoyable than Tim Harford.”- Bill Bryson

“This entertaining, engrossing book about the power of numbers, logic and genuine curiosity”- Maria Konnikova

I’ve set up a storefront on Bookshop in the United States and the United Kingdom. Links to Bookshop and Amazon may generate referral fees.

In the early 1900s, Sir Arthur Vickers keeps the magnificent Irish Crown Jewels safe under lock and key at Dublin Castle. When the jewels disappear, the King rages, the police investigate, and even Sir Arthur Conan Doyle gets involved. No one is ever charged and no jewels are ever recovered. Except, we have a very good idea of who took them, and why the truth has stayed buried. 

For ad-free listening and bonus episodes, video conversations and our newsletter, please consider joining the Cautionary Club.

[Apple] [Spotify] [Stitcher]

Further Reading

This episode relied on three books about the Irish Crown Jewels affair. The classic account, from 1965, is Vicious Circle: The Case of the Missing Irish Crown Jewels by Francis Bamford and Viola Bankes. More recently, contrasting theories on the case are presented in The stealing of the Irish Crown Jewels: an unsolved crime, by Myles Dungan, and Scandal and Betrayal: Shackleton and the Irish crown jewels, John Cafferky and Kevin Hannafin. Brian Lacey’s book, Terrible Queer Creatures: Homosexuality in Irish History, provided additional context.

Ari Adut introduced his theory of scandal in an American Journal of Sociology paper from 2005, A Theory of Scandal: Victorians, Homosexuality, and the Fall of Oscar Wilde, and expanded it in book form in On Scandal: Moral Disturbances in Society, Politics, and Art.

It’s always tempting to blame rising prices on corporate greed. Inflation is painful, and somewhat mysterious, and it’s nice to have someone to blame. Companies do try to maximise profits, which is as close to greed as an emotionless institution is likely to get, and it’s hard to deny that big retailers and the firms that supply them do decide what number to put on the price tag.

But as an explanation of inflation, that won’t quite do. Any theory of corporate greed, no matter how intuitive or satisfying, has to explain why the opportunity to be greedy seems to wax and wane. Companies are sniffing out profit opportunities on a 24/7 basis, but are we supposed to believe that inflation surged in 2023 because CEOs were suddenly struck by the unprecedented idea of jacking up their prices? We can dismiss that notion.

So is there a coherent case that inflation spikes are somehow exacerbated by corporate profiteering? One possibility is simply that corporations use more general inflation as camouflage, raising prices more than they need to because consumers are already expecting trouble, or are confused about what prices should be.

Paul Scanlon of Trinity College, Dublin, has proposed this in his article “A model of greedflation”. One of the costs of inflation has always been that it disorients everyone, making it harder to figure out whether a change in a price, or a wage, is a shift in relative prices or part of some general inflationary surge. It is not absurd to suggest that companies might take advantage of the chaos.

Another possibility, described by Isabella Weber and Evan Wasner of the University of Massachusetts Amherst, is that supply shocks in one part of the economy might act as a signal for rival firms to raise prices in unison without the need to risk prison by explicitly colluding. To conjure a purely hypothetical example, if the Strait of Hormuz were closed to oil tankers, the price and profitability of oil from other sources would rise. Since the price of oil is connected to many other prices, perhaps that would give other companies the signal they needed to raise not only their prices but their profit margins.

But the opposite seems equally plausible: that customers are militant when prices are rising sharply and complacent when prices are subdued. This alternative view suggests that when input costs increase, companies will do everything they can to minimise the impact. It is when input costs are falling that companies fail to pass on the savings to consumers, taking advantage of the fact that customers are relaxed. Of course, when inflation is low, and some prices may even be falling, few people worry about greedflation. That is precisely the point.

So, what does the evidence suggest? Johannes Brinkmann and Nikhil Datta of the University of Warwick recently published an analysis of the impact on petrol and diesel prices of the oil price shock in 2022, following Russia’s onslaught in Ukraine. They found that in the UK, retailer margins compressed: the wholesale price of diesel rose by 39 pence per litre, while retail prices only rose 16 pence. This is the opposite behaviour to that predicted by the greedflation hypothesis.

A natural explanation of this price compression is that retailers feel under more intense scrutiny when prices are rising. Brinkmann and Datta show that searches on the petrolprices.com website increased dramatically when prices did — and that areas where such searches were more common were also areas where the price compression was more intense.

Brinkmann and Datta’s analysis is merely the latest in a long tradition of research describing “rocket and feather” pricing at the pump — capturing the idea that pump prices neither faithfully track the ups and downs of the crude oil market, nor exaggerate them — instead, they shoot up like a rocket but drift down again like a feather. What is more, the quick surge upward reaches prices less lofty than one would expect; it is during the slow descent that retailers make their money.

Fifteen years ago, Matthew S Lewis and Howard Marvel noted that customers spent more effort searching when prices were rising, even though there was little benefit to that search, since most retailers were charging similar prices. When pump prices were falling, there was more variability from forecourt to forecourt and a higher return to shopping around, but most customers did not bother, feeling content that prices were moving in the right direction. (The search heuristic is simple enough to imagine: if you drive past the forecourt and prices have gone up, keep driving and hope to find a better deal; if they’ve gone down, stop worrying and fill the tank.)

Crude oil and fuel markets are transparent and easy to study. What about the broader landscape of prices? Back in 2000, the never-dull Sam Peltzman of the University of Chicago published a survey with the self-explanatory title “Prices Rise Faster Than They Fall”. Peltzman studied hundreds of products, limiting himself to those with a significant identifiable input. For example, the price of peanuts goes some way to explaining the input costs of peanut butter manufacturers. Peltzman found that “in more than two of every three markets examined” prices responded to cost increases more quickly than to cost reductions, the rocket-and-feather behaviour again.

Don’t expect the greedflation question to be easily resolved. Thankfully, it doesn’t have to be: whether you think that companies exploit market power under the cover of widespread inflation, or you suspect that companies exploit market power when inflation is low and customers have stopped paying attention, it’s never a bad time for the competition policy authorities to look for ways to sharpen competition by breaking up dominant firms or improving price transparency.

With wholesale prices rising sharply in the US (to 6 per cent in April), perhaps we shall see the greedflation debate return to prominence. Fine. But it’s when nobody is talking about greedflation that consumers should be most on their guard.

Written for and first published in the Financial Times on 20 May 2026.

Loyal readers might enjoy How To Make The World Add Up.

“Nobody makes the statistics of everyday life more fascinating and enjoyable than Tim Harford.”- Bill Bryson

“This entertaining, engrossing book about the power of numbers, logic and genuine curiosity”- Maria Konnikova

I’ve set up a storefront on Bookshop in the United States and the United Kingdom. Links to Bookshop and Amazon may generate referral fees.

When John Darwin walks into a shop in London, it causes an instant stir. After all, John Darwin has been dead for five years. He claims to have amnesia, but everyone – from the police and the media to his insurance company – suspects he is lying. No one can prove a thing, until a young woman at home with her baby thinks of something everyone else has missed.

For ad-free listening and bonus episodes, video conversations and our newsletter, please consider joining the Cautionary Club.

[Apple] [Spotify] [Stitcher]

Investigative journalist David Leigh and retired Detective Superintendent Tony Hutchinson wrote The Thief, His Wife and The Canoe: The astonishing true story of Anne Darwin and ‘Canoe Man’ John and their jaw-dropping deception. It became the basis for a four-part ITV drama of the same name, along with a documentary – The Thief, His Wife and The Canoe: The Real Story. David Leigh also worked with Anne Darwin to write her book, Out of My Depth.

The Darwins’ case is given another book-length treatment in Tammy Cohen’s Up The Creek Without a Paddle. This script also relied on contemporary news reporting in a range of outlets including The Times, The Independent, the BBC, the Mirror, The Sun, the Mail and The Standard.

For more on the origins of context collapse, see this post from danah boyd; recent research includes From context adaptation to context restoration: strategies, motivations, and decision rules of managing context collapse on WeChat by Pengxiang Li, Hichang Cho, Cuihua Shen and Hangchen Kong.

A cynic might suggest this near silence speaks volumes. Maybe official statistics have little value? That was the radical view of Sir John Cowperthwaite, who was the financial secretary of Hong Kong throughout the 1960s, when it was a rapidly growing, laissez-faire British colony. Cowperthwaite thought the value of official statistics wasn’t merely minimal, but negative: he told the economist Milton Friedman that he didn’t collect economic data, because it would only encourage the Whitehall variety of mandarin to interfere.

In context, Cowperthwaite’s position was understandable: few economies were more at risk of clumsy meddling than Hong Kong, a colonial possession pursuing a libertarian path on the opposite side of the world from soft-left imperial rulers. Still, there are at least two weaknesses in his argument. The first is the hope that ignorance might restrain the interventionist impulses of governments. It might simply make those interventionist impulses clumsier. The second is the unexamined premise that only a government might find official statistics useful.

A report from the US National Academies last year argued otherwise. While governments do rely on official statistics for everything from political representation (often tied to population) to the inflation adjustment of pensions and other welfare payments, many organisations and individuals also rely on trustworthy statistics for anything from deciding where to locate a new storefront or warehouse to directly selling analysis based on government data. The National Academies reckons that the revenue of the “government data-intensive sector” in the US almost doubled between 2012 and 2022, to just shy of $800bn, a direct sign that somebody finds these numbers useful. For context, the total budget of all US statistical agencies and programmes in 2022 was $7.1bn.

But if you want to understand whether a thing is useful, you can always look at what happens when somebody breaks it. Call this the Joni Mitchell principle: you don’t know what you’ve got till it’s gone.

The National Bureau of Economic Research’s new working paper “The Value of Reliable Statistics” comes from Stanford’s Nicholas Bloom, Erica Groshen, a former boss of the Bureau of Labor Statistics (BLS), and two scholars from the American Enterprise Institute, a pro-market think-tank. It studies the impact of one particular fracture: President Trump’s firing last August of Erika McEntarfer, head of the BLS, along with his simultaneous claim that, “In my opinion, today’s Jobs Numbers were RIGGED in order to make the Republicans, and ME, look bad.”

As I wrote at the time, this was a two-pronged attack on the credibility of the BLS. By attacking the institution’s record, Trump was damaging it in the eyes of his supporters, and, by replacing its leader in such a way, he was damaging it in the eyes of his opponents. Bloom and his colleagues do not try to measure the impact of Trump’s actions on the BLS’s capacity; it is the question of its credibility that interests them.

To measure this, the researchers look at the Economic Policy Uncertainty Index, a dataset developed about 15 years ago by Bloom and others. The EPU measures uncertainty about the direction of economic policy by analysing the text in major US newspapers. It spikes when newspapers can talk of little else but how policymakers are causing confusion.

Unsurprisingly, the index sharply rose immediately after McEntarfer was fired. It fell back not long after, but as the researchers note, “even when the underlying rise in uncertainty is more persistent . . . the news cycle moves on”. Based on earlier research into the effect of uncertainty on investment and growth, the researchers suggest that the increase in economic uncertainty that week could have caused $104bn of economic damage and that up to 168,000 jobs could have been lost.

Those are large numbers, but as Bloom and colleagues freely admit, they are not good estimates of the impact of Trump’s words and actions because there were other reasons for the EPU to increase. The first and most obvious reason is that the trigger for McEntarfer’s firing was a large downward revision in the jobs numbers, which would itself have raised uncertainty even if Trump had done nothing. A Federal Reserve governor, Adriana Kugler, announced her resignation on the same day. All three events happened within hours of each other and all three could reasonably have been interpreted at the time as adding to a sense of chaos and pushing the EPU up.

After trying to isolate coverage only of McEntarfer’s firing, the researchers produced a preferred estimate of nearly $20bn of economic damage, resulting from the fear, uncertainty and doubt generated by the ejection and criticism of McEntarfer. It’s still fair to describe this estimate as itself highly uncertain. It is, after all, measuring what the newspapers found newsworthy. Generally, serious newspapers put stuff that matters on the front page, and when the news is about unpredictable economic policy that has generally been a bad sign. But sometimes newspapers get excited about things that don’t much matter; perhaps McEntarfer’s firing was one of those things. It is impossible to be sure.

The estimated damage from the affair, while a tiny sliver of US GDP, is about 25 times the entire budget of the BLS. This, perhaps, is the argument for investing in reliable statistics, and for not undermining them in the hope of fleeting partisan advantage: they do not cost very much relative to what they are trying to measure.

In the US, just over one dollar in a thousand of federal government spending goes to statistical agencies and other statistical programmes. The case for government-funded statistics is that it is worth spending one dollar in a thousand in the hope that the other $999 might be fractionally better used as a result.

Written for and first published in the Financial Times on 13 May 2026.

Loyal readers might enjoy How To Make The World Add Up.

“Nobody makes the statistics of everyday life more fascinating and enjoyable than Tim Harford.” – Bill Bryson

“This entertaining, engrossing book about the power of numbers, logic and genuine curiosity” – Maria Konnikova

I’ve set up a storefront on Bookshop in the United States and the United Kingdom. Links to Bookshop and Amazon may generate referral fees.

This episode is available exclusively to members of the Cautionary Club.

[Apple] [Spotify] [Stitcher]

Many thanks to author Karen Torghele for kindly allowing us an advance view of her soon-to-be-published biography Albert Sabin: The Life of a Polio Vaccine Pioneer. This episode also relied on David M Oshinsky’s book, Polio: An American Story, and The Swine Flu Affair: Decision-Making on a Slippery Disease by Harvey V. Fineberg and Richard Neustadt.

The academic literature on anticipated regret includes Anticipated Regret and Health Behavior: A Meta-Analysis, To feel, or not to feel, is it a question of time? The influence of feedback availability on the magnitude of anticipated regret, Narratives of Regret: How Anticipated Regret and Counterfactual Thinking Can Promote COVID-19 Vaccination Intentions in Unvaccinated Adults, Anticipating pride or regret? Effects of anticipated affect focused persuasive messages on intention to get vaccinated against COVID-19, and An experimental investigation of the effects of anticipating regret and relief on intentions and decisions to get the influenza vaccination.

As Governor of Britannia, Magnus Maximus has a huge army at his disposal, which is just what he needs to secure the Roman imperial throne. But perhaps the impressive general should have looked into the past before focusing on his future. Tim is joined by Paul Cooper, host of Fall of Civilizations Podcast, to explore why powerful civilisations such as the Assyrians, the Han dynasty and the Roman Empire all ultimately collapsed.

Paul Cooper is the author of Fall of Civilizations: Stories of Greatness and Decline.

For ad-free episodes, monthly bonus episodes and other benefits, check out the Cautionary Club.

[Apple] [Spotify] [Stitcher]

Me: “How do you plan to get to the start line?”

He: “I’ve asked ChatGPT. It says Elizabeth Line to Liverpool Street, then the train to Blackheath.”

That didn’t sound right. Was there a train from Liverpool Street to Blackheath? Google Maps and Citymapper suggested getting to Blackheath from Charing Cross or Waterloo.

Me: “Are you sure? I’d suggest the Circle or Bakerloo to Charing Cross.”

He frowned for a moment and pulled out his phone. “No, ChatGPT says that ‘The Circle Line is not a good choice on marathon day. It will be too crowded. There are too many stops and too many steps. It’s a route for tourists, not for runners.’”

I checked Google Maps. Sure enough, there is no train from Liverpool Street to Blackheath. ChatGPT’s recommendation would leave him stranded, trying to catch a bus over the marathon route, then trying to get on to the train from Charing Cross at a busy London Bridge. I told him that sounded like a bad idea. He frowned again and typed another query into his phone. “Oh, you’re right. ChatGPT says, ‘Correction: take the Elizabeth Line straight to London Bridge.’”

Me: “The Elizabeth Line doesn’t go to London Bridge.”

You’ve heard tales of artificial intelligence hallucinations before, but it’s not the AI that fascinates here: it’s the human. 

The route-finding algorithm on Google Maps is a minor miracle. It will solve a complex optimisation problem across multiple modes of transport, taking into account real-time congestion or delays, and it’s been available on smartphones and browsers for years. It is a proven, practical example of AI in action. So on marathon day, when the stakes are high and the clock is ticking, why would anyone turn instead to a fancy word-guessing machine such as ChatGPT?

Perhaps it’s that ChatGPT seems so human. It served up an uncanny impersonation of a friendly and knowledgeable local guide. The Circle Line? Pfft, it’s fine for tourists but you’re a marathon runner: think about all those steps! (It’s true, the creaky old Circle Line does have steps.) 

Part of the bot patter reminded me of clickbait ads: INSURANCE COMPANIES HATE THIS LOOPHOLE! ChatGPT wasn’t just giving a route, but giving a rationale, even explaining why we shouldn’t listen to the lamestream advice of Google Maps. This is the approach of a confidence trickster.

In the introduction to her book The Confidence Game, psychologist Maria Konnikova explains: “The true con artist doesn’t force us to do anything: he makes us complicit in our own undoing . . . we believe because we want to.” One difference between the con artist and the large language model (LLM) is that the con artist knows the truth and is trying to conceal it. One similarity between the con artist and the LLM is that both of them have perfected seeming plausible.

A recent paper in Nature finds that when LLMs are trained to be warm and friendly, they also produce dramatically less accurate answers, “promoting conspiracy theories, providing inaccurate factual information and offering incorrect medical advice”.

That sounds bad. I’d suggest that the reality is worse: the sycophantic AI not only produces mistakes, it persuades us to believe them. 

In 1950 Alan Turing, the mathematician and visionary of the computer age, famously proposed an “imitation game” in which a human judge would communicate through a teleprompter with a human and a computer. The computer’s job was to imitate human conversation convincingly enough to persuade the judge. 

Turing’s test remains intriguing, but there is a longstanding difficulty: the fallibility of the judge. A primitive 1960s chatbot, Eliza, responded like a parody of a therapist (“How does that make you feel?” “Why do you feel sad?” “Please go on.”). People lapped it up; it’s nice to feel listened to. A 1980s chatbot, MGonz, just fired off insults and was perfectly plausible, partly because insults are simple to deliver and mostly because they prompt rage rather than reflection in the human recipient. And Robert Epstein, an expert in the Turing Test, has written entertainingly about how he was fooled into a four-month correspondence with a sexy Russian lady who was, in fact, a 2006-era chatbot. None of these bots had a thousandth of the sophistication of a modern LLM, but they didn’t need it: when humans are sad, angry or amorous, we aren’t very sophisticated judges, either.

We are all going to find ourselves in strange variations of the Turing Test in years to come, and I wonder if we are up to it. And not just us, but those with power over us. As Cory Doctorow, author of Enshittification, is fond of observing: you won’t be replaced because an AI can do your job, you’ll be replaced because an AI salesman convinces your boss that it can. If my journey to the marathon start line is any guide, that salesman will have an easy job.

The capabilities of modern AI are impressive. But what determines whether we use it is not the capability, but the impressiveness. They are correlated but they are not the same thing. There’s a tale about the French poet Jacques Prévert seeing a fellow begging for change on the streets of Venice with a sign that read “Blind man without a pension”.

Prévert stopped to chat to him; not many people were moved to contribute, and Prévert offered to write a new sign.

The next day, he returned to find the man overjoyed. “It’s incredible; I’ve never received so much money in my life.” 

Prévert had written: “Spring is coming, but I won’t see it.” 

The new sign contained no news — in fact, it was less informative than the old. But it told a story. Google Maps was the first sign: it told me where to get my train. ChatGPT was the second sign: it told my companion not just where to go, but how to feel about taking such a clever route.

I left him at Paddington, urging him not to try to take the non-existent Elizabeth Line train to London Bridge. I am not sure I was as convincing as ChatGPT.

I ran the London Marathon in support of the Teenage Cancer Trust – not too late to make a donation.  

Written for and first published in the Financial Times on 6 May 2026.