{"text":[[{"start":5.5,"text":"Two activities have occupied my summer thus far to the extent that it is now difficult to imagine my life without them. "}],[{"start":12.1,"text":"The first was a World Cup pool, designed and contested by members of my family. There were eight of us and we drafted teams in turn, each assembling portfolios of six nations, and whichever portfolio scored the most goals won. The second was conducted in the FT newsroom, a World Cup blackjack game, as in the casino card game, its goal being to choose any three countries that would combine for as close to 21 total goals as possible, without going over."}],[{"start":40.2,"text":"To prepare, I administered millions of hypothetical World Cups inside a computer. Invisible, digitised Messis, Mbappés and Bellinghams silently scored tens of millions of goals at the clip of a few gigahertz, all stored on hard disk. I aggregated these millions of cups, moulding them into piles, sculpting a likeness of the tournament. By feeling its surfaces, I learnt how its groups might unfold, who was likely to play whom in the knockouts, and how I might get the better of my loved ones and colleagues."}],[{"start":71.1,"text":"This was a newly maximalist World Cup, with its 48 teams, up from 32, and a complex system for determining advancing teams and placing them in the knockout stage. At least once, the American broadcaster Fox displayed an incorrect route to the final. All the better, then, to unleash the brute force of simulation. (The tournament in 2030 may feature 64 teams, an expansion that simulators ought to welcome.)"}],[{"start":96.5,"text":"As a species we have modelled and simulated natural phenomena and human affairs for centuries. We built orreries and played war games. In the computer, we generate nuclear weapons and weather, things that are too dangerous or impractical to recreate in real life. The computer simulation is “a process that imitates one process by another process”, as defined by the philosopher Stephan Hartmann. "}],[{"start":121.55,"text":"My processed and chosen blackjack teams — Canada, South Korea and Switzerland — scored exactly 21 goals. My obsession with simulation has now earned me dozens of pounds. This success featured dumb luck too, of course. My simulations had predicted 6.7, 5.6 and 7.0 goals for these three teams respectively (hedging against busting). The true figures were 9, 2 and 10. Even the simulations gave me only a 7.3 per cent chance of getting 21 on the nose. But Dan Ndoye angled one more in for Switzerland against Argentina, and then the Swiss obliged further by getting knocked out."}],[{"start":162.5,"text":"But this is beside the point, and in the family pool I finished fifth. (My simulator had told me that, when the Netherlands were eliminated, my chances of catching my brother wilted to 1 per cent.) The allure was always the simulation itself. My simulations were at the whim of countless choices of modelling and calibration: how to measure team strength, mapping that strength to goals. But whatever I chose, I would incubate a million futures, multiverse World Cups that might have been but never shall be, the aggregation of which can place a probabilistic structure on — get a sense of — even the most complicated tournament that might be dreamt up in a Fifa boardroom."}],[{"start":204.35,"text":"Just now, I simulated a fresh batch of a million 2026 World Cups, as if this summer had never happened. In 0.005 per cent, Curaçao were world champions. In one of those, simulation number 437,824, they played Argentina to 0-0 in the final and won on penalties. Somewhere on my hard drive, that is what happened."}],[{"start":229.95,"text":"All this, I hate to admit, is a handy application of generative artificial intelligence. Rather than merely asking the AI which countries to draft, or what set is likely to combine for 21, an offence to which some of my colleagues have pleaded guilty, the AI can and ought to be used to create the simulating arena, to spread the agar in the Petri dish, even for non- or quasi-coders like me. This tactic is crucial: it transforms a chatbot from a black box to the builder of a clear plastic box, an ant farm in which we watch worlds form."}],[{"start":264.25,"text":"Strip the exercise to its bare essentials. Simulate thousands of random walks, a pixel moving madly around a two-dimensional plane for, say, 10,000 steps, and draw that path. I have built such a tool in the past few minutes, random_walk.js. There is no practical application of these walks, just as there is no practical application of a walk through the park. They are valuable nonetheless. Spread the images of the walks on your screen. They tell stories of their process. This is digital ikebana. They are mesmerising and serve as a kind of in silico meditation."}],[{"start":300,"text":"When studying an ant farm, we are not interested in any individual ant or where it goes, but in the emergence of the tunnels through the earth, the shape of the system. It’s just a welcome side effect, then, that these things can be useful to triumph over your colleagues at football games."}],[{"start":316.65,"text":"Admittedly, this interest may be a personal quirk; I’ve written here before, for example, about my obsession with digital power-washing simulators."}],[{"start":326,"text":"This ant-farm activity, I believe, stands at some angle of opposition to another recent capitotechnical phenomenon, the prediction market. Advertisements for companies such as Polymarket and Kalshi lit up pitchside hoardings and saturated television in the US during the cup. Together about $6bn was traded on their main tournament-winner markets."}],[{"start":348.7,"text":"Prediction markets collapse all future worlds into one number — a price — 59 cents for Spain, for example, just before the final. The sharps on these sites certainly run their own modelled simulations. But the temptation for the broader consuming public is strong to read the price and ignore the swarm of worlds from which it is made. This is like viewing the sculpture through a pinhole."}],[{"start":371.5,"text":"This matters beyond football. Take elections, for example, where a single price collapses all world events, all policy positions, all rhetoric. It collapses all public-opinion polls, which in turn collapse voters. "}],[{"start":384.2,"text":"Philosophers of science, bless them, have struggled with the epistemology of computer simulations. What kind of knowledge-producing thing is a simulation? There has been among them over decades a pendular debate about their nature. On the one hand they are scientific models; on the other they are experiments. A newer strand of thought suggests they are a scientific instrument in their own right. They are. And they teach us not about the ant but about the earth. "}],[{"start":411.7,"text":"Oliver Roeder is a commissioning editor on FT Weekend Magazine. Tim Harford is away"}],[{"start":417.75,"text":"Find out about our latest stories first — follow FT Weekend Magazine on X and FT Weekend on Instagram"}],[{"start":429.75,"text":""}]],"url":"https://audio.ftcn.net.cn/album/a_1790786197_4115.mp3"}