INDIA

The voters humble a government that thought it was invincible, and India's democracy surprises the world and itself

The exit polls promised a landslide. The count delivered a rebuke. Modi returns for a third term, but shorn of his majority and dependent, for the first time, on allies who can say no.

The votes of nearly 650 million Indians, counted on June 4, delivered the result almost no one had predicted: Narendra Modi's BJP lost its parliamentary majority. The party that had campaigned on winning 400 seats, that had governed for a decade with a single-party majority and the confidence that came with it, fell to 240, well short of the 272 needed to rule alone. Modi will return for a rare third term, his coalition holding a working majority, but he does so a diminished figure, dependent now on regional allies who have their own demands and their own capacity to withdraw support, in the messy, bargaining, coalition politics India had not seen at the centre for ten years. The opposition INDIA bloc, written off for dead through the whole campaign, surged to a total that stunned the pundits and revived a two-sided politics many had assumed was finished.

The result was a rebuke assembled from a thousand local grievances the national narrative had missed: unemployment, inflation in the price of onions and cooking gas, rural distress, unease in places about the majoritarian turn of the government's rhetoric, and, in the heartland state the BJP had made its fortress, a quiet revolt no pollster caught. India's voters, poorer and less connected than almost any electorate on Earth, had once again done the thing that makes the country's democracy real despite all its strains: they declined to be told the result in advance, and they cut a confident government down to size, peacefully, with a button. Markets that had soared on the exit polls' promise of a supermajority tumbled on the reality of a weakened one, then recovered. The pundits who had called it a coronation spent the week explaining how they had missed the biggest political story of the year in their own country.

EUROPE

The far right surges, and a French president gambles his country on a snap election

Elections to the European Parliament on June 6 to 9 delivered sharp gains to far-right and nationalist parties across the continent, a rebuke to the mainstream from voters angry over immigration, the cost of living, and the green transition's price. Nowhere was the shock sharper than in France, where the far-right National Rally trounced President Macron's party so badly that Macron responded with an extraordinary gamble: he dissolved the National Assembly and called a snap election for the end of the month, betting that faced with the far right actually within reach of power, French voters would rally to stop them. It is a bet that could hand the far right the government, or block it, and no one, Macron included, can be sure which.

IN BRIEF

Apple gets its AI; the G7 meets; a president's son convicted; a rescue and its cost

Apple unveiled "Apple Intelligence" at its developers' conference on June 10, weaving generative AI through the iPhone and, in a striking concession, partnering with OpenAI to power the smartest requests, a sign that even the most self-reliant company in technology could not build this alone. The G7 met in Italy from June 13, pledging support for Ukraine and wrestling with AI, which the Pope himself addressed. Hunter Biden, the president's son, was convicted on federal gun charges on June 11, a painful counterpoint to Republican claims of a two-tiered justice system. And Israeli forces rescued four hostages from Gaza on June 8 in an operation that also killed more than 270 Palestinians, the rescue celebrated in Israel and the toll condemned abroad.

The Column

The territory the map could not see

Every exit poll promised a landslide; the voters delivered a rebuke. The miss is usually explained as bad polling. The more interesting explanation is that the thing the polls could not see was the thing that mattered, and that the blindness was built into the method.

By Meera Anand

Every exit poll promised the Indian government a landslide this fortnight, and the voters delivered, instead, a sharp rebuke, stripping the ruling party of its majority and humbling a government that had come to believe itself invincible. The reflexive explanation is that the pollsters were simply bad at their jobs, and no doubt some were. But the more interesting and more useful explanation is that the miss was not an error within the method but a limit of the method itself, that the polls failed to see the rebuke coming because the thing that produced it was precisely the kind of thing their method is structurally unable to see, and that this blindness is worth understanding, because we are building an entire civilisation on methods with the same blind spot.

The anthropologist James C. Scott, studying why grand schemes to improve the human condition so often fail, identified the culprit as what he called legibility: the process by which a state, or a model, or a poll, makes a complex reality manageable by simplifying it into the few standardised categories it can measure, and then mistakes that simplified, legible version for the reality itself. The map, in Scott's telling, is not merely a reduction of the territory; it becomes a substitute for it, and the planner who acts on the map acts blind to everything the map left out, which is most of the actual texture of the world. A political poll is a legibility engine of exactly this kind: it reduces the vast, tangled, lived reality of a hundred crore people's hopes and grievances and loyalties into a handful of quantifiable variables, and it can see, with real precision, everything that fits those variables, and nothing at all that does not. And what does not fit the variables, the quiet grievance that people will not voice to a stranger with a clipboard, the local texture that no national category captures, the shame or pride or fear that moves a vote without ever registering as an opinion, is exactly where a democratic surprise lives.

Let me grant the counterargument, because polling is not worthless and the sophisticate's contempt for it is cheap. Polls are, most of the time, remarkably accurate; the aggregate methods have called election after election correctly; and to conclude from one dramatic miss that quantification is useless would be to overlearn from a single case in exactly the way I am cautioning against. The map is not the territory, but a good map is still enormously useful, and a world without polls would not be a world of deeper democratic understanding but a world of pundits guessing. This is true. The point is not that the map is worthless but that the map is a map, that its usefulness and its blindness are the same feature seen from two sides, and that the moments it fails are not random errors but systematic ones, clustered precisely around the reality it was built to simplify away. The poll sees what is legible and misses what is not, and a democracy's most important movements, the ones that humble the invincible, are so often the illegible ones, the shifts that happen below the threshold of what any standardised instrument can register.

Which is why the miss should be read not as a failure of polling but as a warning about the method we increasingly trust to see the world for us. We are building a civilisation that runs on legibility engines, on models and metrics and quantifications that reduce the tangled real to the measurable few and then govern by the reduction, and every one of them has the blind spot the polls displayed this fortnight: they see the legible and are blind to the rest, and the rest is where the surprises live. The Indian voters were not invisible; they were merely illegible, present in the territory and absent from the map, and they delivered their rebuke from exactly the blind spot the method could not cover. The lesson is not to abandon the maps, which would be foolish, but to remember, always, that they are maps, that the territory is larger than any legible reduction of it, and that the most important things about a people are so often precisely the things that no poll, no metric, no model can be built to see. The voters humbled the government this fortnight. They also, quietly, humbled the machines, and the second humbling is the one worth remembering, because the machines are not going away, and neither is the territory they cannot see.

The Ledger

Apple's partnership with OpenAI to power the hardest requests is the fortnight's tell: even Apple, the most vertically-integrated company in technology, could not build frontier AI alone and had to rent it. The dependence on a few model-makers deepens.

Data centers & power

Apple Intelligence promises to run much AI on the device, for privacy, but the hardest queries still go to the cloud, to the data centres, to the grid. The load keeps climbing whatever the marketing says.

Rates

The Fed held again and trimmed its own forecast to a single cut this year, more cautious than markets hoped. The last mile of inflation stays stubborn; patience, once more, extended.

Real estate

US 30-year mortgage 6.95 percent (June 13), back below 7. Small relief, still frozen, still waiting.

India tech

A weakened Modi government, dependent on coalition allies, means a less unilateral, more negotiated politics, which markets initially disliked and then accepted. For the technology sector, the continuity of economic direction matters more than the reduced majority, and that continuity holds.

What we called wrong

A good fortnight to admit it: this paper, like nearly everyone, took the exit polls seriously and braced for a supermajority. The voters proved us and the pollsters wrong together. We log it gladly. Being wrong about a democracy's capacity to surprise is the best kind of wrong.

The Back Page

The pollsters' humbling

Spare a thought, this fortnight, for the Indian exit pollsters, who on the evening of June 1 told a nation of a billion and a half people that the result was a foregone landslide, and were, three days later, comprehensively wrong. It is worth dwelling on not to mock them, though the mockery has been plentiful, but because their failure is a small, useful parable for an age that has come to worship prediction. These were not amateurs. They used large samples, sophisticated methods, decades of experience. They were confident, and they were unanimous, and they were wrong, and the markets that moved on their confidence lost and made fortunes on the correction. There is an old temptation, stronger now than ever in the age of big data and machine learning, to believe that with enough information the future becomes knowable, that human beings in the aggregate are just a system awaiting a good enough model. And then a few hundred million people walk into voting booths and quietly refuse to be the thing the model said they were. The exit polls will be back next election, more confident than ever, and they may well be right, and we will believe them again. But it is worth keeping this fortnight in mind, as a small monument to the stubborn, precious fact that the people, in the end, are not fully predictable, and that every confident forecaster who has ever been humbled by them is a reminder that some part of the human story still refuses to be told in advance.