Something worth carrying to the end of this, unanswered for now: what exactly did the governing party lose on Tuesday?
It did not lose the election. Narendra Modi’s alliance won 292 of 543 seats and he was sworn in for a third consecutive term, which two prime ministers in Indian history have managed. On any ordinary reading that is a comfortable victory.
The belief being tested here is the one that governed every forecast in the weeks before: that this was a formality, that the only question was the size, and that the exit polls converging on a landslide meant a landslide. The record says something more interesting about what agreement among forecasters is worth.
What everybody expected
A large majority for the Bharatiya Janata Party on its own, with talk in the party of four hundred seats for the alliance, and exit polls on 1 June clustering around that figure.
The clustering is the part to notice. Multiple independent pollsters, using different methods, arrived at similar numbers, and that agreement was taken as corroboration.
It is not corroboration. Indian exit polling has a long and well-documented record of missing, and pollsters operating in the same environment with the same broad assumptions will miss in the same direction. Convergence among forecasts is evidence about the forecasters’ shared model, not about the world.
What the record shows
The BJP won 240 seats. In 2019 it won 303. A party that had governed with a single-party majority for a decade no longer has one, and the number required is 272.
The alliance holds the government because its partners supply the difference, which means two regional parties now hold something they did not hold on Monday: the ability to end it.
The Congress party nearly doubled its own total, to 99, and the opposition alliance took 234 between them. That is not a winning position and it is a functioning parliamentary opposition, which India has not had in ten years.
The first test: where it moved
Look at the map rather than the total and the result stops being a national swing.
The losses are concentrated, most strikingly in Uttar Pradesh, the largest state and the one the party’s national arithmetic has rested on. A party that loses ground there has lost ground in the place it could least afford to, and the reasons appear to be local: candidate selection, the effects of a caste arithmetic that the opposition organized around, and specific constituency-level grievances.
National explanations were offered immediately and most of them do not survive the map. This was not a uniform verdict on a decade. It was a set of state-level results that summed to one.
There is a second thing the map shows and it cuts against the simplest reading. The alliance gained ground in places it had never held, in the south and east, where the party’s organization has historically been thin. So this was not a retreat. It was a redistribution, and a redistribution that cost more than it gained because the seats lost were in states that carry more of them.
The second test: what a coalition changes
Governing with partners is not a weaker version of governing alone. It is a different activity.
A single-party majority means legislation passes when the leadership decides it should. A coalition means every contested measure is negotiated with people who have their own state-level interests, their own elections to fight, and the standing ability to withdraw.
The two largest partners are regional parties with a history of changing sides, which is a matter of record rather than an insinuation. Neither has any incentive to bring the government down now and both have every incentive to make that possibility visible.
The practical consequences fall on exactly the measures that require the most political capital: anything touching the constitution, anything touching the federal balance, anything that a regional partner’s own voters would object to.
There is also a resource question that a coalition creates immediately. Partners expect ministries, and ministries are finite. A cabinet assembled to satisfy two regional parties is a cabinet in which several portfolios are held by people the prime minister did not choose and cannot easily remove, which is a constraint on execution rather than on legislation and is the one that bites daily.
The third test: what the campaign said it was for
The alliance’s own slogan named a target of four hundred seats, and that target was doing work beyond enthusiasm.
A two-thirds majority in the Indian parliament is the threshold for constitutional amendment. Opposition parties campaigned explicitly on the claim that the number was being sought in order to use it, and the government denied this.
Whatever the intention was, the arithmetic that would have permitted it does not exist. That is the clearest thing the result did, and it was done by voters in a handful of states who were not, in the main, voting about the constitution.
What an opposition of 234 can actually do
Worth being concrete, because “a real opposition” is the phrase everybody reached for this week and very few specified.
Numbers of that order change parliamentary mechanics rather than outcomes. Committee composition shifts, because committees are constituted in proportion to strength, and committees are where legislation is examined line by line rather than debated. A bloc of that size can force divisions instead of letting business pass on voice votes, which converts a government’s preferences into a recorded list of who supported them.
It also changes the arithmetic of the upper house and of the presidential election, both of which are decided by electoral colleges that include state legislators, and neither of which is settled by Tuesday’s result alone.
None of that stops a determined government. All of it slows one down and makes it legible, and legibility is the thing a decade of large majorities removed.
The answer
Not power. The party governs, its leader is prime minister, and the machinery of the state is unchanged.
What it lost is the ability to act without asking. For ten years the distance between deciding something and doing it was very short, and that distance is now populated with people who have to be persuaded, and who can be seen to have been persuaded, which is a different thing again.
Which is the part that will take years to show. A government that must negotiate in public leaves a record of what it wanted and did not get, and that record is the raw material of the next election in a way that a decade of unobstructed majorities never produced.
