The middle fell out

First, the number that is wrong

Every film here has a position in twenty-dimensional emotional space, so there is a direct way to ask whether cinema is homogenising: measure how far each film sits from its nearest neighbours, and watch that distance over time.

Do it naively and the answer is dramatic. The average film of the 2020s sits eleven percent closer to its neighbours than the average film of the 1950s. That is the number that would make the headline, and it is largely an artefact of counting. There are twenty times as many films from the 2010s in this catalogue as from the 1950s, and the more films there are, the closer the nearest ones will be, whatever the films are like. A city has closer neighbours than a village and it says nothing about the people.

Every figure below therefore comes from equal-sized samples of each decade, so the number of films cannot move the result. Three independent draws agree to within a rounding error.

Three cinemas, not one

Corrected that way, the whole catalogue barely moves: distinctiveness dips into the 1990s and comes back. On that evidence alone you would conclude the homogenisation of cinema is a myth, and that is exactly what this piece originally said.

It was the wrong population. A random sample of everything with a couple of hundred votes is mostly mid-size and small films — independent, international, genre, the long tail. The complaint about modern cinema is not about them. Ask the same question of the two hundred most-watched films of each decade, and of the hundred and fifty most expensive, and the answer inverts.

Distinctiveness indexed to each series’ own starting decade. The catalogue ends the period at 94. The most-watched films end at 89 and are still falling. The most expensive films were flat until 1990 and then fell off a cliff.

The expensive line is the one to look at twice. Through the 1980s and 1990s the biggest films of the decade were as unlike each other as the biggest films had ever been. Then, at the 2000s, that stops: a twelve percent collapse in a single decade boundary, exactly where the modern franchise era begins. The 2020s recover a little, which is the first encouraging number in this piece.

So the variety did not leave cinema. It left the centre. The long tail stayed varied and even recovered after the 1990s — which is precisely why a random sample of the catalogue shows nothing. Small and foreign films are doing the work of keeping cinema diverse while the middle narrows.

It is not just the capes

The obvious culprit is the obvious culprit. Films carrying comic-book or franchise markers were three percent of the most-watched two hundred in the 1970s. By the 2010s they were nearly a third of everything people watched.

Share of each decade’s two hundred most-watched films carrying comic-book or franchise markers.

But they do not carry the finding. Delete every one of them and re-run the measurement on what is left, and the convergence in the most-watched layer softens from twelve percent to nine. Superhero saturation accounts for roughly a fifth of it. The other four fifths is mainstream film in general becoming more like itself — the comedies, the thrillers and the dramas that people actually saw, converging on each other without any help from a cinematic universe.

That is worth saying plainly, because “it is all superhero films” is the comfortable version of this complaint. The uncomfortable version is that the films you would name as the alternative have been narrowing too.

What narrowed: the end of not knowing

An aggregate distance tells you that films converged, not what they converged on. Most candidate answers collapse the moment you hold genre fixed — films look heavier or more violent across the decades largely because the mix of what gets made changed, and that is a finding about the slate rather than about the films.

One reading survives that test and gets stronger under it. Ambiguity: whether a film leaves you unsure what happened, or what it meant. Comparing dramas with dramas, so the genre mix cannot explain it:

Share of well-known dramas scoring 60 or more for ambiguity. Roughly one in three in the 1960s; one in eight today.

This is not a drift, it is a step. Ambiguity holds through the 1960s and 1970s, halves between the 1970s and the 1980s, and then sits flat for forty years — through the indie boom, through digital, through streaming, through prestige television. Whatever happened, happened once, and none of the upheavals since have moved it. Films stopped being willing to leave a question open, and they have not started again.

The films that still refuse to resolve are browsable — and they are not a fringe. They include some of the most admired films ever made, which is the awkward part.

What this is not

It is not a claim that films got worse. Distinctiveness is not quality, ambiguity is a choice rather than a virtue, and some of the least resolved films ever made are also among the most tiresome. It is a claim about range: the films at the centre of the culture differ from one another less than they used to.

The early decades need a warning. A film from 1970 only appears here if people are still watching it, so the old end of every line describes what survived rather than what was released, and what survives skews towards the ambitious. That flatters the past in all three series. It does not explain the 2000s break in the expensive line, because survival has no opinion about which side of 2000 a film falls on, and it does not explain why the three populations diverge from each other — they are all subject to it equally.

This piece originally ran with the opposite emphasis: that the convergence was mostly a counting artefact and cinema had not really narrowed. That was true of the catalogue and false of the cinema anybody watches, which makes it the wrong headline. The correction is the finding.