What a director actually decides

The question

Every scored film here carries twenty readings — how dark, how playful, how tense, how beautiful, how tender. Two films can therefore be placed a measurable distance apart: how differently they feel, as one number.

Which makes an old argument testable. People talk about a director having a signature, a sensibility you could recognise blindfolded. Is that a real property of the films, or is it a story we tell about people whose names we know?

The naive test is easy and the answer looks emphatic. Take any two films at random: they sit 6.09 apart. Take two by the same director: 4.33. An enormous gap, and for about an hour it looked like the whole finding.

Most of it is the genre

It is the wrong comparison. Directors do not work at random: they work in one or two genres, over one or two decades. Horror films of the 1980s resemble each other whoever made them. So the question is not whether a director’s films are more alike than any two films — it is whether they are more alike than other films of the same kind, from the same years.

Against that baseline the effect shrinks hard, and survives. Two films sharing a primary genre and a decade sit 4.79 apart. Two that also share a director sit 3.73 apart, across 4,373 such pairs. The director is worth a further 22%.

Genre and decade close about a fifth of the gap. The director closes a fifth of what is left — roughly 45% as much again as knowing both the genre and the decade.

So a signature exists. It is considerably smaller than the raw comparison suggests, and most of what people hear as a director’s voice is the genre talking.

What they hold on to

The interesting part is which of the twenty readings a director controls. Asked without the control, the answer is comfort, playfulness, violence, darkness and tension — tone and intensity, which is more or less what you would expect.

Every one of those is genre. Hold genre and decade still and the ranking does not merely shrink, it inverts.

Violence looked like the third most directorial reading in the catalogue. Controlled, it is negative: a director’s films vary more in violence than their genre peers do. Romance falls furthest, to −15%.

What holds up is a strange, coherent little group. Ambiguity survives best at 34%, then how cerebral a film is at 26%, existential weight at 23%, comfort at 21%, beauty at 19%.

Read them together and they are not a subject or a mood. They are a set of decisions about how much to withhold: how much is left unresolved, how much thinking the film asks for, how much weight it puts on you, how much reassurance it allows, how much it cares what the frame looks like.

A director does not own what the film is about. Whether there is violence in it, whether anyone falls in love, how tense it gets — those come with the assignment. What they own is how much they explain.

The auteurs are the least consistent

Which leads somewhere uncomfortable. If a signature is measured as films that feel alike, the strongest signatures in the catalogue do not belong to the people with reputations for having one.

Directors working mostly in animation average 3.37. Directors working mostly in drama average 4.90. The tightest filmographies here belong to people making the same film carefully, many times — and the widest belong to the canon.

Of 342 directors with six or more well-known films, 234 — about 68% — make films more alike than their genre and era would predict. The rest are more varied than their peers, not less.

The strongest signatures

How much tighter each director’s own films sit than other films of the same genre and decade.

• Kazuya Tsurumaki6 films · -59%
• Tom McGrath6 films · -55%
• William Lau6 films · -53%
• Chris Renaud6 films · -53%
• Jackie Chan7 films · -52%
• Ron Clements7 films · -50%
• John Musker7 films · -50%
• Hideaki Anno7 films · -48%
• Garry Marshall10 films · -47%
• Nick Park8 films · -46%

The widest range

The same measure, the other way: filmographies that cover more ground than their genre and decade would suggest.

• Robert Wise6 films · +36%
• Billy Wilder10 films · +33%
• Lawrence Kasdan6 films · +28%
• David Koepp6 films · +26%
• George Miller12 films · +25%
• Kevin Smith13 films · +24%
• Darren Aronofsky9 films · +23%
• Richard Linklater12 films · +23%
• Marc Forster11 films · +22%
• Ethan Coen15 films · +21%

What this is not

This measures how a film feels. It does not measure how one looks, how it is cut, how it sounds, or how a camera moves. A director can be unmistakable frame to frame and still make films that feel nothing like each other — and by these numbers, that is exactly what the most celebrated ones do.

Which is probably the better description of what people admire anyway. Nobody praises a filmmaker for making the same emotional object repeatedly. The consistency we actually prize is a consistency of craft and attention, and range is the thing we give people credit for. This measure finds the opposite of prestige because it is measuring the opposite of what prestige rewards.

One limit worth stating: it takes six well-known films to appear here at all, which leaves out anyone early in a career and anyone whose work never travelled. And a director credited on a film is not the only author of it.

The films that sit furthest from everything else in the catalogue, by a different measure: the Cinnamon Hundred.