Non-Submersible Units and the Forgotten Art of Sequence Writing

The Kubrick Quote No One Could Quite Explain

I’m always looking at different ways to structure stories - in words and in the edit. Three act, five act, nine act, Kishotenketsu, Freytag’s Pyramid, Fichtean Curve; I’m curious about everything. Any idea that can help you diagnose why something isn’t working, or to help you get over a blank page is worth investigating.

One phrase I came across a long time ago when looking for ideas about story structure was Stanley Kurbick’s “Non-Submersible Units”. I searched on and off for many years trying to find out where the term was originally from and what it actually meant.

The only place I could find it was in a bit of passing advice from Kubrick given to Brian Aldiss when they were working on AI. It’s often referenced by people from random Kubrick fans to other filmmakers.

But no-one agrees on the actual meaning of what he was talking about. You can’t find “non-submersible unit” anywhere else. It’s a Kubrick neologism, completely dissociated from any film theory.

The quote comes from a Brian Aldiss interview in A Life In Pictures and another from Stanley and Us:

I was always keen on the idea of narrative. My books always have a narrative. That is to say, cause and effect. That's what I like. But Stanley was less interested in that and he said to me “now forget about the narrative“ He said ”what you need to make a movie is six 'non-submersible units.“ That was the phrase he used: 'non-submersible units'.

And:

One of the many sensible and perceptive comments he made over the years was that a movie consists of, at most, say 60 scenes, whereas a book can have countless scenes. So, he said, it's very difficult to boil down a novel to make a film, as he found with The Shining. Much easier to take a short story and turn that into a major movie. 'All you need is six non-submersible units. Forget about the connections for the moment” ... once you've heard this, you see how 2001 was constructed.

Death of the Director: Inventing Meaning from Mystery

Unfortunately Aldiss never said explicitly how 2001 was constructed only that it was obvious if you look.

Originally I took him to mean Kubrick’s own cards, because elsewhere Aldiss describes The Shining as constructed in this way.

So are these the 2001 non-submersible unit breaks?

  1. Dawn of Man

  2. Trip to the Moon

  3. Jupiter Mission

  4. Jupiter and Beyond the Infinite

That’s four sequences which doesn’t track with “six to eight”. They do, however, fail to perfectly connect.

Diane Johnson said about The Shining:

Next came the process of deciding on a structure, that is, which scenes, which additional scenes, and in which order. We used an eight-act structure divided roughly into timed sequences, sketched out by Kubrick: First Day, Day of the Psychiatrist; Arrival; Before the Snow things are going well; Snow (lull); Big Day (argument, radio dead, finds scrapbook, key to room 217, Lloyd, Jack to room 217 [see 216n77]); Night scene, with Sno-Cat distributor cap; last, Elevator, calls to Halloran, last twenty-four hours of terror. He saw the first four sections as lasting forty-six minutes and the rest seventy-six—which of course the film greatly exceeded.

That tracks with Kubrick thinking six to eight in a movie. I have also seen:

  1. The monolith visits humankind in its infancy

  2. An early man discovers technology (Moon Watcher smashes the bones)

  3. The monolith is excavated on the moon by astronauts and sends a message to Jupiter

  4. Humankind send a manned mission to Jupiter to investigate

  5. Advanced technology (Hal) endangers the mission crew

  6. Technology is defeated and the surviving cremember rendezvous with the aliens

  7. The Starchild is born

The Aldiss discussion takes place in the 1980s. Perhaps Kubrick more explicitly developed his ideas after 2001, and the success of 2001s structure informs the idea rather than exhibits it exactly.

However, Ben Wheatley, a big Kubrick fan, interpreted non-submersible units as key images:

“I’m a kind of believer in the Kubrick thing about the non-submersible unit, where you look for the main images for the movie before writing the script and work backwards from that.”

That, alas, doesn’t sit well with Aldiss recounting Kubrick, or Johnson.

However, there’s no way of knowing what Kubrick meant, only what is most productive to you.

Breaking the Enigma: Kubrick's Strategic Thinking

After David Lynch died I read about his screenwriting methods. He recommended Screenwriting: The Sequence approach by Joseph Gulino who was his teacher at the AFI.

In the 1920s screenwriting manuals instructed to work in sequences to the length of a physical reel because there was often a physical stop to change reels. That meant stories of about 15-20 minutes. In the 30s the playwrights came in to write dialogue and they married three acts into the eight reels make a feature. So sequences would run to 8-15 minutes or there about.

At the same time I was skimming through Command and Control about nuclear weapons  on submarines. It suddenly occurred to me.

Submersible.

Non-Submersible.

Oh!

Kubrick was a military nerd. It’s a naval metaphor. He’s talking about the watertight compartments on ships and submarines that lock if there is catastrophic damage to prevent the ship sinking.

He’s saying each section of story, should stand on its own. If one bit fails for the audience, the rest still floats. Sequences don’t need careful connections if they are secure internally.

Six to eight. That’s fifteen minutes per sequence for a 90-120 minute movie.

Kubrick’s talking sequences.

It’s Studio System screenwriting in military drag.

Sequence storytelling turns a 90 page problem into a series of 15 page problems. It’s far more manageable and offers particular liberation if you’re drowning in plot complexity (as Aldiss was).

This is quite different from Robert McKee, Joseph Campbell, Syd Field and so on. It’s far more productive and basic. It says, make sure your story moves along every 15 minutes.

Lawrence of Arabia is discussed at length in the Gulino book. The sequences are as follows, and each has a setup, a goal and a deadline (ticking clocks are always an easy win). And all of the driving elements can be put as a question to be answered, with Lawrence at the center.

  1. Can Lawrence get out of the desert? 11.5 minutes. Yes.

  2. Can Lawrence get to Prince Faisal with his guide? 17 minutes. Yes

  3. Can Lawrence meet Faisal and get out in the retreat? 9 minutes. Yes

  4. Can Lawrence figure out the Arab problem? 14 minutes. Yes

And so on. This awesome MovieWise video covers the same material:

https://www.youtube.com/watch?v=GMG8nm6qWEo

Kubrick developed his own unique angle on sequence screenwriting. Aldiss notes that Kubrick wasn’t interested in plot and takes the compartmentalization of sequence structure to its logical extreme. Focus on dramatic moments, let the viewer figure out the rest. Why bother with the boring bits?

Given that context I would split 2001 into the following sequences, using the main questions drive the story through each part.

  1. Can the monkey survive? (Dawn of Man)

    • Yes. Technology acquired, learn to kill.

    • 18 minutes

  2. What is happening on the moon? (The Space Station section)

    • It’s something bad, but no reveal.

    • ?16? minutes (check for publish)

  3. What is the Object? (The Moon section)

    • The Monolith. What is the signal?!

    • ?16? minutes (check for publish)

  4. How does Discovery work smoothly?

    • It doesn’t.

    • ?20? minutes (need to check for drraft)

  5. Can we fix the relay?

    • No. HAL kills Poole.

    • ?20? minutes (check for publish)

  6. Can Bowman defeat HAL?

    • Yes. Bowman kills HAL.

    • ?20? minutes (check for publish)

  7. Can Bowman Meet the Aliens? (Jupiter and Beyond the Infinite)

    • Maybe!?

    • 20 minutes (check for publish)

Now that still doesn’t quite track with 15 minute sequences, but 2001 is a slow film…

Kubrick thought movies were more like a musical symphony. You remember and enjoy moments and scenes in the way you remember musical themes not the musical movements.

Barry Lyndon is a fascinating example. Effectively there is some highly dramatic scene with some setup, then a bunch of plot is skimmed over by Michael Horden’s voice over, and then another highly dramatic scene with some setup. It’s more like a series of vignettes strung together with a voice over. It’s great.

How to Play Studio Screenwriting Jenga

To construct the script this way are looking to create 6-8 self contained sequences of 10-15 minutes (pages), each with a beginning, middle and an end.

Gulino describes the main tools to keep people interested:

Telegraphing. You tell the audience where the story will go, and you go there. Or not. “Meet you at the Aquaba!” And you follow a few scenes to get there. Or, you fake them out and cut to the man getting arrested and you never get to the Aqaba.

Deadlines. These ones are easy and evergreen. The bomb is ticking. We have three days to get to Vegas. The bus explodes if it goes below 30 miles an hour.

Dangling Causes. A clear direction or an interesting question, withheld. This is similar to telegraphing but usually concerns character objectives. Often comes in dialogue hooks such as “I will steal the jewels!” but more subtly can also be asking interesting questions such as “Who are you?” in Lawrence of Arabia.

Dramatic Irony. The OG of suspense. Two characters talk, the audience knows there’s a bomb under the table. Or, in a RomCom the audience knows two characters are perfect for each other, will they realise this. Hitchcock was brilliant at layering this. Not only between audience and characters, but characters and characters. Hitchcock gets 5 minutes of nail biting from Cary Grant waiting in an empty field in North by Northwest. Similarly, Dario Argento is a world master at shots of people walking slowly in corridors. In Susperia he gets maybe the greatest horror tension and shock in cinema from someone waiting in an empty square.

Dramatic Tension. This is the OG of western drama. Somebody wants something badly, and they’re struggling to get it. (The Ark of The Covenant, Escape from Jurassic Park etc.) Often these are larger arcs but this can be the core of a fractal system, where scenes have similar dramatic structure. For Mamet “There are no characters… There are lines of dialogue meant to be said by the actor,” because “every scene must answer the questions: Who wants what from whom? What happens if they don't get it? Why now?" And (block caps in the original) “ANY TIME TWO CHARACTERS ARE TALKING ABOUT A THIRD, THE SCENE IS A CROCK OF SHIT.”

There are other techniques too, but these are more rare and complex. How you deploy them is up to the writer.

For example, the “Shall we shut down HAL?” scene of Bowman and Poole in the pod in 2001 complicates Mamet’s “crock of shit” dictum. As Hitchcock said, you can make any conversation interesting if you plant a bomb under the table and let the audience know it’s ticking.

That is, there are layers to narrative pleasure:

  • What happens next? (the peripheral interest)

  • How does what is going to happen next change the trajectory of what could happen after that? (the core interest, bigger divergence means more tension)

  • In the context of the larger story how much do I care to see any of these things anyway? (the interest multiplier)

The level of dramatic tension in any moment arises from how much that moment is likely to alter, in an interesting way, what the audience thinks they are going to see next.

Drama is not about scale, but relevance. This is a frequent miss in modern blockbusters.

The thrill is not ”Will the family/city/world/solar system/galaxy/universe be saved?” but “How much does saving x make interesting changes to what I expect see next?” This is why Hitchcock can get more out thrills from a static shot of a door than most blockbuster can squeeze out of $50m in VFX.

The audience are unconsciously (or consciously) calculating the odds of what happens next in terms of their interest in the story. So if the audience doesn’t care to see what happens after the world is saved, then no-one cares that the world is saved.

Kubrick’s non-submersible units are sequences focused on these moments. Kubrick’s push is towards minimalism. Don’t worry about connecting plot points, those can be fixed later if we have to.

This is bottom up writing, and I think much easier to work with than the Robert McKee and Syd Field conceptions of screenplay writing that start with a big structure. This is much more flexible, fast and generative.

In the edit suite and on the page, I have found too much plot is an easy trap to fall into. The audience wants to see exciting things on screen, you need some plot to make that happen but the moments are more important. This also dovetails with Howard Hawks note that “All you need are three great scenes and no bad ones.”

The right amount of plot is the amount that lets you get the maximum juice out of the best dramatic moments that elucidate the theme. Sequences offer a good tool to start solving that problem.

So my suspicion is that non-submersible units are just sequences renamed. He was telling Aldiss to relax and write the old fashioned way. If you write interesting watertight 15 minute sequences, then you have yourself a movie. It’s an easier problem to solve than 120 blank pages.

Kubrick’s interesting argument is that you can push this logic to the edge because plot is not so interesting. The sequences barely need to connect. We only care what’s going to happen next.

Previous
Previous

A Hitchhikers Guide to Large Language Models 1

Next
Next

Editing Controls