Only two people really understand how “Bandersnatch” – the latest episode of Black Mirror – actually works. Producer Russell McLean is one of them. But even he isn’t sure how many endings there are. “I’m not even sure how you’d count,” he says. Assistant editor John Weeks is the other. “There are eight variants of one part of the story depending on which piece of music you’ve chosen,” he says. Weeks and McLean have spent months wrapping their heads around the labyrinthine intricacy of the system that holds the episode together.
The plot of “Bandersnatch” is classic Black Mirror, following a teenager in 1980s London as he struggles to make his own choose-your-own-adventure video game. But this is something completely new: a maze-like hybrid, part video game, part film, in which you are in total control.
At the heart of the episode are two-and-a-half hours of footage divided into 250 segments, hidden behind an elaborate series of decisions. Starting as a seven-page outline, “Bandersnatch” quickly grew into a 170-page script – hand-coded at first – that required Netflix to build its own choose-your-own software to bring it to life. To stream the episode, the company had to work out a way of simultaneously loading multiple versions of each scene so viewers could follow different narrative paths without encountering the dreaded buffering circle.
For all those involved in “Bandersnatch”, the ambition was clear: to ensure that nobody watching knows how migraine-inducingly complex the system that powers it is. But to truly appreciate Black Mirror’s latest dark turn you’ve got to break the whole project apart. It all started in a meeting room in Netflix’s Los Angeles offices in May 2017. Netflix’s vice-president of product Todd Yellin and director of product innovation Carla Engelbrecht Fisher had a proposition for Charlie Brooker and Annabel Jones. They wanted the Black Mirror co-creators to produce the platform’s first ever interactive film for adults.
Brooker and Jones knew exactly how to react: they smiled politely, nodded, then completely ignored the idea. “It would just feel a bit cheap,” says Jones, recalling the meeting. Brooker agreed. “Afterwards I was like, ‘Those things never really work.’ We couldn’t see what the story would be.”
But Netflix had other ideas. By the time of that meeting, the firm had been quietly experimenting with interactive films for a while. Its first choose-your-own-production – a kid’s adventure called Puss in Book: Trapped in an Epic Tale – would be released in June 2017, with three more children’s films lined up for the next year and a half. Netflix had proved it could make interactive films work, but it wasn’t clear where the experiment should go. Was this just a fun gimmick for kids, or the beginning of something that would change the way we experience film on a mass scale?
“Bandersnatch” might provide some answers. Set in the mid-1980s, it puts viewers in control of Stefan – a painfully awkward 19-year-old video game developer who has plans to make a video game based on a choose-your-own-adventure he found in box of his mother’s old belongings. Every few minutes, the viewer must make decisions on Stefan’s behalf, and watch as the consequences of those decisions unfold. It is utterly unlike any episode that has gone before it and yet is unmistakably part of the Black Mirror universe. Only this time the lingering dread that accompanies most Black Mirror episodes hits uncomfortably close to home.
Brooker and Jones thought they had evaded Netflix’s hankerings for an interactive episode, but found they weren't in control of this plot. A few weeks after that LA meeting, they were discussing story ideas and floated the idea of a young game developer tormented by the thought that his own life was playing out like one of his choose-your-own adventures.
“This idea popped up that would only work as an interactive,” Brooker says. “It was a moment where we went, ‘Oh great that’s exciting, it’s a story that would only work in this way.’ Five minutes later we thought, ‘Oh shit, now we’ve got to do that, and it’s probably going to be complicated.’”
He was right. In a few months “Bandersnatch” evolved from that seven-page overview into an episode with millions of permutations. While the decision tree behind Puss in Book is recognisable as a run-of-the-mill flowchart, the “Bandersnatch” schematic is an impenetrable web of decisions and scenes that must be played through to be understood. When he started work on it, Brooker estimated that it would require as much work as two standard Black Mirror episodes. In retrospect, he says, it was more like four.
At its heart is a web of decisions. To begin with these are mundane – choosing between two songs or breakfast cereals – but they quickly get much more serious. Choices will send viewers branching off in different directions, only to sometimes jackknife back. Along the way, the system remembers past decisions, and these will influence things viewers see later in the episode.
Although most of the people behind the episode refer to it as a film, Weeks, who programmed computer games earlier in his career, likens it to a video game. As viewers make decisions they unlock certain parts of the narrative – a little like finding the key to another level in a game. If they die – and there is more than one way to die – viewers can re-spawn and replay the episode by making different decisions and unlocking different parts of the narrative. The logistics of hooking all this together and delivering a seamless piece of television are, for everyone involved, the source of countless late-night headaches.
After Netflix approved the idea, Brooker started working the script into an flowchart, detailing where each decision point would take viewers. As the decisions multiplied, this lo-fi solution swiftly became unwieldy, and Netflix introduced Brooker to the open-source software Twine, popular with writers of choose-your-own-adventure novels.
“Charlie took to Twine quite quickly,” says McLean, who was the visual effects supervisor for most of seasons three and four of Black Mirror. In Twine, Brooker could code the script so it was possible to directly jump between sections, and add the memories that cumulatively shape the film’s different branches. The endings – there are several – are in effect locked behind a series of if-this-then-that statements. To give a simplified example: if you chose W over X, and then opt for Y instead of Z, you may be shown a different ending to someone who picked X and then Y. In reality, the logic behind “Bandersnatch” is considerably more complex.
McLean and Weeks were a couple of weeks into editing when they switched to using Branch Manager – software that Netflix developed to help manage the sprawling network of interconnected clips. Now McLean could play through them and start to see the episode taking shape for the first time. “That was just a case of going through each individual known segment in the Twine document and analysing it, breaking it down, and turning it into the equivalent piece of logic for Branch Manager," says McLean.
The results spider out across three monitors in a cramped, dimly lit editing suite in west London. To get there you have to ascend a series of narrow staircases and pick your way down a craggy hallway in Ealing Studios – the vast production lot that hosted the shoots of Shaun of the Dead, The Theory of Everything, and the interiors of the TV series Downton Abbey.
As Brooker, Jones and McLean got deeper into the process of fleshing out “Bandersnatch”, it quickly became clear that the story was growing in unexpected directions. Linear stories can only grow in one direction – by getting longer or shorter – but with this non-linear story Brooker and Jones had to think about how the show would grow horizontally, through adding or removing paths in the middle. At some points the script was so complicated that it had to be debugged in order to make it run. “It would literally crash,” Brooker says.
“A lot of work was just in the basic structure of it. That took the most time, and then the script was adding all the extra fun and detail and richness of the world into that framework,” says Jones. But giving viewers the option to map their own route through the episode brought up a disquieting aspect of the format: most viewers will only see a tiny slice of the plot. “When you make a film you want someone to see it all,” says Jones. “We’ve tried to think about every way that people can play it, but there are so many different ways that it’s hard to totally understand the emotional reaction that they are going to have.”
Structurally, “Bandersnatch” is composed of three main building blocks. First off there are segments – each one a distinct chunk of footage that is linked to other segments by decisions. Each decision leaves behind memories – the second building block – that are built up throughout the episode from the choices viewers make, and these influence which segments they’ll see later on. Two viewers who make exactly the same choices in all cases except one will have slightly different memories, so may end up following completely different narrative paths.
Finally, you’ve got 41 recaps, which play when the viewer gets to one of the episode’s endings. Shot with old-fashioned tube cameras and played on a TV screen within the episode, each recap presents a quick overview of the decisions a viewer has made to get to that ending. From that screen they can then choose to re-enter the episode at an earlier point – a bit like re-spawning at a checkpoint in a game – or to exit to the final credits scene, knowing that there are still parts of the episode that remain hidden.
Brooker likens watching the episode to a tourist visiting a city for the first time. “You want to feel that you’ve seen enough of a city before you leave to feel satisfied,” he says. Yet he is nonplussed about the prospect of much of “Bandersnatch” remaining hidden to its viewers. “I don’t mind the thought of people not seeing a lot of stuff,” he says, comparing the film to open-world video games where vast areas go unseen by most players. Brooker is a longtime fan of video games – something he mentions so often that it has become a running joke between him and Jones, eliciting a roll of the eyes and a faux-incredulous “Really?” every time he drops in a reference to gaming.
“I just drift off – I probably haven’t finished most video games,” he says. “I’m aware that if you’re creating a game you might create whole vistas that people will never see,” he says. Although there are two-and-a-half hours of total footage in “Bandersnatch”, most Netflix employees who have been testing the film internally reach an ending after watching for between 60 and 75 minutes.
The spidering horizontal structure of the episode also meant that some scenes ended up being a little hard to reach – locked behind an arcane combination of unlikely decisions. Fairly late in the editing process, the team decided to make it much easier to access one ending, and brought another branch of the narrative much closer to the heart of the film. Even so, parts of it may remain elusive to all but the most determined viewers. “There are some things that are really hard to find in it,” Brooker says. “I couldn’t tell you how to get to them.”
Despite spending most of the last 18 months creating this interactive behemoth (while also writing and executive-producing Black Mirror season five), Brooker isn’t quite sure how to refer to his creation. Will viewers watch the episode, or interact with it? Or are they not really viewers at all, but more like participants, or players? Is the episode a successor to the choose-your-own-adventure games of the mid-1980s, or closer to a more traditional episode of Black Mirror?
Brooker and Jones bounce around between these different options, at times referring to "playing" the episode, but usually referring to “Bandersnatch” as an interactive film. Not an entirely new format in itself, but a kind of levelled-up version of normal filmmaking. “To me it feels like it is an additional genre. It has become a whole new form of doing things that can slot in alongside more traditional films,” Brooker says. “It would never supplant doing a linear story, but it’s another tool in your box as a writer.”
For Todd Yellin – the Netflix VP who first floated the idea of an interactive Black Mirror episode – getting the terminology around the episode right is of critical importance. “We debated this a lot,” he says, leaning back in his chair in a meeting room on the ground floor of Netflix’s cavernous Los Gatos headquarters.
Yellin places his hands about a metre apart. At one end of the scale, he says, you have a normal video game. At the opposite end, Yellin says, indicating his right hand, you have a normal Black Mirror episode. He moves that hand a few centimeters to the left. "That’s 'Bandersnatch'.” Yellin is keen not to call it a game because he doesn’t want Netflix users to be wrong-footed when they load up “Bandersnatch" for the first time.
“If we call this a game, [gamers] are going to be expecting a big game world,” he says. But Black Mirror fans who aren’t gamers might be put off if they think it’s too interactive. “We think [interactive film] best characterises it and sets the expectations about what people are going to get when they turn it on.”
Engelbrecht Fisher, who heads up Netflix’s interactive efforts, acknowledges that Netflix isn’t venturing into completely uncharted territory. “We know we’re not the first, but we’re the first to do it at scale,” she says.
The choose-your-own-adventure format – which has its roots in Edward Packard’s 1976 novel Sugarcane Island – has had a few abortive forays into cinema. In 1992, a company called Interfilm partnered with Sony to fit 42 US cinemas with controls that allowed customers to make plot decisions. After a handful of poorly received films, the kit was dismantled and the four films made for the project were shelved. In 2016, the interactive film Late Shift was screened to positive reviews at the New York and Raindance film festivals.
But “Bandersnatch” will be released on another level altogether. At its launch, it will be available in 27 languages to 137 million Netflix subscribers. The only notable devices excluded from running the interactive film at its launch will be Apple TV and Google Chromecast as Netflix didn’t have time to optimise the episodes for those platforms before its release. Regardless, Yellin is confident that the “vast majority” of Netflix customers will be able to access a device that can play the episode.
“This, I imagine, will be the first interactive film that is so mainstream and accessible. That is daunting,” says Jones. But is there a danger that the success – or failure – of “Bandersnatch” will come to be seen as a judgment on interactive films themselves?
Yellin doesn’t think so. “It’s just one story,” he says. Although Black Mirror was his first option, Yellin says he would have taken the idea to someone else if Brooker and Jones hadn’t changed their minds. In any case, he expects the release of “Bandersnatch” to blow the doors wide open when it comes to new ideas for the format. “There are going to be great storytellers who see this as a great opportunity, and we expect a lot of pitches to come rolling in during 2019.”
Although the reception of “Bandersnatch” won’t decide the fate of Netflix’s experiments with interactive films, Yellin and Engelbrecht Fisher will be looking through the data generated by viewers’ choices to work out how to improve future productions. “We expect there will be some people who get great joy out of watching an hour, and then they’re done. And there will be other people who get great joy watching three hours and keep on exploring multiple pathways,” Yellin says. “What we don’t know are how big those different audiences are.”
Engelbrecht Fisher is particularly keen to understand what makes a choice feel good to a viewer. “We think a great choice causes you to care. If we just ask whether you go left or right and there’s no context, you don’t care,” she says. From her experience of interactive children’s films, she knows that most audiences split fairly evenly between choices, but a decision that skews sharply towards one option might become a talking point after the show has debuted. One decision in “Bandersnatch”, she estimates, will split audiences 90/10.
It’s also possible to watch the episode without lifting your remote once. Brooker and Jones have pre-programmed a default path through the episode that will lead all the way to an ending, but to do that viewers will have to choose to ignore decision points.
To design the best way to present these decisions, Engelbrecht Fisher tested prototypes of the decision user interface with Netflix viewers. This helped her team work out how to space those decision points. “One of the things that we would see is that there is a point when you make a choice that the remote starts drifting down and it finds its way between the couch cushions before too long,” she says. Eventually, they settled on putting a choice in the film every three to five minutes – long enough for cinematic storytelling, but short enough that the viewer never feels the decision mechanic has been left behind.
Stylistically, too, the decision points are designed to get the viewer to engage with the decision-making experience. After the two decisions rise up from the bottom of the screen, a timer narrows across the screen letting the viewer know how long they’ve got left to pick a decision. In the background, the occasional use of rising synths ramp up the tension, while the camera focuses on Stefan’s face. Since rewinding or fast-forwarding the episode is impossible, Netflix is testing different navigation controls, giving some viewers the option to skip back to earlier decisions points, while restricting others to the ability to skip forward or back 10 seconds at a time.
So what’s the payoff for Netflix? Interactive films are more expensive and take longer to make than linear narratives, but they might mean viewers coming back for more. “These are made knowing that people will generally want to go back and re-engage,” says Engelbrecht Fisher. “If we didn’t see that happen we would look to see if something was amiss in the story and if we just didn’t tell a good one in this case.”
For Yellin, this is just one experiment among many – albeit one that he’s particularly fond of. “Right now we’re trying unscripted reality TV programmes that you wouldn’t expect to get on Netflix, but you will see them on Netflix if it works,” he says. Will the company put out more interactive films? “I hope so,” he says. “That’s to be seen.”
Years before he started work on “Bandersnatch”, Brooker had an inkling he’d like to try out alternative versions of his films. For season three’s “Playtest”, where a backpacking American agrees to test out an experimental video game, Brooker thought about putting together an even nastier version.
“I remember at one point pitching the idea of nightmare mode where, if you watched “Playtest” a second time, it had a second ending, so I really liked the idea of someone freaking out,” he says. He and Jones also thought about alternative endings for “White Bear” in season two, where a woman wakes up in an unknown house and finds herself pursued across a city by smartphone-wielding strangers.
But when Brooker and Jones started work on “Bandersnatch”, they turned to David Slade, who had directed the season four episode “Metalhead”. One of the most linear stories in Black Mirror, “Metalhead” is entirely shot in black and white and follows a woman pursued across a post-apocalyptic landscape by a robotic dog.
Hoping to tap into “Metalhead's” claustrophobic intensity, McLean contacted Slade in December 2017. Initially, the director found the interactive element one of the least interesting elements of the film. “I was dead against the idea of a video game, I wanted it to be a dramatic thing,” he says. As with Brooker and Jones, his first thought was “no”.
The more he thought about it, however, the more Slade realised that the interactive part of “Bandersnatch” wasn’t supposed to be a gimmick – it was the only way this story could be told. An episode that lets viewers pick their own route through an episode naturally subverts the idea that film-makers have the final say over how a film will be viewed or understood. “The question really becomes: how do you get away from a dominant reading because the idea of a dominant reading is really antithetical to the plot,” he says.
Multi-narrative, non-linear stories are also antithetical to production schedules. Many of the scenes exist in multiple different versions, each one subtly tweaked depending on the decision a viewer has chosen. The difficulties of keeping them consistent, while avoiding continuity errors, make the format a nightmare for directors, editors and actors.
“It’s not just different on paper – the entire process is different,” Slade says. When it came to creating three variants of the same scene, he started by filming the first version all from one angle, then shooting versions two and three from the same angle.
He tried this for the first two days of the 37-day shooting schedule, but soon realised that it was impossible to get a consistent feel when the actors were having to constantly switch between different versions of scenes. “We realised very quickly that we couldn’t do that, and we had to take each scene in a linear order,” Slade says. “So with each scene we had to complete it before we went back and did it again [in a new version]. Because the reactions and where you came from could be different.”
It was also crucial to get the small details right when it came to recreating the humdrum world of 1980s London. “The nostalgia helped normalise how dark we were going to go,” says Slade. When it came to the set for Stefan’s home, the crew visited 59 houses before landing on one with the right feel. “It’s trying to find the right design aesthetic so it’s authentic to the time but it also looks good,” says McLean.
Much of this meant recreating the world of Brooker’s childhood – England in the 1980s. Brooker was raised in Oxfordshire, not London, but like Stefan he had a ZX Spectrum and was obsessed with games and trying to work out how to code. In a way, McLean says, “Bandersnatch” is a Charlie Brooker biopic.
When I ask Brooker whether he agrees, he deflects the question to Jones. “This is your world, and your era,” she says to Brooker. “The number of conversations we’ve fucking had about Spectrum games and the sound a Spectrum makes and ‘It wouldn’t make a sound like that if you were pressing that button there’, and ‘It wouldn’t load anything like that’.”
“The contempt is interesting isn’t it?” Brooker jokes, although he admits that he finds the episode’s inaccuracies slightly painful. Stefan programmes his game in BASIC, which would never really happen, he says, and the error codes that appear on screen are completely made up. “The level of detail that you have revealed to us about your knowledge of this world is shameful,” Jones says.
But by getting the basics right – or at least most of them – Brooker and Jones manage to ground the story in reality despite the unfamiliar choose-your-own-adventure frame. The result is a film that plays with genre boundaries without ever feeling too self-consciously clever. “It’s not like we sat down and thought we’d write a meditation on modern storytelling because we don’t tend to work like that because we’re simpletons,” Brooker says. “We tend to instead think, ‘What’s a funny hook?’ And that was our hook.”
It just happened to be a hook that required seven drafts of a 170-page script, two-and-a-half hours of footage divided into 250 segments, and around eight months of production. In order to put out “Bandersnatch”, Netflix had to invent its own choose-your-own-adventure software and work out how to compress and pre-load multiple versions of scenes so viewers could follow different narrative paths without encountering the buffering circle. The huge file sizes involved in streaming multiple versions of the same scene, by the way, are why the film isn’t available for download.
Brooker says the first time he watched the completed episode of “Bandersnatch” was as close to profundity as he ever gets. “Seeing it work and work really smoothly has been quite odd. I found it almost emotional – as emotional as I get about anything, it was going to be something geeky.”
If Netflix’s bet is right, and it also works for viewers, then this episode of Black Mirror might open the floodgates to a whole new era of film-making. But Brooker isn’t thinking about being the torchbearer for a new kind of narrative cinema. He’s already thinking about the next interactive episode. “It’s interesting that in doing this you learn the ropes,” he says. “Recently I’ve had all sorts of ideas for things that you could do…”
Updated 28.12.18, 13:40 GMT: John Weeks was assistant editor on "Bandersnatch". Tony Kearns was editor.
This article was originally published by WIRED UK
