I finally was able to watch Black Mirror: Bandersnatch last night, after years of the Apple TV Netflix app telling me it wasn’t supported.1As far as I know, it’s still not. I ended up watching it on the built-in app on my “smart” TV. Technological dystopia indeed! I can’t talk about it in any detail without spoiling it, which would be a shame because its best moments are when it does something surprising with the format.
My short “review” is that it’s absolutely worth watching. As an episode of Black Mirror, it’s pretty strong. As a 1984 period piece — specifically to being a young tech nerd in 1984 — it’s really fun. And as an examination and indictment of all the implications that come from interactive fiction, it’s interesting. My major criticisms are that it doesn’t do much that’s actually groundbreaking with interactive fiction, apart from delivering it to a wide audience in a different context. And ultimately, it doesn’t work that well as a coherent narrative.
I think the most remarkable thing about Bandersnatch is that it seems like Charlie Brooker (creator of Black Mirror and writer of this movie/episode) genuinely understands all of the implications of interactive fiction. Which is remarkable because I think a lot of people working in video games and interactive fiction full time still don’t get it.
That’s not to say that it’s a super-accurate depiction of video game development, even in 1984. There are several scenes where a programmer furiously types a few lines of BASIC code and hits the RUN button (?) to see semi-3D graphics or a perfectly-rendered high-resolution title screen, which I don’t think is all that accurate. But that’s creative license, and complaining about it is as pedantic as faulting a car chase for going through the most scenic parts of a city instead of a real-world route.
What Bandersnatch does get right are the fundamentals: the quickly-expanding complexity of branching narratives, the lack of genuine agency on the part of the player, and the lack of stakes in the player’s story.
The Sense of an Ending
My first impression was that the movie was an interesting but ultimately failed experiment. It started out so much better than I’d expected, lobbing a few easy choices at the audience but quickly getting into the good stuff: playing with choice-based narratives in clever — and definitively Black Mirror-like — ways.
But as I got further in, the options started feeling more contrived and the resulting branches less impactful. In other words: like just about every other branching narrative I’ve ever played. After exploring a few very fun dead ends, I eventually got to a “real” ending. I say “real” because it transitioned into the end credits instead of offering me the chance to go back. More than that, it felt very much like a Black Mirror ending: a satisfying conclusion for the main character’s story, but pretty bleak and miserable overall. Childhood trauma had left Stefan feeling he had no control over the things that happened to him, and the entire appeal of choose-your-own-adventures to him was the idea that he could go back in time and undo his greatest regret.
So it was sufficiently foreshadowed and thematically appropriate, but ultimately kind of unsatisfying. It felt like I’d stumbled onto the ending instead of building up to it. As a result, I condescendingly assumed that Brooker went in with a high concept, quickly got overwhelmed with how branching narratives explode in complexity, and had to rush and simplify to turn it into a Black Mirror episode.
But then I read that there are multiple “real” endings to Bandersnatch, in this article on The Wrap. I had seen a couple of the other endings on that list, but they felt more like novelties than real conclusions. But one of them that I had missed specifically addressed my condescending assumption. Stefan tells his therapist that he finally realized the key was the reduce the number of choices, making the player think they had more control over the narrative than they actually did. That finally got him a perfect score from that pitch-perfectly obnoxious 1984-era TV video game reviewer. (And sent to prison, too).
That’s the ending that makes the most insightful conclusion about branching narratives, instead of all the other meta-textual stuff throughout the movie that does little more than comment on it. I can’t tell if it was intentional or ironic — or intentionally ironic, since it’s so tough to gauge intent with black comedy or satire — that it’s an ending that’s only available in a branch that I never would’ve chosen on my own.
To get to the ending that comments on how branching narratives inevitably end up railroading the player, I had to deliberately choose a branch that felt like I was being railroaded. Specifically: the story offered me the choice “KILL DAD” seemingly out of nowhere, when it wasn’t warranted. You can get to that junction point by following different branches, where it seems more justified since Stefan’s paranoia has been getting more intense. But the first time it’s offered to you, it seems like a decision that’s out of character and not in line with anything else. You’d pick it just to be provocative, to see the most dramatic thing happen, even if it doesn’t make sense.
Lack of Choice vs Power of Choice
Possibly the most brilliant decision in the whole project was to have the main characters’ mindsets reflect certain aspects of branching narratives. Essentially, they’re personifications of the limitations of this type of story.
Stefan’s all about player agency, with his growing paranoia that he has no free will, and his compulsive behaviors caused by the player’s choices.
The first part of Bandersnatch cleverly plays around with all the variations on what it means to present the audience a choice in a branching narrative. Your first options are more less cosmetic. Then you get to what feels like a significant junction point — accept the publisher’s offer or reject it. The most obvious and most common option sets up the conceit of the whole movie: Colin tells Stefan he made the wrong choice, and we see a fairly brief sequence leading to an unsatisfying conclusion. Then, Stefan declares he’s going to go back and do it over again, and the story rolls back to let you choose the other branch.
The most clever choices in Bandersnatch follow that. You’re given options that are basically just to screw around with Stefan, to exert your control over him and see how he reacts. But those choices are soon turned around on you. The story offers options that seem arbitrarily limited, as if to remind you that you’re being manipulated just as much as Stefan is. A system that only offers you two choices at each junction point is inevitably going to feel limited, but the result here felt as if it were using the limitations to make a point: you only have the illusion of control.
And my ongoing frustration with both video game developers and players is that they always describe this limitation as a problem to be solved, instead of a constraint to be exploited. Players complain that if their choices don’t result in hugely impactful changes to the narrative, then the choices were meaningless. Meanwhile, developers keep insisting that they’ll eventually overcome the limitation, whether by throwing more developers at it to create tons of content that players will never see; or by throwing more technology at it, as if machine learning-driven AIs will inevitably result in infinitely responsive simulations.
I’ve learned to never be so arrogant as to declare outright that something’s impossible, but I’m skeptical that either of these are the right approach. One of the first presentations I went to as a game developer was from a group of people pitching a narrative system that could inject new events into a story according to a pre-defined “drama curve.” Nothing came of that, as far as I ever heard, but in the 25 years since, I’ve seen the claim made over and over again. Every few years seems to bring another pack of Don Quixotes tilting at the narrative game equivalent of a perpetual motion engine.
But even if it is possible, I still can’t see the appeal. Even if you have artists meticulously crafting the parameters of a system, you’re still losing the direct feeling of connection between author and audience. The most memorable moments I’ve had in games were when I felt like I was actually having a conversation (of sorts) with the developers.
Bandersnatch has a mention of this, too, in a non-interactive moment. As Asim Chaudry’s character is trying the Bandersnatch game for the first time, he quickly encounters the main villain. He immediately chooses to worship the villain, which breaks the game. In the story, that foreshadows the central conflict: if Stefan is losing his free will, he’s in danger of doing things that violate his morality. Like, say, “KILL DAD.”
In terms of game design, though, it’s just a bad choice that shouldn’t have been in the game in the first place. If the player can make a decision that so drastically changes the intent of the story, then what’s the purpose of the story at all? Constraints in a game aren’t a cop-out; they’re what make a game a creative work instead of just a box that echoes back everything the player puts into it.
The best choice-based narratives aren’t the ones that give the freedom to do anything and see what happens. They’re the ones that make the choice moments themselves impactful, because they’re the moments when the author and audience are directly engaging with each other via the work.
One of the simplest examples: there’s an entire branch where Bandersnatch plays around with meta-text, making Stefan aware that he’s in a Netflix special from the future. When he tries to explain this to his therapist, she responds by challenging him, asking if this were entertainment, shouldn’t it be more entertaining? The player is given two choices: “YEAH” and “FUCK YEAH,” both of which result in a fight scene between Stefan, his therapist, and in some cases, his dad.
As a big fan of Garth Marenghi’s Darkplace, I loved seeing Alice Lowe get the chance to go nuts, but the best part of that choice isn’t the scene that follows the choice. It’s the choice itself. It’s the moment when the filmmakers break out of the narrative and address the audience directly.
Consequences vs Context
If Stefan represents the lack of player agency in choice-based narratives, Colin represents the lack of consequences. He says it outright: if you can go back and undo past decisions, to see all the possible outcomes, then every possible outcome is meaningless. If all possible outcomes are “real,” then none of them are.
I said earlier that the first ending I encountered — in which young Stefan goes on the train with his mother and is killed in the crash — was the “real” ending, both because it was the one that resonated most with the rest of Stefan’s story, and because Black Mirror “declared” it the real ending by having it lead into end credits. But finding out that there were other “real” endings with end credits cheapens that.
That’s one of the main problems with exposing the artificial mechanics of the story to the audience: it breaks the suspension of disbelief, so that no outcome is any more resonant than the others. What does the story “mean?” No more and no less than what you want it to mean.
The most common complaint among people playing choice-based narrative games is that they feel the choices are meaningless because they don’t have enough impact on the story. The solution for that isn’t to make more, wildly divergent outcomes, though. The solution is to make the choice feel meaningful while the player is making the choice. After all, the outcome is passive; it’s just the player listening to what the developer has to say. If you’re really trying to engage the player directly, that moment of choice is the moment where you have direct 1-on-1 engagement.
That means making the player fully aware of the complete context of the choice and exactly why it’s a crucial junction in the narrative. No choices that come out of nowhere, like offering to kill a character who’s just been little more than an irritant. And no choices that are wildly out of character, even if they would make for more “dramatic” moments, since it makes it seem like the character development that’s occurred up to that point was arbitrary. If you’ve made a well-designed choice point with sufficient context, the actual outcome is the least interesting part, because the player has already imagined the possible outcomes the moment you offered them the choice.
I’m still not clear whether Colin’s philosophy was a result of his being the 1984 equivalent of one of those “we’re all living in a simulation” douchebags we have to listen to in the 21st century, or if he was specifically aware that he was a character in a choose-your-own-adventure type story.
Regardless, it was interesting that he didn’t become a purely nihilistic influence like you might expect, but was pretty upbeat about the whole situation. There’s a moment in one of the branches in which you have the choice to make Stefan kill Colin or let him live. Whichever one you choose, Colin takes it matter-of-factly — it doesn’t ultimately matter, after all — but if you let him go, he says that he’s glad, because he’s been enjoying this version of his life. And why wouldn’t he? He’s not just a “rockstar” game developer, but he’s got a wife and child. He seems to be eager to put down roots in this version of reality, instead of just giving in completely to paranoia and pointlessness.
Which makes me think that even presented as part of a series that tends to present the most cynical, darkest takes on the effects our technology has on us, Bandersnatch was never really intended to be a solemn examination of determinism. Even with all of its murder, mental collapse, and existential crises. I get a greater feeling of nostalgia for an era of computer games when it felt like there was so much potential, and people really did believe they were creating virtual worlds.
Everyone Will Misremember That
The further away I get from Telltale Games, the less impressed I am with the style of game that they became famous for, post-The Walking Dead season one. Frankly, I think the success of those games — and to be clear, I think the ones that were successful deserved that success — is barely due to the choice-based mechanics at all.
Instead, I think it’s the result of teams of absurdly talented working their asses off to bring the language of cinema to accessible video games. “Accessible” meaning they don’t make huge demands of your time, and they don’t require the dexterity or frustration of cinematic action games like the Uncharted series. It reminds me of the appeal of graphic adventure games, back when you had to put up with some confounding puzzles if you wanted something that had actual story.
There are some really strong choice-based moments in the ones I’ve played, in which they’ve done the work to put you into an untenable situation, and you’re imagining the weight of all the possible outcomes while the timer is running out. But they’re outnumbered by arbitrary choices between “A” and “B” that exist only to move the story along. Not necessarily for a lack of imagination, but just because it’s extremely difficult to construct a finite narrative around several meaningful choices.
In retrospect, their gimmicks should’ve been a clue: if a choice were really meaningful, you wouldn’t need the game to tell you explicitly that you just did something important. You’d have already felt it.
I think the more interesting work in the mechanics of interactive fiction has been in terms of systems. Not necessarily by responsive AIs2Although I’d be happy to be proven wrong, but by leaning into the artificiality of a choice-based narrative, and the need for constraints on the player, by making those constraints visible.
80 Days is a pretty good example: the system is visible to the player, so you generally know what the cost and possible outcomes of each choice are in advance. That game does seem to be more about exploration and discovery than strictly “narrative,” though, and I have yet to encounter a choice in that game that felt impactful instead of just interesting.
Possibly the most ironic thing about Bandersnatch is that it’s the kind of thing you can only do once. It has to be relentlessly self-referential and meta-textual, since the mechanism that drives it is comically simple and artificial. I can imagine its being used for a The Play That Goes Wrong-style comedy, but any attempt at non-self-referential drama would inevitably seem absurd.