Esther and Stanley and Fate

Often we talk about narrative logic when we talk about stories. Does a thing hang together, make sense? Does it have plot holes? How shall we plug them? How can we make it all correct? The metaphor suggests an orderly process, like solving an engineering problem. It implies that there’s some objective physics of how storytelling works, and stories fail when they violate those physics.

It’s a limited metaphor. You can talk about stories in terms of logical chains of cause and effect, but to think of this as an objective measurement is misleading. We complain about minor engineering decisions made in a weapon the size of a moon, in another galaxy filled inexplicably with humans. Logic isn’t what we’re really talking about here. Narrative logic just means whether a thing satisfies you or not. It’s about comfort. And we tend to find effects clearly linked to causes comforting. We want to live in a deterministic universe.

Dan Pinchbeck’s Dear Esther and Davey Wreden’s The Stanley Parable have a lot in common. They both use the Source engine and share a mechanic of walking around a space and triggering voiceover snippets. They also share a concern with stories and how they satisfy, or fail to.

While both games are about storytelling, they approach the theme from opposite directions. A story, traditionally, is a sequence of events that follows a chain of cause and effect. The Stanley Parable is about how story structures mock the idea of free will. Dear Esther is about how people force incomplete and untrustworthy information into story structures. One features a protagonist trapped in a deterministic world, and the other a protagonist trapped in a non-deterministic one. One of these turns out to unsettle players much more than the other.

In Esther, the player explores an island, listening to a voice, and tries to piece together what happened there. Much of the narration concerns a similar trip across the island. The narration itself concerns someone trying to make sense of a tragic event. The narrator wants to understand why Esther died. He visits someone named Paul, supposedly the drunk driver of the car that killed her. Paul caused this event, so the narrator tries to resolve matters with him. This is narrative logic, looking for a comforting collapse of the causal chain. The resolution is not reached, or at least, it wasn’t any of the times I played the game. Later the narration questions whether Paul was even drunk. Maybe the whole event happened for no reason at all.

The game itself shares this lack of predictable logic. At each triggering point, the game randomly selects snippets of voiceover from a few possibilities. The variations are often subtle, but they can introduce different or conflicting pieces of information. Other objects in the world change between playthroughs. Ghosts appear or don’t; streams run in opposing directions; revelations change. The game doesn’t signal these changes, so they remain invisible on a first playthrough. After playing it twice, they are mostly subtle enough to be put down to tricks of the memory. You try to reconstruct a story, but the material you have to work with is inconsistent and ambiguous.

Stanley isn’t interested in how people make stories out of chance events. Rather, it’s concerned with how events define types of stories. Stanley makes use of branching paths, a much more familiar structure in videogame narrative. You walk through one door or another, go up or down in an elevator, and so on, and are rewarded with different endings. Like some other games that deal with player choice, it also takes the time to mock the illusion of free will this creates. If you follow the narrator’s instructions, you end up with a meaningless happy ending with a sarcastic “My Way” playing over the credits. If you disobey him at every turn, he dumps you in a broken Half-Life 2 level and yells at you until you quit.

Choices in between these two extremes produce different outcomes, all of which are unsatisfying in different ways. There’s an “it was all a dream” ending, an ending where nothing happens, an ending where a bomb ticks down and you – a game hero! – are helpless to stop it, one where Stanley is killed by some random death machine, and one where you quit the game before dying. It’s a pretty good taxonomy of terrible endings. Unlike most games with branching narratives, Stanley isn’t trying to give you the ending you want or punish you for being bad. Your choices don’t define your character, they define the kind of story being told.
In many ways, Stanley and Esther are about different things: sorrow versus frustration, people versus abstract structures. But in the end they both deal with how and why people interpret a chain of events as something meaningful. The difference is that Esther is about chance and Stanley is about fate.

Stanley represents the old philosophical debate of whether free will is possible in a deterministic universe. If an omniscient viewer can in theory predict all outcomes based on the trajectories of particles, is free will just an illusion produced by the predictable particles that make up brains? Similarly, if you’re in a game system with a predetermined set of paths to finish, what does choice really mean? Stanley’s world makes a perfect graph: see, you can map it succinctly in Twine. (Playable demake here.) From any given game state, you can know what other states you can reach and how. Not all games make this possible.

Esther, for example, can’t be mapped as a deterministic graph like that^*. Because of the randomly selected elements, you can’t predict future game states based on the current state. Your actions don’t seem to affect outcomes at all, so even having a (necessarily gigantic) graph of the possible random outcomes wouldn’t help you get where you want. If Stanley shows how determinism makes free will meaningless, Esther shows how nondeterminism makes free will impossible.

Both games reflect their mechanical approach to determinism in their fiction. Stanley’s narration explicitly calls out the illusion of free will. Each path you head down has something to say about how your deterministic game world undermines your supposed choices. It’s most direct in the fully disobedient ending, where the narrator angrily points out that complete freedom means ending up in a part of the game with no content. You follow the linear paths because that’s where the stuff is. Players want stuff, yes?

Esther’s story likewise keeps coming back to random chance. Everything started with a car accident. The game repeats a motif of falling (a fall breaks each chapter, and relates to Paul’s ride to Damascus, referenced throughout). Illness afflicts the characters. As in the game itself, random happenings fill the story. Like the player, the narrator tries to string these chance events together into a coherent sequence, pulling themes from biblical sources, coincidence, and the island’s geography. Esther doesn’t address free will at all: in a truly nondeterministic universe, will isn’t useful, because people can’t affect things predictably.

This gets reflected in how interactivity is perceived in the two games. Stanley and Esther use pretty much the same mechanic: you walk around and trigger events. But several reviews of Esther describe it as non-interactive (for example, Tommy Rousse’s “On Ruining Dear Esther”), while Stanley mostly avoids that mess. “Interactivity” is an overloaded term in games writing, but here its use seems to reflect a difference not in what actions you can take but in the feedback you get from those actions. In Stanley, the narrator clearly marks and explains the results of each significant movement. Esther never tells you if any of your actions have an effect on the game system. Indeed, they probably don’t.

One of the ways people use the term “interactive” as in “more or less interactive” is to say: do my actions feel meaningful? Do they seem to determine outcomes in a way I can predict in advance and/or explain in retrospect? Esther doesn’t feel interactive to some players because actions don’t seem to do anything. A branching narrative like Stanley’s automatically feels more like “I’m doing something” just because the path can be traced backwards: I did this, then this happened, then I did this… and so on. Stanley’s use of clear thresholds, like doorways, stairs, and elevators, further marks its pieces of narrative as branching points. You cross a threshold, the rules change: this is a familiar metaphor. A game where those chains aren’t so easy to trace feels more like it’s pulling you along, which is part of the emotional arc of Esther. Out in the open space of the island, thresholds are harder to see and expect. Esther playthroughs technically exist in a bigger possibility space than Stanley’s, but it doesn’t feel that way for many players.

Storytelling always gets wound up in ideas about determinism. Believing in the logic of stories is a lot like believing in fate. There’s an explanation for everything. God has a plan for us all. This plot hole is unacceptable. People like to see the logic chains of messy experience collapse into a neat, tidy graph. We are animals optimized for making plans, which means we must be able to predict the effects of actions we perform. Looking back at things that happened and analyzing cause and effect can exercise this cognitive skill. Stories comfort because they display causes and effects in a sequence that can be analyzed. A prettily designed display of a deterministic universe reassures the audience that they can predict the outcome of their own actions.

I think this helps explain the differing reactions to Esther and Stanley. People found Esther unsettling; much more was written about it, and more involved hand-wringing about “is it a game?” This question is ridiculous, but it usually stands in for “how should I react to this?” Esther’s random mechanics and nondeterministic view of storytelling can disrupt normal story enjoyment. By contrast, critics received Stanley more or less as a joke. Its point seems more obvious. Stanley wrestles with the free will vs. determinism debate, but that’s old news.

How the two games approach their narrators reflects their differences in philosophy. In Esther, the narrator and the protagonist are sometimes hard to tell apart. Players often assume the narration is their avatar’s internal monologue, at least at first. This becomes less clear as you go along, but by the very end the game seems to bind you together again. Stanley explicitly presents the narrator as an antagonist on many paths, and implicitly does so on all of them. In one path, another narrator appears to question the first’s omniscient status.

This difference compliments the cosmology of the two games. In a deterministic universe, individuals might not know where they’re headed, but an omniscient being would. For that reason, determinism evokes heavy-handed god figures, fate, premonitions, and all that. While Stanley lightly references such theological debates, Esther has all the explicit religious imagery. But its driving parable, Paul on the road to Damascus, is appropriately ambiguous. A man is stricken down by something, sees a vision, and everything changes. It’s the kind of story into which nonreligious people can read the desire to attribute miraculous meaning to random events. Maybe Paul had a migraine. It depends on how you interpret it. In a deterministic universe, fate is the very existence of knowable paths. In a nondeterministic universe, fate is a choice people make, among others, to explain events. I think you could sub “god” in for “fate” in those sentences and say pretty much the same thing.

Related to this, both games have faced the charge that they lack coherent universes. Stanley’s endings do not all exist within the same logical story space, as Davey Wreden admits in an interview by Kirk Hamilton. Stanley can’t both be a feverish dream of a man dying in the street and the metafictional hero wandering a Half-Life level. The choices you make within the game don’t just decide what happens to your character, but what kind of universe he lives in. In a way, this takes a more honest view of interactive storytelling than you see in more traditional branchy games like Mass Effect. In other kinds of fiction, the choices a protagonist makes are determined by the kind of story he’s in. If a Shakespearean character chooses whether to humble themselves or persist on a prideful path, they decide whether their story is a comedy or a tragedy. That in turn determines more about the story than its ending. Choices in a story don’t naturally exist in a vacuum. What a character does is linked to what their world is.

In the case of Esther, incoherence comes from random dice rolls. When a story can play out in arbitrary ways, how can it represent the same world every time? The Esther I see differs from the Esther you see, and discussing the game can be frustrating when it feels like you experienced different objects (as Katie Williams describes eloquently). Yet because of the ambiguity in the telling, it’s easier to ignore or resolve this incoherence than in Stanley’s case.

Here’s an example: the nature of Esther and the narrator’s relationship. I’ve read about twenty articles and reviews on Dear Esther. Not all of them name the relationship between the narrator and Esther. Of those that did (for example, Cameron Kunzelman’s close reading), every one labels her the narrator’s wife. I was surprised the first time I read this and fascinated by the fifth time. It stood out to me because I had read Esther as the narrator’s daughter. I think this is because one of the artifacts I found was an ultrasound next to a bird’s nest filled with eggs.

The nest and the ultrasound don’t mean anything, of course. Maybe Esther was pregnant. Maybe those things never appear in your playthrough at all, or you don’t notice them, or you see them as a red herring or a rebirth motif or a reference to a biblical passage I know nothing about. To my knowledge, the text of the game does nothing to clearly mark the relationship either way. As with those who read Esther as a wife, I made the best of deliberately ambiguous information, coupled with whatever was going on in my head.

I think the stability of the wife reading points to something important about how people approach ambiguous information in stories. When not told otherwise, we have a tendency to revert to cliché. Men fall apart when their wives die, and must be redeemed by accepting the loss. That’s a story we tell about men. My version is nearly as stereotypical. Stanley explores how controlling outcomes through design leaves players unsatisfied. But Esther’s ambiguous approach doesn’t solve the basic problem of satisfaction. In the absence of information, we default to the stories we’ve heard before.

Determinism comforts, and nondeterminism unsettles. But there’s something frustrating and unsatisfying about living in a universe that’s entirely mapped. When a universe can’t be mapped at all, it frustrates in a different way. This tension lies at the center of much videogame story design. Is there a sweet spot somewhere in the middle of the deterministic and nondeterministic worlds? Some fans of emergent narrative argue as much, but there’s still a lot of unexplored territory there.

Esther and Stanley make a useful pair because they focus intently on two opposing poles in this design space. They also point to strengths and weaknesses of their approaches. Esther’s nondeterminism creates more possibilities, without needing a designer to define them in detail. At the same time, the information gaps it leaves are, on average, going to be filled with things of average quality. Stanley’s deterministic branching lets the player feel in control, but also makes that control feel pointless. Taken together, these sum up a lot of the current tradeoffs in interactive storytelling.

* Okay, it can, but not in a way that matters.