Story (Robert McKee) and the Expectation Gap in Interactive Story

Story is one of a handful of screenwriting books that turn up constantly in the bibliography of game writing books. McKee himself gives courses — I’ve never been, but I hear they’re very good shows, whether or not they’re good advice. It’s advised at least as often as Save the Cat, and possibly more so.

It is also, I think, more applicable to non-cinema writing than Save the Cat: McKee is interested in structure, and he has a lot of formulaic rules to suggest, but he cares about content as well. At one point he has a speech about the need for emotional truth, and how this can only come from within the author.

There are aspects of the book that aren’t entirely to my taste. In support of his points, McKee often spends quite a while reprinting classic screenplays — he’s particularly enamored of Chinatown — with his own commentary interspersed. I did not generally find his comments to be that much more instructive than the original dialogue, undisturbed. And even when he’s not giving verbatim chunks of screenplay, he spends an awful lot of time summarizing the events of various movies you’ve probably seen. He’s also a bit grandiose with his rhetoric about the great imaginative work of writing.

Then, too, quite a lot of his advice belongs to the “add an appropriate amount of salt” school of recipe writing — warning that too little or too much of something will be bad, but offering no heuristics.

All the same, there is a lot of basic vocabulary about how plots are assembled and how scenes are designed, which this book introduces as well as or better than many another. Personally, I’d be inclined to go for e.g. Wonderbook instead, if you want an introduction to basic structure vocabulary, and you’re not specifically writing screenplays. For most purposes, Wonderbook is more varied and goes deeper than Story.

There are, however, a couple of points — the ideas of expectation gap and of internal subconscious conflict — where I think it’s interesting how those concepts carry over to interactive work.

One of McKee’s core concepts is “the gap”: the distance between the protagonist’s anticipation and reality, between expectation and result. Any scene where the protagonist gets exactly what they expect is a scene not worth shooting, he argues (and I think he largely has a point here):

But this is an observation that is particularly tricky to carry over to choice-based fiction. In order to play, the player generally does need a goal for the scene. Meanwhile, the protagonist not getting what they want often reads as a reason to replay the scene, not a sign of dramatic progress.

To break that down a little further: I often think in terms of advancement, exploration, and idling moves for the player. Advancement moves are moves that push the story forward or commit to important decisions. Exploration moves are those that open new information for the player. And idling moves are those that consist of things like looking and taking inventory in a conventional parser game, or waiting in a real-time narrative, or clicking around UI indicators in a different sort of choice-based game. They often mean that the player is confused, or that they’re at a lull in the pacing.

A McKee-style expectation gap is easy to offer in response to exploration moves because the player is seeking more information, doesn’t know precisely what to expect, and is ready to be both surprised and gratified by finding something out.

And arguably a gap is unnecessary in idling moves because the player is not expecting to change anything or get anywhere. Idling moves are often extradiegetic anyway — not meant to be part of the main storyline. A lack of dramatic tension is okay if you’re not currently in the drama.

The gap gets trickier with advancement moves, where the player is trying to move the plot forward in some decided fashion. If you give the player an option and the result is not at all what the player expected, that feels like a lack of agency. If you give an option and the result is predictable, you’ve lost McKee’s gap.

Often, what you need is a “yes, but” reaction. The player’s choice succeeds, but there is a side effect or a cost that propels the story forward. The player’s immediate intention works, but it makes their ultimate goal even harder to achieve.

“No, but…” also works sometimes — your attempt failed but it’s changed the world in some other way, so it doesn’t feel like the author was simply ignoring the player’s choice. A game full primarily of “no, but…” consequences is going to feel like the protagonist is just bumbling through the plot, which may or may not be the desired effect.

When the “yes, but…” plays out in a painful way, you can get quite a strong audience reaction to it. In my own BEE, one of the sequences that hit players the hardest involved a trip to the hair salon.

(Spoilers follow, but it’s impossible to play a working copy of BEE right now. Alternatively, you can skip to the next section.)

The protagonist of BEE is a girl of about 11, being raised by broke and also rather conservative parents, and at one point another adult, as a treat, takes her to a nice salon to have her hair done professionally for the first time. Later, the player has the chance to ask for a return trip to the salon, and the mother is grudgingly persuaded, as a special treat. But the haircut costs a lot more than the girl realized on the first trip, and her mother winds up spending their hard-saved grocery money to cover it. Mom is humiliated in the moment and stuck in a difficult financial position afterward.

Throughout the story, the protagonist’s desire for independence and a grown-up self-image is in conflict with her desire to follow her parents’ rules and regulations, and the player runs into that dynamic in a lot of the choices. In the hair salon choice, it’s clear that she’s pushing for the independence side at the expense of the parental preference side… but within pretty well-understood and seemingly harmless parameters. The outcome is that the protagonist gets what she wants, but at a price she wouldn’t have been willing to pay if she’d known it in advance.

“Character is unexpectedly charged a hundred bucks at a fancy salon” is pretty low-stakes compared with a lot of situations in games, including many others I’ve written myself. But I got a disproportionate amount of player feedback about how much they’d been affected by that particular moment. I think a few things helped make this work:

• The unexpected consequence didn’t rely on coincidences or accidents
• The fact that it was unexpected illustrated something about the character and her problem: she’s starting to want to explore how adult women present themselves, but her mother is ill-equipped to guide her into that world. A richer mom could have afforded the salon, and a more worldly one would have at least known it was out of their price range
• The player opted into the action, and opted in knowing that they were prioritizing the girl’s independence over family conformity. They just didn’t realize what they were risking when they did. It looked like the risk was damage to the relationship between the girl and her family, not between the family and the world

The first two points would have worked the same way in a non-interactive version of the story, but the third seems to have given it an extra kick for some players.

*

It might seem like introducing a chance of failure is another way to resolve the advancement move problem. If I attack a dragon expecting to kill it, and I miss my roll and die instead, does that count as expectation gap?

I’d argue that from a narrative perspective, the possibility of failure and the existence of those stakes is an important part of the choice framing to start with. But neither the “you die” nor the “you win” consequence fulfills the requirements of McKee’s gap. A savvy player expects that both victory over the dragon and death are on the table. Even choice/consequence pairs with an element of randomness or challenge still need the the plot to build on the win and loss states usefully.

*

Back to Story. McKee also advocates for characters who have a subconscious desire running counter to their conscious one. That’s easy enough — or, at least, within the parameters of ordinary writing craft — to do with non-player characters.

It can be trickier to imply or evoke that kind of internal conflict in a player character, especially if you’re trying to avoid telling the player how to feel. There are certainly ways to set up protagonist interiority: narration that steers us towards understanding the world in a particular way; characters who obviously sabotage themselves in their own activities; protagonists who don’t understand themselves as well as the player character does.

But getting the player to enact that conflict, to choose in a conflicted way, is more challenging.

A few games handle this by splitting the protagonist into effectively two or more characters, the conscious and subconscious self (or perhaps multiple drives), and then handing control of only one of those to the player. Perhaps the player drives the conscious goals while game mechanics of some kind enforce the unconscious ones: Depression Quest does this, ruling out choices that are impossible for reasons of mental health, no matter what the player might prefer to do.

And Shrapnel is an old and now not-much-played game by Adam Cadre in which the player can accept or veto the protagonist’s first dialogue impulse, but if you veto, you have no control over what the protagonist says instead. The protagonist’s subconscious level has intervened.

In games with heavier mechanics, it’s possible to use stats and resources to force the player to acknowledge the protagonist’s needs. In Sunless Sea and Sunless Skies, the player has a quantity of Terror that gradually rises, triggering nightmares and additional problems until they find a way to bring that terror down again. This kind of presentation treats the emotional self as a somewhat unruly aspect of being, one that clutters things up and gets in the way.

For the most part, though, those systems work best with inner needs that are pretty low on the Maslow hierarchy: it’s common to have systems that force the player to eat, sleep, and find shelter regardless of their other goals; it’s less common but possible to make the player manage the need for psychological safety or affection. It’s harder still to imagine how you’d have a game where the player’s need for artistic self-actualization is slowly ticking away in the background, but some of the Sims games (for instance) do approach some of this territory. (They just then aren’t all that narratively driven.)

It’s harder for me to think of games where the player is directing the subconscious urge of a protagonist while the game is directing the conscious or willed action — though Coloratura, with its alien that can influence the emotions of the human players, is arguably adjacent.

Other resources:

• Bluebeard’s Bride, a tabletop RPG in which the players represent portions of the protagonist’s psyche.