The Well-Made (Video Game) Plot, Part 1

Whodunnits, dunnwhats, and well-made plots.

A “well-made plot,” in literary circles, is not simply a good plot, but rather a plot that runs on cause and effect, where the world, the characters, and the narrative arc all behave according to believable causality.  In fact, mere causality is not enough.  You could write a novel where event A leads to event B, which leads to event C and so on — all perfectly rational on the small scale — and end up with an event Z which is unrelated in all but the most mechanical of ways to any previous event other than Y.  This would not be a well-made plot.  Well-made plots must be teleological.  They require causality to function on the large scale, where it sometimes goes by the name of fate.  Finally, on a more pragmatic level, the well-made plot also assumes a certain minimum level of incident.  A novel that describes the experience of drinking a single cup of gas station coffee in bone-pulverizing detail could not be said to have a well-made plot, no matter how rationally it is worked out.  Well made plots need to have weighty narrative goals that are teased, built up to, and achieved. (Note here that the narrative goal is not necessarily the same thing as the protagonist’s goal:  heroes of well-made plots can fail, but as long as their failure is a decisive, causally inevitable moment, it can still be a narrative goal.)

Probably the genre most dependent on the well-made plot is is detective novel, or whodunnit. A detective novel that lacks large-scale causality cannot properly be called a detective novel at all, no matter how many people you have running around smoking cigarettes in fedoras.  There used to even be informal rules about this sort of thing, the idea being that a well-made detective story was one where an observant and intelligent reader could potentially solve the mystery as soon as the detective did.  Which means that you can’t have the detective notice something crucial and keep it to him/herself:  the reader needs to be informed!  All of the Encyclopedia Brown stories are well-made in this sense, even if the solutions are sometimes moronic.  (The Sherlock Holmes stories, incidentally, are not well-made.  Sherlock is forever noticing details which Watson, the narrator, is too dim to pick up on, with the result that neither Watson nor the reader gets the crucial clues until Sherlock has already given the game away.  And for a more extreme example, see Harry Stephen Keeler’s The Riddle of the Travelling Skull, for instance.)

Understanding how well-made plots work in detective stories can help us understand how they work in other kinds of stories, too.  In the mystery novel, the well-made plot is generally understood to be “about” answering the question of who committed the crime:  thus, the whodunnit.  But whodunnits are for the most part also howdunnits and whydunnits.  If we don’t learn the killer’s motive and modus operandi, we’re going to find the story unsatisfying.  And ideally, both of these are also going to be subject to the same kind of large-scale causal logic, and to the same “smart readers can work it out on their own” standard of narrative fair play. Those are the questions that occupy the reader of a detective novel:  who, how, why.  Other questions matter less.  What has been done is necessarily known — a murder — and where and when usually are, although all three of these can be destabilized over the course of the work so that we learn M. Body was actually killed in the hall not the conservatory, or that he was never killed at all.  Mysteries focusing on forensic science do tend to focus on when and where the victim was killed, but only in that it might lead us to the killer.  The information is of no use in itself.  A real whendunnit would have to be written from the point of view of, say, an insurance investigator, attempting to prove that the victim died hours after his policy had lapsed, right?

But actually, that’s not true at all!  Whendunnits do exist, they just aren’t detective stories — or aren’t necessarily detective stories, at least.  Rather, they are stories structured as flashbacks:  stories that begin by showing us that a certain important event has taken place, and then walking us through the circumstances leading up to it.  (Whendunnits are always also howdunnits.)  Nimród Antal’s thriller Vacancy was originally structured like this, before the studio cut it into a more linear and less interesting form.  It was supposed to open with a shot of a wrecked car, pulling slowly out to reveal that it had been driven into a hotel lobby.  Pulling out further, we see police milling about, and a number of stretchers with bodybags.  Then we see Kate Beckinsale giving her statement to the investigator — and then we flash back, and she’s riding in the unwrecked car with Luke Wilson.  That right there is a whendunnit.  We know that sometime before this night is over, that car is going into the hotel lobby.  We also know who is involved, at least in a nebulous way.  But we don’t know when or how.  When and how are the most important question with this kind of plot, just like who and how are the most important questions in the classic detective novel.  And both species of plot can be well made. Both have well-defined causal arcs that start out partially obscured and are completely revealed by the end of the story.  The viewer/reader should be able to work out the solution to both ahead of time.  Not always in a very articulate way, to be sure.  There’s no tell-tale clue halfway through Vacancy that points to the car flying through the window precisely twenty minutes before the end of the film.  But when those dominoes start falling into place, the viewer needs to be thinking “Oh okay, I get it, here it comes, here it comes!” or else the plot has utterly failed to function.

Normal plots – ones that aren’t flashback-structured, like the whendunnit, or post-facto, like the whodunnit, tend to be dunnwhats.  We know who will be doing things, because the main characters are established early on — usually even before we start reading.  Pick up a serious respectable novel off the shelves at your local bookstore, and I will wager you dollars to donuts that the synopsis on the jacket has a lengthy description of the main characters.  But we do not know what they will do. (An important variant, of course, is the exposition-of-character novel, in which what we learn is not so much what the character will do but who they really are on a deep psychological level.  Still, this is usually revealed through a series of critically revealing actions, so the label “dunnwhat” can still apply.)  In the well-made version of this, it’s possible to figure out what the characters will do before they actually do it.

In so far as they aspire to a well-made plot, then, all novels are mystery novels.  There should be information scattered in the early part of the narrative that in some way explains the crucial mystery that gets solved at the end, whether that mystery is who’s been murdering guests at the chateau, or whether the main character will find happiness despite stifling social pressures.  This doesn’t mean you can’t have plot twists!  A little uncertainty on the small scale is necessary to keep things interesting.  But if even retrospectively there’s no way for the reader to figure out what has been dunn, well… that’s a problem.  Not necessarily a crippling one, but a problem.  Take the Harry Potter books:  you keep thinking that Dumbledore is playing a deep game, and that when you’re finally privy to his plan — that is, when Dumbledore finally does the important “what” that his character arc has been building towards — everything will make sense.  But this never happens, which retroactively makes both the character and the whole series less satisfying.

Tension

Despite these similarities, there are two big important ways that these kinds of stories differ from each other. One has to do with the role of the protagonist.  In whodunnits, the reader has to match wits with a detective character who is also trying to solve the mystery.  The detective serves as both a rival and a point of identification for the reader.  We’d like to beat them to the punch, but we understand that they’re on our side.  They feel about their world the way we feel about their world, the same desire drives them as drives us.  In whendunnits, contrariwise, the reader is and must be on their own.  We get to try to solve the whendunnit because we stand outside of the work’s timeline.  Kate Beckinsale doesn’t get to know ahead of time that the car is going through the lobby.  Uma Thurman may hope to Kill Bill, she may plan to Kill Bill, she may even be convinced that she will succeed in Killing Bill, but she doesn’t get to know that “Uma Thurman Will Kill Bill” was the tagline of the movie. The question for us is when/how, the question for her still has to be what/whether.  Whodunnits and whendunnits are on the opposite ends of a spectrum here.  “Normal” plots fall somewhere in between.  In most cases, the main characters don’t themselves know what they are going to do (or, for the psychological variant, who they really are), which puts them in the same shoes as the reader.  But they may or may not care about it.  There’s a very narrow subset of the novel of psychological revelation in which the character is actually trying to figure out who they really are, but in a great many cases they are simply going about their day to day business as a funeral home director, stick-up artist, or heir-apparent to a motorcycle gang.  The actions which are psychologically revealing to the viewers are to the characters completely unrevealing.

The non-psychological dunnwhat, which is more about physical action, tends to work more like a mystery novel.  Arya Stark wants to get home to Winterfell, but she doesn’t know what she has to do to make that happen.  The readers want her to get home to Winterfell, and we don’t know what she has to do to make that happen.  This kind of identification often slips back and forth a little bit — either we have information the character doesn’t, leading to dramatic irony, or the character makes a preparation that the author doesn’t share with us, leading to antidramatic irony — but for the most part it holds fast.  You could notionally imagine a text where a the reader wants to know what the characters will do, but the characters themselves are ambivalent… The Manchurian Candidate, maybe, sort of?  The Neil Gaiman Miracleman?  Weekend at Bernie’s? But as you can see from the examples, this requires some rather special effort on the author’s part.

The second big difference between these narrative forms has to do with the degree of narrative constraint.  Reverend Knox’s very first rule for a well-plotted mystery demands that the killer turn out to be a character introduced within the first few chapters of the story.  Unless these take the form of a phone book, then, a well plotted whodunnit will present you with a sizable but strictly limited number of solutions right off the bat.  Every character who is introduced could be the killer, but only these characters are allowed, which means that — like in a game of Clue — you can narrow down the possibilities over time and get invested in one possibility over another (a bad idea, in Clue, but possible).  For whatever reason, the fewer possibilities there are, the more intense our desire to know becomes.  It’s as if there’s a certain amount of water, flowing through a network of pipes, and as you close the valves on more and more pipes the water pressure in the rest of the system mounts and mounts and mounts.  The climax comes when there’s only one viable choice remaining… but note that this is not sustainable; it’s a moment, not a condition; if you wanted to extend the plumbing imagery, this would be the point where the pipe bursts and starts spraying raw narrative tension all over the shag carpet in your allegorical rumpus room.  So ideally you wait right to the end before letting this happen.  Also, you generally want the rate at which potential murderers are eliminated — at which valves are shut off, like — to increase rather than decrease over the course of the novel.  If there are twenty suspects, and seventeen of them die in the first ten pages of a two-hundred page book, that’s a problem.  It’s also a problem if they’re all viable suspects up until ten pages from the end, at which point, again, seventeen of them die.  You want to eliminate suspects at a steady rate, or on a gently accelerating curve.  And writers really get in trouble when they think they have two viable choices, but one of them is quite clearly not the guy — and then this gets extended for hundreds of hundreds of pages, resulting in, burst pipes, flooded basement, water damage, nasty mildew infestation all up in your crawl space.  (Take the whole Team Edward/Team Jacob business, just for instance.)

Whendunnits like whodunnits are constrained, but in a rather different way.  As soon as the meter drops on the plot proper (after the stakes have been established in the flashback, that is), the crucial event could happen at any time:  if the plot flashes back exactly one hour, then you’ve got exactly sixty minutes in which the events could occur.   But if they happen right away, the tension is basically zero, so nobody’s going to write it like that.  If you wait thirty minutes, that tension will build:  now you have only thirty minutes for the event to happen in!  And if you wait forty minutes?  If you wait fifty?  Any storyteller worth their salt, of course, will save the events for the fifty-eighth or fifty-ninth minute.  Just like whodunnits reach their peak intensity when you have the smallest number of potential criminals, whendunnits reach their peak intensity when the window that the crucial actions can fall into is as small as it possibly can be.  There’s something like an unexpected hanging paradox going on here, of course, but in practical terms this matters little.  We can safely assume that the important events in a whendunnit are always going to happen within the last twenty pages of the book or twenty minutes of the film, but knowing this does not protect us from experiencing the increased narrative drive:  “Oh okay, I get it, here it comes, here it comes!”  Once you’ve figured out the underlying narrative logic, it becomes clear that whendunnits don’t technically need to have flashbacks.  All you need is for the readers to have some knowledge of future events that the main characters don’t have, and for there to be a limited time frame in which these events can occur.   We know that the Power Rangers will destroy the rubber-suited monster of the week, we even know how — that one big diagonal slicing move with the power sword.  All we need to know is whether it’s time for it to happen yet in this episode, i.e. when it will happen.  And one of the interesting corollaries of this is that all whodunnits are also whendunnits.  The detective will figure out who the killer is at some point, because it’s generically required.  As you get closer and closer to the end of the book, the time window (expressed here as a matter of page count) during which this can happen gets smaller and smaller — and as a result, the intensity of the experience gets stronger and stronger.   Very often, these two kinds of narrative drive appear in a single work, so that the shrinking subject pool and the shrinking window of opportunity are neatly superimposed.  It’s not just in detective stories either — one particularly clear example of this is the flashback-structured romantic comedy Definitely Maybe, where we know that Ryan Reynolds has had a child with one of his old ex-girlfriends, but not which one.

Compared to these humming narrative engines, the relatively open-ended structure of the serious novel, in which a vividly realized character merely does… something… falls a little flat, which is part of why these books are typically not page turners.  Adventure stories, dunnwhats that are page turners, get around it by providing constraints on their narratives in a different way.  Rather than setting up a situation where the character has no predefined teleological goals, they set up a very clear goal which may or may not be attainable.  They are not dunnwhats but dunnwhethers.  Frodo will toss the ring of power in the volcano… or not.  Lizzie will marry Darcy… or not.  Tony and Maria will find a place for themselves, somewhere a place for themselves… or not, although arguably in this case we already know that they’re doomed, and it’s just a matter of learning when and how they succumb to that doom.  Narratives of this kind present us, like the whodunnit, with a multiple choice question, and therefore are typically considered more frivolous because real life mostly slaps us with short essays.   Many, many romantic novels, serious or trashy, have their heroines choosing between two or three guys — but in real life, it’s never that neat.  Rather, we face the implied choice between all guys, everywhere, and of course there’s a very real chance that even that queasy pseudo-choice will be made meaningless by factors beyond our control, such as whether the guy we choose is even interested, or whether he’ll get hit by a bus tomorrow.  Some life decisions really are unambiguous choices with meaningful results.  Plead guilty, or innocent?  Accept the marriage proposal, or reject it?  But these moments — which we might describe as narratively exciting life experiences — are pretty few and far between.

So to summarize, before we move on to some actual video games:

•  Well-made plots require a sense of long-term causality, or fate.  Typically they involve a who, a what, a when, a where, and a why, but some of these elements are unknown at the start of the story.  A smart reader should at least technically be able to work out the missing pieces ahead of time.

•  In non-interactive fiction, at least, we can distinguish between plots that primarily that are primarily about finding out who did some known thing, plots that are about finding out when some known future event will happen, and plots that are about finding out what some known person will do.  (There are probably other categories — figure them out for yourselves.  Could you imagine a pure howdunnit? )

•  In some kinds of stories, the main characters care about the same questions that the readers care about.  In others, they almost never do.  This has a pretty big effect on our experience of the work.

•  For all of these categories of plot, constraints on narrative tend to make the narrative more exciting.  Which is not to say better, necessarily…

•  There’s often both a plot and a meta-plot at work, so that on one level, we’re wondering who the killer is, and on another, we’re wondering when the detective will figure out who the killer is.  Ideally both levels of this should be well-made:  if we feel like the detective identified the killer only because the episode was about to end, that’s a non-trivial weakness.

Badly-Made Plots:  Videogames and the Picaresque

Video game plots are almost never well-made in this sense.  On the one hand, you have plotless or near-plotless games, which don’t contain enough incident to qualify.  Tetris is about as causal as it gets, and sure, you can typically work out the ending in advance if you’re paying attention.  But a well-made plot it’s not.  On the other hand, you have plot-heavy games like RPGs.  Here, you run into the “A then B then … Z” problem, where the early stages of the plot aren’t connected to the later stages in any non-mechanical way.  Take as an example, the submersible sand castle sequence from Final Fantasy III née VI).  The main characters show up at a town called Figaro, pursued by the forces of the Evil Empire™.  The bad guys demand that the locals hand the heroes over, and the locals respond by activating the town’s special burrow mode and sinking into the sands.  It’s a pretty neat sequence, and — from the point of view of the gameplay experience — an important one.  It’s not just a random digression, as are the random fight scenes that make up %80 of the actual gameplay.  It’s a major cut scene, an important event in the ongoing plot.  And in the long term, it has no effect on anything whatsoever.  Or rather:  it only matters because it’s a pretext for Edgar to join your party.  And why does that matter…?  This is actually a rather profound question.  In grand narrative terms, Edgar doesn’t matter.  He helps the main character escape in this early sequence, he introduces you to another of the playable characters, and he has a couple of lines in the cut scenes.  But beyond that, he has no function.  We don’t learn anything important about him, he doesn’t do anything significant unless you happen to use him to peg one of the plottier bosses.  But from the point of view of playing an RPG, Edgar is super-important, because he’s a playable character — and one with some fun and useful abilities at that.  (“Flash” will save your ass, early-game.)  He’s important not as a person but as an object.  We care when we collect Edgar for the same reason that we care when we collect Pikachu:  you’ve got to catch them all.

Heavily narrativized games are lousy with characters and events like this.  In the original Final Fantasy, there’s a sequence where you need to feed a titan a ruby.  Why?  Because titans love rubies, and this one’s standing in your way!  In the proto-survival-horror game Alone in the Dark, there’s a point where you need to pacify a dining-room full of zombies by offering them a pot of soup.  Why?  Because zombies love soup, apparently! (Maybe best not to ask what was in it.) And in The Legend of Zelda:  Ocarina Of Time, you spend what I consider to be a frankly irresponsible amount of time, given the larger issues at stake, hunting down chickens for a random villager you meet.  On a local level, all of these tasks and characters are designed to be as colorful, evocative, and entertaining as possible.  But they build to nothing.  They derive from nothing.  They are powered, in narrative terms, by nothing.  Literature does offer a model for this kind of plotting, in which colorful episode succeeds colorful episode in turn until eventually, exhausted, the author calls it a day:  we call these novels picaresques, and their plots are not well-made.  All video games that have plots have picaresque plots.  And as a result, obviously, they can’t build up tension in any of the ways that well-made plots traditionally do.  Because the episodes that make up the picaresque, quasi-serialized plot of the game are disconnected from each other, there’s never a sense that they are building to an end.  There’s always room for one more — often quite literally, as a bonus stage or boss or dungeon or the like.  And as a result, the plot is never going to grapple with, or be energized by, a meaningful sense of constraint.

All of the big narrative questions that figure in traditionally well-made plots can feature in video games.  There are games that revolve around murder mysteries, games with heavy foreshadowing, and even games where you’re trying to find out more about the psychology of the player character.  The old point-and-click adventure Déjà Vu is all three of these, in fact — and hey, I ended up talking about it after all!  But these narrative questions do not matter in the same way in games as they do in other stories.  We fundamentally care about something else.  Video games are only ever dunnwhethers, and rather than caring about the characters’ actions, we care about our own skill.  In this sense, the big overarching question for plotless Tetris is the same as that for plot-addled Final Fantasy, assuming we play Tetris on the B-Type mode where a win condition does exist.  Will I be able to beat the game?  The major constraints we deal with when we play video games — and this should be painfully obvious to anyone who has ever played video games — are the constraints of our own skill.  That’s the struggle, and the primary excitement, that we take away from the experience.

But is that really all there is?

Academic videogame critics sometimes like to separate the game as game (the ludic element) from the game as story (the narrative element).  I almost don’t want to bring this up, because the line I just wrote constitutes the sum-total of my knowledge of academic videogame criticism, and I’m over my head here enough as it is.  But the ludic/narrative division is one neat way to get around the problem I brought up in the last paragraph.  By this way of thinking, the game’s “plot” is its narrative, and responds with a few adjustments to the standard toolkit we use for other kinds of narratives.  The gameplay on the other hand is entirely non-narrative:  it’s a system of rules we can manipulate and master, and nothing more.  So when we care more about beating a game than we do about the psychodrama of the characters, that’s just a function of our valuing the game element more than the story element.  This is certainly a useful division, and there’s probably at least a grain of truth to it.  But I don’t buy it.  Because in my own experience, there is something distinctive and untranslatable about the way that story, specifically, operates in videogames.  My goal here is to tease out what exactly that is.  Today I’ve just laid in the groundwork — hopefully next time, I’ll get a little further.