A story can divided into three acts that work in more or less the same way from story to story to story. This is not a controversial point of view. And honestly, it had not occurred to me that the idea of dividing story-games likewise into three acts would be particularly controversial. Until, that is, the comments started to come in on The Mamet Post.
Ted Braun, the professor of screenwriting who taught the advanced script analysis course I took at USC, was always very careful to use the phrase “division into acts” rather than “three-act structure” when talking about dramatic structure. In considering the question of acts in games, I remember why.
It’s because when you say “three-act structure,” it makes it sound like you’re talking about a blueprint. But division into acts should be much more a diagnostic tool than a blueprint, best used to figure out why a particular story (or game, I argue) smells like ass.
Before going any further, let’s talk specifics: How are these three acts best defined?
The first act asks a dramatic question. It ends as soon as the question has been posed. It has been asserted that this is the moment, in a movie theater, when you feel inexplicably moved to put your popcorn down on the floor.
Dramatic questions posed by first acts are things like, “Can Agent Kujan find Keyser Söze?” and “Can Luke reach Leia?” A first act usually takes up about one-quarter of a story’s overall mass.
Some argue that it is better to divide a drama into four acts, or five, or seven. Apparently such people have hijacked the Wikipedia page on Dramatic Structure. I haven’t done a lot of reading on Freytag’s and other proposed alternatives, but from what I have read and heard, there’s no particularly compelling argument to complicate the division past “beginning,” “middle,” and “end.”
Often, a first act conveys introductory information about characters and situations, but frankly, this is (a) not necessarily, and (b) on par with saying “a first act most usually contains the speech of humans.”
The second act answers the dramatic question posed by the first act. It ends as soon as the question has been answered, one way or the other. Usually, it approaches the question from lots of different angles, and the protagonist fails a lot. In fact — and this is a key, key, key point — the second-act question is often ultimately answered in the negative. That is, the protagonist often fails completely. (Agent Kujan finds the wrong Keyser Söze at the end of the second act.) Note, however, that the hero’s failure won’t necessarily deny us our Traditional Hollywood Ending.
The second act is almost always the longest of the three acts, weighing in at about half of the story’s fighting weight. It is the part of the story a writer is most likely to use the word “slog” to describe.
The third act asks and answers, “Then what?” Leia’s been rescued… now what? Ripley’s escaped Alien Hell… so she’s going to do the only sane thing and jump right on that dropship of Bishop’s right?
In a story that’s any good, “Then what?” conveys the sense that there has been a larger purpose to answering the question explored in the second act. It communicates how the lives of the protagonists have been or may be forever changed, for better or worse. The third act often poses and answers a question that ought to have been asked at the beginning (had we only known more), or escalates to some even-more-critical follow-on question.
The third act usually runs about as long as the first act: one-quarter of total length.
Division into acts. All well and good for traditional stories. What about games, and what about practice instead of theory?
The concept of division into three acts is absolutely useful to game designers, and not just to game designers who work on traditional and computer RPGs. I’ll be following this post with a series of three more, each addressing one of the three acts, and, with any luck, proposing a variety of ways the basic structure described here can be used to make better story-games.
Stay tuned.
I’ll buy the idea that value can be had from looking at games by dividing play into acts but I think that I can count on one hand the number of games that have anything like a third act to them.
Unless, of course, gaining and spending XP counts as the third act.
I’ll have more comments later, after I soak in the whole article, but the “three act” structure originates in the whole “every story has a beginning, middle, and end” thing. Aristotle’s Poetics is no where near as “blueprint” oriented as its critics claim.
As for whether 5 acts, or 4, or 27, are better, that’s inconsequential from a certain perspective. Shakespeare is downright mechanical in his use of the 5 act structure, every act III scene I is VERY important, but even his stories had a beginning, middle, and end.
It should be noted that Television, and film for that matter, actually follow a 5 act trend. Television does it to better coincide with commercial breaks, though arguably they might have timed the breaks with act breaks.
Freytag’s pyramid is one of the roots of my organization for dramatic structure in RPGs, though the branches of my structure wander pretty far from the strictness of Freytag’s design.
Jason, I think the third act in a lot of games happens during the post-game, when players are either asking the GM questions about the fates of this or that NPC, when the GM is revealing a bit of denoument to the players, or when the players are simply hashing out what their characters are doing between sessions.
One story’s Act III can be another story’s Act I, in other words. This is, in my experience, a great way to structure play. Put Act I at the end of a previous game session and start the next session in the middle of things (in medias res), right at the beginning of Act Two. The jobs traditionally associated with Act I are often handled in a single paragraph of read-aloud text in an RPG, anyway. (E.g., “You must slay the evil wizard and rescue the Princess! Your are standing at the door the wizard’s tower and…”)
But the jobs that first and third Acts serve are still important in RPG stories. They’re just not quite the same.
Someone whose name I don’t remember said “My job as a storyteller is to chase the heroes up a tree, and then find a way for them to get down.”
He, whoever he was–I think a movie director. My failing memory makes me think it was Preminger or Hitchcock, possibly Sidney Lumet–was making a point about the Three Act Structure. He was saying, “I divide the story into two sections. Getting everyone in trouble, and then getting them out of trouble.” George Lucas has said something similar. His movies assemble a group, then disperse the group, then bring the group back together.
One of the things I fought for, for about 18 months, on the Mercenaries 2 project, was a First Act. Most video games don’t have one. Most games put all of what would normally be in the first act of a story, into something the player reads, or is hamfistedly exposited to him, so the
To use the simplest example, Donkey Kong, unjustly uncredited as being the first video game with anything approaching a story, begins with a King Kong, fresh from having his serial numbers filed off, stealing Princess McGuffin. Mario must save her.
This means Donkey Kong, like most games, begins with Act Two. The chase to get the girl.
In Mercs2, I refused to draft a version of the story in which the game opens with the bad guy doing something shitty, and the hero vowing revenge. I called this the “Solano Kicks Your Dog” model.
Instead, you work for the bad guy. You actually do a mission (originally 3) for the guy before he stabs you in the back. By the time Act One is over you should, if I’ve done my job right, know who the bad guy is, hate him viscerally, know his minions and the structure of his organization, and know what he wants. And be motivated to get him, find him…kill him.
Since those heady days of 2005, we’ve seen Mass Effect and Assassin’s Creed which both have distinct First Acts. In Assassin’s Creed’s case, it’s pretty poorly executed, but it’s there!
Sounds like Mercs2 uses the Standard Cyberpunk/Shadowrun Model. Employer X hires the PCs to do a job (Act I). At the end of Act I, Employer X betrays/backstabs/doublecrosses PCs. Act II, Employer X does his best to kill PCs. Act III, PCs raid Employer X’s HQ/lair/secret base/private island/etc. for the final confrontation. Not a bad model, actually… the plot is highly predictable, but at least it keeps moving forward.
From a plot-driven standpoint, however, 3 acts is a bit artificial. I like the idea that you can’t divide up the acts until the story is over and you’ve moved into analysis. For a collaborative work that is entirely first draft, I think that fits RPGs as a performance medium fairly well. But I’m having a tough time wrapping my head around “Analysis = Act III”.
Christian:
When thinking about this topic, I think it’s worthwhile to draw a clear distinction between the structural division of a story and procedural division.
Television, indeed, divides one-hour drama into four (or, increasingly, five) acts for the point of inserting commercials, but if you look at the underlying dramatic structure of most of them, a four-act teleplay usually maps such that its first act corresponds to the structural Act I, the second and third map to the structural Act II, and the fourth is more or less the structural Act III.
My guess is that the same kind of correspondence is true of Freytag.
I like the idea that you can’t divide up the acts until the story is over and you’ve moved into analysis.
That’s interesting. The assumption I’ve been working with is that the broad outline of the second act question is almost the only thing you can identify up before the story’s over. That is, if you know, in broad terms, what the overall thrust of the scenario is — its dramatic question — that gives you a necessary framework where you can hang everything else that the PCs want to do. Even if the heroes approach the dramatic question from the opposite side (they decided that they’d rather kill the princess instead of save her, or whatever), you’re still more or less on the spine you identified in the first place, and so your planning and thinking in advance of the game were not for naught.
You’ve got me realizing, Matt, that lots of the games I’ve loved (rather than just liked a lot) have first Acts: THIEF (all three), SPLINTER CELL: CHAOS THEORY, a few others. Notably, HITMAN: BLOOD MONEY, wherein the series found its stride IMO. What is it, then, about the stealth genre that enables first Acts?
SANDS OF TIME has one, too, actually, in addition to bookends. SECOND SIGHT has one, it just doesn’t play out at the beginning… I guess. I was aware that these games all had Act Ones, but I hadn’t drawn any potential connection between the presence of a first Act and my admiration. Hmm.
We could argue that training levels are often the first Acts in videogames, but even I don’t quite buy that. Surely, though, it’s the formula of shooter level structure that tends to bore me even through the action. I’m just now playing F.E.A.R. for real(z), for example, and the larger tale of the game is boring the gumption out of me.
I will think on this more.
Jeff:
I agree that there is a distinction between structural and procedural act structures, hence the Aristotle (beginning, middle, and end) reference in my first paragraph. I really think those that have disdain for the three act structure as a rigid blueprint for a script don’t understand how broad the three act structure truly is. Shakespeare’s five act, procedurally and rigid, structured plays fit easily within a three act narrative structure. They have beginnings, middles, and ends.
It should also be noted that games like Donkey Kong, Karate Champ, and Street Fighter, may not seem to have a first act when in fact they do. Either the first act is implied, all players of DK know that the princess was kidnapped by the Gorilla, or they are brief. As you rightly point out, the second act is where the action is, particularly from a player perspective. Most of the old D&D modules contained the first act stuff in the DM background and provided the outline for the 2nd act as their content, leaving any denouement to the respective DM in his/her campaign.
Games like Feng Shui followed the heavy act II formula, where DC Heroes encouraged the creation of act I’s in the character design process — gotta write that background for xp. Vampire had a similar “background” story to be roleplayed element.
I also think in games, especially in a “campaign” environment, that one adventure’s Act III can be the next adventure’s Act I. Daxter for the PSP is a perfect example of this kind of game narrative.
All that said, the 5 act television screenplay is more than a procedural structure. While the three act assumes a beginning, a middle, and an end (set-up, action, and denouement) the five act follows a introduction, adventure begins, raise the stakes, hero loses it all, denouement formula (the Shakespearian outline though hamlet’s fifth act is very short — mere lines of dialogue depending on where you place the fifth act in analysis).
All of which leads to the comment that clear demarcation of where acts begin and end is better left after the narrative has been constructed. Not necessarily after the final draft, but certainly at some point after the basic creation of the narrative.
A writer oughtn’t write, or possibly design, from a “now here is the beginning of the third act” as a practice of the writing process. Rather, during the editing process — or in the creation of a first draft if working from a murder board — how the story develops is an issue.
There is more to say, but I visit your blog during my 15 minute breaks and there is only so much I can type. Otherwise all posts would degrade into a comparison contrast of Pieces of Eight vs. Cthulhu 500 as examples of game design, oh…and Trojan sports.
Fight on!
Will-
I’m not sure, though it is a good question. Don’t underestimate the power of one person/company innovating and succeeding and then other people ripping them off without knowing why. A lot of development is based off the Cargo Cult mentality.
If we decide for the moment to presume it’s something about stealth games that causes it, and wonder what that might be, I’m challenged to come up with a reason. I think I’d start casting about for templates in movies they’re stealing.
We referenced Red Harvest by Dashiell Hammett frequently because it was the earliest story the other writer and I knew that had the basic template of “I get screwed and decide to kill everyone.” Presumably, however, most video game developers have not read Red Harvest. They may have seen Yojimbo!
It’s telling, isn’t it, that when trying to figure it out, the first thing I do is wonder if there’s a movie they’re all ripping off. 🙂
Christian:
I gotta give you credit for volume of thoughts and words in 15 minutes!
The only thing you said that I’m skeptical of is drawing an equivalency between a third act and a denouement. Perhaps we’re getting to nitpicking about word meaning at this point, but I think an important feature of a third act with real worth is that it has its own dramatic tension: there’s a question to be answered. I’m not just watching to get a murder mystery–style explanation of how it went down earlier on. Instead, there’s something new going on that’s going to tell me what happens to the protagonist in the long run, and at this point I care a lot about him or her, and so I’m dying to continue following the story.
But like I said, at this point, we may be debating the definition of “denouement.”