Chris Crawford on Interactive Storytelling (28 page)

FIGURE 7.3:
PacMan at a junction.

Now consider the situation as it might exist a moment later (see
Figure 7.4
).

FIGURE 7.4:
Same junction, different internal variables

PacMan finds himself back at the same intersection, with exactly the same three choices: up, down, or right. But this time the context is different; the positions of the ghosts indicate that PacMan should go up. Therefore, it’s possible to design a network of choices so that the player faces the same choices over and over, but the choices mean something different each time. PacMan offers players exactly 23 such choices (if you reject turning around in the middle of a pathway and
reversing course). Of those 23 choices, 19 are binary (if you exclude course reversal as a choice), and 4 intersections offer three choices. That’s all there is. Instead of creating a thousand different branchpoints, the PacMan designers created only 23. It’s the context that makes those choices variable in meaning. You can actually quantify that context. There are four power pills and 270 edible dots. Each object has just two states: eaten or uneaten, requiring just one bit of information to store. That’s 274 bits. Next come the four ghosts and PacMan himself; each has a position that could be stored as just 9 bits of information. That’s at total of 45 bits for the five movable characters. Add up everything, and you get a grand total of 319 bits of information. That’s the context of PacMan. It might not sound like much, but in fact it amounts to 10
⁹⁶
differentiable states. Yes, 10
⁹⁶
is a large number—ridiculously large. In other words, the PacMan designers built a system with a tiny number of branchpoints but made it rich by giving it 319 bits of context, which transformed a simple maze into something much more interesting.

Interactive storytelling, however, isn’t a mechanical subject that can be handled with simple geometric mazes and up/down, left/right decisions. The systems built for interactive storytelling can’t directly use a system as elementary as PacMan’s, but storybuilders can use the example of PacMan to guide their efforts. You cannot return players to the same decision point 100 times, but you can return them to similar decision points several times. For example, consider the layout of a romantic storyworld; surely kissing plays a large role in such a storyworld. What’s wrong with presenting a potential kissing situation to players several times during a storyworld? The first time, it’s a clumsy oaf who wants to kiss the heroine; the next time, it’s a smooth operator; the third time, it’s the clumsy oaf who has since turned out to be quite sweet. It’s the same situation in three completely different contexts.

In general, then, branching stories don’t work, even with foldback. If you twist a branching tree system around enough, however, so that lower-level branchpoints can feed back to upper-level branchpoints, you can transform it from a tree into a network. (Computer scientists call it a
directed graph
, which is why computer scientists haven’t made any progress with interactive storytelling.) If you then set up appropriately differentiating contexts, you can have your players move through the network of dramatic possibilities, revisiting each point with a different context each time.

Another commonly used strategy is the
constipated story
. This little jewel offers a story in fragments, with each succeeding fragment earned by successfully completing a game segment (see
Figure 7.5
). It’s rather like watching a movie on DVD, except this DVD requires you to jump through hoops before it shows you the next portion of the story. Sometimes the obstructions are puzzle-like; some are more game-like. The story in these cases is never interactive, however; apparently the designers believe that if they alternate between an interactive game and a noninteractive story fast enough, the game alloys into an interactive story. Right. Despite these problems, purveyors of these software products are always able to find some people willing to pay $40 for them. Most people, however, prefer to pay $15 to get the entire movie without the masochistic encumbrances.

The classic example of this approach dates back to the early 1990s and a game called The Seventh Guest. This game was quite a sensation when it appeared because it featured full-motion video, which was a major technical breakthrough. The gameplay consisted of a series of tricky puzzles; after successfully completing a puzzle, you’d be treated to (Wow!) more full-motion video presenting a little more of the story. Only after solving all the puzzles did you see the resolution of the story.

I described this approach in my discussion of Dragon’s Lair in
Chapter 6
, “Verb Thinking.” The basic idea is to set up a single storyline and make alternative paths available to the player. However, should the player be so insolent as to actually try one of these alternative paths, the game kills the player. “You can have any story you want,” says the designer, “so long as it’s mine.” These products are little more than training mazes for rats who pay for the privilege.

People have been attempting to embellish games with stories since at least 1980. In the early 1980s, stories were put into the game documentation. The story wasn’t part of the game; it served as dramatic context for the game. The
documentation would present some grandly written tale of action, intrigue, and suspense (the most common line being “THE FATE OF HUMANITY IS AT STAKE!!!!!”). Several of my early games used this technique, although with somewhat more nuance.

FIGURE 7.5:
A constipated story.

Typical games of this time had considerable gaps between the story’s dramatic content in the manual and the game on the screen. You’d read grandiose tales in the manual and then confront orange and purple squares buzzing about on the screen.

In the early ‘90s, CD-ROMs became available, and suddenly multimedia was the rage. Designers used this new capacity primarily for the purpose of inserting a story into the game. One example is Wing Commander, a space combat game that included an overarching story. The Wing Commander series of games were quite a fad in the early ‘90s. You would go out and fight the bad guys, and then come back to the space base to experience another step in the unfolding story. The story and the game were independent, so the manner in which you played the game didn’t affect the story development beyond simple winning and losing, and the story development had little effect on the game. This design is what I call an
interleaved story/game
. You played some interactive game, saw some noninteractive story, and then went back to the game. Somehow the alternation between noninteractive story and interactive game was supposed to blur the two together and create a “kinda-sorta” interactive story.

A variation on this design is the
interleaved story/puzzle
, best exemplified by Myst. This game offered a classic adventure-style puzzle built into a kind of hidden story. The product’s popularity was boosted by the use of beautifully rendered 3D imagery. The result was a huge success, but Myst didn’t offer anything like true interactive storytelling. Players deduced the story from fragments picked up during their meanderings through the Myst world. The story itself was completely noninteractive, in no way responding to players’ actions. The most convincing demonstration of this approach’s futility lay in the declining success of the product line. Although each succeeding product in the line offered even higher quality, overall sales dropped off. The Myst concept was a fad, not an enduring solution.

The game Half-Life (1998) produced an even better attempt at an interleaved story/game. The story was presented in the same 3D real-time engine used for gameplay. Previous games had alternated the display between the game screen, with lots of action, and the story screen, with lots of graphics. The dysjunction between the two displays was obvious and intrusive; by integrating story elements into the same display engine as gameplay, this dysjunction was eliminated,
and the interleaving of story and game became seamless. But the change was cosmetic only; the basic architecture of interleaving was unchanged. Half-Life triggered a series of emulators that expanded on the interleaving concept. One of the more notable examples is Grand Theft Auto. Its third incarnation generated lots of excitement, but didn’t change the basic structure.

Three arguments explain the continuing failure to attach stories to games. The first error lies in subordinating the story to the game. For many game designers, the game is the core of the entertainment, and the story exists to embellish the gaming experience. In many cases, the story is nothing more than a crude justification for the game’s mayhem. After all, blasting every living creature you encounter isn’t exactly approved by Miss Manners, so some convoluted backstory must be concocted to make this extreme antisocial behavior morally plausible. Subordinating story to the needs of the game, however, inevitably yields handicapped stories.

The second argument concerns the intrinsic subject matter of the two entertainments. Games are ultimately about things, and stories are fundamentally about people. Game designers go to great lengths to dehumanize victims, so the player spends his time gathering things, tracking things, acquiring things, shooting things, being chased by things, and so forth. The gamer plays with things, never people. But stories always have been and must be about people. So long as games are stripped of human relationships, they can never be compatible with stories.

The third fundamental incompatibility arises from the noninteractive nature of stories that are bolted onto games. Attaching a noninteractive story to an interactive game yields a product that’s neither a good game nor a good story. When jammed together in a shotgun marriage, story and game don’t complement each other; they merely stumble over each other. The public’s appetite for stories is so great that it can tolerate such monstrosities, but nothing that has emerged from these efforts has achieved the goal of interactive storytelling. These storified games remain a design bastard with no promise for the future.

People have been trying and failing to build interactive storytelling software since the 1980s. They keep repeating the same mistakes: branching trees, fold-back schemes, constipated stories, kill-’em-if-they-stray, and storified games. There are so many more interesting ways to fail with interactive storytelling; try to avoid these boring ways to fail.