August 18th, 2020
I developed the encounter system as an adjunct to the Storytron technology. However, I am now working on making it a standalone technology. To do this, I must determine the architecture of the storyworlds operating in the encounter system.
I’ll begin with some architectural diagrams of the three different interactive storytelling systems: text adventures, Storytron, and encounter systems.
A standard text adventure or interactive fiction system starts with a room diagram. The fundamental verbs available to the player move them from one room to the next; inside each room the player is able to execute verbs that obtain props or manipulate components of the room.
The player follows a path through the rooms, often doubling back or retracing his steps. From the player’s point of view, the experience looks like this:
Here, the circles represent not rooms but verbs; the heavy arrows represent choices that the player made, while the gray arrows represent unchosen choices that were available to the player. The player’s path through this tree structure represents the story that the player experiences.
The Storytron technology uses a directed graph of verbs:
The player experiences much the same thing that the player of the text adventure experiences: a series of choices that traverse a tree:
Of course, in both the text adventure and Storytron technologies, there is a crucial additional factor: a set of global variables that change values in response to the player’s actions. The player wins by accumulating the proper set of global values. In text adventures, the correct set of global values usually permits access to a room that comprises victory. In the Storytron technology, the correct set of global values induces other actors to perform actions that grant victory to the player.
Encounter-based storyworld operate on a completely different basis. Their consequentiality graphs look something like this:
In this diagram, the diamonds represent encounters, each of which presents the player with up to five options, which in turn do NOT affect what encounter is selected next. Instead, encounters are chosen semi-randomly (there are a few logical constraints the storybuilder can place on the consequentiality of the encounters). This represents a radical departure from other systems, in which the player’s decision directly determines what follows. In the encounter system, the player’s decisions affect only the relationships that other actors have towards the player. The order in which events occur is not in any manner under the player’s control.
This sounds radical, I’m sure, but I believe that it is more appropriate to storytelling, in which events may not follow an obviously logical course. The protagonist seldom makes a single decision that decides the outcome of the story. Instead, the protagonist in a story makes a large number of small decisions that ultimately lead to the final conclusion. Other interactive storytelling technologies are, at heart, modified game systems. The encounter system is more appropriate to storytelling.
A profound difference between games and encounter systems lie in the path taken to success in each. Games are fundamentally boolean in structure in that the player must successfully overcome each and every challenge in the game. These challenges are provided in a sequence, so that the formula for success in a game is:
Winning = A & B & C & D & E & F…
where each of the letters represents a challenge to be overcome. But encounter systems are essentially arithmetic in style, because the only global effect of the player’s choices are the pValues that other characters have for the player. These values change in small increments. In order to succeed in an encounter system, the player must slowly build up a set of pValues that are appropriate to success. Thus, the formula for winning in an encounter system is more like this:
Winning = A + B + C + D + E + F…
The player can slip up on one of the steps and still succeed. The effect of this difference is profound. In a game, failure at any step is obvious; if you screw up step E, then you don’t proceed to the next level, or you don’t get the flaming vorpal blade you require to conquer the next monster. In playing a game, you keep banging away at that obstacle until you have beaten it. The learning curve for a game looks like this:
It’s a staircase with steps of approximately the same size; if any step is too high, the game is too difficult. If most of the steps are too low, then the game is too easy. The crucial point to note here is that, at each point in the progression, the player gets clear feedback showing exactly how well they’re doing.
But the learning curve for an encounter system might look more like this:
A game designer looking at this learning curve would experience a violent digestive expulsion. This is about as bad a learning curve as you can design! The steps are entirely too big! The feedback is completely inadequate for the needs of the player. Only the fact that you know I must have something up my sleeve prevents you from closing this window and never reading anything by Chris Crawford again.
The something up my sleeve comes from the way we learn. Yes, children have difficulty learning and need constant positive feedback to continue keeping their nose to the grindstone as they learn. Games are designed for children. But the need for constant positive feedback diminishes with maturity (note that I write ‘maturity’, not ‘age’). Christopher Columbus did not get constant positive feedback as he ventured west. There were no signs every hundred miles saying “Only ____ miles left to the New World!” He persisted despite the lack of positive feedback. Thomas Edison famously experimented with 1600 different materials before happening upon carbonized bamboo fibers. Not one of the failed materials gave him any positive feedback. He just had to keep plugging along.
Or consider the physicists who built the first atomic bomb. They started off with nothing more than theory. They had absolutely nothing to confirm their belief that a nuclear chain reaction would work. Moreover, they couldn’t build a small demo version to test their ideas. Nuclear bombs have a “critical mass” below which they cannot work. So they had to spend a billion dollars (about 0.5% of the entire US GDP at the time!) to build their test bomb. Fortunately for them, it worked. But they labored for years without any positive feedback. The learning curve for the Manhattan Project looked like this:
I admit that it’s possible to take this process a little too far. Here’s the learning curve for my work on interactive storytelling:
I continue working on the problem in the hope that somewhere further down the road there will be a working interactive storytelling solution. It’s as if Columbus somehow missed the New World, sailed across the Pacific, somehow missed Asia, and just kept going.
The critical notion that I’m trying to convince you of is that mature people don’t need instant feedback. Depending on their determination and self-confidence, they can keep working towards a goal without lots of positive feedback.
And in fact this notion shows up nicely in cinema. Let’s consider two standard genres, genres so fully developed that they’re formulaic: the action flick for males, and the romantic comedy for females.
The action flick provides a standard little-boy version of progress. The hero punches some bad guys in the face and they all fall down. Then the hero uses some fancy kicks to take down some more bad guys. Then he shoots up some more bad guys, gets in a car chase in which he tricks them all into fiery crashes, then punches-kicks-and-shoots even more bad guys. Meanwhile, he is accompanied by a sexy young woman who is duly impressed with his prowess. The hero eventually confronts the Boss, defeating him in an epic battle that somehow saves the girl from a horrid fate. Presumably the girl will be so impressed with the hero’s prowess and grateful for his saving her that she’ll eagerly jump into bed with him. Note that in the action flick we are fed a steady stream of challenges that the hero overcomes, providing the positive feedback that the audience needs. Note also that the continuing demonstration of masculine prowess gains the reproductive favors of the girl.
Now let’s turn to the romantic comedy for female audiences. Its formula always starts with “boy meets girl” under most inauspicious circumstances. He spills his soup on her or somehow earns her righteous ire. They begin an antagonistic relationship. But as the relationship progresses, each one notices endearing traits in other and softens a bit. Slowly, incident by incident, they undergo a transition from antagonism to friendship to love. The movie ends with them getting married.
Now, the rom-com does indeed show signs of progress throughout the movie. The relationship between the two people clearly undergoes change. But the changes are subtle and the outcome remains uncertain right up to the end. The learning curve’s uncertainly looks rather like this:
Thus, if we are to produce a medium of artistic expression that appeals to a group larger than young males, we’ll need to dump the boolean victory model used in games and action flicks. We’ll need something with more subtlety, more uncertainty, something requiring greater patience and self-confidence from the player. Because the changes consequent to the player’s decisions must be small, we’re definitely talking about numeric models instead of boolean ones.
My encounter-system storyworld, Le Morte D’Arthur, uses an extreme version of this approach. The player experiences several hundred turns, some of which serve only to provide dramatic context and information about the different actors in the storyworld. Arthur makes decisions that affect people’s perceptions of his competence as king. Late in the story, Arthur’s son Mordred declares a revolt, and only then do the cumulative effects of Arthur’s decisions come to bear. Actors will be forced to choose which side they will support. If Arthur’s decisions have convinced the other actors that he is indeed a competent king, then more actors will side with him; if not, more actors will cleave to Mordred. A final battle between Arthur’s and Mordred’s forces determines the outcome of the story.
The player as Arthur gets little feedback on how well he’s doing; he must make his best intuitive judgements as to what is necessary to win the love, respect, and admiration of the population. Initially the decisions will be easy, but as the story progresses, Arthur will face increasingly difficult decisions requiring difficult trade-offs.
