Chris Crawford on Interactive Storytelling (32 page)

Chris R. Fairclough and Pádraig Cunningham of Trinity College in Dublin, Ireland, have integrated Propp’s system into a story engine with impressive results
³
. Their software uses a double layer of intelligence to provide the narrative. The lower layer uses agent-based technology to guide individual Actors in the storyworld; the upper layer uses Propp’s rules in a case-based reasoning system, which applies old precedents to new problems. The software maintains a library of previously solved problems in the problem domain. Upon encountering a new problem, the software searches through its library of precedents and identifies the closest matching precedent, and then adapts the solution to the precedent to solve the current problem. Propp’s ideas are used to supply Actors with roles to play (Villainy, Guidance, Testing of the Hero, and so forth). Actors are capable of gossiping with each other; the authors hope to augment this capability with deception at some future time. Their system isn’t complete as of this writing, so we must wait to evaluate their results.

In 1921, a French writer, Georges Polti, published
The Thirty-Six Dramatic Situations
^4,
which reduced the basic storylines of all literature and theater to a core set:

1. Supplication
   
2. Deliverance
   
3. Crime Pursued by Vengeance
   
4. Vengeance Taken for Kindred Upon Kindred
   
5. Pursuit
   
6. Disaster
   
7. Falling Prey to Cruelty or Misfortune
   
8. Revolt
   
9. Daring Enterprise
   
10. Abduction
    
11. The Enigma
    
12. Obtaining
    
13. Enmity of Kinsman
    
14. Rivalry of Kinsman
    
15. Murderous Adultery
    
16. Madness
    
17. Fatal Imprudence
    
18. Involuntary Crimes of Love
    
19. Slaying of an Unrecognized Kinsman
    
20. Self-sacrifice for an Ideal
    
21. Self-sacrifice for Kindred
    
22. All Sacrificed for a Passion
    
23. Necessity of Sacrificing Loved Ones
    
24. Rivalry of Superior and Inferior
    
25. Adultery
    
26. Crimes of Love
    
27. Discovery of the Dishonor of a Loved One
    
28. Obstacles to Love
    
29. An Enemy Loved
    
30. Ambition
    
31. Conflict with a God
    
32. Mistaken Jealousy
    
33. Erroneous Judgment
    
34. Remorse
    
35. Recovery of a Lost One
    
36. Loss of Loved Ones
    

Although Polti intended this list to be universal, it does reflect the cultural context of his times. For example, Situation #27, “discovery of the dishonor of a loved one,” hasn’t been much in use in the past 50 years. However, Situation #9, “daring enterprise,” covers many of the movies and books now popular.

To give you a better idea of his system at work, here are his categorizations of some famous tales:

Hamlet
: #4 and #13 (vengeance taken for kindred upon kindred, enmity of kinsman)

Romeo and Juliet
: #29 (an enemy loved)

Macbeth
: #30 (ambition)

Don Quixote
: #2 (deliverance)

The Purloined Letter
: #11 (the enigma)

Around the World in 80 Days
: #9 (daring enterprise)

The Ring of the Nibelungs
: #5 (pursuit)

The War of the Worlds
: #6 (disaster)

Many of Polti’s categories are broken down into subcategories. For example, he breaks down Situation #22, “all sacrificed for a passion,” as follows:

A.

1. Religious vows of chastity broken for a passion
   
2. A vow of purity broken
   
3. A future ruined by a passion
   
4. Power ruined by passion
   
5. Ruin of mind, health, and life
   
6. Ruin of fortunes, lives, and honors
   

B.

1. Temptations destroying the sense of duty or pity
   

C.

1. Destruction of honor, fortune, and life by erotic vice
   
2. The same effect produced by any other vice
   

Finally, Polti lists the basic roles necessary to each situation; for #22, he offers The Lover, The Object of the Fatal Passion, and The Person or Thing Sacrificed. I believe this taxonomy could become the basis for an interactive storytelling engine, but at present I have no concrete suggestion to offer.

It might be possible to build an interactive storytelling engine using a scheme that breaks stories down into components or categories. Three such schemes have already been built: Aarne-Thompson, Propp, and Polti. Of these three, only Propp has been adapted for computer use.

“IF YOU CAN’T SAY IT
, you don’t
know
it.” That’s what one of my English teachers used to say, and over the years I have come to realize how profoundly right he was. It’s not just that transcribing an idea from thought into language is the only proof of thought; the thought and the language are deeply intertwined.

Linguists have tussled over this idea for 75 years; it’s formally known as the Sapir-Whorf hypothesis. The strong version of this hypothesis declares that language
determines
the nature of thought.

The weak version claims only that language
influences
thought. The strong version has few advocates, but the weak version has attracted a large following. Those who don’t like the weak Sapir-Whorf hypothesis spend their time attacking the straw man of the strong version. I accept the weak version: that language influences thought, but you can Google the term “Sapir-Whorf” to find more debate than you’d likely care to wade through.