Water Cooler Games
Water Cooler Games served as the web's primary forum for "videogames with an agenda" — coverage of the uses of video games in advertising, politics, education, and other everyday activities, outside the sphere of entertainment.

The site was maintained at watercoolergames.org from 2003-2009, where it was edited by myself and Gonzalo Frasca. It is now archived here in full.
Search Water Cooler Games:  
You are reading an archived version of this article. The original URL was (loading...)
Chris Crawford's Nine Breakthroughs
by Ian Bogost April 12, 2008
categories: Game Design

Games industry curmudgeon and interactive storytelling proponent Chris Crawford spoke at the Game Developers Exchange conference here in Atlanta yesterday. As a part of the talk he explained the "Nine Breakthroughs" that were important to his work on Storytron. I recorded them here, after the jump.

Chris Crawford's 9 Breakthroughs

(1) People not Things
Our focus needs to be on social not physical interaction.

(2) Always ask, "What does the user do?"
Not what does he see, hear. If you know what the user does you know what the software does. "Crawford's First Law of Software Design."

(3) Focus on Verbs
The user talks into the computer. Language is the primary way people transfer infomration. It's built into us. If you want to understand user interface, you need to use the frame of language. Language has verbs. In the future, people will really talk to computers -- not using a command line or a GUI - but a LUI, a linguistic user interface.

(4) A Toy Language for a Toy World
You can't do language on a computer, not easily. The Sapir-Whorf hypothesis: language is closely tied to the perceptual reality of the speaker. Language mirrors reality.

Natural language on a computer would require reality as well. And reality is hard to get inside a computer.

When you create software, you are creating a model universe -- a toy reality. If a toy reality fits, a toy language can fit.

(5) Use an Inverse Parser
If you are going to use a language, but it's not a natural language, how does the player learn it?

A parser makes sense of language. An inverse parser does it in reverse: it starts with the language

Start with "I" for example. There are only so many valid verbs in a context, so provide a menu. Choose "give," for example. If you give, you have to give some thing. Now present a menu of the valid objects to give. Now the only word that makes sense is who you are going to give it to. Another menu. That's an inverse parser.

You don't have to know the language. It presents it when necessary.

(6) Design the Language First
Start from the language, and the language creates the model. The model suggests the interface. The strongest version of Sapir-Whorf: the language defines the reality. You just "fill in the dictionary" -- define what the words are and what they do.

First design the language, then the model, then a user interface that presents that model to the user.

(7) Programmers are not Storytellers
You need an editor, an authoring environment for real storytellers.
Editor, engine, system for building language. A language specific to one story world. An author who uses this defines words, verbs, objects, actors, quantlfiers, attributes, and so forth.

For verbs you have to be able to specify algorithms that define what they do. This creates a new problem, one that has really burned the game industry: programmers are not storytellers. If you want good storytelling, go to a pro.

Real interactive storytelling about social reasoning requires an accessible tool, despite its inherent complexity.

(8) Programming Story Requires a New Programming Language

Sappho programming language designed for artists. Programmers will look at it and ask "how can I gain access to the defense department?" Sappho has strong, hard datatyping. No matter what you do, Sappho constrains numbers between -1 and 1. Actors are blue, numbers are red, stages are orange, props are magenta. The language gives you prompts that help you choose the right element. Syntax errors are impossible. You never type anything in: it's point and click. You build scripts with menus that only present legal options. There are no runtime errors: divide by zero? No problem. We invented a number for that. All calculations are segmented such that if a runtime error is generated, that calculation is "poisoned" and we're not going to deal with it. The options available to an actor may diminish, but the program continues running.

(9) An Infrastructure
It's called Deikto, and it ties it all together. The engine that runs the system. The system is ready in a manner of weeks, Chris promises. Working demos include Chit-Chat, a gossip app, and Balance of Power 21st Century, an update of Crawford's famous geopolitics game with social reasoning as the basis of success. "Interactive Storyworlds," not "Games."

Comments (1)

Man, I've been so close to the Storytron thing it's hard for me to get excited about it. It's been a few months away from being ready for like, three years. Seems like it anyway. I'm fairly confident that they are almost ready though, for real this time, having noted the milestones.

Storytron may or may not be revolutionary in it's platform, but it is a great lesson for content creation platforms in general. I think it will find it's niche among them as well. I know of one dedicated to short form games (Mockingbird), one dedicated to virtual worlds (MetaPlace), one dedicated to serious games (TBA) and then there's Storytron, deddicated to storyworlds. Why not?

I've got a post-mortem of serious game research done with Storytron coming out on SGS pretty soon, check it out.

Patrick Dugan on April 12, 2008 5:39 PM