December 4, 2003

Magic Crayons and More

by Andrew Stern · , 9:32 am

Recently I made some free time to fully read Chaim Gingold’s Masters thesis on game design and interactive narrative, which we briefly posted about last May. I’m going to offer some informal feedback and comments in this post. Frankly I found this thesis as substantial and useful as most PhD dissertations in the field. It could also serve as a game design handbook! Chaim’s really trying to present theory and make it accessible and useable to practitioners, which I found impressive. (This should have been turned into a paper presentation at LevelUp, not just a poster…)

The title, “Miniature Gardens & Magic Crayons: Games, Spaces, & Worlds,” is a bit misleading; at first I expected the thesis to focus exclusively on microworlds, and in a sense it does, but in fact it goes pretty broad with several universal design principles for enjoyable interactive experiences, about more than just “games”.

Some of the thesis defines and describes several common-sense design principles, such as consistency, balance, intelligability, plasticity, safety, etc. Chaim talks about syntonicity — a fancy word for “how we relate to worlds”. The points he makes about the risks of building complex worlds are very good — that they can become snarled and hard to understand, without clear feedback on the effects of player’s actions; how higher resolution but “dead” data makes a world inert.

The thesis has a good deal of discussion of two “masters” of game design, Shigeru Miyamoto (Donkey Kong, Mario, etc.) and Will Wright (SimCity, the Sims, etc.). I found it very useful for the major gameplay and design features of Miyamoto’s work to be identified and analyzed like this. And some interesting design practices of Will Wright are revealed, gleaned from personal conversation, that I hadn’t read elsewhere.

(All of this analysis of well-designed virtual worlds made me think more explicitly about the nature of the (small) virtual worlds I’m personally interested in building. I realize I want to build worlds that are consistent, intelligable and plastic, but not perfectly “safe” — worlds than can get under your skin, and bother you later. Worlds that have some sort of “real” consequence for players, that force you to think twice about acting destructively or recklessly in them, or worlds that at least flirt with that feeling of danger… Perhaps rather than elegant, cleanly designed worlds (“miniature gardens”), I’m interested in messier, more complex, more ambiguous worlds. Well, actually, I do want there to be clean design — but with some messiness, some noise, that allows for some trangressive behavior…)

Chaim’s term “magic crayons” is a nice way of framing the idea of languages for authoring dynamic, computational objects and worlds. He’s trying hard to promote the idea of creating new, powerful, expressive authoring tools for non-programmers :-) — a big, critical issue the field faces, due to get more and more attention over time. In the discussion he is simultaneously paying attention to the programmer of the languages, the author using the language building a work, and player who finally experiences the built work. He compares the activities of each — each of their “possibility spaces”. Following thoughts of Will Wright, he makes interesting points as he relates and compares paint programs to games. I liked the line about “mountain climbing in Kansas.”

At one point he begins breaking down the components of point of view, and discussing them in various games.

In the final 25% of the thesis, following the design principles he identified earlier, Chaim describes how he has begun building his own (still-under-development) magic crayon, “Comic Book Dollhouse,” discussing the design process he went through. (I suppose if this had been a PhD thesis, CBD would have been even more complete, but have taken another year or two of work…) CBD allows players to create storytelling objects and characters, and attempts to collaborate with the player to create new (fixed) sequential art (comic) narratives. The creation of the comics is the primary activity of CBD; the narratives generated are non-interactive.

“CBD, unlike other story generation or interactive storyworld software, emphasizes authorship and story construction as the primary mode of participation for both players and authors. … CBD is intentionally designed to solve the problem of storyworld participation and authorship, and leave the question of generation for later.My hypothesis is that the problem of generation will not be so hard once we have a grasp on the design problem of authorship and player participation.” This a design approach that I bet could work well, perhaps really well.

Chaim gets a bit into the realm of story generation, with CBD’s feature of suggesting alternate future narratives sequences: the system looks at the existing author-created patterns, to learn new ones. The only problem I can imagine with this, is that at first with only a few example stories for the system to look at, it will be suggesting plot fragments of things that were recently authored, making it seem limited or repetitive? This system probably needs to be stocked with a nice set of existing plot fragments (which Chaim suggests in the Sculpt vs. Specify section).

Come to think of it, the name “Comic Book Dollhouse” isn’t perfect. “Dollhouse” does convey the idea of a microworld, but unlike the Sims, from what I can tell in CBD the setting of the stories you can create aren’t necessarily in a house. So maybe something like “Comicbook Doll World” might have been a little more accurate?

Chaim loves the word “inchoate”; I gotta start using that (just like I learned “aleatory” during a previous Fiction and Recombinant Text discussion.) A miscellaneous little suggestion: I think Section 6.3.3.3 needs a diagram of what overviews look like…

I had time to read the thesis but not yet a chance to play with CBD itself (available for both Mac and PC!). As much as I like the ideas that went into CBD, personally I’m afraid I’m not sure I’d enjoy playing it — but I’m sure that’s just me, accustomed to using expressive AI-based languages I helped develop to create interactive stories, so I’m probably a really bad test case, not who CBD is meant for (non-programmers). However, it’s not just that… hopefully this will be constructive criticism: As a story is being made, there’s something depressing about seeing the finished comic on the left of the screen. It feels dead, fixed — which those frames are at that point, yes? (although undoable and modifiable, if the user wants). Perhaps that’s just an interface issue; if we only saw the present and future, and not the past, I would probably not feel that same dragged-down feeling. I’m wondering if seeing the (now dead) past alongside the present and future suggests too much to me that the point of this is to create a static comic for enjoying later; instead I want to enjoy the moment itself, and only later perhaps look at a trace (the static comic) of what happened.

Chaim ends his discussion of CBD with, “Ultimately, I want players of storyworlds to feel they are taking risks, making strategic choices, and solving problems while engaging miniature dramatic worlds, and authors should feel they have a magic crayon for making these worlds.” From what I can tell, the version of CBD described in the thesis (perhaps CBD has advanced since then, I don’t know), doesn’t yet accomplish all of this. Ideally the thesis would have done a little more thinking about how to acheive those things, although those are hard design questions, that will most likely need significantly more R&D to answer.


Creative Commons License The text of this blog entry is licensed under a Creative Commons License.