I Can't Believe I Did That

A Reflection on Some of My Biggest Shattertown Blunders

Installment 1 - Kitchen Sinks and Dramatic Contrast

by Laura J. Mixon

I made any number of decisions that turned out to be bad ones, during my work on Shattertown. In the interests of helping other writers avoid making the same mistakes, here are some notes on the worst of them.

Wow! Look - a Kitchen Sink! Just What I Need!

The first and among the biggest of my mistakes was to try to use every single, pea-pickin' one of the Erasmatron's wide array of features. If there was a button or menu item, I wanted to bring it into play. Things, stages, lies, secrets, deals, nested macros, historical references - you name it; I tossed it in there.

Now, this came about in part because an important component of my job was to give Chris feedback on the features he was putting into the Etron; another part of it was the ooh-a-cool-new-toy-to-play-with! impulse. A third reason was that I simply had no idea when I started just how big and complex the Erasmatron was. I thought I was supposed to use it all. That was what it was there for, right? When Chris would bring a new feature online and Iíd start fiddling with it, ideas for ways it might be put to use in an interactive fictional setting would start popping into my mind, and I’d be off and running.

But just as this is bad design in programming, to throw in every kind of feature just because you can, it's bad design in storytelling. It's ass-backwards. The technology should serve the needs of the story; not the other way around. Idea generation is crucial to the creative process, but so is winnowing-out. Every story doesn't need a kitchen sink. When writing prose, I don't use every single feature that my word processor provides; it's there for when I need it. The same is true for the Erasmatron.

In the prose world we have a saying: "kill your babies." To write a good story, you must be ruthless. You must chop out whatever doesn't advance the story's plot, characterization, or theme, no matter how well-written or really cool it is. You can't afford even a speck of ruth (OK, maybe a speck...).

The Result

After spending several months creating all these threads that were supposed to lead to other really cool stuff, I was suddenly faced with a huge array of dangling ends and odd bits that didn't go together very smoothly. I spent the last few months madly splicing, shoveling, and hacking, trying to smooth out the worst of the incongruities, but v1.0 came out a real mess despite my best efforts - and even now, six weeks later, with version 1.04, the user is often faced with odd consequences fairly frequently (or, as Chris calls them, "consequence-o's"). Shattertown is off-balance and rather choppy, with odd jerks and lurches in the interactions.


While creating, stay focused on the story you're trying to tell. Sure - play for a while, but at some point, you gotta grit your teeth. Ignore those shiny play-pretties that beckon from the edges of your screen, unless they really help you achieve a dramatic effect you need. Yes, you need to be able to play to learn, and that playfulness will serve you well; just don't let yourself be distracted from your story by the capabilities of the technology. If a particular set of interactions doesn't help the user discover something important about the nature of your characters or theme, no matter how clever it is, rip it out by the roots.

Quite frankly, I don't know how avoidable this mistake is, since we're all just beginners at this, we're all going to be learning and groping for good techniques for quite some time, and the way you learn is to fiddle with the knobs and see what happens. Just be aware of it as a pitfall. As too much of a good thing, if you will.

Hey, This is War and Peace, Here - Don't Talk to Me About Simple

This mistake is so basic that you'd think I wouldn't have stumbled into such an obvious trap. But I did. Early on, despite Chris's warnings, I made my decision scripts and verb-to-verb linkages way too complicated.

There was a good reason for this. My inclination as a storyteller is to try to grasp the underlying motivations of all my characters, and bring into play all the aspects of personality, mood, and relationship that would affect each characteristic decision-making. The decision script creation process beautifully accommodate this. But I really went overboard. I reached the point where I was juggling so many variables in my decision scripts that they became confusing and downright useless. (Check out "ChallengeHot" in the editable version of Shattertown, for a case in point.)

After a few months, I found that (a) I could accomplish the effects I wanted without resorting to an anal accounting for every increment of personality, mood, and relationship that might affect each character's decision in a given situation, and (b) due to the nature of the interface, much of that subtlety I'd built in couldn't be communicated to the user, so it didn't enhance the story-playing experience anyhow. After a certain point, it was wasted effort.

The linkages were less of a problem, though not trivial: too many linkages and your reaction equations start getting hairy really fast, as you attempt to direct an interaction down one thread rather than another, via a verb that gets a lot of traffic from different directions (I actually found this aspect of the problem fairly straightforward to deal with, for the most part; you just have to be methodical and do a lot of tests using the thread tester and rehearsal functions - both of these are good at helping you track derailments of one kind or other. And if you get totally stumped, use the decootier). More importantly, though, the reason too many linkages is a bad idea is that with each added linkage you diminish each choice's dramatic impact.

To use an absurd example, if Harry socks Mike in the teeth, and you make it so that Mike can hit back with five foot-pounds-per-square-inch of force, ten, fifteen, or twenty, the user is going to yawn. What's the point? Too many gradations create lots more work for you with no improvement in the quality of the story. The different choices a character has should ideally each be dramatically different and should lead off to a variety of inherently interesting routes.

Of course, just as in the visual arts, there's room for a variety of styles, and for differing hues and shades of interaction. What I'm trying to say is, below a certain threshold, the difference between one choice and another can become too murky and low-contrast to be meaningful to the user, and that is the big risk you run by having too many linkages.

The Result

I spent a lot of time fiddling around with increments of personality and interaction that I might better have spent improving the plot or cleaning up those jarring consequence-o's.


Keep it simple keep it simple keep it simple. (Did I mention you should keep it simple?)