Shattertown Sky

A Post-Partum

by Laura J. Mixon

Almost exactly twenty-eight hours ago, now, I shipped "Shattertown Sky v1.0," the world's very first e-story (as I like to call them) created using the Erasmatron, Chris Crawford's development environment for creating a new kind of electronic interactive fiction.

Now that the crunch is over, my head is filling up with all sorts of thoughts about "Shattertown" itself, as an example of a new narrative form that has some things in common with other forms of entertainment, but some important differences as well. Very much on my mind is what all this might mean for electronic entertainment in general.

This essay is an attempt to sort out this tangle of thoughts, feelings, and philosophical musings - to share my experience, for the benefit of my friends and colleagues in the biz, and to help future storybuilders. I intend future essays that talk more generally about what Erasmatron-based e-fiction is and what all this will mean for the future of interactive entertainment.



StS1.0 comprises:

  • 596 Verbs (i.e., possible interactions)
  • 30 Characters
  • 37 Active Stages
  • 128 Things (82 assigned to characters at story start)

Note that all these quantities are merely potentials at runtime start - they only come into play as the reader makes choices that carry them down particular paths. This makes e-stories highly interactive and replayable, as the reader's experiences can vary dramatically from one playing to the next.

Following is a breakdown of some info on the verbs and their components. This will give you an idea of the basic nature of StS. To understand the chart below, bear in mind that each verb, which is a potential action a character can take with respect to another character, can have up to 16 roles (chart range r, below: 1 to >8), each of which can have up to eight options, or potential responses to the action that is taken (chart range o, below: 0 to 8). A role defines the circumstances under which those options are to be considered, and by whom.

For instance, if Fred kisses a woman he just met, Mary, in a crowded bar, her responses might be to slap his face, kiss him back, or turn her head away. If he were to kiss his wife Ann in a more private place, her options might be to complain of a headache or jump his bones. If Fred kisses his two-year-old son Hercules good night, Herc's responses might be to gurgle contentedly, squall for Mommy, or bonk him with a rattle. Each of these different outcomes would be linked to the verb Kiss using different roles; thus the verb Kiss would have three roles; two of those roles would have three options associated with them, and the third role would have two options.

Each verb also has a value called Verb Import (chart range vi, below: 0-9 to 90-100, in increments of ten), which defines the action's relative importance in the story.
Most verbs have one to two roles; most roles have two to three options, and most verbs have a verb import of less than sixty. The median verb import range is between 30 and 39 out of a possible 100 points. The median number of roles per verb is one, and the median number of options per role is one.


I started StS in mid-October of 1996, and turned in version 1.0 on October 1, 1997, so it took me just shy of a year to finish the work. The bulk of the effort of creating an e-story is creating, linking, and defining all the verb scripts.

At the beginning, I was only creating a handful of verbs a week at most. By January, three to four months into the process, my rate of verb creation had crept up to 10-12 per week, and by May my pace was as much as 25 or more verbs in a 40-50 hour work week. (After August my rate of verb generation dropped off, and the main focus of my efforts went into creating more linkages - roles and options - between verbs, in addition to testing.) In other words, I experienced tremendous acceleration as the project went along, as I began to be able to apply major leverage off my earlier efforts.

The first few months my progress was extremely slow. There were three reasons for this. The first reason was that between October and December of 1996, the Erasmatron was going through a lot of changes - major changes - as Chris made substantial improvements to the design. I was pleased with most, even all, of these changes - but the result was that the system was very fragile, and it also required lots of data file structure changes (meaning down-time, time when I sat around doing nothing at all). A lot of my time was spent dealing with system crashes, data corruption, data transition, and debugging efforts. Storybuilders who follow me won't have to contend with this (other than the odd bug that will still occasionally crawl out of the woodwork). Also, I had no testing tools for the first few months, and didn't have the viewing software, the Erasmaganza, for six months. So I was really flying blind for the first several months, and this was certainly a factor in my slow start.

On the other hand, other storybuilders won't have the software's designer instantly available to hand-hold and patiently explain the minutiae of each component of the system, so there is probably a significant tradeoff here; I wouldn't anticipate too much savings in time due to this factor.

Second, I was in a major learning curve at the time. It takes a month or two of fiddling to really start grokking the Etron, and internalizing its grammar. This understanding of the grammatical logic of an e-story is a watershed - a crucial mental "eureka!" that must occur in order to be able to craft a functional e-story. If it doesn't happen, the storybuilder will merely flail about for a while, and the e-story will founder and sink. Writers who have done computer games will almost certainly make this transition more smoothly than writers coming to this interactive form for the first time. It's my hope, though, that creative sorts from _all_ media can be encouraged to surmount this conceptual barrier. E-fiction will be the better for it. The bigger the creative pool, the better.

The third reason progress was slow initially is simply, I believe, the nature of the beast: creating an e-story is not like prose; you can't just jump in and within a week or two have something to look at. A _lot_ of set-up and laying-of-groundwork is necessary before you can see much of anything going on. In my case, it took me till mid-December just to get the background set up - the characters, relationships, things, stages, and so on - and to create my first 70 verbs or so.

Now, StS is a _big_ e-story - I'm a novelist by nature, and also, I spent a lot of time just trying different kinds of things out to see what the Erasmatron could be used to do; e-stories with a smaller cast and tighter focus won't require nearly so much time. In any event, though, if you're coming to this software from another field such as prose or script-writing, you should be prepared for this big lag between investment and the first inklings of a creative payoff.


The major differences from prose are threefold. First, creating an e-story means using your high-school math. Don't worry if math isn't your strong suit. You don't need calculus or geometry or any of that stuff. It's really basic - adding and subtracting some quantities from each other, a little multiplying and dividing. If you can balance your checkbook or bake a cake from scratch using a recipe, you can do the math. Give yourself time to get it.

The tougher task, frankly, is acquiring the mental discipline and logical rigor necessary to keep things straight as you're going along. Since verbs are reusable in any number of different circumstances, you have to remember how you set things up, upstream from where you are, and how to tie it all back in downstream. Novel writers will be used to this, since keeping track of all the elements of a big, complex story, keeping in mind what all the characters are supposed to be doing and making sure your facts are right, all this requires a similar kind of discipline - only an e-story is a lot more rigid. You can perhaps slip an inaccuracy or logical inconsistency past your editor or copyeditor, but the Etron is a lot less forgiving. Fortunately, the Etron has tools that help you comb out these problems. So this second difference from linear narrative forms such as prose is more a difference of degree than kind.

Third, because the e-story by its nature is non-linear, it's not so easy to pick up where you left off, especially after a while. With a novel or a screenplay, picking the threads back up is fairly straightforward; you can edit what you've already done or you can go to the end and create new stuff. But in an e-story, there is no "end," really. What you encounter when you go back in to work on it some more is a big, three-dimensional blob, with verbs hanging all over the place and you somewhere in the middle, rather than you standing on a line from, say, page 1 through page 85. Everything you do will lead to dozens of other things you could work on, and after a couple months' work you'll find yourself surrounded by so many different things that need or want doing that it can be hard to (a) impose any kind of order on the process; and (b) keep track of where you are.

So, here are some tips that hard experience taught me:

1) Before you get up and take anything more than a brief break, use the Verb Description field to take notes for the verb or sequence you're on. Keep a to-do list right there in the e-story.

2) Use the keys in the Keys display, not only to sort by type of verbs, but as a way of marking incompletions. It's really handy to be able to simply identify the verbs you haven't finished with yet, by creating a "construction in progress" key and marking unfinished verbs that way. You can use this in conjunction with the Verb-Description-as-memo-pad trick for maximum efficiency, and occasionally go back through and do cleanup.

3) When you create a verb, set all the General and Keys info that the verb requires RIGHT THEN, before you move down the line and create a bunch more verbs and links. It's a real pain to have to go back later. Include in your verb description things like exactly what it is the subject is doing to the direct object, and if you used any actor-object, thing-object, stage-object, event-object, or whatever (i.e., a person, place, thing, or prior event that is referred to in the verb's script), be sure to describe each object's function. This is so you can easily keep track of this stuff later, and don't have to keep going back and re-figuring it out. This was one of my biggest and most annoying mistakes.

4) For my next e-story, I also plan to try putting in at least a single line of Tinkertoy Text for each viewpoint for each verb, as I create it (there are six viewpoints per verb). That way I won't end up with tons and tons of uncompleted Tinkertoy text, next time around (I don't know if this will work or not, but I'm going to give it a try).

To summarize, don't get so swept away by the fun of creating linkages and scripts (and it _is_ fun!) that you forget to set your verbs up properly as you create them. Invest a little time in this administrative stuff when you first create the verb, and you won't be setting yourself up for a real mess you'll have to spend tons of time cleaning up later.