The Tool Shapes the Hand of its User

March 22nd, 2005

Alan Kay once told me an adage: "When you’ve got a big enough hammer, everything starts to look like a nail." That was 25 years ago, and I am glad that I took it to heart, for it has served me well. But it seems as if everywhere I look, I see people who don’t appreciate just how important that adage is. So many designers seem to be children of their tools. This is especially true with the less technical designers.

Let’s start with HTML, the simplest programming "tool" of all. It really doesn’t do much: just present text and images, along with the provision for hyperlinks. HTML wasn’t designed to do anything more. We have a generation of web designers who live, breath, and think in HTML. Their imagination of what can be done on the web is shaped by HTML. And that holds us back.

Next up the evolutionary scale is HyperCard -- although HyperCard predates HTML. This is a now-forgotten application that Apple Computer released with far too much hype in the late 80s. It was basically HTML for the Mac, with one gigantic addition: HyperTalk, a wonderful little programming language. HyperCard inspired an entire generation of creative people to build all manner of HyperCard applications for the Macintosh. Sadly, few of them bothered to learn HyperTalk -- had they done so, I think that HyperCard might have evolved into something permanent. As it happened, though, HyperCard lives on today in the work of the Miller brothers, who got their start writing applications for HyperCard. They wrote a number of graphic adventures for children using this tool, and then they jumped over to the Windows environment and got somebody to write a HyperCard clone for them. Of course, this was a much soup-up HyperCard clone, and the result was a game called
Myst, which became a huge hit. Even today, the descendents of Myst still occupy store shelves, and they still operate with pretty much the same architecture of the original games the Miller brothers designed for HyperCard. 15 years later, their designs are still basically HyperCard.

An even more striking example is provided by the "interactive fiction" (IF) movement. Back in the 80s, we called these things
text adventures, but nowadays we use the more flattering term interactive fiction. When text adventures went the way of the dinosaurs, a number of determined enthusiasts hung on, creating their own engines for authoring and playing text adventures. Engines like this are not hard to build -- I recall one written in BASIC for the Atari 800 that was available within a few months of the first release of the machine. There are now a number of competing standards in the IF community.

As befits its ancient heritage, the architecture of IF products is pretty simple. There’s a simple parser capable of understanding a few dozen basic verbs, a few hundred objects, and synonyms for many of these words. There’s an internal map that keeps track of the player’s position. There are connections between the various rooms in the map. There are flags that can be set or reset by the player’s actions. And that’s about it. The modern engines have additional bells and whistles, to be sure, but this is the basic architecture of IF systems.

The appeal of IF derives partly from the ease of learning the basics of IF. The beginner can easily produce a working system in no time, and from there it’s not difficult to master the remaining technical details. This has encouraged a large and vibrant community of IF enthusiasts to converge (facilitated by the Internet).

What’s so striking about this community is the energy with which it has exploited the limited capabilities of the basic IF architecture to produce some impressive works. These people are like kids who start with nothing more than a skateboard and end up developing an impressive array of gymnastic tricks with such a simple device.

The downside is that they’re still messing around with skateboards while the rest of the world is driving rocket-propelled flying Porsches. Why?

I can certainly appreciate the obvious argument that the IF medium permits a clean emphasis on pure text, undistracted by vulgar graphics. I also endorse the argument that IF authorship is accessible to anybody, while creation of a modern game requires millions of dollars. More on this later.

But the point I’d like to hammer at is the way that IF people seem to have allowed their worldview to be defined by the limitations of that medium. Many years ago, some clod at a major IF website published a scathing review of my first version of the Erasmatron. Now, that version was in fact full of problems, and there were plenty of perfectly good scathing reviews that could have been written, but this chap seemed obsessed with the absence of puzzles in the Erasmatron. After all, puzzles are central to most IF, and since the Erasmatron didn’t offer puzzles, it was clearly substandard. And that attitude continues in the IF community. Their design worldview is shaped by their medium.

This is sad, because the basic IF architecture is so limited. It has no general provision for human feelings. (It’s possible for an author to cobble together a kinda-sorta human feelings facsimile, but it’s woefully limited.) It has no general recollection of previous events. It has no understanding of the notion of privileged information, and its NPCs are cardboard cutouts. The IF people seem to be aware of these limitations, and apparently some heroic efforts have been made in this direction, but they’re tack-ons, not fundamental changes.

What we all need is a powerful tool that permits us to capture narrative structure while sill being accessible to sub-superhumans. The Erasmatron is not that tool; it demands too much of the storybuilder. I’m not certain that such a tool can be created.