Let's All Live In Jay's House

I’m insanely jealous. It turns out that Jay Cross lives in a house built by the great architect Christopher Alexander. I’m a huge fan of Alexander’s work. Furthermore, I think anyone who does instructional design should read Alexander’s book. No, I don’t mean The Nature of Order, which is the book that Jay is apparently reading now; I haven’t read that one yet (though I intend to). I mean A Pattern Language: Towns, Buildings, Construction.

To begin with, if ever there was a book that prefigured hypertext, this is it. Written in 1977, it is a honeycomb of short, cross-referenced entries. You can’t really read it cover-to-cover; instead, you browse. You scan. You jump around. You…surf.

But the real value is in the way Alexander changes your thinking about design. In stark contrast to the whole idea of architect as designer and architecture as engineering, Alexander sees the architect as an ecologist and architectural design as the product of natural selection. He points to the distinctive qualities of old buildings throughout the world that have stood for hundreds or thousands of years while other buildings around them were torn down. What qualities do they have in common? What features can we deduce as natural, as creating a pleasurable and healthy living environment? What remains useful to people generation after generation? And what remains beautiful? What are the patterns that emerge?

Then notice how these things fit together. Certain patterns will fit together nicely while others will not. There is a grammar for stringing together architectural elements, Alexander argues. A natural language. A pattern language. If you surf his book and see the pictures, I’ll bet money that you’ll become convinced he’s right. I did.

We should be thinking about our instructional design, our information architecture, and our software user interfaces in the same ways. Despite the newness of computers, there are going to emerge more or less natural ways of interacting with it. The only way to find out what the more natural ways are is to let people build their own systems, move the furniture around, tear things down, and start over again. That’s hard to do with software, though HTML makes it somewhat easier by lowering the knowledge requirements. (Hypercard was probably the first step in this direction.) Likewise, we need course management systems that are highly configurable, allowing instructors to experiment with plugging in new pieces and arranging them in different ways. And above all, we need a pattern language for learning objects.

Perhaps this sounds like a stretch. If so, then it’s not the furthest stretch that anyone has made with Alexander’s work. Software engineers have taken to pattern languages big-time.