From A New Kind of Building?, Maurice Martel “was interested in generating architectural structures subject to spatial constraints (such as a given area in which they need to fit).” So he “settled on a project in which he would run 2D cellular automata on irregular grids determined by arbitrary polygons.” His ultimate goal was to design an actual 3D structure.

Using Mathematica he was able to go from this:

To this:

I think what’s interesting is just how benign the result of the Mathematica-assisted design process. Give certain architects free reign, no restrictions and no context and you could very well end up with more folly than form (or function.)

[…] an illustrated manual on planning and designing for “good” density, and a catalog of more than 250 diverse neighborhoods across the country, noting density in housing units per acre for each site. Four photographs of each location are included—close-up, context, neighborhood, and plan views—to provide an impartial and comparative view of the many ways to design neighborhoods.

The website provides access too all those book images in a simple searchable database. Pick your region, your density and your setting and see what the density looks like. For instance, show me something from the West, low density and I don’t care about the setting. First two results are Beverly Hills and Hollister, CA. Compare and contrast Beverly Hills & Hollister with only 0.2 units/acre to San Francisco’s 222 units/acre. That’s a pretty profound demonstration of “density”:

So what, you have some interesting pictures to show folks. How do they help?

[…] many people have difficulty estimating density from visual cues or distinguishing quantitative (measured) and qualitative (perceived) density. We tend to overestimate the density of monotonous, amenity-poor developments and underestimate the density of well-designed, attractive projects, thereby reinforcing the negative stereotypes.

Some reality to counter the misperceptions in other words.

(As an aside: I’m very bullish on the use of visual explanations to educate these days and I’m not talking about the run of the mill business charts and junk like that. We have smart people putting together projects like Gapminder and Swivel, pushing the notion that data is best when it’s actually used to educate & debunk. Our tools for turning all this great data into visual answers are getting better too. Mathematica 6’s “integrated data sources” enable all sorts of crazy math mashups and symbolic programming visualizations. Might have to redefine what “math” is after Mathematica 6 gets done with it. Shame there’s only 24 hours in a day.)