Ground truthing space syntax

I. Different environments, different models How can we formally model a built environment and capture the fundamental properties relevant to its inhabitants? Space syntax provides a set of rigorously defined techniques for describing spatial configurations like building interiors and urban squares—it appears to be a natural approach to use. Yet whether you chose to use an axial map, a visibility graph, a segment map, a trial run of agents, or any other such computational method in the space syntax toolkit, the question of ground truthing remains. You must make a set of decisions about what environmental features to include in your model, to what level of detail you will describe them, how you will consider variable and changing features, and how you will address error and uncertainty. These questions are critical to the cognitive relevance—not to mention the empirical evaluation—of any resulting model, yet they have received less attention than the mechanics of space syntax. A variety of space syntax techniques are now well-defined in algorithmic terms; still, I can find no guidelines for exactly what environmental features to input into those algorithms. What essentials of the multitudinous world should we record and subsequently abstract into the language of space syntax? Simply put, space syntax analysis begins by dividing open space from closed space. In Central London, the native habitat of space syntax, open space is pavement—that is, roadways, sidewalks, squares—and closed space is buildings. Axes or segments are laid through open space. Isovists expand through open space. Agents move through open space. Consequently, the boundary between open and closed space is key. When modeling the overall structure of a city, particularly one that is dense and built-out, the question of where to draw that boundary isn’t vexing. It may even be avoided, by directly creating an axial or segment map from readilyavailable road-center lines. Building interiors are also straightforward. Not only are architectural plans usually available, but more importantly, walls and other elements of a building’s mass naturally demarcate open space. Smaller than the far-reaching scale of the city and larger than the enclosing scale of the building is the outdoor landscape. These environments are the surroundings that we move through, the vistas that guide and engage us. At this scale, detail is necessary. Buildings lend some definition to the surrounding space, as do planters, bushes and trees, staircases, ramps, benches, streetlamps, signage, and other pieces of street furniture. Certain landscapes may not be as articulated as others. Consider the differences between a parking lot off an exurban strip and a university campus of quadrangles, lawns, and walkways. Different professions work at these three scales: architects craft buildings; urban designers and landscape architects shape the outdoor landscape of city squares, shopping centers, and campuses; and urban planners orchestrate cities and regions. Correspondingly, models will include and be defined by different features depending upon the scale they represent. Applying space syntax to building interiors and city structures seems relatively clear. My interest is in the intermediate scale, and my concern is how we draw the line between open space and closed space at that scale of the outdoor landscape.