Monday, April 16, 2012

Moneyball for Cities

Alert: nerdy, wonky stuff herein. But, I can't resist. I find this stuff fascinating. Below are a series of maps creating using Space Syntax. Space syntax is a theory of urban design and planning sprouting via the convergence of mathematics, systemics, and cities. In many ways, it is urban planning coming full circle from Jane Jacobs, who essentially spawned complexity thinking as she introduced concepts of self-organization in relation to cities.

What space syntax does (and the variety of programs that operate it) is measure the complexity of the network. As the theory goes there is a direct relationship between the objective mathematical measure of complexity, the integration of networks (red is high, blue is low) with a variety of implications on the city including: resilience over time, pedestrian activity, land value, crime or lack thereof. In many ways it is an objective measure of predictability. Critical in both planning and real estate development. Vary the network and you will inevitably vary the system within and all that implies including land value, density, land use arrangement, and so on.

What is most interesting about this stuff however is that while it is entirely objective and quantitative, there is an implied qualitative component to it as well. As I often say, integration begets accommodation. Integration is the connection. Accommodation is what makes the place nicer. As Gil Penalosa presented to Dallas City Council, "it doesn't matter if it is nice (yet). All that matters is the integration of networks." By saying this, Penalosa implicitly understands that the connection is what drives value and the accommodation, or qualitative improvement of the place, is henceforth inevitable. An outgrowth of the underlying system.

Integrate networks and you get investment. Disintegrate networks and you get disinvestment. In terms of moneyball and mapping, this allows us to point out areas that may have too much investment and too much density based on the value of their spatial integration. These would be places where you either 1) wouldn't want to put your money into, or 2) can identify for network improvement to bring the spatial integration value UP to the value of the investment. Think of: Victory and LoMac, here. Lots of density (in terms of height and floor area), very little urbanism (as in spatial integration).

On the flip side, we can see areas of high integration and low value/density. When there are disconnects like these we can see opportunity areas. But first, we have to ask ourselves, what is wrong with those areas. And that is where more in depth local analysis is required. We'll get to a few examples of that below:

Space Syntax model of all of London.

If you squint your eyes enough, in abstract the city begins to look like this (above). There are two levels of hierarchy at work. What we might call the macro- and micro-hierarchies. Macros compete with macros. In this case, London vs. Dallas. The micro are competing amongst themselves within the system, ordering themselves from top to bottom. The top of the food chain being the most interconnected and therefore the most value.

Above is downtown Plano and vicinity. I also included the plan for the Richardson TOD plan if it were to be built out in full as planned just for funzies. As you'll notice, this is a bit less ordered than the London map. There are two primary and disparate reasons for this. One, London is a much older city that has had a thousand years to self-organize. Meanwhile, Plano is only a few decades old. Secondly, and similarly important, this is looking at a micro-scale whereas the map of London viewed the macro containing hundreds of these interdependent micro-systems. The order is easier to see at the macro-scale.

What is interesting about this Plano map is the "hottest" or most spatially integrated area is downtown Plano. Some of the least? Collin Creek Mall and Collin County Community College, which is arranged much like a conventional shopping mall.

The Plano map also includes DART in the network model, however it doesn't fully weigh its value unless you map the entire system (because it only connects regionally). The model above only sees the linkage from the Bush Tollway stop to downtown Plano to Parker station. And even then, the DART red line is one of the "hottest" or largest drivers of value within this map just behind 15th and K.

Downtown Richardson. I haven't progressed as far as with Plano, but the degree of disorder is even more prevalent. I suspect this is related to the current shape of the respective downtowns, Plano vs Richardson.

Here is downtown Dallas. Predictably, the Main Street core area glows the brightest. Two things also to notice: how surprising disconnected uptown is and how interconnected East Dallas is. There are a couple of explanations for this. One reason is the map I worked from. It is wider east-west than it is tall, N-S. This will invariably give more weight in the direction that the model thinks the city grew. Since it doesnt see the rest of Oak Lawn and North Dallas, this current model makes uptown look less interconnected than it really is.

However, there is some truth to that. Uptown, aside from the CityPlace/West Village area and to a lesser extent State-Thomas isn't that well integrated into its surroundings. It puts undo pressure on McKinney with little other choice of route. I interpret this to mean that the density and investment uptown has experienced is more due to its place within the macro-system than its particular spatial integration. Meaning, its current value is due to its proximity to downtown and its location between downtown and North Dallas (the favored quarter in Leinberger's terminology).

But, Near East Dallas shares a similar proximity to downtown. And it is better interconnected. While there has been some haphazard investment and land use intensification the Live Oak and Ross corridors remain tremendously underdeveloped given their potential. Unlike many parts of the city, this area seems to be the least encumbered by disconnecting elements (rivers, lakes, highways, railroads). In fact, it is one large contiguous area begging for investment between downtown and White Rock Lake.

Areas like these that the model sees as high value, but you know in fact to not live up to that potential are places ripe for investment. Except that investment typically has to be a concerted effort between the public and private sides. The private side is risk averse. It doesn't want to pioneer into an area if it won't be supported by the necessary public investment. The disconnection is usually from the public side. Something is wrong with the network. In the case of East Dallas, this is an infrastructural problem. A public realm that is not conducive to pedestrian activity, lack of quality choices for multi-modal transportation, as well as fragmented parcelization.

The above is the beginnings of a map of the entire city. As you can see there are no hot spots. Part of this is due to the incomplete nature of the map. It isn't seeing the full gradient from high to low. But this also pinpoints the issue plaguing downtown Dallas. Instead of red areas (high value, high density), the demand via spatial integration value is smeared across the city as no place is truly, highly interconnected and integrated, making density desirable, valuable, and predictably successful.

And above is downtown Fort Worth, which I thought I'd begin to put together for FortWorthology's interest. This still has quite a bit of work to do as well. Not just adding more of the grid in surrounding areas, but also de-linking the highways. Without doing so, it sees overpasses as interconnected when the roads are actually flying over each other. And by doing so, by making the "flow" of traffic less stop and start, we devalue the individual intersection, convergence points, and diminish the overall network of intersections. And therein lies the value of the city. Meeting points. Not bypasses.