In one section of my thesis, I am reviewing the literature of computational urban modeling approaches starting from 1900. I mainly compare them from the way they are using computational capacities offered by different stages of computational technologies starting by mainframe simulators and then to microsimulations and later networks of communications and distributed models and more recently by what we call urban data streams.
As it shown in the following figure my argument is that all of these computational capacities as well as those modeling approaches, which are based on each of these capacities are co-existing at the moment.
As a result it is easy to find recent papers or projects that are based on methodologies of modeling, which are 100 years or even more old. For example, most of the recent projects under the umbrella of “city science” are following the ideas of Newtonian Mechanics. As a result even though they are data driven or fully computational, the mind set behind them is analytical and rooted in 17th century. They are only opportunist regarding the availability of digital data and numerical methods. If you are interested, a more detailed description of these kinds of discussions can be found in my recent text here. Therefore, in my opinion it is very important to decouple modeling methodologies from time. But nevertheless it is interesting to see that modeling approaches have their own life cycles or in other words to see how they were born one day and how they reach to their peaks and become popular and when they start to decline. One interesting source for investigation of this idea is Google n-gram viewer, which is a collection of lots of n-grams from more than 5.2 million digitized books. It gives you a simple API and you can make lots of queries. Using a simple code I collected all the relative counts of several key words (each referring to one famous modeling framework or concept) and in order to compare their life cycles to each other, I normalized all the values individually between zero and one. Next figure shows the relative importance of each modeling concept from 1900 to 2008 with 3 years smoothing of the original count-values. It should be noted that here we are only interested in the shapes of these graphs, while their exact values, extracted from Google (the percentage of their counts in relations to all the other possible n-grams) are not comparable as they might be in different scales. Further, I sorted them in a way that similar behaviors are closer to each other. The following figure shows the initial results. As you can see there are categories of modeling approaches that are living together as they are following the same mind sets. For example, those models or concepts such self organization, fractals and chaos theory that are in the domain of complexity theories have the same birth and death more or less.