Simulation Results

We run the simulation model for four different experimental blueprint settings (p = 0.2, 0.8; q = 0.2, 0.8), i.e. four extreme scenarios in a contingency table of high-merging-low-splitting (p = 0.2) and low-merging-high-splitting (p = 0.8) versus high-progressive-low-conservative merging (q = 0.2) and low-progressive- high-conservative merging (q = 0.8).

We study the effect of these differently structured transformation blueprints in several spatial configurations of regions: a circle, a string, and a cluster (see Fig. 5). We run 50 cases per scenario and, per region, we compute the average complexity of artifacts produced and determine the complexity in the 95 %, 50 % (median) and 5 %-percentile case. In the present work, the complexity of an artifact is the number of tiers in the artifact tree. Figure 6 illustrates how an artifact of advancedness t = 4 is constructed by using transformation that uses two inputs of advancedness 3, each taking two inputs of advancedness 2, etc., until at the but-one-lowest tier each eight artifacts takes two raw resources. In this case the depth of the tree and thereby the complexity is 5.