The metrics of street network connectivity: their inconsistencies

The concept of street connectivity has been gaining increasing appeal among researchers, planners, and planning authorities. In response, many connectivity metrics have been developed in an effort to understand better street network connectivity. This paper will study the effectiveness and consistency of three mainstream metrics - the Connectivity Index, Intersection Density, and Street Density - with respect to differences in study area and geometry. While these metrics are intended to be applied incrementally, this paper reveals that the metrics often fail to do this successfully. By controlling for many variables - including block size, block geometry, right-of-way size, network size, and network geometry - actual behaviors of these metrics deviate substantially from their intended behaviors. The metrics are non-linear functions of both study area and geometry and are ultimately inconsistent and unpredictable. In other words, each metric will yield inconsistent readings based upon the amount of area studied or the arrangement of the study boundary drawn. This has two major consequences: (1) the metrics will not produce the results desired as they are applied to incremental development; and (2) the metrics can be easily gamed by a developer privy to the information found within this paper. Neither of these outcomes is desirable in helping to better understand and potentially regulate street connectivity.