Observation of continental drift sparked a paradigm shift in our understanding of Earth's interior and surface activity: the lithosphere is tessellated by piecewise-rigid tectonic plates that embed the continental lithosphere, and it is the plate generation/subduction process that induces the observed drift. Underlying this first-order kinetic description lies a more complicated picture: increasing data suggest just 85% of Earth's surface appears to be rigid and the remainder is comprised of diffuse regions of high deformation often near plate boundaries, as well as complicated plate geometries and poorly defined boundaries that complete enclosing perimeters. Unsurprisingly, for any geodynamic study that involves dynamic plate generation, having a consistent way of identifying robust candidate plates is paramount. Here, we present a systematic method that utilizes the Random Walker algorithm to identify plate boundaries (in both real and model data sets) by reformulating the problem of identifying plate boundaries in terms of the combinatorial Dirichlet problem for a discrete connected graph, yielding a solution accompanied by a probability array that quantifies the algorithm's confidence in its reported solution. I shall discuss the method development and early results obtained from applying the method to both GPS surface velocity data and mantle convection model output.