We measure the flow kinetics of a polyethylene extruded through the exit of a sapphire capillary tube in order to understand the nature of sharkskin, a surface roughness in the extruded material. Optical velocimetry shows that sharkskin can occur under a variety of polymer/wall boundary conditions; stick, slip, or oscillating stick/slip, demonstrating that the flow boundary condition is not the direct cause of sharkskin. Downstream of the exit, high-speed video microscopy reveals two distinct material failures during each sharkskin cycle, the first is cohesive and splits the material into two regions, the second one occurs at the polymer-wall interface. Upon modification of the surface with a polymer processing additive (PPA), we confirm strong slip at the wall and a suppression of sharkskin, but find that sharkskin does return at sufficiently high flow rates. The extensional strain rate at the onset of shark-skin is significantly higher in the case with PPA than that without. We then empirically define a "reconfiguration rate" and find it is comparable at the onset of sharkskin for the two surface conditions, We use data in the literature to;how that the reconfiguration rate also predicts the relationship observed between the onset of sharkskin and the capillary radius.