Standard benchmarking provides the run times for given programs on given machines, but fails to provide insight as to why those results were