Introduced in the early 2000s by computer scientists Daniel Spielman and Shang-Hua Teng, this is a general paradigm to bridge the gap between the average-case and worst-case performance of algorithms.