The minimum variance method (MVA) has been used for over 3 decades to analyze directional discontinuities in solar wind magnetic field data. We recount how MVA was formulated and show that it does not do what many expect. We demonstrate a robust alternative for distinguishing tangential discontinuities from rotational discontinuities (RDs). We also stress that averaging magnetic field data to time resolutions of a minute or more (common with National Space Science Data Center data sets) obliterates RDs.