The comparative method is a relatively well-defined tool that has been employed successfully in the classification of languages for two centuries. In recent years, there have been several proposals about the classification of the Austronesian languages that violate basic principles of method. Because some of these have been advanced by scholars who are well established in other branches of linguistics, they have acquired an influence that is out of proportion to their scientific merit. This paper addresses three of these proposals: the Austronesian-Ongan hypothesis of Juliette Blevins, the Quechua-Austronesian hypothesis of E. M. Kempler-Cohen, and the higher phylogeny of Austronesian and the position of Tai-Kadai by Laurent Sagart. By carefully delimiting the analytic operations that belong to the comparative method and those that do not, it is shown that each of these scholars makes use of illicit operations to justify inferences about the classification of Austronesian languages, whether this involves claims about relationships that are external to the family or internal to it.