Predicates of personal taste (PPT), which include words like 'tasty,' 'fun,' and 'cool,' have been at the centre of a set of lively, interdisciplinary debates for the past 20 years. These debates cover an impressive, and sometimes daunting, range of issues, including: disagreement, meaning, context-sensitivity, subjectivity and objectivity, truth, aesthetic and gustatory taste, evaluation, speech acts, and so on. Over the past 20 years, philosophers and linguists have engaged with these issues in developing many interestingly different accounts of what PPT mean and how we should evaluate assertions involving PPT. However, there is a foundational, methodological question that participants in the PPT debates have yet to adequately address: what sorts of evidence should be used in evaluating an account of PPT? In recent years, certain philosophers and linguists have pointed out that since accounts of PPT turn on empirical hypotheses about natural language, a proper evaluation of such an account must involve clearly articulated empirical diagnostics. In this talk, I will explore one of these diagnostics, which involves retraction. After setting out the major analyses of PPT, I'll consider the most influential explanation of why retraction data matter in the PPT debates, which comes from John MacFarlane. I'll identify two significant problems with MacFarlane's explanation and offer what I take to be a better explanation. I'll then draw on this explanation in reframing the significance of extant empirical data on PPT and retraction.