The concept of extravagance, i.e. the maxim "talk in such a way that you are noticed" (Keller 1994, Haspelmath 1999), has recently been invoked to explain multiple language chage phenomena, including e.g. the development of the English progressive (Petré 2017) or German quantifier/degree-modifier constructions (Neels et al. 2023). In this talk, I give a brief survey of corpus-based research on the role of extravagance in language dynamics and change. I will argue that while this line of research has contributed significantly to our understanding of individual language change phenomena, it is always in danger of almost inescapable circularity, and should therefore be complemented with other approaches including experimental studies. I will present a first attempt to assess the role of extravagance in language change with the help of artificial language learning.