There has been in recent years a huge surge of interest in the so-called extended Wigner’s friend scenario (EWFS). In short, a series of theorems (with some variation in detail) puts pressure on the ability of different agents in the scenario to account for each of the others’ measured outcomes: the outcomes cannot be assigned single well-defined values while also satisfying other reasonable physical assumptions. These theorems have been interpreted as showing that there can be no absolute, third-person, ‘God’s eye’ description of our reality. The focus of this paper is the strongest of these no-go theorems, the ‘local friendliness’ theorem of Bong et al. (2020, Nature Physics, 16, 1199–1205), which gives earnest consideration to the possibility of a measurement that unitarily reverses an entire lab system, including a conscious agent, thereby erasing the agent’s memory. The purpose of this paper is to begin the philosophical conversation regarding key questions concerning this process: Are the events in the lab merely ‘erased’, or do they in some sense not exist at all? What would it be like to be unitarily reversed? Should an agent care about any experiences they have inside the lab before they are reversed? This analysis employs a parallel case of memory erasure, to which this case can be contrasted, arising in the context of drug-induced amnesia as a result of administering anaesthesia during medical procedures (Carbonell, 2014, Bioethics, 28(5), 245–254). I argue that the consequences of unitarily reversing an agent are much more dramatic than simply memory erasure—the set of events themselves, and the personal timeline of the agent, leave no record at all inside or outside the lab. I consider the ramifications of this for the picture of reality that arises from the EWFS.