According to epistemic orthodoxy, receiving a defeator can undermine the justification to believe a hypothesis provided by some piece of evidence. For example, suppose that a reliable testifier tells you that p so you come to have a justified belief that p. According to the orthodoxy, this justification could be undermined by contrary testimony or evidence that the speaker is in fact just guessing. Further, defeat is standardly used to answer Harman’s dogmatism puzzle: if one initially knows that p, and then gains misleading evidence that not-p, why isn’t one entitled simply to dismiss that evidence since one knows it is misleading? According to the “defeat solution”, the misleading evidence undermines one’s knowledge and this is why one isn’t entitled to just dismiss it. Nonetheless, this epistemic orthodoxy has come under attack recently. Lasonen-Aarnio has argued that this notion of defeat is not in fact motivated by Harman’s dogmatism puzzle. Others have worried that defeat is incompatible with a Bayesian framework. In this paper, I defend the notion of defeat from these recent attacks.