How confident do you need to be in order to know? In this talk, I'll suggest that the answer is: not very confident at all. More specifically, I'll defend a picture on which you know the answer to a question just in case you make a good guess about it and the guess you make is non-luckily correct. Combined with a popular story about guessing (Holguín (2022), Dorst & Mandelkern (forthcoming)), this picture implies that knowing the answer to a question is compatible with assigning it any credence other than 0. I'll argue that this view offers the best explanation of a range of linguistic data and that some of the more obvious challenges it faces can be dealt with once we get clear on the norms governing assertion. I'll conclude by saying a little about what it says about higher-order knowledge.
