Something has been bothering me about the Many Worlds Interpretation. Proponents of it (e.g. Sean Carroll) often claim that it does away with the observer, or at least the paradox-inducing status the observer has with Copenhagen. However, my impression is that Many Worlds instead displaces the interpretation of probability, and gives the observer, if anything, a more mysterious role than in Copenhagen. Many Worlds seems to take the probabilism Copenhagen ascribes to the wave function and transposes it onto the observer. In Many Worlds, every possibility of the wave function is realized, so the probability becomes about the observer, and which "world" the observer ends up in. Is this a correct way of thinking about Many Worlds? This doesn't seem to me to do away with the observer, or make its status any less confused than Copenhagen does.