I argue against the claim -- advocated by Albert Einstein, Bernard Williams, and Ted Sider, among others -- that a description is objective only if it says how the world is in itself. Instead, I argue for the claim -- inspired by comments of Niels Bohr -- that a family of descriptions is objective only if they co-vary with their respective descriptive contexts. Moreover, I claim that "there is a shared objective reality" simply means that it is possible to satisfy this kind of covariance requirement.