We establish three new fundamental limits regarding our knowledge of the quantum state of the universe. Suppose the universal quantum state is a typical unit vector in a high-dimensional subspace $H_{0}$ of Hilbert space $H$, such as the low-entropy subspace defined by the Past Hypothesis. We show that:

(1) Any particular observation is incapable of identifying the universal state vector or substantially reducing the set of possibilities.

(2) For any reasonably probable measurement outcome and for most pairs of vectors in $H_{0}$, that outcome will not appreciably favor one vector over the other.

(3) Bayesian updating on any measurement result, unless it is extraordinarily improbable, has a negligible effect on the initial uniform probability distribution over the states in $H_{0}$.

These findings represent the most stringent epistemic constraints known for a quantum universe and are derived from a typicality theorem in quantum statistical mechanics. Importantly, these results apply to any quantum system (regardless whether it is the universal state) if its wave function at some time is typical in a subspace of high dimension. We close by considering how theoretical considerations beyond empirical evidence might inform our understanding of this fact and our knowledge of the universal quantum state.

Joint work with Roderich Tumulka. Paper version: https://arxiv.org/abs/2410.16860v2