Abstract: Superconducting quantum computers, once scaled up, could solve problems intractable to even the largest classical supercomputers, but better superconducting qubits are needed before this can occur. Superconducting qubit coherence is currently limited both by cryogenic low-power dielectric loss and by large temporal fluctuations due to strongly-coupled defects. In this talk, I discuss how single qubit performance is generally understood, models and proposed sources of decoherence, and new research disentangling decoherence channels in superconducting circuits, all of which is needed to build better, more scalable superconducting qubits.