The traditional approach to fault tolerance for quantum computers is to choose a quantum error-correcting code and design a universal set of fault-tolerant “gadgets” that control the propagation of errors through the circuit. However, a new possibility has emerged in recent years, in which the qubits are encoded in a quantum error-correcting code which is constantly changing. Instead of limiting error propagation, we can handle errors by trying to track down both the position and time at which they originated, and then compensating for any subsequent error propagation when we correct the errors. The whole protocol should then be viewed in a way which puts space and time on a more equal basis. I will present what we know so far about this new viewpoint on fault tolerance.