With further scaling, the error rates on "logical" qubits can hopefully be driven down sufficiently that we can run arbitrary-length quantum computations on a quantum computer, instead of ones with a limited number of steps.

→ View original post on X — @jeffdean