Why 'no normal' could be the best kind of 'new normal' for p