When disruption theory is useful (and when it isn't)