The residency system itself is a relic and represents a sort of trapped labor system.
What I can't quite figure out is patient outcomes... morbidity appears to have increased (very slightly) as resident working hours were reduced (possibly more risk from handoffs than tired MDs?) - how does the EU compare in this regard? Somebody must have done a study...
Anyway, some form of on-the-job training seems reasonable here. But the current residency system in the US definitely appears broken, for several reasons.