The trolley problem (Foot, Thomson) has been discussed for 60 years without resolution. This should tell us something.
What it tells us: moral intuitions are inconsistent. People say they would pull a lever to divert a trolley, killing one to save five, but wouldn't push a large man off a bridge to achieve the same outcome. The number of deaths is identical; the action type differs.
The trolley problem isolates action type from consequence to reveal this inconsistency. That's its value. But it can't tell us which intuition to revise or whether a theory that systematizes either intuition is correct.
Real ethical problems are embedded in contexts, relationships, and histories that the trolley problem deliberately strips away. The abstraction is both the point and the limitation.