Overdetermining causation and prevention

Overdetermination seems to work differently for prevention and positive causation.

Suppose Timmy the turtle is wearing steel armor over his shell, because it looks cool. Alice shoots an arrow at Timmy’s back from the side, which glances off the armor. Let us assume that arrows shot at the back of an unarmored turtle from the side also harmlessly glance off the shell. Then we have two questions:

1. Did the armor prevent Timmy’s death?

2. Did the armor cause the arrow to glance off?

My intuition is that the answers are “no” and “yes”, respectively. You only count as preventing death if you are stopping something lethal. But I assumed that an arrow aimed at the back of an ordinary turtle from the side glances off the shell and is not lethal. On the other hand, it is clear that the arrow glanced off the armor, not the shell, and so it was the armor that redirected the flight.

Why the difference?

I think it may be this. There is a particular token flight-redirection event f₀ that the the armor caused. When you cause a token of a type, you automatically count as having cause an event of that type. So by causing f₀, the armor caused a flight-redirection event, a glancing-off.

However, it does not seem right to say that in preventing an event, one is causing a token non-occurrence. There would be too many non-occurrences in the ontology! Prevention is prevention of a type.