The solution to the problem of actual causation - i.e. determining what caused an effect in a specific scenario - put forward by Halpern and Pearl recently has received a lot of attention. It forms the basis for many other approaches within the dominant tradition of counterfactual theories of causation. However, their solution runs into a number of difficulties for a certain type of examples exhibiting so-called switching causation and early preemption. We discuss these in the light of the core concept of counterfactual dependency, and offer a comparison with the recent definition of actual causation formulated in CP-logic. We argue both that for this type of examples the CP-logic definition provides better anwers, and that it does more justice to the fundamental intuitions underlying counterfactual dependency.