Basically, the paths work like this:
The camera will be set up at each vertex on the camera path. So, if you have 23 edges in the path curve, there will be 24 camera positions (one per vertex).
If the target path has the same number of vertices as the camera path, the target will be at each vertex. Otherwise, the target point will be interpolated along the path so that there are the same number of target points as camera points. For example, if the target path had only 3 edges (4 vertices), the whole path length would be divided mathematically into 23 equal segment lengths, and the target points would be set at equal distances along the 3 edges.
Hope that makes sense...