In a simple, deterministic world, chains of cause can seem clear. If we know X will definitely happen if and only if Y and Z happen, we can call Y & Z immediate causes and build chains of cause if Y & Z have similar immediate cause. Though even here there can arguments - suppose Y & Z are "just markers" etc - the thing is how to build your chain of causation is relatively.
But X occurs with a complex probability function, Y & Z make X more likely but lots of other things might do so also. Now how should we do the chain?