The three-reference frame example is the easiest, because you can start with a frame where two events, A and B, happen simulateously. A reference frame (say, a spaceship), flying along a line in the A to B direction will observe A happen, then B. A ship flying the opposite direction will experience the opposite, B then A.
So whose observations were correct? All them are perfectly valid. The problem is if we allow A to cause B, in which case the B then A frame has the effect happen before the cause.