I have a question about "full duplex mode." When I was first introduced to this concept, it was explained to me in the following manner. On a dedicated circuit or network segment communication can take place in both directions simultaneously. Both nodes have transceivers with differentiator circuitry. They transmit over the full bandwidth of their shared media. Some of the bits collide while others pass. The resultant scalar wave front, which is composed of reflected and passed bits, returns/continues to the transceivers. The differentiation circuitry compares what was sent with what was received to determine what was actually just received from the other node. If my understanding is correct, it is awe inspiring to realize the speed at which this takes place! It means that the two nodes are communicating by sending a single packet across the media at a time and waiting for the partially reflected/partially transmitted packet to return/arrive before sending the next packet! Is that essentially correct? If it is, I can understand the physics of how this could be possible (i.e. voltage signals colliding and being reflected). As amazing as that sounds, especially considering the synchronization required, I don't see how it could work over fiber. According to our book, 10 Gb Ethernet operates only over fiber! Perhaps I need a refresher on basic physics, but I thought that when light collided with light it created an interference pattern, not a reflection. I would think that even in the somewhat constrained environment of single mode fiber, the collision of light would create a dispersion pattern that wouldn't be decipherable at either transceiver. I can see how it could work with electricity, but not with light. The only way that I could see for full duplex to work over fiber is if the signals are sharing the bandwidth (i.e. multimode fiber) instead of colliding. Am I wrong?