Why does CSMA/CD require a minimum frame size?

Question

I read this somewhere - An Ethernet frame has a minimum size of 64 bytes because anything that is shorter than the 64 byte minimum is interpreted by receiving station as a frame resulting from a collision. This allows the receiving station to inform the transmitting station that a collision has resulted, hence it backs off and begins a re-transmission at an arbitrary time. I am able to comprehend the overall idea. Could you please explain it ?

Solution

The minimum frame size on Ethernet is so that by the time the beginning of a frame gets all the way across a maximum-width network, if a collision is detected there on the far side of the network, there's still enough time for the jam signal (collision detection notification) to make it all the way back across the network while the transmitter is still transmitting, so that the transmitter will receive the jam signal and know it got collided with so it will have to retransmit.

Anything below 64 bytes is just considered a runt (a buggy transmission) and ignored. Runts are not collisions and do not trigger retransmissions.

But this is all pretty much moot as nobody runs hubs anymore, since Gigabit Ethernet requires switches. So today's networks are always full duplex and CSMA/CD is a thing of the past.