Ethernet is really like blind guys coming to an intersection. If nobody is coming they can go on through but if there is another car in the intersection there is a "collision" and everyone is in the penalty box for a while. The adapters time out at different rates until everyone stacked up gets through. In a small network with a few users that is not a problem but the more users trying to get on, the better chance for another collision.
The use of "switches" vs "hubs" helps this problem by holding the bus until the current user is done.
The token is like the train in your example but if nobody is at the trolley stop it goes around pretty fast so you don't wait long. You do get to send your packet when you get the token and it will get through.
