I've been reading about how Tor and I2P use cryptographic addresses to send messages to other nodes, but how does this work? I looked through the documentation for each and neither is very clear on this. For two machines to communicate, you ultimately need the IP addresses at some level, right?
EDIT: Removing my second question as it is unrelated to my original. What I'm getting at is: if these anonymous networks are overlays built on top of the existing internet infrastructure, then they must be using TCP/IP protocols for nodes to communicate, right? So do these cryptographic addresses somehow translate to IP addresses? I'm not really looking for a Tor-specific or I2P-specific answer, just a general answer on how this works in overlay networks. I don't understand when the Tor/I2P documentation mentions that nodes communicate without knowledge of the others' IP addresses, only knowledge of their cryptographic address.