I'm implementing a DHT based on Kademlia following this paper.
The protocol described in this paper uses nonces to protect against man-in-the-middle attacks. The disadvantage is that latency gets doubled by effectively introducing a second round trip time.
Now my question is, are man-in-the-middle attacks against p2p systems where each peer knows quite some peers and no long-term connections realistic? Was the nonce exchange added for purely academic purposes to propose a "fully secure system", or does it actually prevent a real risk?