Ok, so ICMP is easy to turn off. But why couldn't this technique be done over UDP or TCP? The server could send a UDP packet to a known "fake" address and the client could spoof its source address to penetrate the nat.
Because the ICMP message is a case where NATs are already programmed to allow arbitrarily-sourced traffic to pass. The point is that the trick discovers the client's address, so they can establish direct 2-way comms.
