This also makes it harder for an adversary to ban all I2P routers.
A problem Tor has is that someone observing network traffic can make their own connections to the IPs and ports they see, and start speaking the OR protocol to it, and if it’s a Tor relay it will respond accordingly. The Great Firewall has been observed doing this. This is worked around by bridges using pluggable transports that only respond if sent a message that contains a shared secret obtained out of band (such as from the bridgedb). Does I2P do anything to prevent probing like this?