As Philip says the original design of BGP assumed that external neighbors would be directly connected. My understanding of the logic for setting TTL to 1 for EBGP neighbors is that it was done as a way of reducing risk. If you are directly connected to the EBGP peer then TTL of 1 was enough for the traffic from the peer to get to you. And there was an assumption that you could trust this traffic since it came directly to you from the peer. But what about the situation where the link connecting you to that peer stopped working? What if that peer forwarded the BGP packets to some one else, who forwarded them to some one else, who forwarded them to you? Should you still trust that traffic? Setting TTL to 1 is a way to say that if it did not come directly to me then I do not trust it and will not use it.
Obviously in todays networks there are sometimes situations where you are not directly connected to your external peers. And in those situations you need to specify the multihop parameter.
HTH
Rick
HTH
Rick
