Page 5 is a third party negotiated connection. Page 4 is an attempt at direct connection. So in the third party negotiated connection the tables get built differently than when the two clients try to hit each other directly. That''s the entire point of page 5''s existance, that there is a difference in the NAT tables.

AIM file transfer may use TCP connections. TCP through two-way NAT is... not really feasible. (There are people who try, but it''s all a bunch of guesswork that often doesn''t work)

AIM uses TCP because that''s less likely to be blocked by corporate firewalls. It tries first on its own port, and then on a bunch of known ports (80, 53, 443, etc) to see if it can get through to the greater net somehow. UDP is usually blocked by corporate IT staff, because it''s mostly used for games.

tieTYT is right, its not consistant..

on page 3 it says:

"NAT keeps the original source address and adds it to an in-memory table (the NAT table), along with the new source-address information."

Which it sticks to until page 5 where they made the mistake...

The public IP address in the NAT table should idd be the IP of the router... Although it does seem a bit pointless because that information is not really needed, the NAT already knows its own IP address.