Well, you can read the rfc's yourself online. Books are so about the money, the slow-witted and yesterday, last millenium even! I short reference volume is handy in case you want to know how many bits in the IPV4 IP Protocol field (9). There are several different directions advanced study can go. Making tcp/ip perform well is one area. Using udp/ip is a whole second area of expertise. Mbone/multicast is another very special and interesting area. Routing protocols are another busy area, usually leading you into the supporting media like Ethernet, bridges and such. IPV6 opens a world of possibilities that have to be run down, to see if they came to life, and how and where the switchover is made in the current Internet. Understanding DNS and the current vulnerability being worked is another layer. How to make them work is one thing, but how to hack them or make them secure is a whole other world. The Transport Area Working Group (tsvwg), Internet Engineering Task Force (IETF) has been making all sorts of complex high reliability and high performance toys. You can learn a lot by writing some of the pieces, or studying code, like tcp_relay.c or the various open source clients, servers. Throwing money at a problem is usually a way to avoid the effort of study. Meditation is also useful, thinking about how something works at the edge of the envelope, anticipating situations. Why does a third tcp/ip download session only slightly improve net usage, and the next does nothing? There is always a lot of room at the top.