Right away I must state that I did not read "The TCP/IP Guide" (TTG) cover-to-cover. I doubt anyone will, which raises interesting issues. This review is based on the sections I did read and my comparisons with other protocol books.
Protocol books should be divided into two eras. The first is the "Stevens era," meaning those written around the time Richard Stevens' "TCP/IP Illustrated, Vol 1: The Protocols" was published. For six years (1994-2000) Stevens' book was clearly the best protocol book, and it taught TCP/IP to legions of networking pros. The second is the "modern era," beginning in 2000 and continuing to today. TTG fits in this group.
I question the approach taken by TTG. The book contains extremely basic information (what is networking, why use layers, what is a protocol, etc.) and extremely obscure information (PPP Link Control Protocol Frame Types and Fields, SNMPv2 PDU Error Status Field Values, Interpretation of Standard Telnet NVT ASCII Control Codes, etc.). If TTG were an introductory book, it wouldn't need the obscure material. If TTG were a reference, it wouldn't need the introductory material. I think beginners would be scared by this book, although the tone and explanations are suitable for those with a real dedication to learning. (Note: TTG features 88 chapters, 14 of those are 8 pages or less.)
For beginners, a better introduction is Jeanna Matthews' "Computer Networking: Internet Protocols in Action." Matthews' book is shorter (273 pages), more direct, and packet-example-based, meaning it ships with a CD-ROM of traces that readers can analyze as they read Matthews' commentary. The lack of examinations of packet traces is one of my biggest problems with TTG. If TTG aims to be comprehensive, it should have looked at real traffic using Ethereal/Wireshark instead of staying at the specification level.
For intermediate readers, Eric Hall's "Internet Core Protocols: The Definitive Guide" is a great look at the building blocks of networking, albeit without IPv6 or application protocols. Hall's book is also packet-oriented, with examples for each concept.
For expert readers, "Troubleshooting Campus Networks" by Priscilla Oppenheimer and Joseph Bardwell is outstanding. J. Scott Haugdahl's "Network Analysis and Troubleshooting" and Kevin Burns' "TCP/IP Analysis and Troubleshooting Toolkit" are also excellent. All three show packets.
Those with some networking experience looking for a thorough (but not packet-example-based) examination should definitely read Adrian Farrel's "The Internet and Its Protocols: A Comparative Approach." Farrel demonstrates deep subject matter expertise by showing similarities and differences between protocols. He also covers protocols like MPLS and SCTP that are ignored by TTG.
So what could Kozierok's TTG have done differently? First, the book should be split into three volumes. Volume 1 should cover all of the core protocols (ARP/RARP, IP, IPv6, ICMP, ICMPv6, UDP, TCP). Those of us already familiar with those protocols or already in possession of other books on the same subject could safely ignore Vol 1. Vol 2 should be the first of two volumes on application protocols. Vol 2 could cover all of the standard application protocols well-documented elsewhere (DHCP, DNS, SNMP, TFTP, FTP, HTTP, SMTP, POP3, IMAP4, Telnet, SSH, SSL/TLS). Many people could ignore Vol 2 too. Vol 3 should cover protocols not well-documented elsewhere, but important, like SMB/CIFS, NFS, NTP, various flavors of P2P, VoIP, and instant messaging and IRC.
If a book like TTG is going to devote a chapter to Gopher (which I probably haven't used in 10 years), it should cover SMB, Microsoft's file sharing protocol (also known as CIFS, incorrectly called "NetBIOS" by some). TTG covers NFS instead, saying NFS is "the most common [network file and resource sharing protocol] for TCP/IP." Given Microsoft's domination of the desktop, SMB is ubiquitous. (A few of my recommended books address SMB.)
TTG is still a monumental effort. I did find several sections very helpful. The discussion of bit masking (set bits with OR, clear with AND, invert with XOR) in Ch 4 was clear. Ch 28's explanation of NAT terminology, such as using DNS with bidirectional (inbound) NAT made sense. I agree with Gordon Shephard's review, including the comment about TTG's PPP chapter being unique.
I'm not sure exactly who should read TTG. I would strongly consider it if your networking shop has no other TCP/IP books and you work with people of varying networking skills.