Dale Hatfield's 2002 report [pdf] on issues affecting implementation of wireless E911 is a precise and thoughtful piece.  It's astonishing that with that report in hand (and with knowledge of what VoIP service providers were working on themselves [VON Coalition 2005 white paper pdf]) an independent agency would demand that E911 be implemented by "interconnected VoIP" services in 120 days.

Translated:  Hatfield presented E911 for wireless carriers as an overwhelmingly complex undertaking.  He says, "I did not fully appreciate the complexity of the task facing the Nation until I undertook this inquiry.  There is complexity in every direction."

Which directions was Hatfield talking about?  Well, there are thousands of emergency call centers (PSAPs).  The wireless carriers were using several different technologies across several generations of equipment and many different frequency bands.  The wireless industry was having a tough economic time.  And, perhaps most importantly, the PSAPs were underfunded and sometimes unsophisticated, and the local telephone companies were standing right in the middle between the PSAPs and the wireless carriers.  Standing there controlling access to relevant databases, equipment, routing decisions -- and charging whatever they wanted for it all.

When you take all of that complexity, all of those thousands of inputs, and move it into the nomadic world of IP telephony, it has to be exponentially more complex to create a seamless 911 system for internet users. 

Hatfield, judging from his 2002 report, never expected such swift movement on mandatory VoIP 911.  He says:

Clearly, the long term network architecture and other issues associated with the movement towards VoIP could be addressed by the Advisory Committee or other entity with overall system engineering responsibilities that I recommended . . .   As I envision it, that entity would work with the Commission and the various wireless, wireline, and Internet standards groups to facilitate the necessary exchange of information to reach the necessary consensus to ensure a seamless E911 system in an increasingly IP-oriented national infrastructure.

As far as I can tell, there hasn't been time for this facilitation, information exchange, and consensus development to happen, and no one is telling the local telephone companies to charge reasonable rates or work more efficiently.  So we've got an overwhelming and unbelievably complex systems engineering problem for which the emotional stakes have been ratcheted up unbearably ("people will die if we don't mandate this"). 

Most attempts at solving most highly complex systems engineering problems fail dramatically, particularly where (as here) there is a broken legacy system in place.  The odds against anyone being able to comply with the FCC's order are steep indeed.

As Hatfield said, "There is complexity in every direction."