The following text is copyright 2003 by Network World, permission is hearby given for reproduction, as long as attribution is given and this notice is included.


Can AT&T learn to be stupid?


By Scott Bradner


AT&T has been in the press quite a bit of late.  In addition to its effort to derail, in an altruistic effort to support the public interest I'm sure, MCI's escape from bankruptcy (and, incidentally from a pile of debt), AT&T announced that they were going to spend a lot of money over the next few years on their network.  Then, a few days later, AT&T's chairman said that they were already testing a Vonage-like VoIP service that would run over anyone's networks.


According to the announcement, AT&T is planning to spend $3 billion or so to replace their existing network with a new-fangled IP and MultiProtocol Label Swapping (MPLS) one with the aim of getting rid of the remainder of its once extensive circuit-switched voice-centric technology over the next few years.  AT&T's CTO said it could take a decade to finish switching the global network.


A major focus seems to try to attract multi-site corporations by having the best edge-to-edge network in the business.


At the same time, AT&T wants to branch out from its managed VoIP service, which it now offers to enterprises in 40 countries, to a consumer-oriented voice-over-the-Internet service similar to the service offered by Vonage.  As is the case with the Vonage service, customers would not have to be getting their Internet connectivity from AT&T to be able to pay AT&T for the new phone service.


I hope that AT&T realizes something about networks by offering the consumer VoIP service.  I hope AT&T realizes that, just like it needs the local ISPs to stay out of the way so that AT&T can offer the consumer VoIP service, it needs to stay out of the way on the networks it uses to provide IP service to customers.  AT&T, Vonage and many others can offer their services over the Internet because the Internet does not know, or care, what applications are running over it.  But historically AT&T, and the other traditional telcom companies, do not understand this.  (See "The Rise of the Stupid Network" at for some history.)  


AT&T has to understand that after spending all that money upgrading its network to offer IP service that it has to then stop being helpful.  Its network should not be "content aware" or otherwise try to figure out what is running over it.  The customer might tag some packets to request special handling but that should be up to the customer.  That means, among other things, that others will offer VoIP services over the AT&T network without AT&T getting a piece of the action and that customers will run their own peer-to-peer applications, including voice-based ones. (See


It's not easy for a traditional telecom company to purposely be stupid.  While they may have adequate experience in doing so when they do not know what is going on, itŐs another thing entirely to understand when not to make something smart.  I wonder if AT&T is up to the challenge.


disclaimer:  I'm not sure Harvard would be up to the same challenge but I did not ask.  Thus, the above is my own view.