Sometimes, an apparently oxymoronic policy actually displays some hidden good sense. Take Sprint. It killed off its ION project for universal end-user access to IP services, then boasted a week later that its local-carrier division would be completely converted to an all-packet network by January 2003.

Schizophrenic? Not really. What Sprint was saying was that residential users and many small businesses are not quite ready to move to all-packet services, but that the backbone should move to IP for sheer cost of transport.

Then we have Qwest. Joe Nacchio promotes the value of quickly moving to packet backbones in both interexchange and local-carrier businesses, then halts all construction and procurement contracts without notice. Yes, Qwest is in trouble, but why put all asset development on ice?

What Qwest did makes some sense in terms of shareholder concerns. The established carriers know they can reduce transport costs by eliminating circuit switching in the network, yet that entails laying more fiber and buying more broadband switching equipment. The shareholders hear about the "fiber bandwidth glut" and demand that capital spending on such equipment be eliminated.

Qwest tried to make up the difference by accelerating DSL access to end users through upgraded loop carriers, only to find that the take rate for DSL is going down, not up. Some might place the blame on Qwest's decision to bind DSL to acceptance of Microsoft's MSN. But the real problem may be that in a recession, 95 percent of Americans don't care about broadband access in the home. It's a luxury, and we're at war.

And that's why so many investors hate broadband these days. They don't realize that once the carriers convert the infrastructure to IP, the services come cheap. They only see the up-front costs involved, and the public's lack of interest.

Early this month, Level 3 CEO Jim Crowe said that the broadband explosion still will happen. The question is: How can carriers minimize IP conversion costs to get through the next brutal year?