In the mid-1990s, smarty-pants Internet Protocol overlay carriers argued that long-distance providers could no longer use voice as an amortization tool. Once data growth surpassed voice growth, they said, it was only a matter of time before even voice would be carried in IP packets. Funny how most of those carriers are now long gone. Their arguments were correct-albeit for the wrong reasons.

Voice-over-IP certainly isn't a big threat to circuit voice, at least not yet. But just as some savvy customers have replaced their home wireline phones with cellular as their primary service, residential local-wireline phone customers are canceling their long-distance accounts in favor of pay-as-you-go phone cards. They figure that once they run out of minutes on their cellular plans, it makes more sense to meter themselves on phone-card options than to sign up for a monthly long-distance plan.

Like the wireless subscribers without a home phone, the home phone customers without an interexchange carrier (IXC) represent a distinct but growing minority. And it's a scary trend for IXCs.

When IXC buildout strategies started crashing in early 2001, the most cynical analysts said the packet-transport problem was so severe it didn't make sense for most IXCs to reorganize under Chapter 11. Better for two-thirds of them to go away, the analysts argued, and for survivors to pick up fiber assets at bargain rates. If dependable voice revenue starts disappearing as everyone goes phone-card crazy in 2003, this trend could accelerate. For long-haul transport, that means fiber networks can be upgraded based only on broadband packet aggregation in business and consumer markets, not on voice multiplexing. But it also has implications for local and regional markets.

Sonet's mini-renaissance could prove short-lived. If circuit-switched voice becomes something to pick up at 7-Eleven along with Doritos and a six-pack, then the last struts will be kicked away from the upgrade of time-division multiplexed systems. Which IXCs are prepared to face that? For that matter, which local incumbents and super-metro carriers are prepared to justify new Sonet buildouts in three years' time?

Packet may win not by design, but by default.