In October 2000, during the waning days of the Clinton administration, the U.S. Department of the Navy awarded the biggest information-services contract in the history of the American federal government.

It was called the Navy Marine Corps Intranet, and it was designed to replace thousands of incompatible information systems operated by sea-service shore installations with a single, integrated network. NMCI became the biggest "intranet" in the world, supporting 350,000 computers and 700,000 users.

It isn't surprising that the Navy Department was the first American military department to push for an integrated internal network. The U.S. Navy and Marine Corps have led joint-force thinking about the military applications of digital networks since the dawn of the information age. It was the Navy that first embraced the notion of network-centric warfare, and the Navy that made the most aggressive steps to reorganize its war-fighting posture around the synergies afforded by connecting all members of the force.

Having provided a path that other services could follow into the new millennium, though, the Navy has now begun to lose its way. Navy planners want to balkanize implementation of the successor to NMCI in a scheme that looks doomed to fail. The new system is called the Next Generation Enterprise Network. It is one of several overarching networks that are supposed to complete the integration of sea-service information systems circa 2016.

The basic idea is to segment the operations of the next-generation network into a series of functionally related services such as data storage, user support and information security that would be provided by best-of-breed industry teams, with the overall system being integrated and managed by a dedicated U.S. Navy office.

That sounds reasonable enough, but there is little basis for believing the U.S. Navy is capable of efficiently integrating the efforts of a half-dozen unruly teams as they implement the pieces of a continuously evolving network. To make matters worse, the Navy Department proposes to have the new network up and running on Oct. 1, 2010, even though it hasn't yet described a complete acquisition approach, much less conducted the competitions needed to select providers.

The U.S. Navy has adopted this improbable plan largely in reaction to its frustrations with the way NMCI unfolded. The Clinton administration elected to outsource the whole network to Electronic Data Systems, which left EDS not only managing dozens of suppliers but also owning the hardware and intellectual property.

EDS then misjudged how hard it would be to implement the network, causing delays and disagreements. As a result, even though NMCI has worked well for years now, every time there is a problem sailors blame the contractor.

Everyone in the U.S. Navy agrees the service needs more control over the successor to NMCI. That's fine, but Navy leaders ought to be asking themselves whether system integration of the world's biggest intranet is really one of their core competencies. (It isn't.) They also ought to be asking whether a segmented system will respond as well to the kinds of cyberattacks the military has been experiencing as NMCI has. (It won't.)

The sad truth about relying on networks is that we have shifted military competition to an arena where many enemies can play, so we better be damned sure we can count on the network before we build our war-fighting posture around it. The Navy's dis-integrated approach to implementing NGEN looks like an invitation to disaster.

(Loren B. Thompson is chief executive officer of the Lexington Institute, an Arlington, Va.-based think tank that supports democracy and the free market.)