Many companies have been building P2P systems for a number of years, even before Napster made P2P famous. In fact, P2P has been around for a long time under other names. Distributed computing is the most accurate - it captures the essential idea the best.

The idea is simply that just as client/server replaced mainframe-centric architecture, client/server will be replaced by an even more decentralized architecture. In this decentralized architecture, every organization is a "peer" of every other organization. It provides services to some and requests services from others. There's no central omnipotent server surrounded by dependent clients. Instead, there's a distributed network - a federation - of independent servers that connect to each other on an ad hoc basis to accomplish the tasks of the moment.

The Web, as originally conceived, was a distributed system. All Web servers are inherently equal. However, the Netscape IPO in mid-1995 ushered in a retrograde period on the Web, which lasted until the dot-com meltdown in mid-2000. During this period, venture capital inflated hundre

What does this have to do with P2P? Well, we have to approach the subject sideways. The Consortium Trading Exchanges (CTEs), because of their industry dominance, have to tread very lightly. They're now scrutinized by antitrust authorities and must overcome the natural skepticism of the smaller companies in that sector. Unlike the Independent Trading Exchanges (ITEs), which can be closed, proprietary, and otherwise exclusive, the CTEs must be open, standards-based, and inclusive.

To gain the trust and acceptance of the thousands of smaller companies that comprise each sector (the automotive industry includes over 70,000, for example), the CTEs must be as accommodating, respectful, and nonintrusive as possible. And, when you think about it, that implies a P2P architecture.

A P2P architecture treats each participant as an equal partner. It respects their existing way of doing business (workflow and legacy computer systems). It allows each company to control how it transacts business with other participating companies. It allows each company to interact with its fellow buyers and sellers on a level playing field.

To be explicit, a lot of people in IT see this level playing field as the Internet (HTTP/FTP/e-mail) plus industry-standard messages (XML) plus a lightweight membership/authentication/security umbrella (PKI/LDAP). I feel that in the next few months all of the highest-profile CTEs will announce that they'll support such a simple, open backbone for their trading exchanges.

This belief is grounded in practical experience. The Open Applications Group Inc. is an industry group dedicated to the exchange of fundamental business documents using open standards. For five years, through seven revisions of the OAGI specification, the 80-odd member companies have been defining electronic versions of all the critical business documents currently exchanged by fax or, at the high end, EDI.

At the beginning of 2000, OAGI turned a crank and produced XML versions of these 170 business documents. The effect has been dramatic. Since that date OAGI members have been doing live interoperability demonstrations in a series of private events for companies such as Ford and Covisint. During a four-hour presentation, 21 vendors - including Oracle, Peoplesoft, WebMethods, and Extricity - worked through three different real-world purchasing scenarios. These scenarios had been designed by major customers such as Ford, Lockheed Martin, and Lucent.

Since this event, several of the leading CTEs have quietly announced that the OAGI documents will be the backbone of their exchanges. Once the overlap between OAGI, RosettaNet, and ebXML are resolved in the first half of 2001, I expect the remaining CTEs will make similar announcements in short order.

Why will these CTEs endorse a single standard? Because tens of thousands of manufacturing firms are participants in the automotive, aerospace, and other sectors simultaneously and they won't want to support multiple overlapping exchange formats. They have only one fax machine and it can talk to other companies anywhere in the world, so why should they accept multiple electronic exchange formats?

These new trading exchanges will doubtless have certain centralized services for adding (and deleting) participants, for maintaining a trusted audit log of transactions, and for daily reporting and analysis of transaction volumes and patterns. Perhaps they'll also have a central financial reconciliation capability to move electronic funds between buyer and seller bank accounts.

In 10 years or so we'll have built a lightweight P2P network for businesses to exchange legally binding business documents without ever committing the documents to paper. This will reduce the friction in commerce and reduce the effort and cost associated with each transaction. In a third-generation solution, it may even allow businesses to buy and sell products automatically - but that's a story for another day and, probably, another decade.