There once was a theory in networking that networks were smart and the things that attached to them were dumb.
One generally picked up one of these dumb gadgets and talked into it, and someone somewhere else could hear you talk.
We called it a "telephone."
That paradigm served us well for over a century. Its only problem was that it carried sound, perhaps digitised sound, and anything we did with it was a re-use, often inefficient, of technology designed for that one purpose.
Modern computer networks are built on the principle that flexibility and intelligence are best reserved to computers around the network's periphery, with just enough intelligence in the network to make it all work.
This allows new applications, such as the web, to be created and changed according to market needs without having to change the network itself.
The direct implication of this change is there are myriads of programs and devices used in the Internet: message switches (called switches and routers), computers of various sizes and purposes, firewalls, network address translators, network disk drives, servers for various infrastructure applications, and so on.
There are few, if any, companies that can deliver the breadth of hardware and software wizardry the Internet deploys, and none that can do it all well.
In the end, vendors and users are forced to work together to define and deploy the products, and to agree together on how they will communicate. In commodity industries, vendors compete to sell identical products on price; in specialty industries, unique products are sold retail to the end user by brand name.
To paraphrase Benjamin Franklin, in this industry we who vastly differ hang together or we hang separately; our life's blood is the flow of application information in IP datagrams through semiconductors, optical fibre, electric wire, and wireless connections.
Given this, wouldn't it be nice if there were a place where network operators, academics, vendors, researchers, and motivated users could sit down, discuss their needs and capabilities, and create solutions for the Internet that each agreed was in their interest?
There is a place, of sorts: we call it the Internet Engineering Task Force, and it is nothing more than a set of mailing lists and triennial meetings where people mandate, pontificate, discuss, disagree, and in the end forge agreements they can live with.
The agreements they make we call "standards", recorded in numbered documents called RFCs.
Not all RFCs are standards; some are April Fool's jokes, some are white papers, some are documentation of experiments nobody wants to repeat - but those that are define the data transported on the Internet.
Fred Baker is chairman of the Internet Engineering Task Force and a Cisco Fellow.
More information about the IETF can be found at http://www.ietf.org.