Reinventing The Internet

Who is he? Internet pioneer Robert E. Kahn is chairman, CEO and president of the Corporation for National Research Initiatives, a nonprofit organization established to "provide leadership and funding for research and development of the National Information Infrastructure."

August 27, 2001 (Computerworld)

Robert E. Kahn co-invented TCP/IP and managed the development of the Arpanet - the forerunner of the Internet - at the U.S. Defense Advanced Research Projects Agency in the early 1970s. For those accomplishments, President Clinton awarded him a National Medal of Technology in 1997.
Now, as president of the nonprofit Corporation for National Research Initiatives in Reston, Va., Kahn is deep into a "reinvention of the Internet one layer up." He has developed a framework for interoperability of heterogeneous information systems that aims to make digital information a "first-class citizen" on the Internet.
Kahn's architecture contains "digital objects" - data of any type, plus a long-lived identifier called a handle. The objects can reside in any kind of storage system, or repository, accessed by a Repository Access Protocol, which enforces rights and permissions to the data within. Kahn recently told Computerworld's Gary H. Anthes how taking a look at the past can guide us into the future.

In congressional testimony three years ago, you urged the federal government to help the U.S. maintain leadership in e-commerce. How's that going? We've made small progress, compared to what's possible. We've had very little recognition at the federal level about the importance of pilot projects. For example, one project could be in authentication of information. When you get something off the Net, there's no way to know if it's accurate, and there's no one party to provide the standard for that. The government could establish a way to verify information going to the public, and whatever they do might then be a good template for the private sector as well.

Is the Arpanet a model for government/industry cooperation? The model I always thought was the right one - what we put in for the Internet from Day 1 - was something that started with total government control. We - I, and then [TCP/IP co-inventor] Vint Cerf and I - ran everything for a while, and over the years, we devolved little by little to the private sector, to the point today that government has very little role to play.

But the government-sponsored standards for openness and interoperability built into the Arpanet live on in the Internet today, right? Yes, but more and more, we are seeing the need for generalized standards for doing the critical functions [like authentication] that everyone wants to do. But most organizations want to create their own and see if they can create a monopoly.

Like Microsoft? People focus on Microsoft and say, "What do we do about the operating system?" I don't think that's the issue at all. Most people are going to be on the Net in the future, and as nets get faster and faster, [users] will be able to download software from anywhere. If they have the right protocols, they can do plug-and-play just like that. So you don't need to buy an operating system anymore; you can just suck all this software in.

You also told Congress that digital information should be a "first-class citizen in the Net." Is it? No, I think it's still a third-class citizen. Everything in the physical world is an identifiable thing with some notion of ownership. It has identity and value. But very few things on the Net have those attributes. If you take a piece of data off the Net, it's not clear who owns it, what the value is, what you can do with it, what you can't.

Is that what you're trying to address in your digital library initiatives, in which you say you're developing ways to "uniquely and persistently identify, manage and track" digital information? Yes, among other things. Today, if you want to find a letter you wrote 12 years ago, what would you do? You wrote it on a 5.25-in. floppy, and it's in an old machine in your basement. It's lost.

You've said that the repositories that are part of your digital library concept aren't literally databases. What do you mean? A repository is an interface specification that we are hoping people will buy into, because that automatically gives you interoperability. People say, "We don't need your repository because we already have a database." It's like they said in the early days of the Internet: "We don't need the Internet because we already have our own networks." What I've been developing is really a reinvention of the Internet, one layer up.