A clean slate for the Internet

Spam? How quaint

Indeed, some of McKeown's clean-slate projects have a 15-year planning horizon. "Changes will be largely in terms of security," he says of the Internet 15 years out. "Terms like , , and are just consequences of the Internet architecture; they are not things that will inherently happen. We may have new terms for new ways that people will try to trick us, but those old terms will be quaint terms associated with the previous Internet."

Not everyone agrees with that assessment. "The argument often is that you need a clean-slate Internet because the current Internet has a lot of problems -- a lack of security, viruses, spam, that kind of thing," says Internet pioneer Robert Kahn, a co-inventor of TCP/IP. "But most of those are not issues that a brand-new structure is likely to solve all by itself."

Kahn, who's now CEO of the Corporation for U.S. National Research Initiatives, points out, for example, that what is deemed to be spam or even pornography can vary depending on who is viewing it, and he notes that the application of technological solutions for such problems introduces other social issues, such as concerns about privacy and censorship. "The reason we don't have security in the current Internet isn't solely a technical matter," he says.

And it's worth noting again that not all of the NSF's FIND and GENI projects deal solely with technology. "We are suggesting that a lot of the security projects should have a social collaborator," Mankin says. "You are not going to change society, but you can help with some changes to technology."

Kahn says clean-slate proposals may involve changes to the networks themselves, to computers, to operating systems or to major classes of applications. "You don't do these things lightly," he says. "The implications are pretty profound across the board."

Given the difficulties, does it make sense to seek a clean slate? "Absolutely," Kahn says. "If there are really good ideas -- that's the heart of it."

A few FIND projects

- Design for Manageability in the Next Generation Internet: University of Wisconsin

- Designing Secure Networks From the Ground Up: Stanford University, University of Wisconsin

- Enabling Defense and Deterrence Through Private Attribution: University of California, University of Washington

- An Experimental Protocol Stack for Cognitive Radio Networks: Rutgers University, University of Kansas, Carnegie Mellon University, Blossom Inc.

- A Framework for Manageability in Future Routing Systems: University of Pennsylvania; University of Massachusetts, Amherst; University of Minnesota

- Future Optical Network Architectures: MIT