Advances in IT over the decades have come mostly in small increments -- Release 2.3 yields to 2.4, transistors shrink a few more nanometers, Ethernet gets another speed boost, bugs are fixed, and algorithms get tweaked. That kind of evolutionary approach has served users well, boosting speeds, capacities and application capabilities by many orders of magnitude.
But such incremental improvements are no longer sufficient to keep the Internet viable, according to a growing number of researchers. In fact, they say, the Internet is at the tipping point of overwhelming abuse and complexity.
The most sanguine of observers say that even if the Internet is able to avoid some kind of digital Armageddon brought on by spammers, hackers, phishers and cyberterrorists, it nevertheless will drown in a flood of mobile gadgets, interactive multimedia applications and Internet- enabled devices, including phones, cars, home appliances and radio frequency identification tags.
Indeed, researchers say, it is time to rethink all the old notions from the late 1960s and 1970s when the Internet was in its infancy. While few think it is possible to literally start over, there are a number of so-called clean-slate research programs that start with the premise that anything is possible and no option is too far out to consider.
Nick McKeown, a computer scientist at Stanford University, heads up one such program. He says the Internet is "broken" in at least two places -- security and mobility.
"Ten years ago, we were saying the Internet would change the world," he says. "In a decade or two, we'd be doing air traffic control and remote surgery over the Internet. But if air traffic control were on the Internet today, I wouldn't fly. Same with telesurgery."
And it isn't just a problem of security and reliability, McKeown says; the Internet is getting crushed by complexity. He points out that the original Internet design was based on the idea that users were immobile and connected to the Net by wires.
"But if the user is moving around, you end up with a whole lot of hooks and kludges to keep track of the user," he says. "There have been various proposals for a mobile IP, and they are all awful. They barely hold together now, but all the routing mechanisms will just break when there are many more mobile devices."
McKeown and his colleagues have developed a prototype network called Ethane, which centralizes security rather than putting it all around the network in firewalls, virus scanners and the like. With Ethane, all communications are turned off by default. A host joining the network must get explicit permission from a centralized server before it can connect to anything except that server. And the server won't grant permission unless it is able to determine the location and identity of the requestor.