Free WWW 3: Wonder by Robert J. Sawyer

Webmind was buying time. Hume gripped the steering wheel tighter and sped on into the darkness.

eight
     
    Masayuki Kuroda was leaning forward in his chair now, looking at the face of Anna Bloom on his screen. “The Americans have a technique that does work to scrub most of Webmind’s packets,” he said into the little camera at the top of his monitor. “Now all they have to do is get the Ciscos and Junipers of the world to upload revised firmware that would cause their routers to reject all packets with suspicious time-to-live counters.”
    “Oh, I don’t think you have to worry about that,” Anna said.
    “Why not?” asked Masayuki.
    “Most of the routers on the Internet are running the same protocols they’ve been using for decades,” she replied. “The reason is simple: they work. Everyone’s afraid of monkeying with them. You know the old adage—if it ain’t broke, don’t fix it. Plus there are thousands of different models of routers and switches; you’d need a different upgrade package for each one.”
    “Oh,” said Masayuki.
    Anna nodded. “In 2009, an Internet provider in the Czech Republic tried to update the software for routers there,” she said. “A small error he introduced propagated right across the Web, causing traffic to slow to a crawl for over an hour. Can you imagine the lawsuits if Cisco or Juniper mucked up the whole net—if, say, the new firmware had a bug that caused it to delete all packets, or modified the contents of random packets?”
    “Well,” said Masayuki, “obviously, they’d test—”
    “They can’t,” said Anna. “Look, before Microsoft rolls out a new version of Windows, they have tens of thousands of beta testers try it out on their individual computers, so that bugs can be found and fixed prior to the release going public—and, still, as soon as it does, thousands of additional bugs immediately come to light. You can test router software on small networks—a few hundred or even a few thousand machines—but there’s no way to test what will happen when the software goes live on the Internet. There’s no system anywhere on the planet that duplicates the Internet’s complexity, no test bed for running large-scale experiments to see what would happen if we changed this or tweaked that. The Internet is a house of cards, and no one wants to send it all tumbling down.”
    “What about the Global Environment for Network Innovations?” asked Webmind’s disembodied voice.
    “What’s that?” asked Masayuki.
    Anna said, “GENI is a shadow network proposed by the American National Science Foundation in 2005, precisely to address the need for a test bed for new ideas and algorithms before they’re turned loose on the real Internet. But it’s years away from completion—and unless it ends up having a Webmind of its own, there’ll be no mutant packets acting like cellular automata on it to perform tests on.”
    “So Webmind is safe?” asked Masayuki, sounding relieved.
    Anna raised a hand, palm out. “Oh, no, no. I didn’t say that. If the US government wants to bring you down, Webmind, they’ve got an easy way. That test they did to see if they could eliminate you: it was doubtless only phase one. You said they used an AT&T switching station?”
    “Yes,” Webmind replied.
    “Proof of concept, and with AT&T equipment.”
    “That’s significant?” asked Kuroda.
    Anna made a forced laugh. “Oh, yes indeed. AT&T has a secret facility that nobody speaks about publicly; employees in the know just call it ‘The Room.’ It has multiple routers with ten-gigabit ports, and, quite deliberately, a significant portion of the global Internet backbone traffic goes through it. Of course, the NSA has access to The Room. Had his small-scale test succeeded, Colonel Hume doubtless would have modified those big routers to scrub your mutant packets. They wouldn’t necessarily get them all, but they’d take out a big percentage of them. Of course, if you hit