Wednesday, October 24, 2007

Artificial life

From Slashdot. I don't know if there is AI (machine intelligence) involved, but it would be easy to set up that way...

At Foo Camp I took a poll of some leading security researchers and the majority thought that the probability of a major Internet failure (e.g., 100M people without access for several days) due to botnets or worms was close to 100 percent over the next few years.

Storm Worm Strikes Back at Security Pros

Posted by ScuttleMonkey on Wednesday October 24, @01:25PM
from the skynet-worm dept.

alphadogg writes "The Storm worm, which some say is the world's biggest botnet despite waning in recent months, is now fighting back against security researchers that seek to destroy it and has them running scared, conference attendees in NYC heard this week. The worm can figure out which users are trying to probe its command-and-control servers, and it retaliates by launching DDoS attacks against them, shutting down their Internet access for days, says an IBM architect."


Anonymous said...

"majority thought that the probability of ... 100M people without access for several days...was close to 100 percent over the next few years."

Close to 100%. Hmmmmmm. That would imply fair odds of, say, 99 to 1. I'll take the don'ts. At 75 to one for $10k on my side, a real overlay for the "experts."

Any takers? .... No? I didn't think so.

Idiotic statistical prediction! It won't happen, at least not at the probability specified. Too bad the people who make these kind of statements are never willing to lay real money on them.

Steve Hsu said...

I'm not sure they're that crazy -- close to 100% could mean 80% or 90%, as these guys are computer scientists, not probabilists or market predictors. But you are right that a real futures market would force a better calibration of the prediction.

Estonia is a good example of what might happen in the future. Also, there have been widescale disruptions of the DNS system in the recent past.

Of course, by raising the alarm and pushing for security improvements, they lower the prob. that their prediction will come true.

Blog Archive