DARPA-funded Linux security hub withers

Nobody showed up

Two years after its hopeful launch, a U.S.-backed research project aimed at drawing skilled eyeballs to the thankless task of open-source security auditing is prepared to throw in the towel.

Initially funded by a research grant from the Pentagon's Defense Advanced Research Projects Agency (DARPA), the Sardonix project aspired to replace the loosely-structured Linux security review process with a public website that meticulously tracks which code has been audited for security holes, and by whom.

As conceived by Oregon-based computer scientist Crispin Cowan, Sardonix was to attract volunteer auditors by automatically ranking them according to the amount of code they've examined, and the number of security holes they've found. Auditors would lose points if a subsequent audit by someone else turned up bugs they missed.

Cowen hoped that the system would produce the same cocktail of goodwill and computer-judged competition that fuels other successful geeky endeavors, from the distributed computing effort that recognizes top producers in the search for new prime numbers, to the "karma" points awarded highly-rated posters on the news-for-nerds site Slashdot.

In the end, though, nobody showed up.

"I got a great deal of participation from people who had opinions on how the studliness ranking should work, and then squat from anybody actually reviewing code," says Cowan, chief research scientist at WireX Communications.

The project's DARPA funding ran out nine months ago, and the website lingers as a mostly-abandoned husk. The only code audits on the site were performed by a handful of graduate students directed to the task by David Wagner, a computer science professor at U.C. Berkeley.

Cowen believes Sardonix was a casualty of security community culture, which he says rewards researchers who find clever or splashy holes in a program, but not for making software more secure. "The Bugtraq model is: find a bug, win a prize -- a modest amount of fame," says Cowen. "Our model is: review a whole body of code, eventually finding no bugs, and receive a deeper level of appreciation from people who use the code.

"It seems the Sardonix lesson is people don't want to play this game, they want to play the Bugtraq game."

Copyright © 2004, 0

Sponsored: Network DDoS protection