unmediated: Immunizing the Internet

« catwav: concatenating two .wav files under Unix | Main | Could I Run a TV Station on Linux? »

October 5, 2006

Can computer crime be beneficial? That’s the question asked by a provocative note, “Immunizing the Internet, or: How I Learned to Stop Worrying and Love the Worm,” by an anonymous author in June’s Harvard Law Review. The note argues that some network attacks, though illegal, can be beneficial in the long run by bringing attention to network vulnerabilities and motivating organizations to address problems.

I don’t buy the note’s argument, but there is a grain of truth behind it. Vendors and independent analysts often disagree about whether a vulnerability is real or could ever be exploited in practice. One thing I’ve learned over the years is that the best (and often the only) way to resolve that debate is to demonstrate an exploit. If you can do something, people will accept that it is possible.

Our recent e-voting study is a good example. Diebold can’t seriously argue that malicious code can’t sway an election, because we have a working demo that we have shown on national TV and in front of congress.

Even when the vendor is willing to acknowledge reality and work constructively to fix a problem, a working demonstration is useful in helping the vendor cope with the problem — and in helping the good guys within the vendor organization neutralize any internal minority that wants to deny the problem. Showing the vendor a working demo can be the first step in a constructive problem-solving relationship.

(To be clear: You can build a working demo and show it to people without revealing to the public every detail of how to build the exploit. How much information to publish about a demonstration exploit is a separate issue from whether to build it in the first place.)

But some sorts of problems can’t be demonstrated without breaking the law. For example, Diebold apparently claims that there is no way to tamper with the upcoming November election in (say) Maryland. I’m convinced that claim is false, but the most direct, obvious way to prove it false would involve actually tampering with the election, which of course is unthinkable.

The note’s reasoning would imply that the penalty for tampering with the election might be reduced, especially in cases where the tampering is engineered to be obvious and to cause minimal damage, for example if it added 10,000 write-in votes for Homer Simpson to a statewide race where a candidate was running unopposed. Though such an attack would be instructive, it would still be wrong and would deserve serious punishment. If the legal lines are drawn in the right places, and if the punishment otherwise fits the crime, then we shouldn’t let attackers off easy just because their attacks were instructive.

Originally posted by Ed Felten from Freedom to Tinker, remediated by yatta on Oct 5, 2006 at 4:19 PM

October 2006

Sun

Mon

Tue

Wed

Thu

Fri

Sat

unmediated

October 5, 2006

archives