Diebold Disses Democracy

The more important computer security becomes, the more likely it is to be written into the law of the land. On the face of it, maybe that's a healthy trend. However, technologists may be surprised by how far things can get off track when the law embraces bad security ideas for no apparent reason. It's not always pretty, as security problems with electronic passports and electronic voting clearly demonstrate.

Academics like Avi Rubin and Dan Wallach have been warning about security problems with electronic voting machines since 2004, but unscrupulous vendors have shrugged them off by denigrating the researchers and claiming that academics know next to nothing about how elections are really run.

The worst culprit in the debate has been Diebold, the manufacturer of the AccuVote-TS voting system. The AccuVote-TS system is one of the most popular touch-screen voting systems, also known as direct recording electronic (DRE) voting systems. It is also among the most deeply flawed from a security perspective. Without fail, security experts claim that this system should not be used in any general election.

In September, Princeton professor Ed Felten and his graduate students released a paper describing how a real Diebold Accuvote-TS can be attacked. Instead of hypothesizing or theorizing about software vulnerabilities, the Princeton team built real exploits. (Follow the link above to see a video they released that shows them tampering with a demonstration election that followed standard voting procedures.) This is serious, and something needs to be done about it now.

Politics as Usual
The voting problems in the presidential election in 2000 served as the impetus for the Help America Vote Act of 2002 (HAVA). HAVA provides funds and sets standards for voter registration and election systems. Because of HAVA, even rural counties (including Clarke County, Virginia, where I live) received funding to purchase electronic voting machines.

One of the main ideas behind HAVA is to upgrade voting technology from the punchcard and lever systems, responsible for the infamous "hanging chad," to more modern systems. This is a fine and admirable goal. The two technical possibilities include optical scan systems and DREs. Though DREs have come under serious fire for security issues for years, the politicians responsible for their procurement and distribution have dug in their heels (along with Diebold).

This is not a problem with one political party or the other. In fact, both major parties have taken positions on both sides of the issue. The fact of the matter is that DREs are a political hot button that makes political hay in both directions.

The real problem is that politicians don't seem to be relying on technical advice from outside experts as much as they might. Perhaps the Princeton e-voting hack will open some eyes. Or maybe it will be politics as usual.

Hacking an Actual Voting Machine
The Princeton team obtained a real Diebold DRE and analyzed it for security vulnerabilities. In their third-party study they took into account standard voting procedures followed in real elections.

They concluded that the machine is vulnerable to extremely serious attacks. For example, an attacker who gets physical access to a machine or its removable memory card for as little as one minute could install malicious code. Malicious code on a machine could also steal votes undetectably, modifying all records, logs, and counters to be consistent with the fraudulent vote count it creates. An attacker could also create malicious code that spreads automatically and silently from machine to machine during normal election activities – a voting-machine virus.

The Princeton team is well respected for its security expertise. Ten years ago, Felten and his students were responsible for bringing serious Java security flaws to light (in fact, Ed and I wrote the books Java Security and Securing Java about that work). The difference is marked in this situation – back in 1996, Sun Microsystems, Netscape, and Microsoft responded to security criticism by fixing the JVM; by contrast, Diebold digs in its heels and spouts complete security nonsense!

Diebold's response has thus far been laughable (from a security perspective) and completely off the mark (from a responsibility perspective). Why, they even claim their system is secure because it uses certain kinds of cryptography! It is high time for Diebold to throw in the towel and agree to fix the mess they have made.

Our democratic government is based on representative government elected by the people. Making elections fair and accurate is critical. Secure voting equipment validated and assured by third-party analysis is not a luxury, it's a necessity.

— Gary McGraw is CTO of Cigital Inc. Special to Dark Reading