Home > Security

Computerworld - The issues of electronic voting system security and reliability first gained public attention in July 2003, when researchers at Johns Hopkins University in Baltimore published a detailed study of the software code used in the Diebold AccuVote-TS voting machine. The report's charges of a startling number of security vulnerabilities and a complete lack of adherence to industry-standard software design and testing methodologies put Diebold Election Systems and its direct recording electronic, or DRE, systems at the center of the debate about e-voting security and reliability.
Working under the direction of professor Aviel Rubin, computer science graduate students found the following:

• No encryption protection for dial-up modem connections used to upload unofficial polling-station results.


• A lack of encryption protection on the smart cards that voters used to cast their votes and poll workers used to gain administrative access to the systems.


• A complete disregard for disciplined software change-control and testing processes.


• A questionable choice in the decision to use the programming language C++. Most experts agree that it's easier for inexperienced programmers to introduce glitches and security holes using C++ than it would be if they were using other languages, such as Java.

Although denounced as inaccurate and unrealistic by Diebold and IT industry lobbying groups that favor the use of electronic voting systems, the Johns Hopkins study was followed by two other independent studies, by Science Applications International Corp. and RABA Technologies LLC, that confirmed many of Rubin's findings and uncovered additional concerns.
In fact, the January 2004 study by the RABA Innovative Solutions Cell outlined how researchers were able to guess the passwords for the smart cards used by poll supervisors, allowing them to reinitialize voter cards and vote multiple times.
The team also exploited a 6-month-old unpatched vulnerability in the Microsoft Corp. software used on the back-end Dell Inc. server, that allowed them to upload, download or execute code on the back-end vote tabulation server. That server used an unencrypted modem connection to report local poll results at the end of an election.
RABA also found gaping holes in the procedure by which precincts uploaded votes to their local election boards. As a result of an incomplete implementation of the Secure Sockets Layer protocol, the team was able to conduct a "man in the middle" attack and use a remote laptop to receive the election results. An attacker who did that would "be able to acquire the name and password to access the server," the report concluded. "With this name and password in hand, the attacker could upload modified results to theserver -- all in real time."

Read more about Security in Computerworld's Security Topic Center.