- By Michael Hardy
- Nov 09, 2003
Here are some of the alleged security weaknesses Johns Hopkins University researchers found in Diebold Inc. subsidiary Diebold Election Systems' electronic voting software and the company's responses.
System security flaws: Hopkins researchers found that if electronic voting code has security flaws, unscrupulous voters or malevolent insiders — including election officials, the developers of the voting systems and the developers of the embedded operating system on which the voting system runs — could exploit those flaws.
Diebold's response: It would take a conspiracy to exploit such security holes. The electoral process is designed to prevent fraud.
Tamper-resistant results: Hopkins researchers found that election results could be tampered with if transmitted between the machines and a home base. Diebold's code does not authenticate the far end of the transmission or check the data's integrity, so the results could be intercepted and changed en route.
Diebold's response: The machines do not transmit anything over public networks. At the discretion of local officials, they may transmit unofficial results over a dedicated network after polls close.
Incorrect cryptography: University researchers found that Diebold systems use cryptography incorrectly in some cases and not at all in others where the need for it would seem obvious and necessary.
Diebold's response: The researchers analyzed an old version of the code. The systems have passed "rigorous functional tests and reviews."
No change-control process: Researchers found that the voting systems seem to lack a change-control process, which could allow a developer to insert malevolent code into the software without being detected.
Diebold's response: The company does have a system to detect such things. "Making such changes would require a conspiracy among nearly all the developers in the company, in addition to members of the quality assurance group," Diebold officials said.
Hacker vulnerabilities: A hacker could forge the smart cards that voters use to cast ballots, allowing one person to cast multiple votes. Furthermore, if extra votes are detected, in the absence of a paper record, there is no way to know which are false.
Diebold's response: The system uses cards with a distinct configuration, not generic, readily available ones. If someone did create false cards, it would be a high-risk way to cast a small number of additional votes. And the extra votes would be detected during a reconciliation process that compares the number of votes cast to the number of voters who signed in throughout the day.