Home > Computing > Researcher applying game theory to identify…

Researcher applying game theory to identify electronic election tampering

Why it matters to you

Future elections might be safeguarded against electronic tampering if a new game theory algorithm works out.

The election process lies at the heart of free and open political societies, allowing citizens to elect their leaders and in many cases directly influence the laws that are written and enforced. Free, open, and honest elections are therefore vital, and election tampering and fraud have been concerns for as long as elections have been held.

Today, electronic voting machines have opened up a new avenue for subverting elections, one that trades outward physical violence for hidden attacks that can be just as damaging. In response, one researcher is using game theory to develop an algorithm that can identify potential tampering, Phys.org reports.

More: Recount on the way amid hacking fears, questions about electronic votes

Yevgeniy Vorobeychik, an assistant professor of computer science and computer engineering at Vanderbilt University is looking first at why attackers target voting machines, and then at the required methods and potential solutions. As a gaming theory expert, Vorobeychik is working on developing an algorithm that can monitor voting machines during the election process or audit them afterward and prior to certification.

Vorobeychik and his team are testing the algorithm and extensions of it against data gathered during the 2016 U.S. election. His theory suggests that anyone looking to subvert an election would be careful to target only individual machines or those placed in specific districts, creating near ties in those districts where the opposition’s candidate would otherwise likely win the vote.

As he puts it, “The new extension more realistically models the attacks on voting systems that would actually happen. It’s easy enough for humans to just work with a list of districts in order of importance to, say, a presidential election. It’s harder to figure out how to randomize that list to best determine which districts would be targeted. Turns out it helps a tremendous deal to have a computer.”

The algorithm was tested against the 2002 French presidential election, where nothing of concern was discovered. Now, the algorithm is looking at results from Michigan in the 2016 U.S. presidential election which, according to the researchers, had a “suspicious swing.”

Essentially, the algorithm allows computers to do the tedious and labor-intensive work of pulling random districts that might be attractive for tampering and check for discrepancies. In addition, the process by which the algorithm would conduct its analysis would be unpredictable, which would be a better match for the techniques that could be used by attackers.

According to Vorobeychik, “With game theory, you can systematically address attacks and their consequences. If there are a million people who voted illegally, you want to know that and mitigate it. How you deal with that is going to be up to the authorities, but they need to detect it first.” By using game theory and assuming the worst-case scenario of an agent that uses a similar algorithm, Vorobeychik hopes to cut off election tampering at the pass.