Democratic societies depend on trust in elections and their results. Throughout the 2016 presidential election, and since President Trump’s inauguration, allegations of Russian involvement in the U.S. presidential campaign have raised concerns about how vulnerable American elections are to hacking or other types of interference. Various investigations – involving congressional committees, the FBI and the intelligence community – are underway, seeking to understand what happened and how. There are many potential problems with elections: Voters can be individually coerced or bribed into changing their votes; the public can be misled about important facts, causing them to draw inaccurate conclusions that affect their votes; and the physical – and electronic – process of voting can itself be hacked. Without conducting a full, vote-by-vote manual recount, which is impossible because many voting machines leave no paper trail, how can we be sure an election was conducted fairly and not interfered with?
My research, as a scholar of game theory applied to computer security, has highlighted how combining two approaches can help solve this vital problem. First, my collaborators and I use game theory to think like an attacker – imagining that we want to influence the outcome of an election and determining the best way to do so. Then, we use our expertise in computer security – including an understanding of the value of randomness – to inform our design of an audit process that maximizes our chances of catching someone conducting that kind of attack.
Sampling ballots to recount
An important way to ensure public confidence in electoral results is to audit the machines’ vote counts. This is best done by checking the numbers each machine reports at the end of an election against paper records made in real time as voters cast their ballots throughout the day. But even if every machine did keep a paper record – and many don’t – doing a simultaneous manual count could cost tens of millions, or even billions, of dollars.
It’s much more efficient – and just as mathematically accurate – to conduct a selective audit, examining a small sample of the voting results to identify evidence of tampering. But that leaves open the question of which districts to audit.
Full Article: Using randomness to protect election integrity.