The dangers of automated voting
It’s encouraging to see the Telegraph’s coverage (Dec. 27) of the dangers to our democracy of automated voting systems.
Electoral integrity depends on openness and independent verifiability. New Hampshire election law allows parties and candidates to appoint "inspectors of election," who have various functions. But if even one step between marking the ballot and reporting the totals isn’t visible to them, verifiability is irretrievably broken. As you point out, what happens inside our electronic ballot scanners is a mystery to the election officials, the partisan inspectors, and everyone else.
Years ago I looked into applying some industrial safety technology to a scanner design that could be conclusively proven to be both logically correct and incapable of running if a component failed. It turned out to be possible but not practical. The polling station volunteers would have needed the skills of mechanics and technicians to detect tampering or misconfiguration.
Today, a different approach may be within reach. Document scanners with sheet feeders are now inexpensive enough for anybody with a laptop computer. Massively peer-reviewed open source operating systems and software development tools have reached full maturity, leaving no place for bugs and back doors to hide. Add an equally open ballot-counting application program, perhaps developed as an academic thesis project, and inspectors could run a count on their own machines as soon as the ballot box is opened. There would be no single point of failure and no single target for insider manipulation.
And if there’s a discrepancy? The human eyeball remains the gold standard. New Hampshire’s election officials and inspectors know how to count ballots.
John A. Carroll