The (Ugly) Secret of Oregon's Elections

by Virginia Ross
Printed in the Oregonian on Thursday, May 17, 2007

With each election, Oregon voters can celebrate the flexibility and convenience of casting their votes by mail-in paper ballots. We can toast the fact that those ballots can be hand counted to confirm the outcome. But before breaking out the microbrew, we should ask ourselves: "Are they?" Hand counted for confirmation, that is.

Well, the answer is "no." The truth is, all of our paper ballots are counted exclusively by optical scanners. And while that fact alone shouldn't dampen the festivities too much, this fact should: All ballots in Oregon are counted using proprietary software whose specifications are considered trade secrets and are not disclosed to officials or the public. So, as in other states, Oregon outsources to private corporations such as ES&S, Sequoia and Diebold the most crucial public function in our democracy -- the counting of our votes.

So celebrate your paper ballots all the way to the mailbox, but understand that there are no people actually counting them, just machines and secret software. Software experts can tell you that the use of proprietary code alone to count votes is akin to asking the author of the software to count the votes in private and hand us the results.

If we must use trade-secret software to determine the outcome of our elections, then we should at least be auditing the results. An important bill in the Oregon House, House Bill 3270, would do just that, mandating the best available accuracy test to confirm our election results: random sample hand counts of the actual ballots.

The bill features a rigorous statistical method vetted by experts over the past two years. It would verify the outcome of randomly selected races to within a 99 percent level of confidence. Since the total hand count would involve fewer than 50,000 ballots statewide, it would also be highly cost-effective.

Oregon's existing ballot checks are inadequate. Currently, routine logic and accuracy tests are used to check whether optical scanners can read and count small stacks of mock ballots correctly. But whether by mistake or on purpose, couldn't secret software easily tabulate real ballots differently than small test stacks?

Other checks include automatic hand recount procedures (triggered by close finishes) and optional hand recount procedures (available to candidates, officials or parties). The former is rarely used, and the latter often burdens the apparent loser with high costs or political risks.

Oregon's election system should not rely on expensive, optional, rarely triggered or politically risky recount procedures as its chief means to verify results. In a democratic republic, elections conducted in public view and verified as accurate are the only legitimate means by which citizens transfer power to the government. Oregon should verify our election outcomes with routine, scientific, cost-effective, random sample hand counts using the actual paper ballots cast.

Then we'd truly have something to celebrate.

Virginia Ross is a Portland attorney who serves on the board of the Oregon Voter Rights Coalition.