This essay was published in the Ann Arbor News on October 19, 2003.

Voting improvements needed
But just eliminating punch cards won’t solve all the problems

I very much agree with your editorial (Oct. 2) calling for action on updating and improving voting systems, and I second your call that the president name his nominees to the Election Assistance Commission.

However, I think your deep concern over punch card voting and “squinting” at ballots is somewhat misplaced. Yes, we should move to newer voting systems. But there are a number of other, far more critical, issues to be addressed — including how we choose those new systems.

The punch card was invented by Herman Hollerith more than a century ago, as a way to facilitate tabulating the 1890 U.S. Census; the cards matched the dimensions of the paper money of the time. Today’s punch card ballots are still that same size and shape.

Punch card voting came to Michigan in the 1970s, and in comparison to other voting systems of the time, it worked well. In particular, punch cards performed far better in recounts than the systems they replaced.

Paper ballot counting and recounting generates endless arguments about whether the X crosses inside the square. Lever-handle voting machines are subject to malfunctions that can invalidate hundreds of votes. By contrast, at least in Michigan, punch card election results have been very solid in recounts. Ambiguous ballots (”hanging chads”) are extremely rare, and the state has clear rules for resolving those cases. Normally a recount changes the original result by at most a vote or two.

Why did Florida have such a bad experience in 2000? The biggest reason was the lack of maintenance on equipment. Apparently in Florida it was not standard practice to empty the chads from the voting devices; the machines were often packed with old chads. That made it difficult to punch cleanly through the card, and resulted in a great many “pregnant” or partially punched-out chads.

Florida also had the “butterfly ballot,” which exposed what a poor user interface the punch card voting device is. Since the lists of candidates and the arrows are in a different plane than the punch card itself, anyone who is above or below average height will experience parallax, that is, the arrows and holes don’t line up.

But the biggest problem with punch card voting is the centralized handling and ballot counting they require, which entails risks of loss, mishandling, error, and fraud. Better voting systems provide for vote counting in the polling place — like the optical scan system Ann Arbor uses now.

A secondary problem with punch cards, along with many other computerized voting systems, is the use of secret, proprietary software to do the counting. Voters and election officials should demand that vote tabulation software be open to scrutiny by all interested parties, to ensure the absence of “back doors” and Trojan horses. In computer security, peer review is always preferable to trade secrets.

Proposals to send votes over the Internet should be rejected. There is no practical way to simultaneously ensure voter authentication and a secret ballot in the online world. Nor should we use “touch screen” voting machines with internal counters. We should have a tangible record of each vote, to act as a check on any automated system.

Further, let’s not confine our focus to the narrow issue of how voters mark or punch their choices. Errors in counting individual votes are dwarfed by gross errors in tabulation. Examples from elections in recent years abound. Here are just a few:

• Marquette County’s precinct returns were entered into a spreadsheet; because of an error in the summing formula, the certified totals didn’t include all of the absentee voter counting boards.
• The city of Northville’s entire vote for statewide offices was included in Oakland County’s certified totals, even though the south half of the city was also counted in Wayne County.
• Amboy Township, in Hillsdale County, certified vote totals that left out a large number of “yes” votes on a statewide ballot proposal.
• Certified totals from the city of Wyoming, in Kent County, implied that an improbable 7,500 voters (40% of the total) had neglected to vote for Governor.

“Pregnant chads” get ample news coverage, but problems like these — no doubt caused by fatigue, poor training, awkward systems, and inattentiveness by election officials — affect vastly more votes.

To achieve fair, accurate elections for all offices is within our reach, but it requires more than just getting rid of punch cards.

Lawrence Kestenbaum is a former Washtenaw County Commissioner (2000-02) and Ingham County Commissioner (1983-88). He has participated in many recounts as an election worker, challenger, or attorney. He’s the author of two articles on voter authentication for the ACM Risks Digest, and creator and Web master for PoliticalGraveyard.com. He’s also a staff member at the University of Michigan Institute for Social Research.