Holes in punch-card system noted long ago

By Jim Drinkard, USA TODAY

GAITHERSBURG, Md. — In the early 1970s, the National Bureau of Standards was a place where engineers puzzled over what made buildings collapse, or how to convert the nation to the metric system. But in his office overlooking the bureau's grassy campus in suburban Washington, Roy Saltman wondered about something else: Could Americans trust the new computerized voting equipment being put into use around the country? Despite a lack of interest from his bosses, Saltman wangled a $150,000 grant from another agency and set about studying election technology. He labored in obscurity, churning out reports recommending that the government collect data on voting problems so it could remedy them. Nothing happened.

A quarter of a century later, Saltman's concerns finally are getting some attention. The Senate Commerce Committee opens hearings today on election reform, and the Senate Rules Committee follows suit next week.

Among Saltman's forgotten reports was a 1988 warning about pre-scored punch-card voting machines — the now-infamous Votomatics used in Florida. He wrote that they were hard to use and vulnerable to error and said they should be banned. Thousands of copies of his report were shipped to voting officials across the country — but were ignored.

"It has always puzzled me why my report never got a wider acceptance," says Saltman, now 68 and retired. His 1988 report is gaining wide circulation, and he has been called as an expert witness before task forces and forums on what ails American elections. "It takes a crisis to move people, and it shouldn't have," he says.

At a recent debate on computer voting technology in Washington, Bryn Mawr College computer scientist Rebecca Mercuri held up a copy of Saltman's 1988 report. "Those of us familiar with this document knew about these flaws all along," she said.

When Saltman set out to study voting systems, little information was available, so he tracked down reports of local election foul-ups, interviewing the officials involved to identify what went wrong. That sounds elementary, but there was no central national agency to oversee elections — a situation that hasn't changed.

Today, two tiny federal entities have elections as their mission: the Federal Election Commission's Office of Election Administration, a five-person bureau that is an underfunded fifth wheel in an agency mostly concerned with campaign-finance issues, and the Defense Department's Federal Voting Assistance Program, which mostly assists military personnel with absentee voting. The two agencies' combined annual budgets are about $5 million, one-fifth of the amount the federal government spent in 1999 to help other countries run elections.

Money is the issue

So the answer to Saltman's question, it turns out, is simple: money. County governments that run elections are always short of cash. Political reality puts a higher priority on building schools and filling potholes than on replacing voting equipment that is used, at most, a few times a year.

"This problem is about money. It always has been, and it continues to be," says Paul Craft, Florida's top state elections systems expert. "Punch cards have been used in Florida for 20 years because they are cheap." In the aftermath of its election disaster, the state is evaluating whether to spend $20 million to convert to a statewide optical-scan voting system.

Saltman's first report, published in 1975, analyzed 14 elections that used computers to tally votes. He found that management problems — failure to test equipment or follow proper procedures — were a far bigger problem than technical failures. Standards for performance and reliability of election computers were "strikingly lacking," he wrote. And election administrators had no way to share information about their bad experiences, dooming them to repeat mistakes that had tripped up others.

He also found the first signs of potential problems in punch-card voting machines. The machines, most sold under the brand name Votomatic, were an inexpensive way to automate vote-counting, a prospect especially attractive to election officials in major cities. They used a standard 3-by-7-inch IBM card with 228 pre-scored holes. It was inserted into a plastic frame with holes that guided a voter using a stylus to punch out the hole corresponding to the desired candidate.

Punch cards had been around since the late 1800s, when they were developed to help tabulate Census data. The pre-scored cards for voting were adapted from cards used by utility meter readers to gather billing information. Their first significant election use came in the mid-1960s; problems soon followed.

Saltman turned up a report on Detroit's troubled first punch-card voting experience, in a 1970 primary election. It found "design inadequacies of the voting device" that led to voters invalidating their ballots by voting for more than the prescribed number of candidates — the same problem blamed for much of Florida's bad experience in November. The ballot cards themselves were "too frail" and tended to "absorb moisture and thereby cause computer jams."

All of these failures were troubling. "An election is like the launch of a space rocket," Saltman wrote. "It must work the first time."

A decade later, again relying on a grant from an outside source, Saltman undertook a follow-up study. In it, he found excessive problems with pre-scored punch-card voting methods. It was almost impossible for such systems to yield consistent results because of problems with the technology and "variability in the ballot-punching performance of real voters," he wrote.

He concluded: "It is recommended that the use of pre-scored punch card ballots be ended."

Before the advent of personal computers, punch cards were the primary method of entering data into computers. Today, voting remains the only vestige of punch-card technology. Still, the warnings of Saltman and others went mostly unheeded for more than a decade.

Panned in Pennsylvania

Saltman was not the only voice warning about the dangers of punch-card voting. In Pennsylvania in 1980, the state asked a board of examiners to evaluate the Votomatic for possible use in the state. "I thought it was a joke," recalls one of the examiners, Michael Shamos, a computer scientist at Carnegie Mellon University in Pittsburgh.

The company that wanted to sell the system to Pennsylvania counties pointed to its approval in other states and insisted the technology was trouble-free. But the examiners recommended rejection. Shamos says it was "vulnerable to tampering in a multitude of ways."

The Pennsylvania secretary of State ignored the panel's recommendation and approved the machines for use in the state, but few counties adopted them, steered away by the examiners' negative assessment.

A few states learned from their bad experiences with punch cards. Massachusetts, Oklahoma and Wisconsin abandoned the technology after foul-ups sapped the public's confidence.

Ironically, one of the worst experiences came in Florida, in 1988 — about the same time as Saltman's warning. Democrat Buddy MacKay lost a Senate race that year by less than 1 percentage point, about 33,000 votes out of 4 million cast. In the state's largest county, Miami-Dade, about one in five voters failed to vote in that race. MacKay blamed punch cards.

As the nation embarks on a re-examination of its voting systems and contemplates spending hundreds of millions of dollars for new equipment, some of Saltman's other findings might be cautionary.

In 1975, Saltman observed that few states had taken responsibility for overseeing their election laws, leaving the responsibility to local governments, which in turn were handicapped by lack of money, training and standards. That has changed little in 25 years.

His words then were prescient: "The unanswered question, of course, is whether the states and local communities will have the good sense to reform their election practices where needed. If they do not, Congress will surely be tempted to do it for them."

Elections, he wrote, should be treated like any routine financial transaction. Americans make bank deposits, withdraw money and charge purchases with little worry under systems that have internal controls to prevent errors and fraud. But most election offices lack the expertise to implement such procedures, Saltman says, and the federal government provides almost no guidance.

The government collects data on health threats, traffic safety and air safety, and it watches out for dangerous products, he notes. "But when it comes to democracy, which is almost as important, we don't do a damn thing."

© Copyright 2001 USA TODAY, a division of Gannett Co. Inc.