Touch-screen machines: riskier than hanging chads?
By Jon Mills
MY VIEW Tallahassee Democrat 27 September 2004
Computers are perfect. How else can opponents justify the resistance to individualized records in current touch-screen voting machines?
With no records, there can be no recount, and with no recount in touch-screen counties, the specter of unequal treatment of voters, so painfully played out in the 2000 presidential election, could again rear its ugly head.
Actually, computers aren't perfect. In 2000, the Florida presidential recount revealed virtually every way an election's result could be made inaccurate:
• Undervotes. This means that no vote was recorded. Where the punch-card ballot was used, those ballots were evaluated to determine whether the chad was hanging. There are two underlying causes of undervotes, the machine not recording a vote, or the voter, either intentionally or unintentionally, failing to record a vote.
• Overvotes. This problem recognizes that the old Chicago advice to "vote early and often" is illegal. You get one vote. And, if you vote more than once in the same election, your vote won't count at all. Here, human error is the culprit, with a voter voting for two candidates running for the same office.
• Machine malfunction. Florida 2000 provided several examples. Punch cards and hanging chads have come to personify mechanical and design defects. These defects allow persons to vote twice, vote for two candidates, and sometimes prevent the voter's intent from being ascertained at all (also known as the hanging or pregnant chad). Also, the distribution of malfunctioning machines was not equal. Majority black precincts saw 10 percent of their ballots discarded, compared to 3 percent in other precincts.
Enter the touch-screen ballot. These computers allow voters to record their vote simply by pressing the screen. This information is then stored as numbers indicating the overall tally for an individual candidate. Advocates of this device claim it more accurately counts votes, preventing the problems of 2000. But there are no perfect computers. Although touch-screen ballots should make overvotes impossible, a Division of Elections report recognized that undervotes occurred more with touch-screen ballots than with optical-scan ballots (where one fills in a bubble to indicate their vote). In fact, in the 2002 primaries, touch-screen ballots undervoted 0.92 percent of the time, compared to a 0.33 percent undervote rate with optical scan-ballots.
Touch-screen ballots not only undervote more often than optical scanners, but they also lack any mechanism to either ascertain voter intent or recount votes. Consequently, in the event of a recount, there are no undervotes recorded on touch-screen ballots to be counted, while undervotes recorded in counties using optical scan ballots can be viewed. This inability to recount touch-screen ballots causes two distinct problems. First, if there is computer error or hacker infiltration and votes are lost or tampered with, how are we ever going to know? And if we do know, what do we do without any record? Second, state law requires that, in the event of a margin of victory under 0.25 percent, the undervotes and overvotes of the race's entire geographical area must be manually recounted.
If the votes of Floridians using optical-scan ballots are recounted, and the votes of Floridians using touch-screen ballots aren't, there is a major problem. The U.S. Supreme Court's basis for ending the 2000 manual recount was the unequal treatment of Florida voters because of differing recount standards. You would think Florida, of all states, would be in the forefront of making every vote count. Instead, we are now faced with a system that would recount some counties' votes, while relying on the perfection of touch-screen ballots in other counties. Of course, if the election isn't close, there is no manual recount. And if there is no manual recount, Florida will have dodged a large bullet. But all signs point to another close election.
That is why some paper trail anonymously indicating voter intent is necessary. Nevada voters, for example, cast their primary ballots this year on touch-screen machines that gave paper receipts through ordinary printers. DeForest B. Soaries Jr., chairman of the U.S. Election Assistance Commission, said voters "actually enjoyed the experience."
Administrative Law Judge Susan Kirkland invalidated an attempt by the Florida Department of State to amend its rules to exclude touch-screen ballots from manual recounts. She stated that this rule ran "contrary to the plain language" of Florida's recount statute, which requires manual recounts of all overvotes and undervotes in close elections.
Floridians deserve and are legally entitled to a recount standard that assures every vote is accurately and equally counted. If this election is close, Florida and Florida's "perfect computers" will be under the microscope.