From the
Miami Herald
Touch-screen voting fails to ease fear of more blunders
Once hailed as the solution to Florida's much-maligned voting system, electronic voting machines have introduced a host of potential new problems, reformers say.
ELECTION 2004
Aug 17, 2004
By JOE MOZINGO and ERIKA BOLSTAD
Since Miami-Dade and Broward counties purchased the iVotronic touch-screen voting system almost three years ago, a gubernatorial primary was botched, the cost of running an election has soared and the notion that every vote counts seems less a certainty than blind faith.
Now as more states around the nation have turned to this electronic equipment, voters have become increasingly concerned that the machines could be manipulated to skew an election, or that a glitch could cause votes to be lost.
And for obvious reasons, they fear it will happen in South Florida.
Election supervisors say they have done everything possible to smoke potential glitches out of the electronic cracks. Testing of touch-screen machines this year has been far more extensive -- and open to the public -- than ever before.
But reformers still are demanding officials do more.
How many times can the state of Florida say `oops' ?'' said Bobbie Brinegar, President of the League of Women Voters, Miami-Dade County.
...
A ROUGH BEGINNING
The machines made their big debut in the September 2002 primary. The event was a fiasco in Miami-Dade and Broward, where last minute changes and poor poll-worker training left untold numbers unable to vote. Officials are confident those problems have been fixed.
Reformers say there are key measures election officials should take to prevent reliving an ugly piece of history. Foremost among them is election-day testing of the machines.
They point to states like California, which performs such testing. Experts say the so-called parallel testing is the best way to make sure the new technology doesn't contain a flaw that could taint an election.
''It's the best thing we know how to do,'' said Douglas Jones, a computer technology professor at the University of Iowa, hired as a consultant to oversee Miami-Dade's elections.
The concept is easy: Randomly selected machines are taken offline and used only for testing. Participants, whether they are citizens or officials or observers, pick the candidates they want. Someone takes notes or videotapes them touching the choices.
At the end of the day, officials simply tabulate the votes recorded on the test machines and see whether they match the voters' choices.
California conducted parallel tests in eight counties during its March primary. Participants cast about 4,300 test votes. The equipment recorded the votes as cast with 100 percent accuracy.
TESTING CRITICIZED
Florida conducts advance tests of its machines. Last week, Miami-Dade, Broward and Palm Beach counties tested 418 machines, finding they counted with 100 percent accuracy.
But experts say testing on election day is critical. Theoretically, a devious programmer could design and hide a malicious bit of code in ways difficult to ferret out.
''Imagine yourself as a programmer developing the Fraud-O-Matic,'' Jones said. ``One of your jobs would be to make it absolutely honest when it's being tested, and absolutely dishonest when it's not.''
To avoid such a scenario in Miami-Dade, County Commission Chairwoman Barbara Carey-Shuler ordered Elections Supervisor Constance Kaplan to look into parallel testing in the primary.
Kaplan has worried that it would confuse voters to see people cast votes that are not actually counted. She is now considering such tests.
...
The Florida Secretary of State's office thinks advance testing is sufficient. ''Our machines in Florida go through the most rigorous certification process in the nation and then they go through the testing,'' said spokeswoman Jenny Nash.
...
A more insidious problem, in many critics' view, is that no one really knows how well the machines have worked in other elections because election supervisors have not taken full advantage of what little data there is to ensure accuracy.
The iVotronics record not only the vote totals, but data that could be used for an audit of the system's performance. Yet officials have not regularly analyzed this data. Miami-Dade even lost it -- then found it.
...
MYSTERIES LURK
The audit data contains records of every vote and every action taken on an individual computer. By looking for discrepancies between the audit data and the results, one can uncover potential problems. Prodded by Carey-Shuler, Kaplan now says she will begin doing this.
Even with the audit data, voters must sometimes rely on conjecture when mysteries surface -- such as the 134 empty ballots cast in Broward County for the House District 91 race this January. With no other race or issue on the ballot, the question is why 134 people would go to the polls, sign in, pull up a ballot, decide not to vote on the race -- but still cast the ballot.
And there is also the concern of extra votes, especially in Miami-Dade, where the process of running tapes that show the machines are vote-free takes place the day before election day. Reformers fear someone could illicitly cast votes overnight. But Jones, the expert, discounts this concern. The machines are locked and sealed overnight, and poll workers are instructed to make sure the machines have no votes when the polls open.
All in all, Jones said, Miami-Dade is ahead of the country in its preparations. ``Even if Miami-Dade doesn't do everything I hoped for, they are working seriously to set the standard.''
© 2004 Herald.com and wire service sources. All Rights Reserved.