oftware tweak restores Broward 'yes' votes for Amendment 4

By Jeremy Milarsky and Rafael A. Olmeda   South Florida Sun Sentinel
Staff Writer
Posted November 5 2004

FORT LAUDERDALE ? A computing error discovered and corrected by Broward County officials early Thursday morning helped change the course of a statewide gambling issue.

Fixing the software glitch restored more than 32,000 "yes" votes cast for Amendment 4, contributing to the amendment's passage while sparking a demand for a recount. The demand was quickly rejected by state and local officials.

The glitch in Broward County's otherwise problem-free computer vote count was the result of human error: A technician failed to adjust the software to allow a greater vote tally than the limit set for smaller precinct counts, Supervisor of Elections Brenda Snipes said on Thursday.

The adjustments came in the hours before Broward County's canvassing board submitted preliminary election results to Florida's Secretary of State. The board, made up of Snipes, county Mayor Ilene Lieberman and County Court Judge Jerry Pollock, still has to finish its count of thousands of provisional and overseas ballots and a final count is due Nov. 13.

Candidates and committees have until that date to request any recount.

Total vote counts released on Thursday showed fewer voters turned out for this week's election than officials had predicted.

Broward County reported more than 700,000 voters cast ballots in the election, a turnout rate of 68 percent. Snipes had expected a 70 percent turnout. More than 175,000 of those people 1 in 4 voters cast ballots early at one of Snipes' offices around the county, and almost 100,000 voted by paper absentee ballot.

Of the total, only about 2,800 people cast so-called "undervotes" in the presidential race, declining to pick any candidate. In one case, a voter who filled out a provisional ballot ed George W. Bush for president, but crossed off running mate Dick Cheney and wrote in "John Kerry." The board treated it as an undervote.

Miami Dade County reported 775,465 people voted this year for a 73.24 percent turnout rate. Palm Beach County reported a total turnout of 454,427, for 62.29 percent.

Officials watching results at the Voting Equipment Center near downtown Fort Lauderdale first noticed the counting glitch when they saw vote totals on a Wednesday-night report of election results were actually lower than totals in a report printed hours earlier, despite the fact that officials had been counting paper absentee ballots all day.

"Clearly, it is a concern about the integrity of the system when the votes go down when more ballots are being fed into the system," Lieberman said.

Here's what went wrong, according to county officials:

While Broward County uses touch-screen machines for individual voting, the canvassing board has five machines designed to count paper absentee ballots. The computer software that generates a report of vote totals is designed to accept only 32,000 votes for each precinct.

Florida voters who opted to vote by absentee used several different ballot types and used ballots printed on multiple pages. Every one of those pages except the second contained localized races, such as a contest for a congressional seat, and therefore would stay under the precinct limitation.

But page two was the same statewide, so Broward's computer system counted each contest on that page state amendments 4 through 8 as four different precincts. When the absentee vote total for "yes" or "no" on each one of those amendments exceeded 32,000, the machine stopped counting them.

Once corrected, the county found more than 64,000 "yes" votes had been cast for Amendment 4, not just 32,000. Passage of Amendment 4 means county commissions in Broward and Miami-Dade counties can ask their voters in a referendum if they want to allow slot machines at seven tracks and jai alai frontons.

Lieberman said officials with Elections Systems and Software of Omaha, Neb. the company that manufactures Broward County's voting machines and software told her they noticed the software problem two years ago, after the fall 2002 election.

Hood spokeswoman Jenny Nash said the glitch could have been prevented if county technicians had prepared for it.

"When they're uploading the information, they have to set up a number of artificial precincts to accept all the ballots," she said. "It was a human error problem."

Broward Elections Supervisor Brenda Snipes later agreed, saying an ES&S technician had failed to add the artificial "precincts."

A spokeswoman for ES&S, Becky Vollmer, confirmed that the company knew about the problem two years ago but said technicians were working with county officials to prevent it from happening again and plan to ask state officials to certify improved software next year.

Lieberman sent a letter to Hood on Thursday asking for more information about the glitch and for state officials to approve any new software if they haven't done so already.

"This glitch needs to be fixed immediately," she said.

An attorney representing groups opposed to Amendment 4 asked the canvassing board for an official recount. However, board attorney Ed Dion said such a recount was not warranted because the software glitch only affected the report showing vote totals, not the actual counting of votes.

Officials said a similar glitch could not happen in Palm Beach County, which uses voting machines from a different company, or in Miami-Dade County, which uses similar machines but different software.

On the undervote, 2,817 Broward voters indicated no choice in the presidential race. That represented a .44 percent rate. Of those presidential undervotes, the bulk were recorded for people using touch-screen voting machines on Tuesday.

The undervote rate for those voters fell by nearly half compared with the March Democratic presidential primary, where almost 1 in 100 Broward voters showed up at the polls for a single-issue election and cast blank ballots.

Early voters fared a bit better. The county reported a .28 percent presidential undervote rate for those voters, and .41 percent for voters who cast paper absentee ballots.