How Florida's Voting Machine Failed (Again)By David F. Carr | Posted 2002-10-10 Email Print
Re-Thinking HR: What Every CIO Needs to Know About Tomorrow's Workforce
The first fiasco in the Florida vote was over paper. The second was electronic. How hard is it to get votes right?
Hanging chads are nothing compared to how untested technology can ruin an election day.
On Sept. 10, the state that produced the 2000 presidential election fiasco was back with a more thoroughly technological version. Having decided to switch to a computerized voting system, the state of Florida may have wound up teaching a primary-day lesson on how not to deploy a new technology.
Florida's largest counties, Miami-Dade and Broward, installed electronic voting devices on an unprecedented scale, spending $24.5 million and $17.2 million, respectively, for touch-screen machines from Election Systems & Software of Omaha. Yet the technology backfired badly enough to raise doubts about whether Tampa lawyer Bill McBride had really won the Democratic nomination for governor over former U.S. Attorney General Janet Reno. Some of the worst problems occurred in areas where she had the most supporters.
The political firestorm that followed obscured some underlying causes. In Broward, the county surrounding Fort Lauderdale, Supervisor of Elections Miriam Oliphant got most of the blame as a newcomer to this elected post whose inexperience contributed to poor planning. Some polling places didn't open on time because the workers she hired failed to show up, while others closed promptly at 7 p.m. despite an emergency order from Gov. Jeb Bush to stay open until 9 p.m.
Yet the touch-screen device chosen by Broward, ES&S's iVotronic, was actually picked by the county over Oliphant's objections. ES&S had an existing relationship with the county and made the low bid. But Oliphant favored Sequoia Voting Systems because it had more experience with touch-screen voting. Miami-Dade and 10 other counties also chose ES&S's iVotronic. Yet, on voting day, counties like Palm Beach that picked Sequoia's AVC Edge reported fewer problems.
Anyone planning to deploy a computerized device with a unique user interface and setup requirements should pay attention to this cautionary tale. Internal company politics can be just as obstructive, and sometimes rules decisions on vendor selection. Also, electronic voting systems have parallels with other specialized computing devices that are increasingly important to corporate systems: the kiosk in a mall or a cafeteria that must be usable without training; the inventory-tracking device on the receiving dock; the ATM that must be so reliable consumers will trust it with their money.
The unbending deployment deadline in this case was one of several extreme requirements. By banning punch-card ballots and mandating adoption of either touch-screen or optical-scanning systems in time for the 2002 elections, the Florida legislature forced counties into implementing new technology, ready or not.
"They talked to vendors, and they talked to other politicians, but they didn't talk to systems specialists," says Emil R. Phillips, Jr., head of systems development for Miami-Dade elections.
Unlike Broward's Oliphant, Phillips believes ES&S's iVotronic was the right choice, offering key features (such as an audio ballot for blind voters) at the best price. But it was a mistake to make poll workers, rather than technicians, responsible for starting and shutting down the machinesparticularly given the rushed preparations, he says.
ES&S was the vendor behind the older Votamatic system, in which voters manually poke holes in a computer punch card. Although this system had been used successfully for years, in 2000 problems developed when the punched-out bits of cardboard, the chads, built up in the hole-punch machines, preventing some voters from recording their votes correctly. Yet with proper maintenance and testing, that old technology might well have performed better than the touch screens this time around.
The iVotronic had never before been deployed on anything like this scale.