A vote glitch casts doubt Keep testing touch-screen details

The latest touch-screen "glitch" report detailing inconsistencies in a certain data procedure involving iVotronic machines is disturbing.

Though the flaw appears to be limited in scope, it inflames doubts about "black box" balloting, in which votes are recorded electronically rather than on paper.

The glitch is troubling because it escaped detection during Florida's certification process. Voters should rightly wonder: If regulatory safeguards failed to catch this problem, what other surprises may lie ahead?

The anomaly, discovered by Miami- Dade officials in 2003 but only recently aired in the news, concerns the ability to obtain an accurate record of the iVotronic's vote-by-vote "event log," contained in the memory of each machine. After an election, the staff extracts this data and stores it for audits and future reference.

There are two state-certified ways to gather this information from the iVotronics (the system used in 11 Florida counties, including Sarasota and Charlotte). The Miami-Dade experience shows that the faster of the methods in which the vote images are "read" off flash memory cards can be vulnerable to the glitch. When Miami-Dade analysts tested this alternative on several units following close elections last year, they found an inconsistent record of the votes. (We were unable to verify a precise description of the problem, due to conflicting interpretations among analysts, county officials and spokeswomen at the state Division of Elections.)

The other method used in Sarasota has no history of glitches, according to the Division of Elections. By this means, which is too time-consuming for larger counties, the event log from each machine is downloaded through a port onto a laptop computer. (The data can then be stored on CDs.)

The iVotronic manufacturer, Electronic Software and Systems Inc., reportedly is working on a cure for the faster method. That's good to know, but there isn't much time for staff in the counties that use this method to learn and implement a new procedure by the Aug. 31 elections.

This frustrating situation reinforces concerns about the unknown vulnerabilities of touch-screen voting. It's all the more troubling because this anomaly, limited though it may be, involves the fundamentally important record of actual votes for which there is no "paper trail."

Providing one by Election Day is impractical. Long-range solutions must be developed, however. In the meantime, intensive, independent testing of touch-screen systems should be accelerated. Overconfidence in the technology seemingly endemic among elections officials is a luxury that democracy can ill afford.