Monday, January 08, 2007

E-Voting Blues

Could 18,000 Floridians get confused and push the "finished" button before voting for their Congressman? Or fail to push the correct button to cast their vote when they had made all of their choices? Such scenarios could explain the anomaly of 18,000 undervotes in Florida's 13 Congressional district election. They are possible, if you credit the joint report of several citizen action groups who studied the reports of E-voting problems in the 2006 mid-term election. Their report, based on 1022 incidents in 314 counties, shows that voting is getting more difficult instead of easier as promised when electronic voting equipment is used. Equipment malfunctions ranged from individual machines to the malfunction of 800 machines in Pennsylvania's Westmoreland County. There were human failures too. Many voters, not accustomed to dealing with computerized equipment, had difficulty interfacing with e-voting machines. Poll workers also experienced problems troubleshooting malfunctioning machines, following complex or inadequate procedures, counting machine tallies, or verifying the accuracy of machine results.

The majority of problems reported were from touch screen (DRE-direct record equipment) machines. There were 59 reports of vote flipping with Diebold touch screen equipment. Diebold was the company that supplied the voting equipment in the disputed Ohio presidential election of 2004. In Sarasota, a poll worker said that voters who voted for the Democratic candidate lost their vote when the review screen appeared and they were forced to re-enter their vote to get the Diebold machine to accept their choice. Voters were concerned that every voter might not have made the extra effort to correct the malfunction upon review. Diebold models that produced paper tally records either produced tallies that did not match the screen or were unavailable for the voter to verify in 47 cases.

Optical scanning equipment performed better than DREs beside having the advantage of a paper ballot available for audits and recounts. However, there were scanning problems in 24 states, 79 counties with the ES&S M100 machine causing the most reports. The Help America Vote Act of 2002 mandates a recording error of less than 1 vote in 500,000[Sec 301(5)(a)]. In an Arkansas race for mayor, a man voted for himself. His wife said she also voted for him, yet the ES&S iVotronic machine only recorded 36 votes for the other candidate. America needs more help to get it elections functioning properly.

Equipment failures also had a substantial impact on voter participation. In 444 cases of DRE equipment failures, there were 221 reports of long lines or voters leaving without voting. In Denver, Colorado a media report stated that an estimated 20,000 people did not vote because of delays due to Sequoia ePollbook malfunctions. Voters in Florida, Ohio and other states reported standing in line for hours before they could vote. Making election day a national holiday would certainly take some of the pressure off pollworkers and voters.

With so many different modes of failure on different types of equipment, our election process must be systemically flawed and in need of a comprehensive fix. Comprehensive solutions using only the most reliable systems--including paper ballots--should be recommended to the numerous local election authorities responsible for conducting elections and purchasing equipment. As the system exists now, its suffers from a lack of uniformity, enforceable standards of reliability, and experienced, trained election workers. Vendors of electronic equipment play much too large a role in determining how America votes. They are selling a lot of expensive machinery, but not helping America vote.