Standards to stimulate e-voting?

Many voting citizens, whether they consider themselves red, blue or green, have been asking that question since the 2000 election shed light on how inconsistent, and often low-tech, the voting systems are in the United States.

Standardization of data fields, interoperability between counties and states, and an unwillingness on the part of local municipalities to embrace change are some of the major obstacles, according to panelists here at the Voter Identification/Registration Conference, hosted by the CalTech/MIT Voting Technology Project at the Massachusetts Institute of Technology on Thursday and Friday.

Politicians and election officials have been scrambling to put together high-tech solutions, but the potential for voter fraud, coupled with the exposure of security flaws in e-voting systems and voter databases, has slowed down the conversion.

The panel put aside legal issues concerning e-voting machines and, instead, concentrated on how technology could be used to ensure that electronic voter registration and identification is valid and consistently maintained.

Ann McGeehan, director of elections for Texas, Kim Brace of Election Data Services, and Thad Hall an assistant professor at the University of Utah who co-wrote "Point, Click, and Vote: The Future of Internet Voting," shared their experiences and examples of why, six years later, there is still no system in place.

One of the most basic problems across the states, Brace said, is trying to match and verify data when there is no standardization for reporting voter registration rolls or for constructing data fields. Some states keep both active and inactive voters on the rolls, some states retain only active voters, and some leave the decision up to individual counties. The differences lie not just between states, or between various departments of motor vehicles and voter registration systems, but between counties within the same state.

"These driver's license files are not as good as everybody thought they were," Brace said.

Names, for example, which should be broken out into first-, middle- and last-name fields, appear as one name field in many data sets, according to Brace. Suffixes like Jr. and Sr., and the modern use of hyphenated or two-word last names, has also added to the confusion. Some counties even collapse street address, town, ZIP code and state into one address field rather than breaking them out.

McGeehen has been overseeing the implementation of the Texas Election Administration Management (TEAM) System, a Web-based voting system that Texas counties can access through the Internet. TEAM, scheduled to be in place by the end of 2006, will electronically enable counties to add local election information to a state-provided ballot, have the ballot certified and send voting results.

The TEAM site allows citizens to confirm registration, get directions to their polling locations and find out what's on their ballot. If a voter moves within the state, TEAM will cancel the voter in the former county and register them in the new county, McGeehan said. But before any information can be imported into the system, the state has had to clean up disorganized data from existing motor vehicle and voter registries.

"In some counties, the data is 'dirty.' For example, my husband and I are registered at the same address and yet registered in different precincts," McGeehan said, referring to her own state's problems. "TEAM has stronger address standards that match the postal standards, so it will not allow for mistakes like that. But we can't deploy until mistakes like that in old data systems are brought forward, cleaned up."

In addition to the issue of bad old data, currently, there are no specific standards or guidelines for data exchange. HAVA, the Help America Vote Act" of 2002, did not include specific standardization recommendations.

For this reason, many states are taking a wait-and-see attitude before spending the money on a new system that may not be interoperable with others down the road. Like the high-definition DVD wars or the former VHS vs. Betamax debate, no one is willing to spend money on new technology because they are waiting to see what the winning standard will be, Hall said.

Currently, OASIS (short for Organization for the Advancement of Structured Information Standards) recommends something called Election Markup Language, or EML. The IEEE Standards Association, meanwhile, recommends XML (Extensible Markup Language). One of these two standards, if adopted, would enable states to have data exchange.

In Texas' TEAM system, all data is converted to an XML standard called EDX. While 254 Texas counties now favor TEAM's voluntary standard, 27 mostly urban districts have chosen to stick to their own systems and remain offline, McGeehan said.

According to all three of the panel members, getting districts to relinquish control over their current voting system is not just a political struggle for power. Many municipalities are reasonably concerned about spending money to implement a new system now and then having to change it down the road.

"If there were transaction standards, states would know how to handle the fields for names, and when a voter registered, you could transmit data from the state they moved from," Hall said. They could have fields asking, "Where were you last registered?" and then use that data as an affirmative reason for allowing states to legally move people from the old roll. It would allow a voter registration file to be updated in real time," Hall said.

"I am confident that three or four years from now, everyone will come online. Urban districts--they want to retain local control," McGeehan said. "They want to see how the new state system works before they put all their eggs in that basket."

Hall pointed out that this sort of massive undertaking of cooperation and standardization has already been successful in another arena.

"They did it in health care under HIPAA (the Health Insurance Portability and Accountability Act). Prior, there were 450 formats," Hall said. "Under federal law, they required one standard. Shockingly, they were able to get all hospitals and doctors and insurance companies on one standard in six years. It lowered administration costs, enabled instantaneous transmission of claims, improved security--since you didn't need third parties (to translate data)--and it improved health care management because it was easier to locate fraud."