Election Updates

Several states are either developing legislative election reform packages for the 2006 legislative sessions or have issues on the ballot in November. I have excerpted from several articles below on this issue and we will continue to update these and other reforms over the next 6 months, when legislative activity is at its peak.

Wisconsin is considering legislation whereby voter registration groups could no longer pay employees based on the number of voters they sign up and the state would require training for county clerks and poll workers. One key legislator said that lawmakers left the photo ID provision out of the package while they pursue a constitutional amendment that would require it. He said calls to end same-day registration in Wisconsin would also be considered separate of the package.

In Ohio, a group called the Ohio First Voter Education Fund has hired the ad agency that brought you the Swift Boat Veterans for Truth ads against Senator John Kerry to fight election reform in Ohio. The reform package in Ohio, which will be voted on by citizens this November, consists of four constitutional amendments. The four proposed amendments, backed primarily by Democrats and unions, would make it easier to cast absentee ballots, reduce caps on campaign contributions, change the way the state and federal legislative districts are drawn and create a bipartisan election oversight commission.

In California, the San Francisco Chronicle reported that Proposition 77, which takes the process of drawing legislative boundaries out of the hands of lawmakers and gives it to three retired judges, is trailing, with 50 percent saying they will vote "no" and just 36 percent saying they will vote "yes."

Today, the GAO issued a 120 page report, Views of Selected Local Election Officials on Managing Voter Registration and Ensuring Eligible Citizens Can Vote (abstract or full-report). This report is a follow up to a June 2005 report on voter registration. The report abstract discusses the difficulties in addressing the volume of voter registration applications that are submitted before the election. The abstract states:

Local election officials representing all but 1 of the 14 jurisdictions GAO surveyed after the November 2004 election reported facing some challenges processing voter registration applications and took steps to address them. Processing applications received from voter registration drives sponsored by non-governmental organizations posed a challenge to election officials in 12 of the 14 jurisdictions, while half of the officials reported challenges receiving applications from other external sources, such as motor vehicle agencies. Challenges occurred in processing these applications for reasons such as incomplete or inaccurate information on voter registration applications. Half of the officials reported that their offices faced challenges checking applications for completeness, accuracy, or duplicates, citing, among other things, insufficient staffing to check the applications. Steps taken by election officials to address these and other challenges included hiring additional staff to handle the volume of applications received and contacting applicants to get correct information.

All but 1 of the 14 election officials reported that, using various sources of information, they removed names from voter registration lists during 2004 if, for example, voters had moved, were deceased, or were ineligible due to a felony conviction. To help ensure names of eligible voters were not inadvertently removed from voter registration lists, officials reported contacting voters to confirm removal, matched voters’ identifying information (such as name and address) with address changes provided by the U.S. Postal Service, and matched voter registration records with felony or death records. GAO reported in June 2005 about problems officials in these same jurisdictions experienced verifying voter information with death or felony information from existing data sources.

GAO’s survey showed that all 14 election jurisdictions permitted citizens to cast provisional ballots during the November 2004 election. HAVA gives states discretion to implement provisional voting based on state voter eligibility requirements. According to the election officials surveyed, about 423,000 provisional ballots were cast in 13 of the 14 jurisdictions, and 70 percent of those votes were counted. Also, 8 of the 14 jurisdictions reported challenges implementing provisional voting, in part, because some poll workers were not familiar with provisional voting or staff did not have sufficient time to process provisional ballots. To address these challenges, election officials in these jurisdictions said they hired extra staff or provided training to poll workers.

Fellow international election observers Sean Greene and Dan Seligson of electionline.org (here shown in a lighter moment in Buenos Aires' Cafe Tortoni) have written a comprehensive overview of the Buenos Aires e-voting pilot project in this week's electionline.org newsletter, "Voting Machine Prototypes Put To Test in Argentina."

If you are interested in reading more about this interesting pilot test, please read Sean and Dan's piece. They have done a through job of covering this pilot project.

I'm now in the midst myself of compiling a report based on my qualitative analysis of the project, and am struggling with trying to organize the hundreds of photos that I took during the pilot test. I hope to distribute my report soon, and will of course let readers of Election Updates know when it is available.

Here is a picture, if you are interested, showing what the Buenos Aires e-voting pilot project looked like on election day, in one of the school voting sites. Voters checked in at the table on the left, moved forward to learn about the voting system they were using (booths number 1-4), then proceeded to the second row of booths (far right side of the picture) to cast their mock vote in this particular location. See the two guys in suits, in the upper left of the photo? The one on the left (dark hair, partly behind the head poll site judge in this e-voting pilot site) is Sean Greene, on his left is Dan Seligson. Here is a close-up of a woman undergoing voter education. Last, here is a shot of a man with his son in this same pilot location using one of the electronic voting machines. Many more photos coming to this location soon!

But one editorial comment here is in order. As Thad and I have been arguing for some time, we think it is imperative that election officials engage in a thoughtful and rational process of voting system development and testing. This is just not something that we generally in the United States have done a good job doing, and we are increasingly seeing example of other nations (for example, the United Kingdom and Switzerland) developing widespread research and testing programs before their adoption of new election administration procedures and voting technologies. Argentina, through the Buenos Aires pilot project, is joining that elite club of countries or localities throughout the world who are trying to use a research-based approach for improving their election process. We hope that election officials, researchers, and advocates in the United States learn from the research that our colleagues abroad are providing, especially how to use a research-based approach for improving elections.

The Estonian Internet voting trail has ended and the government has posted a series of PowerPoint presentations about the trial. The first PowerPoint presentation gives an overview of the electoral system in Estonia. Here, several facets of the electoral process are quite interesting.

• Elections to the Presidency, the national parliament, and the European Parliament are on different schedules. The President serves a 5-year term but the national parliament is on a 4-year term.
• There is a clear governance structure at the National, county, and polling station levels for overseeing the elections. (Also, Estonia has about 650 polling places, or about 8 times fewer than exist in Los Angeles County).
• Estonians disenfranchise individuals in prison and also seem to have a form of felony disenfranchisement, but it is discretionary, not compulsory.

Procedurally, the election is different as well.

• Critically, the state registers voters. Voters are sent their polling place cards 20 days before the election.
• The state does early voting, but much more broadly than we do on the U.S. For the 13th-9th days before the election, there is one early voting site per county.
• For the 6th-4th days before the election, every polling place is open for early voting!
• Voters can also vote outside their polling place of residence in early voting.
• Internet voting was allowed during the advanced voting period.

The second PowerPoint presentation examined the Internet voting system used in the election. We reviewed this in a previous blog entry, but it would likely be of interest to those who want to know what a very high-level Internet voting architecture description looks like.

The final PowerPoint examines e-voting in Estonia from a social and legal perspective. Several findings are quite interesting.

• First, the penetration of the Internet in Estonia is not as high as it is in the U.S. 41% of Estonian households have a computer, with 82% of these computers connected to the Internet. By contrast, 61% of households in the US have computers and 88 percent of those computers are internet-connected.
• Not surprisingly, support for e-voting is highest among those who use the Internet daily and lowest among those who do not use the Internet. Only 13 percent of Internet users are against e-voting, compared to 26 percent of non-Internet users.
• A survey of Estonians in both 2004 and 2005 found that support for poll site voting declined from 2004 to 2005. However, those who were sure they would prefer to e-vote stayed the same. (The switch was in two questions giving both options but preference for one over the other. There e-voting scored well in direct comparison to poll site voting).
• Internet voting has very high support (64-66 percent) from those 15-34, but poll site voting has similar support (64-72 percent) from those 50 to 74.

In an editorial today in Roll Call, EAC Commissioner Ray Martinez wrote about the need for all electronic voting software to be filed with the National Software Reference Library at NIST. We have reproduced the opinion piece below. When Mike and I were in Travis County, we watched the county conduct hash code tests, comparing the voting system software they were using with the software on file at NIST. Unfortunately, not all voting systems currently support hash code testing because the software is not on file.



How to Ensure Accurate, Secure Voting Systems
October 25, 2005
By Ray Martinez III,
Special to Roll Call

Preparations are now under way for another busy midterm election. While much has been written since last year’s presidential election about the need for additional election reform, the Help America Vote Act of 2002, passed by Congress with overwhelming bipartisan support, is poised to deliver further improvements in the administration of elections in 2006.

Among the changes spurred by HAVA is the greater prevalence of electronic voting systems throughout the country. In 2004, approximately 30 percent of registered voters were able to record their candidate preference by using a computer-based voting system — a significant increase from only 12 percent in 2000. And yet the change to electronic voting machines has raised concern among some as to the accuracy and security of these systems.

The Election Assistance Commission, created by HAVA, will soon vote on a series of recommendations to improve the voluntary standards by which we assess the overall integrity of electronic voting systems, including the security capabilities of these systems.

One proposal the EAC will consider stands out as a significant step in the right direction toward ensuring greater security of electronic voting machines.

Since 2000, the National Institute on Standards and Technology, a well-respected government agency staffed by scientists, technicians and engineers, has operated the National Software Reference Library. Containing a collection of more than 7,000 software products in a secured room, the NSRL provides law enforcement personnel, including the FBI, with important data used to identify unknown and suspicious files on computer systems and to meet the need for court-admissible evidence in the identification of software files.

Realizing the potential use of this application to the election process, last year the EAC called on all voting systems vendors to voluntarily submit their proprietary voting system software to the NSRL to create a similar repository for state and local election administrators. The voting system vendors agreed.

Today, the NSRL contains proprietary code and software for most types of electronic voting systems used in this country. In the coming weeks, the EAC is expected to vote on a series of proposed final standards which, among other things, would require that all voting system software, including installation programs and third-party software, be deposited with the NSRL upon completion of a national voting system certification process, in which 41 states currently participate.

This means that a local election administrator will be able to verify that the operating software installed in the election management systems used in that local jurisdiction is exactly the same as the software for that particular system that was certified by an independent testing authority and deposited with the NSRL. Additionally, any irregular or suspicious files could be identified when a local election administrator utilizes the NSRL.

To be sure, more work remains to be done to fully utilize the capabilities of the NSRL. For example, critics note that while the NSRL does provide verification means to compare software installed in a voting system with the relevant software contained in the NSRL, such verification is not as effective if it is performed internally by the voting system, and it also can be difficult to manage when last-minute patches are added. The experts at NIST, along with election officials, are working to overcome these concerns.

And yet, despite these obstacles, even the most avid critics of electronic voting systems concede that the use of the NSRL in the election process is a practical step toward providing added security measures for electronic voting machines.

Like election administrators, the EAC will continue to seek innovative solutions to improve the mechanics of our great democracy. The EAC can bring information and resources previously unavailable to state and local election officials so that all Americans can have greater confidence in the fundamental fairness of our election outcomes. The NSRL stands ready to serve as but one example of such success.

Ray Martinez III is one of four members of the Election Assistance Commission.

Copyright 2005 © Roll Call Inc. All rights reserved.

The recent GAO report on electronic voting has a great figure in it that is very important in the ongoing discussions about e-voting. Figure 1 on page 7 of the report notes that elections are a multi-stage process starting with (1) voter registration, then (2) early and absentee voting, (3) election administration/vote casting, and (4) vote counting and certification.

At each point, an election process--one of the four items noted above--is affected by two interactions. First, there is a human interaction, when people--voters, election officials, observers, and third-parties--enter the process. Second, there is a technological interaction--with voter rolls, ballots, and ballot counting systems (human or mechanical). As Mike noted in his blog from yesterday and on Saturday from Argentina, the legal and cultural norms of a community also interact here as well, but the GAO focuses on the human, technological, and process interactions.

It is this interaction that makes elections complex and difficult to make perfect. As election reforms are proposed--from expanded absentee voting to paper trails and vote centers--it is important to consider how each change affects both the technological, process, and human complexities in elections. Reforms that ease the stress on any given point in the system--for example, by spreading voting out over multiple voting channels (early, absentee, precincts), allowing for voter registration problems to be addressed easily (e.g., election day registration), and improving vote counting and canvassing (e.g., manditory post-election auditing) may be reforms that make elections less problematic in the long-term.

Since arriving in Buenos Aires last week, the local election officials here have kept the observing team quite busy. Other than some time off on Saturday afternoon, it has been a whirlwind visit, especially on Sunday, election day here in Argentina. Our agenda is given here for those interested in seeing what it looked like.

The primary purpose for our visit was to evaluate and examine the e-voting pilot project in Buenos Aires. The e-voting pilot project had been planned for Sunday's elections in Buenos Aires, to be conducted in 53 different voting locations in yesterday's elections. However as the day progressed, we heard that in at least 9 voting locations the local judges had decided to not allow the pilot project to proceed in their location. Still, the initial report we heard late last night was that approximately 14,000 voters had participated in the pilot project. Here's a photo of a streetside sign ( slightly blurry closeup here) advertising the pilot project in Buenos Aires, one of the various ways that the project was advertised in the city.

Most of our time in the past few days has been spend winding our way around Buenos Aires in a tour bus, though on election day we were broken into smaller groups, and spend the day in a smaller and much less comfortable ride! But as Buenos Aires is a beautiful place, and as our mission involved a number of interesting people, we had a very good time traveling from event to event, polling place to polling place.

I'll cover the e-voting project in more detail later, it was very interesting to observe and I believe that we --- and the local team who conducted the project --- learned a great deal about how e-voting might work in Buenos Aires.

But here are some important details about how the voting process works in Argentina, including some of the insights I learned over the past few days.

First, in Argentina voting is compulsory, and elections like these are conducted throughout the day on Sundays, primarily (in Buenos Aires at least!) in public schools. Voters show up to the polling place, and scan posted lists of registered voters for their name and specific voting location in the polling place. Here are voters looking for their names on lists posted outside a public school, "ESC No. 25 Bandera Arg.-EMEM No. 6 Padre C. MUJICA", located at PJE EEUU DEL BRASIL y Av.Ant.ARGENTINA (this is the information as given to us for the location). Note the pink and blue sheets --- the men are listed on the pink sheets, the women on the blue sheets. In many of the polling locations we visited, men and women voted separately, sometimes with the women voting on the first floor of the school and the men voting on the second floor.

Next, here is a typical school voting site in Buenos Aires (located in the same school). Note the "voter bill of rights on the walls" ("A Los Votantes"). The voter approaches the appropriately numbered "mesa" (table) and the polling place workers check off the voter's name against the list of eligible voters for that location. The voter's identification book is then stamped to show that they voted. The voter then stands in line in from of a school room door, and one-by-one they enter the school room. Inside the school room are the ballot lists for each party, for each race. The voter selects the lists he or she wants to support for each race, puts those in an envelope, and brings the envelope back outside and deposts the envelope in the cardboard box on the table. At the close of voting, the workers here count the ballots for each race, and provide the initial count to the tabulation center in each jurisdiction.

According to what we were told, working in a polling place is sort of like jury duty in the United States; workers are randomly selected, sent a letter instructing them to show up for service, and that is how they recruit polling place workers. I did not get any data on how effective this practice is, but would love to receive any data or research on this procedure that anyone has.

But, like in the United States, polling places in Buenos Aires have huge variability in layout and in the physical facilities. We went to another polling site on Sunday that was in one of the most massive spaces I've ever seen used for voting. Here are two different photographs of the interior of this polling place ( first shot and second shot), showing what I counted as 25 separate polling places!

This massive polling place differed from the others, in that here the voters were not going into a classroom to get their ballot papers. Rather, once the voter checked in at the right station and got their book stamped, they went behind a small plywood screen located next to the wall. We were given the opportunty by a very cooperative poll site judge to actually walk behind the screen and see the ballot papers, just as a voter would see them. Here is the best photo I could get of the ballot papers, showing two of the panels very well. Again, the voter would take the ballot papers for the party-office combination he or she wanted to support, put those in an envelope, and deposit the envelope in the ballot box on the reception table.

Based on seeing the existing process, and talking with people who adminster these elections or who participate in them, this existing process has some potential deficiencies. As we were observing regular polling operations in this voting location, we noticed that there were people being escorted by election officials or poll workers to behind these screens where the ballots were located. We later determined that these individuals were party representatives, who are checking to make sure that their party ballots are in sufficient supply for voters to use. Given that these party representatives can access the ballots, there is the very real possibility that they could engage in a number of strategies to help their own party's chances for election by moving their party's ballots to a more accessible and easy-to-find location, putting their party ballots on top of the opposition's ballots to make it difficult to find the opposition's ballots, or even removing entirely the opposition's ballots. Of course, voters themselves could do the same things when they are behind the screen or in the classroom by themselves.

As far as I am aware, the precise extent of these problems is unknown. But one of the hopes is that by using different types of voting systems, including optical scanning and electronic voting devices, that they could improve the process in a number of different ways --- one of them being to make it difficult if not impossible for some of these problems with paper ballots to exist in future elections. Of course, as data becomes available from this e-voting pilot project we will be in a much better position to gauge the extent to which new voting technologies and procedures can improve this process, and at what cost.