Election Updates

I submitted another short paper on voting system threats to the NIST "Threats to Voting Systems" workshop organizers yesterday, on "Potential Threats to Statewide Voter Registration Systems." As many of the threat reports that were submitted for the NIST workshop were focused primarily on ballot casting and tabulation, I thought that it was important to point out that both the new HAVA-compliant statewide voter registration lists might introduce potential new threats to voter registration systems that had not necessarily been discussed. This follows the short paper that I sent to the NIST organizers earlier this week on threats to precinct voting.

In my discussion of potential threats to statewide voter registration lists, I discuss four general potential threats:

1. Authenticity of the registration file.
   
2. Secrecy of the registration file.
   
3. Integrity of the registration file.
   
4. Potential voter registration system failures.
   

By way of a conclusion, I point out that there are no standards for the new HAVA-compliant statewide voter registration systems, and thus there is no system in place for testing and certification to these standards. Of course, standards, testing and certification won't necessarily mitigate or eliminate all threats, but they could certainly help if devised and implemented correctly.

This morning electionline.org is highlighting a story in the Seattle Times, which reports that 18 ballots were apparently not counted in their recent September primary election. Apparently of these 18 uncounted ballots, 16 were found in a box (uncounted) and two absentee ballots were not counted simply becasue they were sent in some non-standard envelope which election workers seemed to fail to realize contained ballots (and thus were not counted).

What I found most interesting about this report was the quote from King County Vice Chairman Pete von Reichbauer: "I hope we don't get to the stage where we have to bring in outside observers to our election like a Third World Country."

What would be wrong with bringing in outside observers --- even if they were not having any administrative problems in King County? Why would having outside election observers make King County somehow like a "Third World County" in terms of election administration? If you want to hear more about the importance of outside and unhampered election monitoring --- in the United States --- listen to my latest podcast.

There is a new report just out from electionline.org, providing the results of a survey of state election directors of their ballot recounting procedures. As the report correctly notes, understanding ballot recounting procedures is increasingly important --- as ballot recounts are becoming more complicated (as there are a multiplicity of voting methods now used in most states, most typically early, absentee, provisional, and precinct voting) and more common.

One of the more interesting findings in the report is that ballot recounts appear to be very common in local level elections: "In responding to electionline.org, eight states described how recounts at the local level happen on a fairly regular basis" (page 5). For example, the report provided an estimate that there may be between 50 and 100 recounts after spring elections every year in local-level elections.

Additionally, the report found that some important differences across states in how voter-verified paper audit trails may be used in recounts:

In states that specifically use electronic voting machines with a voter-verifiable paper audit trail, (VVPAT) most but not all will use the paper ballot in a recount. Fourteen states have specified in their VVPAT laws that if there is a recount, the paper ballot, not the computer record, shall be used. In three states, California, Idaho and Nevada, the electronic record, not the paper ballot, will be used in a recount. Two states --- Maine and Missouri --- reported regulations are still in development. The remaining states have rules about paper ballots but do not use DREs or in the case of Hawaii do not have recount procedures in place.

Last, in addition to Hawaii, the report states that Mississippi appears to have no provisions for ballot recounts.

There is a lot more in this report about recounts ... thanks again to electionline for providing valuable data like this for those interested in election procedures and election reform!

Earlier this morning Rick Hasen pointed on Election Law to a story in the Sacramento Bee (registration required for content) regarding plans by California Secretary of State Bruce McPherson to set up a new Office of Voting System Technology Assessment. McPherson's office has also put out a press release talking briefly about this initiative. The new office is described as "a permanent, professional and fully staffed voting system technology and testing unit and will serve as a one-stop-shop for voting system certification."

As best as I can tell, in addition to providing some hints about the creation of this new office, the one new tidbit released yesterday is that Secretary McPherson "will not consider voting systems unless the vendors or products have ... (point 10) Agreed to volume testing to simulate Election Day use." Apparently Secretary McPherson is planning on continuing the type of testing done recently in San Joaquin County, and apparently also conduced last week in San Diego County (I've not received any information on how the San Diego County test was conducted, nor what the results were --- if anyone has any information on this, please pass it along!).

I've submitted one threat analysis for the workshop tomorrow on threat assessment, on a broad class of threats that I call "precinct voting denial of service." My short paper ("Precinct Voting Denial of Service" goes into much more detail on the various types of threats in this class of attacks, but here is a short teaser description:

The basic approach is that a perpetrator attacks precinct voting, regardless of voting system, on election day in an effort to disrupt the process sufficiently to produce an effective "denial of service" attack. The perpetrator, based on an analysis of past elections returns, would target selected precincts that are highly likely to cast votes in a certain direction. For example, if the perpetrator wished to sway the election for party X, he or she would target precincts that have very heavy concentrations of party Y supporters. In a close election, especially in lower-level races, such an attack could either sway the outcome of an election to party X or could throw considerable doubt and distrust into the announced election outcome."

In the paper I provide a series of examples of these type of attacks, some of which have occurred, some of which are potential threats.

This looks like an interesting workshop, both based on the threat analyses posted so far on the NIST site and because of the strong response they have received to attend the workshop; I've heard they are going to have a very large group (I've heard numbers in the range of 150 and possibly more?). Clearly NIST is pursuing an important topic here, and they have gotten us focusing research energy on developing better threat assessments for our electoral process, from start to finish.

Soon the academic journal Publius is going to release a special topic issue on "The States and the Help America Vote Act." Here is the table of contents of Publius volume 35, number 4:

• "HAVA and the States", by Sarah F. Liebschutz and Daniel J. Palazzolo (who are also guest editors of this special topic issue)
  
  
• "State and Local Government Organizations and the Formation of the Help America Vote Act", by Dan J. Palazzolo and Fiona R. McCartney
  
  
• "Implementing HAVA¹s Voter Education Requirement in Florida's Decentralized Election System", by Susan MacManus
  
  
• "Rational and Pluralistic Approaches to HAVA Implementation: The Cases of Georgia and California", by R. Michael Alvarez and Thad E. Hall
  
  
• "The Effect of HAVA on Late-to-Innovate States: External Influence on Election Reform in Arizona and Illinois", by Glen S. Krutz
  
  
• "The Implementation of HAVA in New York: From Antiques to High Tech", by Sarah F. Liebschutz
  
  
• "The U.S. Election Assistance Commission: What Role in the Administration of Elections?", by Robert S. Montjoy and Douglas M. Chapin, Jr.
  

This special issue is bringing some of the experts on HAVA's passage and implementation with some public policy experts.

A very early, pre-publication version of my contribution with Thad is available, though the paper that will be published has been polished a bit since that version was posted.

We'll let you know when the special issue of Publius is out!

There are an increasing number of papers and threat analyses available on the NIST website, for the workshop they are holding this Friday on voting system threat assessment. While I'm unable to attend what is shaping up to be an important workshop, MIT colleagues Ron Rivest and Ted Selker will be active participants in this workshop. Some of the papers and threat analyses are quite interesting, and as time permits, I'll have comments about some of them soon.

The New York Times ran the following story today on the dramatic population shifts in Louisiana in the wake of Hurricane Katrina.

Here is a direct quote from the story:

The numbers available now, however, are staggering. About 1.5 million people were initially evacuated from the damaged regions, roughly 1 million have applied for hurricane-related federal aid, 30,000 are in out-of-state shelters, 46,400 are in in-state shelters and 932 people have perished in the storms. Officials are unsure how many people are staying in hotels or with family and friends.

The story mainly focuses on the potential impact of these demographic shifts in the long run, with an eye to the 2010 census and the redistricting that will follow it in Louisiana. Of course, we've been focused more on the short-term impact of these population shifts, wondering whether the people who have left the heavily-damaged regions will be allowed to participate in upcoming elections, and how that participation will be administered. These estimates, though, do point how daunting the task of trying to administer any elections in the near future might be, say through widespread voting-by-mail.

Our fifth podcast is now available, focused on the issue of poll watching, or election observation. You can download the mp3 file here. You can also subscribe to our podcast by following the link on the left.

Hot off today's wire services: go to this url to see the original wire report.

Election Officials from Across the Country to Attend Mock Election in Vermont

SOUTH BURLINGTON, Vt., Oct. 4 /PRNewswire/ -- Representatives from the Defense Department and Los Angeles County will be among the election officials gathered in Vermont this week to consider a vote-by-phone system that was
first introduced by the state in August. They will be joined by members of the U.S. Election Assistance Commission, the federal commission responsible for helping to improve elections, at the first live demonstration of a voting
system that allows voters to cast their ballots using a regular telephone.

Officials from the Defense Department's Federal Voting Assistance Program are investigating whether the technology might be a good way to meet the needs of military and overseas voters. Election officials from Los Angeles County,
the most populated county in the country, are also considering offering voting by phone to their constituents.

"This is the best technology available right now, so I'm not surprised that the Defense Department and others are giving it a good look," Vermont Secretary of State Deb Markowitz said today. "The vote-by-phone system meets the broadest range of needs and has the most flexibility for new, future applications."

Vermont is the first state to announce that it will offer telephone voting in 2006, when it will make the new system available to every disabled voter in the state. According to IVS, the Kentucky-based voting services company that produces the system, several other states are also considering the technology.

According to Markowitz, Vermont will use system to meet accessibility requirements established by The Help America Vote Act, which gives states until January 1, 2006 to provide at least one voting system at each polling place that is disabled-accessible and may be used by visually-impaired voters in private and without help. The state may consider expanding its use of the system to include military, overseas and home-bound voters in the near future.

Vermont will hold the vote-by-phone demonstration and mock election Wednesday, October 5, in which 15-20 of the state's counties will participate. Election officials, advocates and members of the press are invited to observe
the election from the Sheraton Hotel in South Burlington.

Election Assistance Commissioner, Donetta Davidson; Director of the Federal Voting Assistance Program, Polli Brunelli; James Dickson, Vice President for Governmental Affairs with the American Association of People with Disabilities; Jennifer Collins-Foley, president of the Pollworker Institute; Gary Mudd, vice president of the American Printing House for the Blind, John Garguilo, of the National Institute of Science and Technology and election officials from Los Angeles, San Diego, Solano and Sonoma Counties in California will attend.

System creator IVS will also make mock election ballots available via a toll-free number beginning October 1. Anyone who would like to experience voting by phone may call 866-979-3900 to cast a mock ballot.

The vote-by-phone system was developed in cooperation with prominent disability groups, and has earned the praise of both national and state disability leaders.

At this time of year, when there are many elections going on around the nation, we often see many different examples of efforts by election jurisdictions to inform voters about how to use their voting technologies. Here are three online video examples.

The first is from Los Angeles County, California, showing how to use their early voting equipment (Diebold Accuvote). This is short and simple, and has some really jazzy tunes associated with it.

The second is from Orange County, California. It is a relatively lengthy video demonstration of their Hart eSlate voting system. Note the very cool Orange County scenes at the beginning of the demonstration video!

The third is from New York City, demonstrating how to use their lever voting machines. This one is dark and a bit antiquated, and has some interesting narration.

It would be a great research project, if someone were to download and archive these video presentations, and then do analysis of them to determine which are more effective at educating voters how to use voting technologies, and to then see if we could develop best practices for these educational tools. Undoubtedly, voters will be increasingly turning to presentations like these to learn more about the voting technologies they use on election day, which means that we really do need to strive to develop and utilize the most informative and usable online voter education tools possible.

If readers have other examples and favorites, please pass them along. We certainly appreciate examples at either extreme!

I have to admit --- I'm an avid user of "instant messaging" (IM), as many of my friends and colleagues know (and some of my students have learned), IM is a great way for us to communicate quickly and cheaply.

We are not the only ones who use IM. In a recent study, the Pew Internet and American Life Project found that in the United States:

• An estimated 53 million American adults use IM.
  
• On a typical day, 12% of Internet users IM with others (about 13 million using IM daily).
  
• Of those 18 to 27 years old, almost half said they used IM more frequently than email.
  

And these results are about a year old!

The question that numbers like these inevitably raise is how can this technology (possibly) be used to mobilize voters? Can it be used to get people to the polls? Can it be used for voting?

An example of how IM might be used for voter mobilization comes from a recent study presented at the American Political Science Association meetings. This study, by Sandra L. Suarez of Temple University ("Mobile Democracy: Text Messages, Voter Turnout, and the 2004 Spanish General Election"), examines how text messaging appears to have had a dramatic effect on voter mobilization in the recent election in Spain, and thus pointing out some of the potential of technologies like IM for getting people out to vote.

While Suarez does not present in her paper any conclusive data documenting a precise linkage between IM and voter mobilization in this election, she does present a series of smoking guns. First, she notes that "According to news reports, there were 40 percent more SMS messages (on election day) than on the average day. Information about the investigation (of the terrorist bombing in Madrid) and the protests was reportedly the topic of conversation among SMS users" (page 13). Second, the governmental party was upset in this election because the opposition party "received the support of almost 3 million more votes than in 2000. About half of the 3 million new (opposition party) voters came from former absentee voters, who tend to be young (ages 18-29) as well as left leaning. The remainder came from voters who became eligible to vote for the first time in 2004 (there were 603,711 new eligible voters), former (government party) supporters who changed their vote, and the so-called `tactical vote' from ... the former communists" (pages 12-13). Third, Suarez argues that the IM contacts, circulated repeatedly immediately before the election, constituted something like the personal get-out-the-vote stimulation that others (like Green and Gerber) have focused recent research on --- a stimulus at least sufficient to mobilize enough young and new voters to make it possible for the opposition party to defeat the incumbent party.

Thus, while Suarez's work does not give any generalizable analysis of how IM might necessarily be used for voter mobilization, it does show that technologies like IM, in certain situations and for certain populations, might sometimes be used to get people involved in the political process. We clearly need additional research, but more about that below.

As to voting, one set of studies that have been undertaken by the UK Electoral Commission have directly studied the use of text messaging for ballot casting. Based on these studies, the Electoral Commission concluded in it's 2002 report, Modernising Elections:

• "Anecdotal evidence in Liverpool, both from interviews with voters and from call center staff responding to queries, suggest that those using text-message voting were generally younger voters ... Electors clearly took advantage of the new multi-channel voting arrangements to choose the voting method with which they were most familiar and comfortable" (page 46).
  
• "For disabled people, SMS text messaging was in many ways the most simple of the electronic voting pilots. All the user had to do was send a string of numbers to an election phone number using their mobile phone. They would then receive a reply if their vote had been verified. Some disabled people, especially those with hearing impairments, use text messaging to communicate regularly and may have found this system especially useful" (page 49).
  
• "...the Commission is not aware of any substiantiated allegations of fraud or malpractice, and there is no evidence to suggest that the procedures provided for by the scheme led to any increase in peronation or other electoral offences, or in any other malpractice in collection with elections. Nevertheless, if public concern about the possibility of fraud were to grow, it could potentially undermine confidence in the use of electronic systems" (page 49).
  
• "These (electronic voting) pilots involved substantial investments of public money ... In the short-term, the pilots did not provide good value for money with regard to the cost per vote" (page 50).
  

Thus, these pilot projects in the UK show that technologies like IM appear to appeal to certain groups of voters, like younger and disabled voters.

So what do these studies imply for voter mobilization in the United States? First, we do need to be aware of the simple fact that technologies like IM are used much more frequently in Spain and the UK than in the US. Thus, at least in the very near future, the potential utility of technologies like IM for mobilizing voters in the US might be more limited than in other nations where IM is much more common.

But, technologies like IM may have great potential to mobilize those who use these technologies, and we ought to be studying their potential in exactly the sorts of field experiments that have recently become popular for studying other types of get-out-the-vote efforts. The important question for this research is whether IM messages might work as Suarez argued, having the effect of a personal contact to mobilize a voter, or whether it might be more like a telephone mobilization message, which Green and Gerber's research generally finds to not have the same impact on mobilization as personal and face-to-face contacts.

Third, specific technologies like IM can be used to help mobilize some populations of voters, for example, young and disabled voters. Research should focus on how these technologies can target specific populations, especially those that may not already be highly likely to vote.

The potential here is interesting, because IM messages aimed at getting recipients to vote could be sent, and then forwarded to people on the receiver's "buddy list", thus multiplying greatly the potential effect of IM-based mobilization strategies. IM get-out-the-vote messages could contain factual information about where to vote, or links to where that information is located. This technology may open up interesting new opportunities for organizations seeking to mobilize voters, for political parties and candidates for running for office, and for election officials.

I just received the following notice of what should be an interesting talk by Paul Herrnson, for those located near the University of Maryland:

Professor Paul Herrnson will deliver his Distinguished Scholar Teacher Award lecture,

Beyond the Hanging Chad: The Promise and Performance of Electronic Voting Systems

Wednesday, October 26, 4:00-5:00 p.m.
Marie Mount Building Room 1400
University of Maryland

Herrnson’s research, funded by the National Science Foundation and the Carnegie Corporation of New York, provides a comparative analysis of six electronic voting systems, including a prototype developed at the University of Maryland. The analysis, based on field tests and laboratory experiments involving over 1,500 voters, demonstrates that despite the fact that most voters expressed satisfaction with the systems, some shortcomings exist. Substantial numbers of voters felt the need to ask for help when voting and were unable to cast all of their votes as they had intended.

Factors related to the “digital divide” accounted for some, but not all of the differences. Certain types of ballot designs also increased the complexities of voting. Herrnson’s presentation will include a PowerPoint presentation of different voting machines, ballot designs, and voter responses.

The event is free and open to the public. Reception immediately to follow. All are invited.

Please RSVP to Mel Freitag, Center Coordinator, at mfreitag@capc.umd.edu or 301-314-2735 by October 14.