Optimizing the Impact of the Internet to Assist Overseas and Military Voters
From Trust The Vote
U.S. citizens face significant obstacles to voting when residing outside of the U.S. or serving in the U.S. armed services. The challenges include obtaining proper information, voting materials, instructions and mailing addresses, as well logistical challenges encountered with postal mail systems. The right to vote and the implementation of the overseas and military voting program falls under the Uniformed and Overseas Citizens Absentee Voting Act (UOCAVA) and has recently been augmented through amendments in the Military and Overseas Voter Empowerment (MOVE) Act.
Contents |
The Ballot Transit Problem for UOCAVA Voters
Among the several challenges involved in overseas and military absentee voting, one of the most widely cited is in the transit time required for overseas vote-by-mail.
In the best case, voters who requested a ballot well in advance of an election, ballots are mailed to voters approximately 45 days before the election. For many voters, there is not enough time for the ballot to be transported to the voter, received, completed, and mailed back to arrive in time for election officials to count the ballots. In the results of the Overseas Vote Foundation 2008 Post-Election Survey, 52% of voters who wanted to vote but couldn’t claimed this was due to late arriving ballots. The Pew Center on the States performed a study of military and overseas voting, and reported (in No Time to Vote: Challenges Facing America’s Overseas Military Voters, January 2009) that half the states do not give their military and overseas voters time to vote and have their votes count, and that 9 more states’ voters are at risk or must forgo ballot anonymity to fax or email their marked ballot. Even for those voters that should have enough time in a 45-day window, the actual experience may vary with a number of factors such as late ballot revisions, slow overseas mail delivery, changes in locale of deployed military voters, and more.
The Ballot Transit Solution: Internet Voting?
To address this transit-time problem, Internet voting is one approach frequently discussed as a possible alternative to vote-by-mail. This term "Internet Voting" can mean many different things and must be looked at in context. When considering UOCAVA voting, one typical meaning of Internet Voting is a process that includes:
- The use of the public Internet to deliver a blank electronic ballot to voters, which they receive directly on their own personal computer;
- The ability to access the ballot online, and vote online from their computer; and
- Submission of the completed ballot via the Internet to election officials in the U.S.
A common form of this scheme allows voters to use their PC Web browser to access a Web site that has an i-voting Web application, which the voter uses to access, mark, and cast an electronic ballot.
This approach has some significant difficulties in the areas of privacy, anonymity, security, and barriers to fraud. As a result, this method cannot create parity with local voting or vote-by-mail on any of these crucial aspects. However, there is an alternative method that avoids these difficulties, that delivers some of the benefits of Internet voting, and also fully addresses the transit-time problem for a great many UOCAVA voters. That alternative is the use of the Internet to electronically deliver to a UOCAVA voter their vote-by-mail ballot to be printed, marked, and returned by post or express mail to election officials in the U.S., at the address provided with the printable blank ballot. Electronically delivered blank ballots can speed the “time to vote” factor substantially. Given the time-frames documented in the Pew report, there is ample time for a paper ballot to make a one-way journey from the UOCAVA voter to election officials, in time to be counted. The below sections describe the difficulties of Internet voting for UOCAVA voters, and how electronic distribution of blank paper ballots not only avoids these difficulties but also may provide valuable experience in how to address them.
The Top Three Challenges for Internet Voting
Leaving aside election law and policy issues related to Internet voting, there are several technological challenges. A common thread for all of them is the technology/policy issue that might be referred to as “parity”: one goal for Internet voting might be to have rough parity between the risks and benefits of Internet voting, as compared with the risks and benefits of vote-by-mail voting or in-person voting. Even trying to define such parity – much less turn the definition into technical requirements – is challenging because U.S. elections are locally administered, with a great variety of administrative practices and voting methods used across the country. One solution would not offer parity for all the election jurisdictions in the country.
One very significant technical challenge for Internet voting is the integrity of the technology used to implement the i-voting process. In the “home voting” version of Internet voting (the one most commonly considered in the context of UOCAVA and MOVE Acts), there is a very significant risk from security vulnerabilities of a voter’s home PC being used as a voting device. For any Internet-connected home PC, there is a significant risk of the PC being compromised, or becoming compromised, by malicious software; likelihood is significant, and detection is difficult for most technology consumers. As a result, it would be difficult to gain some assurance that a given home PC, being used for i-voting, has two critical properties:
- The i-voting software running on the PC is free from tampering;
- The voter information and marked ballot information are free from interception and modification on the PC, before being sent over the Internet.
Without these assurances, a voter’s i-voting experience could be abused to invisibly modify their votes or mis-appropriate their right to vote.
A second key challenge is the authenticity of the i-voting process. That process must include a check-in step, like that in a polling place, to determine that the voter is eligible, and has not voted already in the current election. Doing this check-in over the Internet requires that each UOCAVA voter have some electronic credentials that allow them to use the i-voting system. Therefore, a pre-requisite is the secure distribution of such credential to the voters, without undue risk of their right-to-vote being mis-appropriated by others. There are several technical approaches, but each has varying risk, feasibility challenges, and policy issues. For example, suppose the existing UOCAVA vote-by-mail paper ballot distribution mechanism is used to distribute a userid and password to each registered UOCAVA voter. Just one policy issue would be whether it is acceptable for a voter to voluntarily share this password (or fall prey to phishing) so that others could use that voter’s i-voting capability.
A third key challenge is maintaining the anonymity of votes cast by i-voting. In every online voting system, there are central systems that authenticate voters (ensuring that each voter votes only once) and that record votes. Government IT staff, and staff of outsourced IT services and/or the election support staff of vendors of online voting software, all have access to these systems and their data. Therefore, they all have the capability of breaking data management policies that are intended to avoid vote data from being tampered, or voter and vote data from being inspected to determine how a voter voted. There can be policies and mechanisms that are intended to maintain separation of votes from voter identity, but such policies can be disregarded, and such mechanisms can be broken.
This anonymity issue is a basic issue with any form of remote voting, but more serious with Internet voting. In paper vote-by-mail there is a crucial point where an election official validates the voter’s identity and eligibility based on information on an envelope, and then opens the envelope to remove the ballot. Just at that moment, when the ballot is visible, but before it has been separated from the envelope, anonymity is at risk. With i-voting, there is an analogous risk, but with key differences. With i-voting, the voters are trusting not just election officials to maintain anonymity, but also IT staff and anyone who can gain access to the systems used to run the i-voting software. Further, i-ballot anonymity violation can occur for several voters at once, while paper ballots have to be manually and individually inspected at the point of vulnerability. Lastly, the point of vulnerability may be indefinite with i-voting because of the ability to retain logs and records of voter check-in.
In addition to these three key challenges, it is very much worth noting that there is further set of technological security issues that arise simply from the use of networks and computers to serve the i-voting application and store the vote data. The system’s integrity – that of the stored vote data, and of the i-voting server software and computers – are at risk from the IT staff, from those with datacenter access, form those with remote access, and from adversaries who might gain illicit remote access via security vulnerabilities. And just as those systems could be abused, they could also be targeted by denial-of-service attacks that prevent voters from accessing the i-voting system. But even leaving aside these more typical challenges – which face any Internet-based service – Internet voting still has the specific challenges of i-voting client integrity, voter authentication, and ballot anonymity.
Digital Distribution of Ballots
All three of these challenges are side-stepped by the approach of focusing solely on the initial step of Internet voting – the digital distribution of unmarked ballots. In this approach, the distributed ballots are PDF files suitable for printing and hand-marking.
The main challenge is connecting each UOCAVA voter with the correct ballot for them to vote. This problem does not arise with current vote-by-mail (VBM) because a specific paper ballot is sent to a particular voter by the appropriate local election authority. In a digital distribution scheme, it becomes the responsibility of the voter to find the correct ballot.
There are technological means for assisting voters in this process, which have already been developed and demonstrated in analogous manner by the efforts of the Voting Information Project (a collaboration between Google and Pew). From a voter’s perspective, they can use a “widget” on a Web site to provide a street address, and obtain the location, map, and directions for the polling place for voters resident at the provided address. Behind the scenes, the VIP technology looks up the address in a database that was compiled from data provided by local election officials. The address is mapped to a precinct, and to the precinct’s polling place. Then the address of the polling place is used in a conventional map search. One method of ballot distribution can be an extension of this technique, in which election officials provide not only the polling place for a precinct, but PDFs of the ballot(s) for the polling place for the upcoming election. Again, from the voter’s perspective, the experience is familiar and simple: enter an address, and get a PDF of the appropriate ballot, as well as a link to instructions for mailing back the marked ballot.
This is only one approach among many, using Web-based pull and/or email-based push, or existing Internet services and/or new services operated by election authorities, and other variations. However, the above example shows how the 3 challenges above do not apply. First, because voting is by paper ballots rather than electronic ballots, there is no digital risk to the physical marked ballot data. Second, because the blank ballots are public documents, there is no need to authenticate people who wish to download blank ballots. (There may also be some public benefit to providing an electronic equivalent of sample ballots that are surfaced-mailed to voters.) Third, because voting is by paper ballots as in existing vote-by-mail, there are no new technological threats to anonymity.
Additional Benefits
There are some potential benefits and valuable lessons to be learned from experience with Internet-based blank ballot distribution. Perhaps the most immediate relevance is implementation of the MOVE Act, which does not call for Internet Voting, but does call for automation of several electoral processes related to overseas voting – including ballot distribution – while retaining the auditable, verifiable, re-countable paper ballots of existing vote-by-mail. Experience with blank ballot distribution can provide guidance on how provide – and integrate – the automation of several of these processes, including voter registration, vote-by-mail ballot request, change of address, sample ballot distribution, etc. Appropriately applied, Internet based technologies can modernize and accelerate overseas and military voting processes while they preserve the integrity of the ballots and the voting process. In addition, the integration of these services, and experience in interaction with UOCAVA voters using them, may provide the basis for creating a strong authentication means for subsequent interaction with UOCAVA voters.
Summary
Among the several challenges for overseas and military absentee voter, one of the most widely cited is in the transit time required for overseas vote-by-mail. Internet voting is an often-discussed approach to solving this problem. Yet Internet voting – and especially the “home voting” variant for voters abroad – has three intrinsic challenges that are difficult to address today: i-voting client integrity, voter authentication, and ballot anonymity. A different and less challenging approach is available: Internet-based blank ballot distribution, which side-steps these three challenges. Addressing the transit-time problem specifically, this approach is a solution that that would be effective for all UOCAVA voters who have access to express mail or reliable postal service.
Usage of Internet-based blank ballot distribution, in addition to assisting UOCAVA voters, could also have significant benefits in building public confidence in the use of the Internet for voter assistance, and in building experience and learning about the improving the effectiveness of Internet-based voter assistance.
