New Standards for Elections:
A forum on technical and nontechnical requirements for voting systems

Summary and report

Voting equipment standards are an important new development in the US:

New standards for voting equipment are required by the Help  America Vote Act (HAVA) (2002) which requires nationwide standards for deployment of improved voting equipment by January 1, 2006.  Over the past three years, the IEEE Standards Association Voting Systems Standards (SCC38) Project on Voting Equipment (P1583) developed a detailed functional description for precinct-based balloting and ballot counting equipment. This document, IEEE-STD P1583 "Draft Standard for Evaluation of Voting Equipment", is (at time of writing) approaching its second attempt at approval within the IEEE Standards Association process. Its first circulation resulted in over 1000 comments that have considerably improved the document, but numerous areas of contention and disagreement still exist.  The Technical Guidelines Development Committee (TGDC) of the NIST Election Assistance Commission (EAC) was expected to adopt these IEEE standards for voting equipment as final Voting System Standards.

• Merle King, Vice-chair, IEEE Voting Equipment Standards Project 1583; Chair, Dept of Computer Science & Information Systems and Exec.Director, Center for Election Systems, Kennesaw State University, Kennesaw GA -- P1583 origins and directions.
  • P1583 originated in voluntary Voting Systems Standards (VSS) for accessibility and usability, created by FEC to assist states in compliance with Americans with Disabilities Act (ADA). VSS has evolved since the early 90s.
  • There is a national election operations "community " that has cooperated to varying degrees around these standards:
    • NASED - Natl Association of State Election Directors
    • NASS - Natl Association of Secretaries of State
    • Vendors
    • Federal Election Commission
    • etc
  • HAVA created FEC's Election Assistance Commission (EAC) and charged NIST to assist in setting firm requirements, not just voluntary standards, with Technical Guidelines Development Committee (TGDC). IEEE-Standards Assoc is named as a participant in TGDC.
  • IEEE P1583 incorporated and evolved from VSS; reliability/accuracy/security issues drew new participants from academic and activist communities
    • There has been significant disagreement between old and new participants
    • Scope of the effort -- esp. on policy issues vs narrow specifications -- has been a major issue
    • It has not been possible to resolve these disagreements and release a draft to ballot
  • NASS recently wrote an open letter to Congress requesting that EAC not be re-authorized in 2006 (when it expires)
• Rebecca Mercuri, Notable Software (Philadelphia PA), Fellow, Harvard Radcliffe Institute; active member, IEEE-STD P1583 committee - P1583 status, issues
  • HAVA's origins are in 2002 election audit problems, rather than Florida 2000
  • HAVA funds for equipment procurement have to be released and spent before any HAVA standards are in effect
  • P1583 is a poor starting point for standardization of election systems
  • There is real conflict in this elections operations "community" over election reform
• Stan Klein, Stan Klein Associates (Rockville MD); active member, IEEE-STD P1583 committee - Voting system reliability
  • Reliability figures specified in P1583 are absurdly low; error rates allowed are high
  • Established from duration of common testing practice rather than real requirements
  • Observed failure rates of equipment are also high
  • Much of the problem with DRE accuracy and failure seems to be in touch-screen misalignment problems.
• David Aragon, VoterMarch.org; active member, IEEE-STD P1583 committee  - What standards can and can't do, and what people might be hoping that this one will do.
  • Equipment standards are a poor substitute for election reform
  • Engineering participants should not confine judgement to narrow specification issues, but consider broader goals of fair and clean elections
  • Citizen input and participation is essential; closed voting equipment "community" needs transparency
  • Transparency not a specific feature -- should pervade the standard
    • Functional specs should state how, and by whom, function can be validated in actual use (not only at certification)
    • What expertise, training or privilege is needed? Can the voter do it?
• Ronald Rivest, Viterbi Professor of Computer Science and Engineering, MIT; Chair, NIST/EAC TGDC Subcommittee on Computer Security and Transparency -- TGDC status and direction, election security and transparency requirements of HAVA
  • TGDC was formed very late and has really just started work
  • P1583 adoption was expected to speed work but not complete it
  • Discussions now are at a considerably higher level than P1583 -- include debates over policy issues
  • April 9 and Jan 1 deadlines do not seem to be considered crucial
• Ben Adida: co-author of EVOX, a large-scale online voting system prototype; PhD student, MIT Lab for Computer Science, member of the Caltech-MIT Voting Technology Project , with a research focus on universally-verifiable voting -- Voting System Performance Rating (vpsr.org) (cofounder)
  • Voting Systems Performance Rating is a new group of academic computer science participants in the debate
  • Intends to research policy and equipment issues and establish methods of testing and measurement
  • New "community" organized in a way familiar to academic comptuer scientists -- like Internet Engineering Task Force (IETF)
    • Open to all -- free membership
      • Participation of all stakeholders encouraged
      • Includes vendors, activists, government as well as technology, political science researchers
    • Timeline does not seem to be related to HAVA or other Federal processes including 2006/2008 elections

New election technologies, esp. cryptographic election protocols,  may provide provably clean elections, but are new, complex, and not fully tested in practice, and raise new questions of election transparency, while also raising the possibility of a new standard of technical confidence in election practice.

• Ron Rivest, Ben Adida – Cryptographic election protocols tutorial
  • Tools for possible future election systems, now in research stages
    • Public key cryptography
    • Shared secret
    • Zero-knowledge proof
• Publication of complete election transcript is an essential feature -- permits verification of final tabulation by any voter
• Many technical issues still open
• Transparency issues
  • for technical citizen, even at high-school student level, transparency may be much improved
  • for the average citizen transparency may be very difficult.
• Dick Johnson, consultant, active member, IEEE-STD P1583 committee  - Stem to Stern Security for the Networked Voting Process
  • Transfer of voting workstation and precinct results to tabulation centers is error prone
    • Still often by variations on "sneakernet" technology - a physical medium stores intermediate results for transportation and storage
  • Private networking technology can improve reliability and confidence in results
    • Could include centralization of voting system making the workstation a very simple "thin client"
  • Network management on private networks allows observation of operations by technical pollwatcher
• Chuck Gaston, SAVIOC Voting Systems, active member, IEEE-STD P1583 committee  -- How standards can block improvements [Powerpoint]
  • "Secure and Accurate Voting on Inexpensive Old Computers"
    • demonstrates that the basic electronic voting system problem requires very little in the way of computer technology
    • uses abundant surplus/obsolete PC equipment, only floppy drive storage required
      • stores operating system, voting system software and many ballots on initial floppy;
      • final floppy replaces all ballots except the selected one with all the voting results.
      • replication and distribution of initial and final floppies (which can be verified by hash code check) provides security, transparency and reliability
      • prescriptive standards threaten to make this simple approach illegal.

New expectations for the democratic process are growing in the US, and demand is rising for true national standards for election operations.

• Ted Selker, MIT Media Laboratory, Caltech/MIT Voting Technology Project - Election operations errors in practice
  • Research into voting equipment shows that human factors (voter) are the most important problem in electronic voting systems
  • Observations of election operations shows there are also large problems in training and human factors (pollworker)
  • Broad ignorance and even resistance to proper election operations practice
  • Paper records custody issues are a special problem
  • Training and professionalization of election operations is needed
• Teresa Hommel, wheresthepaper.org, Chair, Task Force on Election Integrity, Community Church of New York Unitarian Universalist -- Oversight of Elections or Techology [MS Word]
  • Standards for technology cannot ensure election integrity because elections are not about technology.
  • Effective observation of elections should not require computer expertise.
  • Legitimacy of elections cannot rest on trust in computers, nor examination of computers rather than ballots.
  • Voter-verified paper ballots are half of a solution
  • Public observation of 100% of vote counting is the second half.
  • Uniformity of technology cannot provide uniformity of integrity.
  • Technologists have been asked to create computers that will persuade people to give up their right to observe the vote recording and vote counting and to passively turn over our elections to computers.
  • Technologists must be more conscious of the political context of their work.
  • Mechanical lever machines are the hardest technology to tamper with, and the easiest to administer
  • Paper ballots marked by hand (or ballot-marking machine for voters with disabilities or non-English languages) can result in perfect elections if there are sufficient observers of all ballot-handling and counting procedures
  • Computers are a trouble-prone technology for elections, because election professionals and observers don't know what the technicians or computers are doing.
• Lillie Coney, Coordinator, National Committee for Voting Integrity, Electronic Privacy Information Center - Voter Privacy, the Secret Ballot, and Centralized Voter Registration Databases
  • Voter registration is public record
  • Registration rolls management was a major issue in Florida 2000 and 2002/2004
    • Purging of felons by broad-net identification of similar names/addresses resulted in large numbers of incorrect removals from registration rolls
    • Racial bias in records handling
  • Privacy of secret ballot is absolutely required
    • Long-term security of published transcript data in future crypto election protocols is a concern
  • Registration database raises national ID system issues
• Arlene Ash, statistician who testified in the Martin County Absentee Ballot fraud case in Florida following the November 2000 elections. Chair of the Subcommittee on Electoral Integrity, American Statistical Association's Scientific and Public Affairs Committee. Author of “Why We Must Question Our Elections,” at www.truthout.org , which has a very useful “voter rights” section - Statistical checks on electoral integrity, "spoilage" and election technology · [MS Word]
  • Standard statistical methods can identify anomalies in the electoral process and the factors (such as, race, geography and voting equipment) associated with “errors.”
  • Statistics can rule out, as well as raise, suspicions of large-scale tampering. ·
  • Disenfranchising voters, especially African-Americans, by manipulating voting lists, “spoiling” their ballots, and other means is a shameful, continuing feature of US elections – (see, e.g., “Voting Irregularities in Florida During the 2000 Presidential Election,” www.usccr.gov ).
  • But John Ashcroft's Department of Justice did not punish well-documented electoral crimes in 2000, and few Democrats and no Republicans joined Representative John Conyers, ranking minority member of the House Judiciary Committee, in investigating, or acting on, wide-spread evidence of wrongdoing in Ohio 2004. ·
  • My truthout article (see above) illustrates statistical evidence of problems associated with voting machine equipment in 2004, and argues that identifying electoral problems and punishing fraud is necessary following every election, regardless of whether the winner is in doubt.
  • Routine, timely, and public release of precinct-level analytic files and standard reports is a necessary first step. ·
  • Technologists who seek to improve US elections should understand that
    • 1) much (most?) voter disenfranchisement occurs outside the voting booth, and
    • 2) it is hard to improve upon the virtues of the publicly-observed, counting and casting of paper ballots. ·
  • The American Statistical Association is searching for ways to partner with other groups to bring statistical expertise to bear in improving US elections.
    
• David H. Harris, Jr, Executive Director of the Lawyers Committee for Civil Rights under Law - Towards national standards for elections [MS Word]
  • States-rights rules for election practice have proven ineffective in delivering fair and honest elections
  • Minority communities remain effectively disenfranchised in many states
  • National standards for election practice are required
    • Tehnical equipment standards are a only a part of the problem

DISCUSSION moderated by Carol Rose, Executive Director, ACLU of Massachusetts