Myth Breakers (E-Voting Machines)
Myth Breakers (E-Voting Machines)
Myth Breakers (E-Voting Machines)
Essential Information for Those Entrusted with Making Decisions about Election Systems in the United States
Newspapers reported serious voting problems starting with the first day of early voting in the 2004 General Election and continuing through Election Day and beyond. Voters' selections changed in front of their eyes on the touch screens of paperless voting machines. Electronic poll books failed to work properly. Tabulation equipment began subtracting votes after accumulated totals reached 32,000. Voting machines lost votes, miscounted votes, and mysteriously added votes. Machines broke down, froze up, paged through ballots backwards, and skipped past important races.
Preface
The first edition of this compendium originated as a result of an email conversation I had with an election official. I had become more and more bewildered by the resistance to a voter-verified paper audit trail (VVPAT) on election equipment. Discovering that she was in favor of a VVPAT, I asked her why so many were resistant. Her reply was enlightening. She pointed out that, with the recent eruption of information and opinions on electronic elections, it's extremely difficult to tell which "studies" are legitimate and which are rumor. Then she said, "Add in a dose of politics the comments about Diebold Execs and Bush ties and you have a real mess. Make sense?" It did. Most of the officials responsible for making election system decisions don't have time to do the extensive, time-consuming research required to learn the enormous variety of information that has become available about voting systems. Those in one state may not be aware of the problems election officials in other states are encountering with the equipment they are using. For example, how many officials responsible for our elections know such tidbits as these? Hinds County, Mississippi had to hold its November 2003 election all over again because so many of the paperless electronic voting machines (Direct Record Electronic DRE) broke down that they couldn't determine the will of the voters. Neglecting to keep the DRE batteries charged between elections cost Arapahoe County, Colorado over $100,000 in battery replacements just before a recent election. If it takes an hour to do the Logic and Accuracy testing on one DRE, San Diego county would have to spend 1275 person-days testing before every election in order to comply with California law.
This slightly renamed "revised edition" has been updated to include additional facts that have emerged since the first edition was completed. Information is always essential to making wise decisions. That premise is the basis for this collection of facts, which is sort of a consumer's guide to voting systems. As the events surrounding elections have become more publicized especially after the 2004 General Election it has become clear that electronic elections bring with them many, many problems. This booklet is simply a collection of relevant information; it is by no means an exhaustive work on the issue. I offer it in the hope that it will help those reading it to make wise decisions regarding our election systems. Respectfully, Ellen Theisen VotersUnite! [email protected]
Preface
iii
Acknowledgments
Thanks to the following people, whose contributions to this document were invaluable: John Gideon, Information Manager at VotersUnite.Org, for his support and review of my work, for the constant flow of news articles that provided much of the content, and for additions he suggested. Robert Kibrick, Legislative Analyst at VerifiedVoting.org, for his support and review of both versions, for suggesting additional material that added to the scope of the work, and for providing the basis for much of the information, especially in the sections on "Potential Problems Inherent in Electronic Devices," "Hidden Costs of DREs," and the comparison of optical scan equipment to DREs. Pokey Anderson, Investigative Journalist in Houston, Texas, whose ideas and editing added polish to this second edition. Warren Stewart, election data researcher for National Ballot Integrity Project, for edits that added clarity. Joan Krawitz, Co-founder of National Ballot Integrity Project, for her careful and thoughtful suggestions. Bo Lipari, founder of New Yorkers for Verified Voting, for his review of the first edition and valuable suggestions. Susan and Gray Multer, concerned citizens in New York state, for editing the first edition and making valuable suggestions. Joan O'Rourke, Registrar of Voters in Fairfield, Connecticut for her review of the first edition and enthusiasm in support of my efforts.
iv
www.VotersUnite.Org
Contents
Preface Overview 1 Facts about The Help America Vote Act (HAVA)
HAVA Does Not Require the Use of DREs HAVA Does Not Prohibit Punch Card Systems HAVA Preserves States' Right to Use Paper Ballots HAVA "Audit" Requirement is Not a Meaningful Recount HAVA Allows Partial Replacement of Old Systems HAVA Preserves States' Rights to Establish Voting Equipment Standards
iii viii 1
1 1 2 2 3 3
4
4
4 5 7 8 9 9 10 11 11 12
13
13 13 14 14 14
17
17 18
19 19 20 v
The Myth of Pre-Election Testing The Realities of Conducting Logic & Accuracy (L&A) Testing on DREs If DREs Fail the L&A Test ... How's it Working?
20 22 23 24
25
25
25 26 27
27
27 28 30 32 33 34
Administration Complexities
Chain of Custody Complexities Increase Exponentially with Electronic Data Increased Problems from Human Error in Electronic Elections Violation of Ballot Secrecy with the Use of DREs Running an Electronic Election - a Ballot Judge's Summary Running an Electronic Election an Election Official's Report
35
35 36 38 38 39
Management Complexities
Saving Money by Combining Precincts May Decrease Voter Turnout Potential DRE Problems Inherent in Electronic Devices Touch Screen Misalignment Power Surges or Static Electricity Discharges Electrical Outages and Inadequate Battery Charges Maintenance Challenges Rapid Obsolescence and Toxic Waste Disposal
41
41 41 41 42 42 43 43
44
44 45 46 46 47 48 48 www.VotersUnite.Org
49
49 50 50 51 52 53 53 53
54
54 56 58 59 59 59 60 60 60 60 61 61
Afterword
62
Contents
vii
Overview
Election transparency is the fundamental basis of election integrity.
In transparent elections, all the processes of handling and counting ballots are completely open to public view. Nothing is hidden, nothing is secret except, of course, each individual's voting choices. Election fraud and miscounts have occurred throughout history, and they will continue to occur. Transparency is the only way to minimize them, but with electronic voting, transparency is eclipsed. Electronic processes that record and count the votes are not open to public scrutiny. Courts have ruled that election software is a trade secret, so even a losing candidate with a computer consultant cannot view it. With electronic voting, the most important and vulnerable election processes storing and tallying the votes are performed in secret, without public oversight. These processes were not developed by government officials charged with ensuring election integrity, but by anonymous software engineers, hired by vendors and not publicly accountable for the results of their work. One would expect overwhelming benefits to accompany this sacrifice of transparency and the resulting loss of public control over election processes. That's the myth. Ironically, overwhelming disadvantages accompany the sacrifice. The logical question is "Why make the sacrifice?" It's a question more and more people are asking. The facts presented in this document dispel many of the myths surrounding electronic voting. It is crucial to lay these myths to rest quickly, for as long as they are held by decision-makers, our democracy is at risk. Here is a summary of some of the myth breakers presented in this document No federal law requires us to record and count votes electronically In recent elections, electronic voting machines have: - failed to count votes - tallied votes incorrectly - given voters the wrong ballot - broken down during elections - handed votes to the wrong candidate - disenfranchised voters - reversed election outcomes The use of DREs increases: - the time required for pre-election testing - the complexity of election procedures - the need for "ballot box" security - the potential for human error - the cost of storing election equipment between elections - the capital cost of election systems DREs are not the only way to provide independent voting for the disabled Some of the stories of successful electronic elections are not based on facts Some of the staunchest defenders of DREs have received donations from vendors Page 1 4 5 9 9 9 11 10 22 25 35 36 50 53 44 56 60
http://www.fec.gov/hava/law_ext.txt
2 Direct Recording Electronic voting machine. Votes are recorded on electronic media rather than paper. 3 http://www.essvote.com/HTML/home.html; see page 46 of this document 4 http://www.vote-pad.us; see page 46 of this document 5 http://www.openvotingconsortium.org/; see page 47 of this document
The HAVA Requirement For a Voter Verified Paper Record. Darryl R. Wold; July 23, 2003 http://www.verifiedvoting.org/resources/documents/HAVA_Requirement_for_VVP_Record.pdf http://www.co.snohomish.wa.us/documents/Departments/auditor/elections/results/general/ DREAgreement.pdf
www.VotersUnite.Org
VotersUnite! Election 2004 Problem Log. http://www.votersunite.org/electionproblems.asp?sort= date&selectstate=ALL&selectproblemtype=Machine+malfunction Computer loses more than 4,000 early votes in Carteret. Charlotte Observer. November 4, 2004. Associated Press. http://www.charlotte.com/mld/observer/news/local/10099907.htm N.C. agriculture commissioner candidate appeals new statewide election. North County Times. January 4, 2005. By Gary D. Robertson - Associated Press. http://www.nctimes.com/articles/2005/01/05/election2004/20_40_441_4_05.txt Certify Troxler. Winston-Salem Journal. December 21, 2004. http://www.journalnow.com/servlet/Satellite?pagename=WSJ%2FMGArticle%2FWSJ_ColumnistArticle &c=MGArticle&cid=1031779782822&path=!opinion&s=1037645509163
10
11
www.VotersUnite.Org
Palm Beach County, Florida. November 2004. Sequoia DRE Battery failure causes DREs to lose about 37 votes.
Nine voting machines ran out of battery power and nearly 40 votes may have been lost. ... The nine machines at a Boynton Beach precinct weren't plugged in properly, and their batteries wore down around 9:30 a.m., said Marty Rogol spokesman for Palm Beach County Supervisor of Elections Theresa LePore. Poll clerk Joyce Gold said 37 votes appeared to be missing after she compared the computer records to the sign-in sheet. Elections officials won't know exactly how many 12 votes were lost until after polls close.
Mecklenburg County, North Carolina. November, 2004. Microvote DRE Results show nearly 3,000 more votes than voters.
According to election-office data downloaded by the Charlotte Observer, 102,109 people voted early or returned valid absentee ballots. But unofficial results show 106,064 people casting early and absentee votes for president.14 Officials suspected that some results may have been counted twice. But they were wrong. A news release from the Mecklenburg County Board of Elections shows that some candidates gained votes in the manual recount of the paper tapes printed by the machine.15 The machine or the accumulation software simply tallied wrong.
Lancaster County, Nebraska. November, 2004. ES&S Optical Scanner.16 Optical scanners double-count ballots.
As the optical scanners read the election-day ballots, they occasionally added votes. While County Election Commissioner David Shively explained that the software was reading ballots twice, ES&S referred to the misread as a mechanical problem. Inexplicably, both Shively and the Nebraska deputy secretary of state for elections, Neal Erickson, agreed that "the malfunctions were not the type that taint vote counts." When the officials tested the six machines, it became clear that two were not correctly counting the ballots. That came as a surprise, Shively said, because all were tested late last week and performed well. After consulting with ES&S, Shively decided to use the two absentee-ballot machines to speed up the election-day counting, but the problem was apparently contagious. The double-counting problem began plaguing almost all the machines.
12 13 14
Nearly 40 Votes May Have Been Lost In Palm Beach County. Associated Press. November 2, 2004. http://www.votersunite.org/info/previousmessups.asp County retallies early-vote results. The Charlotte Observer. Nov. 4, 2004. By Richard Rubin and Carrie Levine. http://www.charlotte.com/mld/charlotte/news/politics/10094165.htm Board of Elections Audits Early Voting Results; Revises Unofficial Results Released by the Mecklenburg Board of Elections. November 4, 2004. http://www.votersunite.org/info/mecklenburgnewsrelease.pdf Problem machines spur call for recount. Lincoln Journal Star. November 14, 2004. By Nate Jenkins. http://www.journalstar.com/articles/2004/11/14/election/doc4189b9c7f14bf764391458.txt
15
16
Bernalillo County, New Mexico. November, 2004 Over 8,000 phantom votes appear in the canvass report.
The New Mexico certified election results reported 2,087 phantom votes (more votes than ballots cast) for president statewide. These phantom votes were concentrated in Bernalillo County. The official canvass report shows 187 precincts in Bernalillo County reporting presidential phantom votes a total of 1,239 votes.17 For example: Precinct 558 reported 178 early voting ballots and a total of 319 votes for president. That's 141 phantom votes, nearly as high as the number of ballots. Precinct 512 reported 166 ballots, with 318 votes for president. In that precinct, Bush alone received 206 votes, 50 more than the number of ballots.
In the presidential race and 14 down-ticket contests examined and shown in the chart below, a total of 8,656 phantom votes were reported in the certified canvass report.
In October of 2004, Bernalillo County Clerk Mary Herrera admitted that phantom votes had been added to several elections over the past two years. She also said her vote-counting experts have always found the phantom votes before they were added to the final tally.18 But not in this election. New Mexico Secretary of State Rebecca Vigil-Giron says phantom votes are not possible, pointing out that her independent auditors didn't find irregularities like this.19 Nevertheless, they are present in her certified canvass report.
17
Bernalillo County Canvass of Returns of General Election Held on November 2, 2004. State of New Mexico. http://www.sos.state.nm.us/PDF/Bernalillo.pdf Brief Summary of Bernalillo County Election Data. January 5, 2005. By Ellen Theisen. County clerk say phantom votes won't be a problem. KRQE Albuquerque. October 26, 2004. http://www.krqe.com/expanded3.asp?RECORD_KEY%5BLargeHeadline%5D=ID&ID%5BLargeHeadlin e%5D=7425. Vote Recount Fight 'Is Not Over'. Albuquerque Journal. December 24, 2004. By Andy Lenderman. http://www.abqjournal.com/elex/278376elex12-24-04.htm
18
19
www.VotersUnite.Org
Orange County, Florida. November, 2004. ES&S Optical Scan System Vote tabulating software omits counting 8,400 votes.21
The precinct results posted on the Orange County elections office Web site showed that Democrat John Kerry beat Republican President Bush by 9,227 votes in Orange County, but the posted results were off by 8,400 votes. The margin was actually only 827 votes. The cause of the error, Orange officials said Thursday, was a software program that could not tabulate more than 32,767 votes in a single precinct. A similar discrepancy affected vote totals posted online for the U.S. Senate race between Republican Mel Martinez and Democrat Betty Castor.
Guilford County, North Carolina. November, 2004. ES&S DRE System Vote tabulating software changes two outcomes in Guilford County.
In Guilford County, ES&S early voting machines also had capacity problems. The totals were so large, the tabulation computer threw some numbers away. Retallying changed two outcomes and gave an additional 22,000 votes to Kerry.22 Ken Carbullido, Vice President of ES&S Product Development, explained the problem to Guilford County. In very technical language, he wrote that when the vote totals reached 32,767 (32K), the system began subtracting from the totals.23 The 32,767 capacity limitation at a single precinct level is a function of the design and definition of the results database used by ERM [Election Reporting Manager]. The data storage element used to record votes at the precinct level is a two byte binary field. 32,767 is 2 to the 15th power, which is the maximum number held by a two byte word (16 bits) in memory, where the most significant bit is reserved as the sign bit (a plus or minus indicator). Additionally, ERM precinct count level data is stored in a binary computer format known as twos complement..... In the letter, Mr. Carbullido admitted the company knew about the problem but had not advised the county.
20
Gambling vote glitch mars tally. Miami Herald. November 5, 2004. By Erika Bolstad And Curtis Morgan. http://www.miami.com/mld/miamiherald/10103931.htm Distrust fuels doubts on votes: Orange's Web site posted wrong totals. Orlando Sentinel. November 12, 2004. By David Damron, Sentinel Staff Writer. http://www.votersunite.org/article.asp?id=3803 Winner so far: Confusion. The Charlotte Observer. November 5, 2004. By Mark Johnson. http://www.charlotte.com/mld/observer/news/local/10104576.htm?1c http://www.votersunite.org/info/GuilfordESS.pdf
21
22
23
Snohomish County, Washington. November, 2004. Sequoia DRE Voters find vote-jumping difficult to correct.
Voters in at least four polling precincts in Snohomish County said they encountered problems with the electronic voting machines. When they touched the screen to vote for a candidate, voters said an indicator showed they had selected the opposing candidate. Those voters told KING5 News it took at least four attempts before the indicator showed the correct candidate.25
Bernalillo County, New Mexico. October, 2004. Sequoia DRE Votes for Kerry jump to Bush.
When the same problem occurred in Bernalillo County, New Mexico, it took some voters as many as three times to get the machine to register their votes for Kerry instead of switching the selection to Bush. Kim Griffith voted on Thursday over and over and over. She's among the people in Bernalillo and Sandoval counties who say they have had trouble with early voting equipment. When they have tried to vote for a particular candidate, the touch-screen system has said they voted for somebody else. It's a problem that can be fixed by the voters themselves people can alter the selections on their ballots, up to the point when they indicate they are finished and officially cast the ballot. For Griffith, it took a lot of altering.26
24
When the Right to Vote Goes Wrong. TrueVoteMD. November, 2004. http://www.truevotemd.org/Election_Report.pdf Scattered reports of voters being blocked and machine malfunctions. November 2, 2004. KING5 News. http://www.king5.com/topstories/stories/NW_110204ELBelectronicvotingproblemsLJ.1aac5fda.html Some Early Voters Say Machines Mark Incorrect Choices. ABQJournal. October 22, 2004. By Jim Ludwick, Journal Staff Writer. http://abqjournal.com/elex/246845elex10-22-04.htm
25
26
www.VotersUnite.Org
Orange County, California. March 2004 Incorrect access codes gave voters incorrect ballots.
Poll workers struggling with a new electronic voting system in last week's election gave thousands of Orange County voters the wrong ballots, according to a Times analysis of election records. In 21 precincts where the problem was most acute, there were more ballots cast than registered voters. At polling places where the problem was most apparent because of turnouts exceeding 100%, an estimated 1,500 voters cast the wrong ballots, according to the Times' analysis of official county election data. Tallies at an additional 55 polling places with turnouts more than double the county average of 37% suggest at least 5,500 voters had their ballots tabulated for the wrong precincts.28
Carroll County, North Carolina. November, 2004. ES&S Optical Scan System Vendor mis-programming caused the miscount.
The chip supplied by ES&S for the election was incorrectly programmed and miscounted the votes for the JP District 2 race between Rocky Whitely and Duane Coatney. Once ES&S supplies a new chip for the optical scanners, the county will rescan the ballots for that contest. 30
27
The Vexations of Voting Machines. Time Magazine. May 3, 2004. By Viveca Novak. http://www.time.com/time/archive/preview/from_redirect/0,10987,1101040503-629410,00.html 7,000 Orange County Voters Were Given Bad Ballots. Los Angeles Times; March 9, 2004; By Ray F. Herndon and Stuart Pfeifer. Reproduced at: http://www.votersunite.org/article.asp?id=1476 Fidlar admits election blip. Quad City Times. November 13, 2004. By Tory Brecht. Reproduced at http://www.qctimes.com/internal.php?story_id=1039447&t=Local+News&c=2,1039447 Computer glitch blamed for miscount in JP voting. Star Tribune. November 10, 2004. By Anna Mathews, CCN staff writer. http://www.greenforesttribune.com/articles/2004/11/10/news/s1.txt
28
29
30
Lake County, Illinois. April, 2003. ES&S Optical Scan System Vendor mis-programmed again.
The problem was caused by a programming error that failed to account for "no candidate" listings in some races on the ballot, Clerk Willard Helander said Thursday. As a result, election results were placed next to the names of the wrong candidates in four different races, including in Waukegan's 9th Ward. ... Helander blamed the problem on Election Systems & Software, the Omaha company in charge of operating the county's optical-scan voting machines. She said a company official told her the programmers were unaware the county would have "no candidate" listings on its ballot.31
Mahoning County, Ohio. November, 2004. ES&S DRE Machines broke down in 16 precincts; others needed calibration.
Many problems plagued the ES&S iVotronic touch screen voting machines in 16 of the 312 Mahoning County precincts. Some of the machines malfunctioned. Others had problems with the personal electronic ballot cartridge placed into the machines before each vote to count the ballots ... Also, there were 20 to 30 machines that needed to be recalibrated during the voting process because some votes for a candidate were being counted for that candidate's opponent. About a dozen machines needed to be reset because they essentially froze.33
31
Returns are in: Software goofed Lake County tally misled 15 hopefuls. (reproduced) Chicago Tribune; April 4, 2003; By Susan Kuczka, Tribune staff reporter. http://www.vote.caltech.edu/mailarchives/votingtech/Apr-2003/0096.html When the Right to Vote Goes Wrong. TrueVoteMD. November, 2004. http://www.truevotemd.org/Election_Report.pdf Errors plague voting process in Ohio, Pa. Vindicator. November 3, 2004. Vindicator staff. http://www.vindy.com/basic/news/281829446390855.php Back up! We need back up! Roanoke.com. November 11, 2004. By Brian Gottstein. http://www.roanoke.com/columnists/gottstein/13719.html
32
33
10
www.VotersUnite.Org
Broward County, Florida. October, 2004. ES&S DRE Break downs require voters to come back the next day.
Hundreds of voters showed up to vote early at Howard Forman Health Park, so many that a decision was made to keep the voting facility open until 11 p.m. Some people waited in line from early in the day until after the sun went down. Unfortunately, for a group of about 50 people, the waiting did not pay off. A mechanical problem with the voting machines caused election workers to close down polling place. The group of 50 frustrated voters will have the opportunity to be first in line to vote today. Poll workers took down their numbers and names and will move them to the head of the line. For one couple, it may not be enough. They were voting on Sunday because they planned to leave on vacation today. Now they will have to choose to cancel their trip, or give up their chance to vote.34
34
Voters Turned Away After Waiting Hours. WPLG Local 10. November 1, 2004. http://www.local10.com/news/3878344/detail.html Voting problems minor, but frustrating. San Antonio Express. November 11, 2004. By Tracy Idell Hamilton, Staff Writer. http://www.mysanantonio.com/news/metro/stories/MYSA110204.online. votingproblems.1a8d060b.html; Reproduced at: http://www.votersunite.org/article.asp?id=3650 Parish by parish list of voter machine problems called in by viewers. WWLTV.com. 10:27 AM CST November 2, 2004. http://www.wwltv.com/local/stories/wwl110204electionmishap.18e9b314.html Few Glitches Reported in Early Fla. Voting. The State. October 19, 2004. By Jill Barton, Associated Press. http://www.thestate.com/mld/thestate/news/politics/9952991.htm
35
36
37
11
Volusia County, Florida. November 2004. Diebold Optical Scanner Memory cards were inspected in the summer, failed in the fall.
Memory-card breakdowns in six machines left political contests in limbo for hours. The county had the memory cards inspected by Diebold in the summer of 2004 in preparation for the busy election season. Ion Sancho, the elections supervisor in Leon County, said officials with Diebold told him that the new, higher-capacity memory cards tend to have more glitches than older cards.40
San Diego County, California. March 2004 Diebold Precinct Control Module Encoders allow multiple votes ... or none.
At least one voter was able to vote twice on her "smart card," and at least 250 polls opened late because poll workers were unable to start up the encoders. Hundreds, perhaps thousands, of people were turned away many of them disenfranchised because they were unable to return to the polls at a later time that day. 41 Later reports estimated that this problem delayed the voting at 40% of the polls and may have occurred at as many as 80% of the polling places.42 While these problems were originally blamed on poll workers, a report43 released on April 12, 2004 by Diebold Election Systems shows that 186 of 763 encoders failed on election day because of hardware or software problems or both, with only a minority of problems attributable to poll workers. Diebold also admitted that tabulation errors during the October recall election were due to software bugs. The following example shows that the problem has not yet been fixed.
38
Early Voting Begins In Colorado. October 18, 2004. By Steven K. Paulson, Associated Press Writer. http://news4colorado.com/campaign2004/local_story_292161858.html Editorial: Put It on Paper Election snafu points up problems for all-electronic voting. Dallas Morning News. November 11, 2004. http://www.dallasnews.com/sharedcontent/dws/news/city/collin/ opinion/stories/111204dnccocollinvote.9eeb3.html Computer glitches slow Volusia results: County officials ask the machines' supplier to investigate why memory cards failed Tuesday. Orlando Sentinel. November 4, 2004. By Kevin P. Connolly, staff writer. Reproduced at http://www.votersunite.org/article.asp?id=3694 Poll workers, voters cite tied-up hotline, poor training, confusion. Union Tribune; March 7, 2004; By Jeff McDonald and Luis Monteagudo Jr. http://www.signonsandiego.com/news/politics/20040307-9999-1n7vote.html Correspondence, written report regarding Touchscreen voting system used for the first time March 2, 2004 by the County of San Diego. From: Walter F. Ekard; Chief Administrative Officer http://www.signonsandiego.com/news/politics/county/20040310-1315-report.html Diebold reports multiple problems: Registrar wants reason for e-voting. Tri-Valley Herald; April 13, 2004; By Ian Hoffman, Staff Writer. http://www.verifiedvoting.org/article.asp?id=1839
39
40
41
42
43
12
www.VotersUnite.Org
Twiggs County, Georgia. November 2004. Diebold Precinct Control Module Unexplained encoder problems.
Twiggs County voters arrived at the polls today to find they could not cast their votes on the computerized voting machines. The voting machines were down in all five precincts this morning because of an encoder problem from 7 a.m. until about 9 a.m., according to Twiggs Chief Registrar Linda Polk.44
44
Twiggs voters face problems at polls. The Courier Herald. November 2, 2004. By Jason Hobbs. Reproduced at http://www.votersunite.org/article.asp?id=3647 County Responds to Voting Machine Problems. Austin Chronicle. October 22, 2004. By Lee Nichols. http://www.austinchronicle.com/issues/dispatch/2004-10-22/pols_feature18.html Errors plague voting process in Ohio, Pa. Vindicator. November 3, 2004. Vindicator staff. http://www.vindy.com/basic/news/281829446390855.php
45
46
13
5) Report Shows 300 Registered Voters For Every Precinct in One County
LaPorte County, Indiana. November 2004. ES&S reporting system
A software flaw in the reporting system showed 300 registered voters in every precinct. Since the column that gives the turnout depends on the registration numbers, the county had to wait for a software patch from ES&S before they could determine voter turnout.49
47
Primary Election Runs Into Problems. HITV 4. The Hawaii Channel. September 24, 2004. http://www.thehawaiichannel.com/news/3760175/detail.html Errors plague voting process in Ohio, Pa. Vindicator. November 3, 2004. Vindicator staff. http://www.vindy.com/basic/news/281829446390855.php Back up! We need back up! Roanoke.com. November 11, 2004. By Brian Gottstein. http://www.roanoke.com/columnists/gottstein/13719.html
48
49
Voter turnout still not known. Herald Argus. November 4, 2004. By Kori Kamradt, County Reporter. http://www.heraldargus.com/content/story.php?storyid=5304
14
www.VotersUnite.Org
Election 2004
A Partial List of Incidents Reported in the News
Grays Harbor Co. WA. ES&S Unity election management system added votes to the totals. [1] Snohomish Co. WA. Votes changed to the opponent on Sequoia touch screens. [2] Nye Co. NV. Sequoia e-voting machines had mis-feeds and printer jams. [3] Boulder Co. CO. Hart/ Intercivic optical scanners failed to read several thousand ballots. [8] Franklin Co. IN Fidlar tabulators gave straightparty Democrat votes to Libertarians in 9 precincts. [16] Mahoning Co. OH 20 to 30 ES&S evoting machines registered votes for the opponent. [17] AND One precinct reported negative 25 million votes. [18] Mercer Co. PA Unilect machines displayed ballot pages in the wrong order. [20] AND recorded 51 pres. votes for 289 voters. [21] Maryland. Diebold touch screens registered votes incorrectly, skipped pages on the ballot. [22]
Sarpy Co. NE ES&S Unity software added 10,000 phantom votes. [10] Lancaster Co. NE ES&S scanners detected some ballots twice. [9]
Franklin Co. OH Danaher e-voting machine gave Bush 3,893 extra votes. [19]
Pike Co. AR. Optical scanner failed to count 692 of 4,083 votes. [15]
Mecklenburg Co. NC. MicroVote e-voting Craven Co. NC. Sandoval Co. NM machines report Wichita Co. TX. ES&S ES&S software Votes changed to the 106,064 votes software failed to doubled the opponent on Sequoia for 102,109 record 6,900 of totals in nine touch screens. [6] voters. [27] 26,000 votes for of the 26 Four Parishes in LA. president. [11] precincts. [29] Sequoia machines Palm Beach Guilford Co. NC. Totals Bernalillo Co. NM Votes malfunctioned in Co. FL. reached 32,767 and ES&S changed to the opponent over a dozen Sequoia touch software began subtracting on the Sequoia touch precincts. [14] screens lost 37 votes. [28] screens. [7] votes. [23] Collin Co. TX. Diebold Lexington Co. SC. touch screens locked Sacramento, CA. Nine Officials couldn't retrieve up, wouldn't release of 712 precinct optical 200 ballots from an ES&S Orange & Broward Co. FL. vote data. [13] scanners broke down, e-voting machine. [25] Totals reached 32,767 & had to be replaced. [4] Travis Co. TX. Pressing ES&S software began Twiggs Co. GA. Diebold touch "Enter" after a straight-Dem subtracting votes. [24] screens were down in all five vote changes vote to Bush precincts when polls opened. [26] on eSlate machines. [12]
Utah Co. UT Punch card tabulators failed to count 33,000 votes. [5]
Carteret Co. NC. Unilect e-voting machine lost 4,400 votes. [30]
15
[1] Gray's Harbor County re-count boosts Gregoire. KGW News. Nov. 16, 2004. By Rebecca Cook, AP. http://www.kgw.com/sharedcontent/APStories/stori es/D86D7FA80.html [2] Scattered reports of voters being blocked and machine malfunctions. KING5 News. Nov. 2, 2004. http://www.king5.com/topstories/stories/NW_11020 4ELBelectronicvotingproblemsLJ.1aac5fda.html [3] Some snags reported in early voting in rural Nevada county. Las Vegas Sun. Oct. 26, 2004. Associated Press. http://www.lasvegassun.com/sunbin/stories/nevada /2004/oct/26/102610184.html [4] Long lines, no ballots for some. Sacramento Bee. Nov. 3, 2004. By Will Evans and Christina Jewett. http://www.sacbee.com/content/politics/ca/election /story/11304757p-12219794c.html [5] Missed votes due to error in machine. Salt Lake Tribune. Nov. 16, 2004. By Mark Eddington. http://www.sltrib.com/utah/ci_2454128 [6] Concerns rise on early voting. The Rio Rancho Observer. Oct. 26, 2004. By Eric Maddy, staff writer. http://www.observeronline.com/articles/2004/10/26/ news/story2.txt [7] Some Early Voters Say Machines Mark Incorrect Choices. Albuquerque Journal. Oct. 22, 2004. By Jim Ludwick. http://abqjournal.com/elex/ 246845elex1022-04.htm [8] Printer played role in Boulder voting woes. Rocky Mountain News. November 10, 2004. By Berny Morson. http://rockymountainnews.com/drmn/election/articl e/0,1299,DRMN_36_3317633,00.html [9] Problem machines spur call for recount. Lincoln Journal Star. November 14, 2004. By Nate Jenkins. http://www.journalstar.com/articles/2004/11/14/ele ction/doc4189b9c7f14bf764391458.txt [10] Countinghouse Blues. Too many votes. WOWT6 News. http://www.wowt.com/news/headlines/ 1161971.html [11] Waiting in Wichita. Times Record News. Nov. 4, 2004. By Robert Morgan. http://www.timesrecordnews.com/ trn/local_news/article/0,1891,TRN_5784_3303816,00.html [12] County Responds to Voting Machine Problems. Austin Chronicle. Oct. 22, 2004. By Lee Nichols. http://www.austinchronicle.com/ issues/dispatch/2004-10-22/pols_feature18.html [13] Editorial: Put It on Paper. Dallas Morning News. Nov. 11, 2004. http://www.dallasnews.com/ sharedcontent/dws/ news/city/collin/opinion/ stories/111204dnccocollinvote.9eeb3.html [14] Parish by parish list of voter machine problems called in by viewers. Nov. 2, 2004. WWLTV.com. http://www.wwltv.com/local/stories/wwl110204elect ionmishap.18e9b314.html [15] Election study finds widespread ballot-counting problems. Scripps Howard News Service. Dec.20, 2004. By Thomas Hargrove. http://www.knoxstudio.com/
shns/story.cfm?pk=MISCOUNT-ELECT-12-20-04&cat=AN
[16] Glitch causes Franklin Co. recount. IndyStar. Nov. 11, 2004. http://www.indystar.com/ articles/3/1938804433-093.html [17] Back up! We need back up! Roanoke.com. Nov. 11, 2004. By Brian Gottstein. http://www.roanoke.com/ columnists/gottstein/13719.html [18] Ohio Is Set to Reckon With Outstanding Ballots. Los Angeles Times. Nov. 9, 2004. By Ralph Vartabedian and Henry Weinstein. http://www.latimes.com/news/nationworld/politics /scotus/la-na-ohio9nov09,1,1213424.story ?coll=lanews-politics-supreme_court [19] Computer error at voting machine gives Bush 3,893 extra votes. Akron Beacon Journal. November 5, 2004. Associated Press. http://www.ohio.com/mld/ beaconjournal/news/state/10103910.htm?1c [20] Back up! We need back up! Roanoke.com. Nov. 11, 2004. By Brian Gottstein. http://www.roanoke.com/ columnists/gottstein/13719.html [21] Democrats' leader decries voting glitches. Vindicator Sharon Bureau. November 6, 2004. By Harold Gwin. http://www.vindy.com/ basic/news/288078640794824.php [22] Maryland e-voting controversy continues in presidential race. Washington Technology. November 3, 2004. By William Welsh. http://www.wtonline.com/ news/1_1/daily_news/24878-1.html [23] Nearly 40 Votes May Have Been Lost In Palm Beach County. Nov. 2, 2004. Associated Press. http://www.votersunite.org/article.asp?id=3659 [24] Distrust fuels doubts on votes. Orlando Sentinel. Nov. 12, 2004. By David Damron. http://www.orlandosentinel.com/news/orllocvotetotals12111204nov12,1,2450210.story [25] Lexington town leaders criticize slow vote count. The Roanoke Times. Nov. 4, 2004. By Tim Flach. http://www.thestate.com/mld/ thestate/news/local/10094010.htm [26] Twiggs voters face problems at polls. The Courier Herald. Nov. 2, 2004. By Jason Hobbs. http://www.votersunite.org/article.asp?id=3647 [27] County retallies early-vote results. The Charlotte Observer. Nov. 4, 2004. By Richard Rubin and Carrie Levine. http://www.charlotte.com/mld/charlotte/ news/politics/10094165.htm [28] Letter from ES&S to Guilford County Board of Elections. Nov. 8, 2004. http://www.votersunite.org/info/GuilfordESS.pdf [29] Election problems due to a software glitch. Sun Journal. Nov. 5, 2004. By Sue Book. http://www.newbernsj.com/SiteProcessor.cfm?Templ ate=/GlobalTemplates/Details.cfm&StoryID=18297&Se ction=Local [30] Glitch could force state to vote again. The Charlotte Observer. Nov. 9, 2004. By Michelle Crouch. http://www.charlotte.com/mld/charlotte/10133265.htm
16
www.VotersUnite.Org
50
Dr. Jones is an Associate Professor of Computer Science at the University of Iowa, has served on the Iowa Board of Examiners for Voting Machines and Electronic Voting Systems since 1994, and has consulted with the ACLU (Illinois Chapter), Miami-Dade County, and the Brennan Center for Justice on voting related issues. Testimony of Michael I. Shamos before the Environment, Technology, and Standards Subcommittee of the U.S. House of Representatives Committee on Science. June 24, 2004. (highlighting added) http://www.house.gov/science/hearings/ets04/jun24/shamos.pdf Dr. Shamos is the Co-Director of the Institute for eCommerce at Carnegie Mellon University. He has served as an examiner of electronic voting systems and consultant on electronic voting to Pennsylvania, Nevada, and Delaware. He is a strong proponent of paperless systems.
51
52
17
18
www.VotersUnite.Org
53 54
National Association of State Election Directors NASED Qualified Voting Systems. 12/05/03 Current. http://www.nased.org/ITA%20Information/NASEDQualifiedVotingSystems12-03-12-04.pdf National Association of State Election Directors General Overview for Getting a Voting System Qualified. Page 5. http://www.nased.org/ITA%20Information/NASEDITAProcess.pdf
55
19
Dr. Shamos' comments on the qualification process: Even if there were suitable standards, it is a significant question how to assure the public that a particular machine meets them. The current process of qualification testing by Independent Testing Authorities certified by the National Association of State Election Directors (NASED) is dysfunctional. As proof I need only cite the fact that the voting systems about which security concerns have recently been raised, such as Diebold Accuvote, were all ITA-qualified. Some of these systems contain security holes so severe that one wonders what the ITA was looking for during its testing. One may wonder, but one cannot find out. The ITA procedures are entirely opaque. ... I find it grotesque that an organization charged with such a heavy responsibility feels no obligation to explain to anyone what it is doing. It is important to note that the ITAs test a machine's design against the federal standards. There is no process (except perhaps the vendors') for testing and qualifying individual machines. So, while a design may be qualified, the individual machines used in elections are not.
Major Issues And Questions Addressed By The Task Force. California Ad Hoc Task Force; March 2003 http://www.ss.ca.gov/elections/taskforce_report_2.htm
20
www.VotersUnite.Org
Dr. Shamos' comments on pre-election testing: The machines may work according to specification but if they have not been loaded with the appropriate set of ballot styles to be used in a polling place they will be completely ineffective. The ballot styles to which Dr. Shamos refers are constructed through a process called "ballot programming." Ballot programming is done uniquely for each election; it produces a ballot definition file (BDF). BDFs are unique for each election; they define all the contests and candidates for each precinct. Data in the files tell the underlying voting machine software how to interpret a voter's touches on a screen or marks on an optical scan ballot (including absentee ballots) and how to record those selections as votes. In other words, BDFs turn votes into electronic vote data. BDFs are the primary focus of the testing that county officials do before each election. As Dr. Shamos indicated, the BDF is a key component of the voting system, yet it is never subjected to an outside review. The lack of independent oversight of these files is a major security vulnerability. If BDFs are incorrectly prepared, the wrong candidate could be elected. Programming election data is a very complex process, especially in counties with hundreds of different ballot styles. Some election districts lack the technical expertise to prepare BDFs, and instead depend on the vendor or outside programmers for the preparation. Others prepare the BDFs themselves. In both cases, however, BDFs undergo minimal testing and no independent audit before being used to determine the results of an election. Little wonder that many serious election disruptions have been caused by ballot definition errors. Since such errors are initially noticed because of suspect results, it is highly likely that other BDF errors have gone unnoticed. Some may have affected election outcomes. All of the proven ballot definition errors occurred on optical scan equipment and were caught by a manual recount of the ballots. In Chapter 2, "Election-Specific Programming Miscounts Votes" on page 9 gives three examples from the 2004 general election. Here are five examples from previous elections: In New Mexico, 67,000 absentee and early-voting ballots were counted incorrectly. 57 In Texas, a difference in ballot data on different machines resulted in miscounts in 18 races.58 In Florida, 2,642 Democratic and Republican votes were counted as Republican.59 In Texas, victories for two commissioners were initially given to the wrong candidates.60 In North Carolina, 5,500 party-line votes, both Republican and Democrat, were uncounted.61
57
Human error is cause of N.M. election glitch. Government Computer News; November 20, 2000; Vol. 19 No. 33. http://www.gcn.com/vol19_no33/news/3307-1.html Glitch affects 18 races Problems in counting early votes could alter some election outcomes. Dallas Morning News. May 8, 2002. Ed Housewright, staff writer. 6/2/04 Conversation with Barbara Montpetit, Supervisor of Elections. Original reference was from Black Box Voting, Chapter 2. Sometimes the old ways are best. The Bradenton Herald, 17 September 2002. 06/03/04. Conversation with Scurry County Elections Director, who said the problem was caused by the ES&S 650 chip with the ballot programming on it, and that they had to get a new one from ES&S. Original reference was from Black Box Voting, Chapter 2. Houston Chronicle, 8 November 2002; Ballot glitches reverse two election results Winners' may be losers. The News and Observer; November 12, 2002; By Wade Rawlins and Rob Christensen. Reproduced at: http://66.102.7.104/search?q=cache:iy0f4rgd7oMJ:www.ncdot.org/news/ dailyclips/2002-11-12zz.html+%22%27Winners%27+may+be+losers%22+wayne&hl=en
58
59
60
61
21
Since DREs (paperless electronic voting machines) provide no way to conduct an independent audit of the results, BDF errors on DREs are virtually undetectable. Given the known problems with BDFs on optical scan voting systems, it is reasonable to assume that similar but undetected errors have also occurred with DREs. January 2004, Broward County, Florida.62 The 134 blank ballots recorded by DREs are still unexplained. Did 134 voters go to the polls, with only one contest on the ballot, and cast blank ballots as election officials assume, or did a BDF error cause votes to be lost as has happened on optical scan machines? November 2002, Georgia. Dramatic upsets in the races for Governor and U.S. Senator could not be investigated because the voting machines were DREs. The outcomes may have been correct, but how do we know the BDF was not flawed?
If BDFs are flawed, a hand recount of the original ballots is the only way to correct the results. When optical scan ballots are recounted by the machine (by running them through the optical scanner a second time), the results typically match the initial tally. But while odd results might suggest a flaw in the BDF of a paperless system, even if further testing proved the flaw, there would be no way to recover the intent of the voters. Accurate election results require accurate BDFs. Some counties have hundreds of ballot styles, and each one must be programmed correctly since a human error in any definition could be magnified by the number of voters using that ballot. Pre-election testing is completely inadequate. Optical scanners are tested by running a small set of test ballots hardly enough to test every possible combination for every ballot style. Testing on DREs may involve simply pressing each button on the screen to make sure they all work correctly. Testing has failed to detect the many election data errors that have disrupted many optical scan elections. If an error occurs during an election, a new BDF is created and used to tally the final result, but without a manual audit of the votes, there is no way of knowing if the new BDF is correct. It could simply give results that are within reason and therefore unquestioned. The extreme complexity of election definition data, the complete lack of security procedures used to create them, the hopelessly inadequate testing: these problems raise serious questions about the accuracy of electronic vote counting on both DREs and optical scanners.
Electronic vote recount stumps Broward officials. Sun-Sentinel; January 10 2004; By Jeremy Milarsky and Lisa J. Huriash, Staff Writers Ballot counting resumes in San Bernardino County after four-hour delay. The Press-Enterprise; March 3, 2004; By Sharon McNary. http://www.pe.com/breakingnews/local/stories/PE_News_Local_websnafu03.9ff35.html
63
22
www.VotersUnite.Org
In a county with multiple DREs at each polling place, the time and effort required for adequate L&A testing increases significantly. For example, to comply with California law, the Registrar of San Diego County must test 10,200 DREs before every election. If it took an hour to test one DRE, San Diego County would have to spend 1275 person-days testing before every election. In addition, securing the tested components until election day becomes a significant challenge, and, in fact, the components in San Diego were not secured before the March 2004 primary election. They were sent home with poll workers after the training sessions, and the poll workers transported them to the polling places.64 The following information is adapted from the Computer Professionals for Social Responsibility (CPSR) description of testing DREs: 65 Each "ballot" of a logic and accuracy test for a DRE must be entered into the DRE by hand by voting the test ballot on the DRE itself. DREs present two special problems for logic and accuracy testing: 1. The process is time-consuming, so a small number of ballots is used often too small to catch important mistakes. Thus, in order to minimize the cost associated with testing, the effectiveness of the testing may be sacrificed. 2. It is difficult to correctly generate a series of test ballots on a DRE without a single error. It is much more likely that an election worker will make a mistake in entering test ballots than that an actual voter will make such a mistake, because the voter only has to remember one set of votes: the votes they wish to make. The DRE tester has a much more difficult problem. Consequently, election workers must conduct DRE logic and accuracy tests with extreme deliberation and caution, as even a single error requires that the entire logic and accuracy test be repeated. In practice, this results in logic and accuracy tests that are smaller yet, to the point where the test is testing for little besides a stuck button or a completely nonfunctional DRE. Consequently many DRE counties use the automated tests provided by the vendor, but these tests do not simulate election-day voting, nor do they run through the same program code used in a live election.
64
65
23
It is also likely that it would be too late to print paper ballots for the entire county to use for the election. It might be necessary to use uncertified software and hope it isn't flawed. Forced into such a corner, several California counties used Diebold's uncertified Precinct Control Modules in the March 2004 primary election. Some results are described on page 12 of this document.
How's it Working?
For the answer, refer back to the 30 detailed examples in Chapter 2, or the sampling of problems shown on the map on page 15. Or read through the 125+ machine malfunctions briefly described in the Election 2004 problem log at VotersUnite.Org.66 Or browse through the files of malfunctions sorted by vendor.67 Virtually all of the systems referenced on these pages and in these files were NASED-qualified, state-certified, and subjected to pre-election testing. For example, in Lancaster County, Nebraska, two out of six qualified, certified, and fully tested optical scanners miscounted votes in the 2004 general election: When the officials began testing the six machines, it became clear that two were not correctly counting the ballots. "That came as a surprise, [County Election Commissioner David] Shively said, because all were tested late last week and performed well." 68 In Pennsylvania's 2004 general election, too, the pre-election testing was to no avail: Mercer County's director of elections said it was a computer software glitch that caused touch-screen voting machines to malfunction in about a dozen precincts Tuesday. ... Election workers in Mercer County raced to take paper ballots to polling places in the Shenango Valley after a series of computer errors. "I don't know what happened," said James Bennington, who had been assured Friday that all 250 of the county's touch-screen units had been checked and rechecked.69 Sometimes, however, the tests fail. But what happens next, at least in New Mexico, is up to the discretion of the election directors. In the 2004 primary election in Dona Ana County, New Mexico, optical scan machines failed the pre-election testing and were used in early voting anyway. In pre-election testing, counters that track the total number of ballots passed through the machine showed incorrect numbers. The counters in four out of five machines were incorrect, showing as many as 20 or 30 votes more than the actual number of ballots tested. Yet the machines were used in early-voting anyway.70
66 67 68
http://www.votersunite.org/electionproblems.asp http://www.votersunite.org/info/messupsbyvendor.asp Problem machines spur call for recount. Lincoln Journal Star. November 14, 2004. By Nate Jenkins. http://www.journalstar.com/articles/2004/11/14/election/doc4189b9c7f14bf764391458.txt Errors plague voting process in Ohio, Pa. Vindicator. November 3, 2004. Vindicator staff. http://www.vindy.com/basic/news/281829446390855.php Company denies problem with voting program. Clovis News Journal. June 3, 2004. By Jack King: CNJ Staff Writer. http://cnjonline.com/engine.pl?station=clovis&template=storyfull.html&id=6358 I followed up with Susanna Martinez, the district attorney mentioned in the article. She provided the additional details in this description.
69
70
24
www.VotersUnite.Org
Software Complexities
A Word about Source Code and Programming
While the vast majority of Americans use software programs, most of them know little about how the software is created. Since much of the debate surrounding electronic voting involves discussion of software, the following information is included for those unfamiliar with software. Many technologists are calling for open source code in voting machines. But election officials and legislators may not be clear on what source code is or why open source code is important. Source code is the list of instructions that cause the computer to display screens, record votes, tally votes, and perform all other functions both visible and invisible. So, for example, when the voter presses the VOTE button, that action triggers a list of instructions for the machine to follow internally. Currently, only the vendors know what those instructions are, and the courts have ruled that those instructions on how to handle votes are a trade secret. In other words, courts have decided that the competitive advantage that secret source code gives to the vendors has a higher priority than the right of the public to know how their votes are being counted. "Open" source code means the instructions would not be a secret. Anyone would be able to look at them. Although the instructions are written in programming language, and few election officials or legislators may be able to understand it, there are many programmers who would understand the language. Nevertheless, finding bugs by reading source code is not an easy task, especially when you consider that complex programs contain thousands, sometimes millions, of lines of instructions and are often not well documented. Here is an example of C++ source code that sorts a set of numbers.
void Sort( INT* ItemArray, INT ArraySize ) { INT Temp; INT i = 0; INT j = 0; INT Offset = ArraySize; BOOLEAN InOrder = FALSE; do { Offset = ((8 * Offset) / 11) ; Offset = ((Offset == 0) ? 1 : Offset); InOrder = TRUE; for (i = 0, j = Offset; i < (ArraySize - Offset); i++, j++) { if (ItemArray[i] > ItemArray[j]) { InOrder = FALSE; Temp = ItemArray[i]; ItemArray[i] = ItemArray[j]; ItemArray[j] = Temp; } } } while (!( (Offset == 1) && (InOrder == TRUE) )); }
Having secret source code is comparable to having vendors write up the procedures they use to count ballots and then refusing to show anyone -even legislators and election administrators -- what those procedures are. And then allowing the vendors to go into a locked room, carry out those secret procedures, and return to public view only to announce the results of the election. Having open source code is comparable to having vendors show everyone the procedures they wrote up, and then allowing them to go into a locked room, carry out those procedures without anyone watching, and return to announce the final results. 25
A low-battery message is written to the memory inside the iVotronic. This is normal. The iVotronic tracks all events, including "low-battery" events.
Here's the bug: The software writes the "low-battery" message BEFORE it moves to a new, blank space in the memory. So, the low-battery message overwrites the previous event message, causing the data to be garbled. Fortunately, they tell us, this bug doesn't overwrite any vote records, just event log records.
Simply reversing the order of the instructions fixed the bug. Similarly, when a vendor makes a "small upgrade" to voting software, it could wreak havoc with the operation of the system.
71
Count Crisis? Elections official warns of glitches that may scramble vote auditing. Miami Daily Business Review. May 13, 2004. By Matthew Haggman. http://www.law.com/jsp/article.jsp?id=1084316008117 For details about both bugs and their interaction, go to http://www.votersunite.org/info/auditbug.asp
72
26
www.VotersUnite.Org
Viewing data
Changing data
Electronic records can be altered undetectably by someone in the next room or a thousand miles away while paper records cannot. Size of data Electronic records have almost no size at all compared to paper records. Data for thousands of ballots will fit on a cartridge the size of a credit card. The same number of paper ballots would fill a grocery bag or two.
One aspect of a computerized voting system has no easy equivalent in the world of paper ballots: the counting process itself. Not only is the vote data electronic, but the counting process is also. This means that election directors' procedures developed to safeguard and count paper ballots cannot be fully translated into procedures for safeguarding and counting electronic vote data. In fact, election directors have no procedures for counting electronic data. All those procedures are electronic processes contained in the software; they are trade secrets of the election equipment vendors.
27
Computer Programmers
Election Officials
Provide Correct Ballot to Voter Collect Ballots
Computer Programmers
Election Officials
Computer Programmers
Design Ballots
Furthermore, while county officials have always overseen registration and voter sign-in books in the past, registration is becoming computerized, in some cases outsourced to private companies.
"A common practice for local election officials is to let election companies run their election make up their ballot, set up their machines, and even count their tallies. This is a dangerous practice." ~ Ted Selker, Cal Tech/MIT Voting Technology Project. 73
Vendor technicians frequently do the ballot programming, which determines how marks on a paper ballot or touches on a screen are translated into electronic vote data.74 Carroll County, North Carolina. November, 2004. A ballot programming error by ES&S caused the ballots to be counted incorrectly. ES&S will supply a new chip for the Model 115 optical scanners and the county will rescan the ballots. 75
Touch to Vote: More Americans to Vote on Electronic, Touch-Screen Systems in November. ABC News. July 18, 2004. http://www.abcnews.go.com/sections/WNT/Politics/e-voting_040718-2.html An ES&S representative told me that ES&S does ballot programming for 1200 counties in the U.S. Computer glitch blamed for miscount in JP voting. Star Tribune. November 10, 2004. By Anna Mathews, CCN staff writer. http://www.greenforesttribune.com/articles/2004/11/10/news/s1.txt
73
74 75
28
www.VotersUnite.Org
Vendor technicians even reprogram ballots directly from the county's equipment. Miami Dade County, Florida. April 2002. An ES&S technician opened the ballot program on the memory cards to change a header. At the same time, he bumped the first candidate to the last position. "When the technician saved the edit, a prompt most likely popped up on the monitor asking him if he was sure he wanted to change the order of the names. The technician ignored the prompt and confirmed the change."76
Vendor technicians test and prepare the equipment for the elections. Napa County, California. March 2004. "Prior to the election, a Sequoia technician ran test ballots through the machine to calibrate its reading sensitivity, but failed to test for gel ink."77
Vendor technicians provide technical supervision during an election. San Diego County, California. March 2004. Diebold provided 200 "Rovers" who were each assigned to monitor a set of polling places in their designated area. 78
Vendors release election results to the media. Walker County, Georgia. March 2004. "A Diebold computer technician began providing incorrect numbers to news organizations. The botched returns were fed to the media for more than two hours after the polls closed before the problem was corrected." 79
Vendor technicians retrieve data from memory cards. Collin County, Texas. November 2004. When election officials could not retrieve the results of the 63 ballots held on a Diebold memory card, they sent the card to Diebold experts in Canada, who retrieved the data.80 Memory cards are electronic ballot boxes. So, a ballot box was mailed to Canada, violating all chain-of-custody procedures.
Vendor technicians investigate when election equipment breaks down. Snohomish County, Washington. September 2004. Twenty touch-screen voting machines broke down during the primary. Sequoia technicians spent a week in Everett testing the machines that broke down, as well as the ones that did not. They did not uncover the cause of the problem. 81
76
Technician's Error, Not Machines, To Blame In Dade Election Mix-Up. The Miami Herald. April 4, 2002. By Oscar Corral. [Purchase through Miami Herald online archives: http://www.miami.com/mld/miamiherald/archives/] Lost E-Votes Could Flip Napa Race. Wired News; March 15, 2004; By Kim Zetter. http://www.wired.com/news/evote/0,2645,62655,00.html Correspondence, written report regarding Touchscreen voting system used for the first time March 2, 2004 by the County of San Diego. From: Walter F. Ekard; Chief Administrative Officer http://www.signonsandiego.com/news/politics/county/20040310-1315-report.html Ballot card problems delayed election returns (Georgia). Walker County Messenger; March 4, 2004; Eric Beavers. Reproduced at: http://www.votersunite.org/article.asp?id=1417 Editorial: Put It on Paper Election snafu points up problems for all-electronic voting. Dallas Morning News. November 11, 2004. http://www.dallasnews.com/sharedcontent/dws/ news/city/collin/opinion/stories/111204dnccocollinvote.9eeb3.html 20 voting machines broke down. Everett Herald. October 6, 2004. By Jerry Cornfield, Herald Writer. http://www.heraldnet.com/stories/04/10/06/loc_voting001.cfm
77
78
79
80
81
29
Vendor technicians troubleshoot and supervise file management activities on the central tabulator during an election. Morris County, New Jersey. June 2004. The tabulation system was unable to read the data from the touch screen machines because the C drive was full. The vice-president of Sequoia led County Clerk Joan Bramhall's technicians through a process that deleted unnecessary information on the drive and "refreshed" the computer. Then they were able to read the cartridges.82
Vendor technicians, unsupervised and unquestioned, repair election equipment. Santa Clara County, California. November 2003. Following the election, Sequoia sent over a group of technicians to make adjustments to voting machines that experienced battery problems. For three weeks, the workers, employed by a Sequoia subcontractor, took apart the machines, removing their circuit boards and making adjustments. Santa Clara County officials didn't know the name of the subcontractor and hadn't verified the identities of the workers it hired when the San Jose Mercury News made an inquiry. They also hadn't documented the changes being made to the machines. To find out such information, "you'd have to contact Sequoia," said Assistant Registrar of Voters Elaine Larson.83
Vendor representatives have even been known to conduct entire elections. Greenwood County, South Carolina. August 2004. "As part of the federal grant agreement, ES&S will conduct the countys first election using the iVotronic, said Connie Moody, director of voter registration and elections for Greenwood County. The equipment will not go into use until November, she said, when county election staff functions mostly as observers and consultants."84
82
Montville and Chatham mayors ousted. Star-Ledger. June 9, 2004. By Lawrence Ragonese and Kristen Alloway. http://www.nj.com/elections/ledger/index.ssf?/base/news-3/108676553355551.xml (paid archives). In a follow-up with Ms. Bramhall, I was told that the Sequoia VP had done the troubleshooting. Electronic voting's hidden perils. Mercury News. February 1, 2004. By Elise Ackerman. http://www.mercurynews.com/mld/mercurynews/news/special_packages/election2004/7849090.htm Local residents will get to test voting machines. Index-Journal. August 19, 2004. By Wallace McBride, senior staff writer. Reproduced at: http://www.votersunite.org/article.asp?id=2554 E-Vote Machines Drop More Ballots. Wired News. February 9, 2004. By Kim Zetter. http://www.wired.com/news/evote/0,2645,62206,00.html
83
84
85
30
www.VotersUnite.Org
Indiana officials may not have been investigating North Carolina elections, so I called to tell them what I knew. Sure enough, they were very glad to find out. Later I read in the news that, after some heated sessions with the election commission, ES&S admitted the earlier version "might not tabulate the votes."86 In other cases, essential information is not communicated as it should be because the vendors fail to follow through on their promises to inform their customers of problems. The ES&S iVotronic audit bug was first publicized in May of 2004.87 In July, ES&S said they were informing their customers about the bug. ES&S spokeswoman Jill Friedman Wilson said the company will continue to communicate with customers about the glitch and the patch. ''We will be working with every customer, and to the extent the solution applies to the equipment they use, help them with it,'' she said. 88 But ES&S failed to inform its other customers. John Gideon, my associate at VotersUnite.Org, contacted election officials in Marion County, Indiana and Bexar County, Texas, both of whom use the flawed machines. Both were completely unaware of the audit bug. I wonder how many of the election officials who need to know actually do know that: The ES&S iVotronic audit bug referenced above may impact their ability to audit their elections. The Sequoia WinEDS system used in Bernalillo County has added phantom votes in every election since it was installed.89 The Unilect Patriot DRE machines have to be specifically set to record more than 3,005 votes. Otherwise, they accept them from voters and throw them away.90 The MicroVote system used in Mecklenburg County, North Carolina tallied the votes incorrectly, and the county officials have no explanation for it.91 The Sequoia Veri-Vote system (DRE with a voter-verified paper trail) failed to print the correct ballot during a demo in Sacramento in the fall of 2004.92
86
Election Commission Bails Out Voting Machine Maker In Time for May Primary. March 11, 2004; http://www.wishtv.com/Global/story.asp?S=1706282&nav=0Ra7LTW3 Count Crisis? Elections official warns of glitches that may scramble vote auditing. Miami Daily Business Review. May 13, 2004. By Matthew Haggman. http://www.law.com/jsp/article.jsp?id=1084316008117 Voting machines to receive digital fix. Miami-Herald. July 10, 2004. By By Mary Ellen Klas. http://www.miami.com/mld/miamiherald/9120544.htm County clerk say phantom votes won't be a problem. KRQE Albuquerque. October 26, 2004. Reproduced at: http://www.votersunite.org/article.asp?id=3421 Unilect vote device causes uproar. Oakland Tribune. November 24, 2004. By David Morrill. Reproduced at: http://www.votersunite.org/article.asp?id=3977 Board of Elections Audits Early Voting Results; Revises Unofficial Results. Released by the Mecklenburg Board of Elections. November 4, 2004. http://www.votersunite.org/info/mecklenburgnewsrelease.pdf Wrong Time for an E-Vote Glitch. Wired News. August 12, 2004. By Kim Zetter. http://www.wired.com/news/evote/0,2645,64569,00.html?tw=wn_tophead_2
87
88
89
90
91
92
31
California Diebold
An audit of Diebold Election Systems voting machines in California has revealed that the company installed uncertified software in all 17 counties that use its electronic voting equipment. ... three counties, including Los Angeles, used software that had never been certified by the state or qualified by federal authorities for use in any election. ... The extent of the changes that Diebold made to upgrades of its software is still unknown.93
93
E-Voting Undermined by Sloppiness. Wired News; Dec. 17, 2003 ; by Kim Zetter http://www.wired.com/news/evote/0,2645,61637,00.html?tw=wn_tophead_2 http://www.votersunite.org/info/Uncertified4.0d.pdf http://www.votersunite.org/info/Downgradedto40.d.pdf Election vendor under fire: Repeated mixups raise issues of trust, reliability. Indy Star; March 4, 2004; By Andy Gammill. Reproduced at: http://www.votersunite.org/article.asp?id=1412
94 95 96
32
www.VotersUnite.Org
Maryland Diebold
Black Box Voting reported that Diebold has used uncertified software in Maryland. According to Diebold employee emails, in the run up to the 2002 election, certified software was installed, but then overridden by uncertified software - before the election making the vote count illegal.97
Software is Uncontrollable
The "National Association of State Election Directors General Overview for Getting a Voting System Qualified," a handbook for election equipment vendors, specifically states: In order to maintain its status as a NASED Qualified system, the hardware and software must be identical to the hardware and software tested by the ITAs. Should it differ even slightly, it would not meet the definition of NASED Qualified and may render the system in noncompliance with state's certification process, so it is incumbent upon the manufacturer to keep their systems current through the ITA process. Vendors know that in most states it's illegal to install unauthorized software. Yet they do it sometimes quite casually. For example, testifying before the California Voting Systems and Procedures Panel regarding the widespread failure of precinct control modules in the March 2004 primary, James Dunn, a Diebold technician, explained that the batteries were refusing to remain charged during the testing before the election. When they discovered discharged batteries, they worked frantically to get them to hold a charge, even to the extent of installing whatever software they thought might work. Mr. Dunn said: We did the software settings with a smart card insert. And they brought out new versions, took the old ones from us, gave us the new ones, and said start using this software. What versions, I'm not sure of.98 Election officials also know that it's illegal to install unauthorized software, but their lack of technical knowledge causes them to rely heavily on the vendors for technical advice, and some of them clearly don't understand the serious potential consequences when software versions "differ even slightly" from one another. As explained on page 25, a software program is created by writing lines of instructions and then running the instructions through a compiler, which converts it into a program that can be executed on a computer. As the iVotronic audit bug illustrates, even one small change to one line of the code can impact the operation of the software, often in unexpected ways. However, many election officials do not understand the integrated nature of software and yet they have the authority to make decisions. For example, in a letter from Washington State Election Director Nick Handy to all Washington State Legislators, he explained why six counties in Washington would be using unqualified software in the 2004 general election: In addition, the software modification only occurred to the "pick a party" feature in the primary. This feature will not be activated in the general election.99 Had he been a software programmer, he would have known that a compiled software program is a single, integrated system component, not separate elements working independently.
97
From Campaign for Verifiable Voting in Maryland to Karl S. Aro, Department of Legislative Services, Annapolis, MD; http://www.truevotemd.org/2003-12-22_Karl_Aro_Letter.pdf Meeting: State Of California Secretary Of State Voting Systems And Procedures Panel. Wednesday, April 21, 2004. Transcript:, page 80. http://www.ss.ca.gov/elections/vsptranscript0421.pdf http://wwwvotersunite.org/info/ToWALegFromNickHandy.pdf
98
99
33
Even if election officials attempted to adhere strictly to their state laws regarding election software, they would have no way of knowing whether the software versions they are using are the ones approved by NASED and their state certification boards. Only by running a cryptographic checksum or hash against systems files is it possible to tell if a software program is identical to the version it is purported to be. 100
http://www.chuckherrin.com/evotingquestions.htm Photo of the wireless port on the Diebold DRE: http://www.truevotetc.org/Photo+Gallery/63.aspx Activists Find More E-Vote Flaws. Wired News. September 22, 2004. By Kim Zetter http://www.wired.com/news/evote/0,2645,65031,00.html
103
34
www.VotersUnite.Org
Administration Complexities
Chain of Custody Complexities Increase Exponentially with Electronic Data
Election Equipment
When votes are cast on paper ballots, they are placed into a ballot box. Carefully developed custody procedures have been in place for decades to ensure the security of the paper ballots before they are counted and afterward in case a recount is needed. Custody procedures become significantly more complex when electronic equipment is used in elections. All election equipment must remain secure before and after the election DREs, optical scanners, control modules, ballot cards (see page 36), and all other software and hardware used in the election process. Procedures for protecting electronic voting equipment are just now being developed. There has not been time to refine them. The current procedures are complex and, even if they were adequate, they are often not used in accordance with the law. Each DRE and precinct optical scanner is a ballot box that must be empty before the election begins and safeguarded to make sure it remains empty. Each one, after being tested and zeroed out, must remain secure until election day.104 Each DRE, precinct optical scanner, and control module also contains a portion of the electronic counting process the software. Once officials are certain that certified software is installed, every single DRE and precinct optical scanner must be secured to ensure that the software contained on it is not illegally altered between elections. Each DRE also contains the software that selects and builds each voter's ballot from the information contained on the access card. That software, too, must be safeguarded from tampering before the election. When machines break down during an election and are removed from service, the chain of custody must be maintained.
Election equipment is particularly vulnerable to tampering after it has been tested but before it is used. For example, San Diego county allowed precinct workers to take the electronic machines home with them after the training program (up to a week before the election) so they would not have to pick them up the morning of the election. 105 As Pam Smith of VerifiedVoting.org pointed out:106 In spite of the vulnerability of Diebold's electronic voting system, the registrar sent computerized voting machines, cards, keys and card encoders to be stored in poll workers' homes before the election, secured only by easily removed stickers and flimsy plastic zip-ties. In one precinct observed by SAVE-Democracy's poll watchers, these security stickers had never even been placed over the memory card ports ---- where votes are stored ---- as they should have been. Poll workers were given extra zip-ties to hold the machines and key-card pouches closed. These were not inventoried and apparently were not even inspected, so no one knows if machines were tampered with.
104
Officials cannot directly ensure that the memory cards have no ballots cast on them. They can only trust the report from the software, which could contain errors or malicious code. (See page 27.) County ordered to offer voters paper ballots. North County Times; February 12, 2004; By Gig Conaughton. http://www.nctimes.com/articles/2004/02/13/news/top_stories/2_12_0422_53_18.txt Electronic voting was a fiasco. North County Times; February 12, 2004; By: Pamela Smith - Commentary http://www.nctimes.com/articles/2004/03/16/opinion/3_15_0422_26_59.txt
105
106
35
107
Electronic S.J. ballot cards lost. The Stockton Record (San Joaquin County); March 4, 2004; By David Siders. Excerpt reproduced on http://www.calvoter.org/news/blog/2004_03_01_blogarchive.html Interview with Freddie Oakley, Yolo County (CA) Clerk-Recorder. May 2004. By Pokey Anderson for Pacifica.
108
36
www.VotersUnite.Org
Neither should voters be disenfranchised by their inability to use an information technology system. However, they have been. For example: Palm Beach County, Florida. October 2004. Sequoia Edge DRE.109 After a woman finished voting, she realized the touch-screen hadnt given her the option to vote on the two referendums for Boca Raton or for state House District 87. She was given the wrong ballot because the computer was programmed for the wrong ballot, but she can't re-vote. Orange County, California. March 2004 110 Poll workers, responsible for giving each voter a four-digit code to enter into the voting machines, gave thousands of voters the wrong ballots. Adams County, Colorado. October 2004.111 Many Denver voters left in frustration when poll site officials were unable to connect their laptops to the central voter registration database to verify the voters. Roanoke, Virginia. November 2004. WinVote. 112 Some ballots were voided because voters left machines before pushing a flashing red button to record their votes. Even election directors continue to encounter new and unexpected problems as they learn to use their electronic election equipment. Elko County, Nevada. December, 2004. Sequoia Edge DRE.113 A month after the election, it was discovered that 271 votes had not been retrieved from the memory cartridges and were never counted. I spoke with Elko County Clerk Win Smith and discovered that the voting machines had been left in test mode on election day, so the upload process had disregarded the votes. Problems had occurred with those machines and cartridges on election day, but Ms. Smith and her staff didn't understand their significance, since they had received no training on using the machines before the election. Bernalillo County, New Mexico. October 2004. Sequoia Edge DRE.114 County Clerk Mary Herrera acknowledged that "phantom votes" (more votes than voters) had appeared in the last three elections, ever since she "upgraded" to the new version of WinEDS (Sequoia's election management system). She claimed her staff has always managed to remove the bogus votes before the final tally. Refer to page 6 for an overview of the 8,656 phantom votes reported by Bernalillo County in November 2004.
109
Balloting Blunders. Boca Raton News. October 30, 2004. By Dale M. King. http://www.bocaratonnews.com/index.php?src=news&category=Local%20News&prid=9999 7,000 Orange County Voters Were Given Bad Ballots. Los Angeles Times; March 9, 2004; By Ray F. Herndon and Stuart Pfeifer, Staff Writers. Reproduced at: http://www.votersunite.org/article.asp?id=2389
110
111
Early Voting Begins In Colorado. October 18, 2004. By Steven K. Paulson, Associated Press Writer. http://news4colorado.com/campaign2004/local_story_292161858.html Back up! We need back up! Roanoke.com. November 11, 2004. By Brian Gottstein. http://www.roanoke.com/columnists/gottstein/13719.html
112
113
Missing votes found in machines. Elko Daily Free Press. December 8, 2004. By Dave Woodson, Staff Writer. Reproduced at: http://www.votersunite.org/article.asp?id=4147 County clerk say phantom votes won't be a problem. KRQE Albuquerque. October 26, 2004. Reproduced at: http://www.votersunite.org/article.asp?id=3421
114
37
115
Some Early Voters Say Machines Mark Incorrect Choices. ABQJournal. October 22, 2004. By Jim Ludwick, Journal Staff Writer. http://abqjournal.com/elex/246845elex10-22-04.htm
116
When the Right to Vote Goes Wrong. TrueVoteMD. November, 2004. http://www.truevotemd.org/Election_Report.pdf Jaime Manzano, Bethesda, Maryland
117
38
www.VotersUnite.Org
Testing of Equipment
All 10,200 touch screens and 1,700 Precinct Control Modules underwent acceptance testing at the Registrar of Voters.
118
Correspondence, written report regarding Touchscreen voting system used for the first time March 2, 2004 by the County of San Diego. From: Walter F. Ekard; Chief Administrative Officer http://www.signonsandiego.com/news/politics/county/20040310-1315-report.html
39
Field Support
The Registrar of Voters recruited 26 Supervising Troubleshooters, who were available in the field from 5:30 a.m. until the polls closed on Election Day. These Supervisors were coordinating and working with approximately 200 Rovers who were supplied by Diebold, each assigned to monitor a set of polling places in their designated area. Rovers began making their rounds at 5:30 a.m. as well.119
Outreach/Public Education
The Registrar of Voters made a significant effort to educate the public on the use of the new system. A public education campaign was developed with the help of the contractor. This included a web site, educational brochures and other written materials, and an instructional video. Additionally, teams from the Registrar of Voters demonstrated the equipment at shopping malls and at community meetings throughout the county during the months preceding the election. More than 60 demonstrations were conducted, reaching more than 5,000 voters.
Election Day
Each polling location received four to eight Diebold TSx touch screen voting machines, based on the number of registered voters, and a Precinct Control Module (PCM). The system used in this election uses an encoded card to give voters access to their appropriate ballot on the touch screen machines. These access cards are encoded by the PCM. The encoded card is then inserted into one of the voting machines to activate the appropriate ballot for each individual voter. Early on election morning poll workers at each polling site removed the PCM from its sealed case and set it up. At approximately 40% of the sites, poll workers found that the machine did not display the expected login screen. Some of the more computersavvy poll workers were able to maneuver through a series of screens until they found the specific login screen upon which they had been trained. Other poll workers did not, as they had only been trained with the expected screen. Therefore, they were not able to perform the card-encoding function. Without the ability to encode the electronic ballot cards at those polls, voters could not vote. There were no back-up paper ballots at the polling locations. Provisional ballots were also electronic. Therefore, many poll workers could not open the polls for voting at 7 a.m. ... At 7 a.m. 64% (1,038 of 1,611) of polling sites were operational. By 8:00 a.m., 88% (1,419) were open and by 9:00 a.m., 98% (1,580) were open. Before 10:00 a.m., 21 more polls were open. Nine additional polls were open after 10 a.m. and the one final poll opened at 11:05 a.m. Rovers were available to help starting at 5:30, and some estimates indicate that the problem may have occurred at as many as 80% of the precincts.
Preliminary Findings
Most aspects of the March 2 election went very well. ... However, there were problems that must be addressed. Most importantly, there was a significant and unexpected problem, which resulted in the delayed opening of 573 out of 1,611 polling places. This inconvenienced many voters, some of whom returned later or went to another polling site, and some who were unable to return at all to vote. There is no method to accurately measure how many voters were unable to vote.
119
All vendor technicians are paid by the county for their services and their travel, adding significantly to the cost of the election.
40
www.VotersUnite.Org
Technical
Approximately 40% of the PCM devices failed to "boot-up" to the correct screen when turned on by the poll workers. Diebold Election Systems, manufacturer of the voting machines has made a preliminary determination that the problem experienced with the PCM devices was caused by an unexpected discharge of the internal battery. This loss of power caused an unfamiliar screen to come up for poll workers upon start up ... The possibility of this large-scale hardware problem was not anticipated by the manufacturer.120 However, it was determined to be a possibility on a smaller scale and 26 supervising troubleshooters were armed with the remedy, as they were for other potential issues that might arise. Technical support in the field was not consistent in that some precincts received support, and others never received a visit from their roving support person.
Management Complexities
Saving Money by Combining Precincts May Decrease Voter Turnout
Some counties are combining precincts to reduce the number of polling places. The reason for combining is two-fold. When county officials first consider the cost of the new electronic voting machines, they realize how much it is going to cost and try to save money by buying fewer machines. One way is to combine precincts, increasing wait times as well as voter confusion. After the first election with the new machines, they realize that they now need twice the number of poll workers they did before they used the complicated new systems. So they consider combining more precincts to cut down on the cost of poll workers and training. Every time a polling place is closed, voters must go somewhere else to vote. This causes some confusion and, in many cases, forces voters to travel further to vote. Comfort levels go down, and some citizens don't bother to vote.
The manufacturer's official claim is contradicted by the testimony of James Dunn, a Diebold technician, who told the California VSSP that battery problems plagued the pre-election testing. Meeting: State Of California Secretary Of State Voting Systems And Procedures Panel. Wednesday, April 21, 2004. Transcript:, page 77-80. http://www.ss.ca.gov/elections/vsptranscript0421.pdf
121
Much of this section is based on information compiled by Robert Kibrick, research astronomer for the University of California Observatories/Lick Observatory; see http://voting.idlecircuits.com/
41
122 123
http://www.elotouch.com/products/accutec/accben.asp
Getting zapped for vote. Clerk's recall election $100,000 pricier after batteries for voting machines lose juice. Rocky Mountain News; February 5, 2004; By Jim Tankersley, http://www.rockymountainnews.com/drmn/election/article/0,1299,DRMN_36_2631038,00.html
42
www.VotersUnite.Org
If such precautions are not taken (and it is likely that they will not, given the level of manpower required to carry them out), the batteries in these voting machines will not provide the number of hours of backup power advertised in the vendors' specifications. Thus, should a power outage occur during an election, it is likely that many of these voting machines may become unusable because their batteries will not have sufficient charge to keep them operating during the outage. For example, in Solano County, California, on March 2, 2004, Diebold TSx voting machines were daisy-chained together and plugged into a single outlet. While this is handy for charging batteries during storage and keeping cables to a minimum during elections, it places a high load on the circuit. The overloaded circuit breaker shut off, and the machines kept running on batteries. Since it affected only the single circuit, poll workers didn't notice the problem. After three hours, the batteries ran down and all the machines shut off. The poll workers called the Fairfield headquarters and were told to hand out provisional ballots until the circuit was reset and the machines were re-started.124 In those cases where a battery has been fully discharged and is unable to take a charge, if the power goes out, the voting machine may shut down without any warning. If this occurs while voters are in the midst of casting their votes, they will be left in limbo, not knowing whether or not their vote has been cast. They will have no way of finding out until power is restored, and it is unlikely that they will be able to remain at the polling place waiting for that to happen. The votes of such voters will thus likely be lost.
Maintenance Challenges
As touch screen voting machines age, like all electronic equipment, they will inevitably develop maintenance and reliability problems. This will require more technically savvy election workers at each polling place, that is, workers who have the skills needed to troubleshoot and respond to such problems. Already, many localities are having difficulty finding adequate numbers of temporary workers to operate their polling places. While some elections officials may argue that touch screen machines will be easier for poll workers to deal with than optical scan paper ballots, it is simply not true that a malfunctioning touch screen machine will be simpler to deal with than a marking pen that has run out of ink.
124
Registrar says electronic vote went smoothly. Times-Herald; March 4, 2004; By Matthias Gafni, TimesHerald staff writer. http://www.verifiedvoting.org/article.asp?id=1428
43
However, even a VVPAT cannot prevent voters from receiving and accidentally casting an incorrect or incomplete ballot a problem that has occurred in many recent elections (see pages 9 and 14 for examples). Nor can VVPAT avoid the ballot secrecy violations that occur when a voter must call a poll worker to help with a malfunctioning DRE (see page 38). Many Americans, discouraged by the multitude of problems that have come to light as election equipment has come under recent scrutiny, are advocating the exclusive use of optical scan equipment or even a return to the simplicity and verifiability of paper ballots, hand counted.
125
An America where votes disappear. April 4, 2004. http://www.oregonlive.com/news/oregonian/ susan_nielsen/index.ssf?/base/editorial/1080997112176372.xml http://verifiedvoting.org/article.php?list=type&type=13 S.202 http://www.leg.state.vt.us/docs/legdoc.cfm?URL=/docs/2004/acts/ACT094.HTM California, Missouri, Montana, New Hampshire, Nevada, Oregon, Vermont, West Virginia http://www.votersunite.org/info/groups.asp
44
www.VotersUnite.Org
130
Optical scan ballots are printed on card stock. Ballots can be available in different languages as needed. Voters cast their votes by filling in the "bubbles" or connecting arrows on optical scan ballots. Voters use lightweight, portable voting booths to afford privacy while voting. These are available in heights to accommodate standing voters as well as ADA-compliant wheelchairheight versions. 131 Once voters complete their ballots, they insert them into the precinct-count optical scanner. Overvoted ballots are rejected, and the voter gets a fresh ballot. Blank ballots produce a warning, and the voter gets the opportunity to correct the ballot. After a ballot is successfully completed and accepted by the optical scanner, the votes on the ballot are counted into the scanner's memory, and the scanner deposits the ballot into a locked ballot box. At the close of the polls, the optical scanner produces a printout of all of the vote totals, the totals are sent to election central, and the locked ballot box is transported to election central in case the ballots were needed for any subsequent recount or audit.
Adding a single ballot-marking device at each polling place fulfills the HAVA requirements for accessibility to the disabled. Two major studies of voting systems determined that precinct-count optical scan systems outperformed DRE voting machines in terms of residual voting errors and cost per voter. 136
130
Examples: ES&S Model 100 precinct-count optical ballot scanner; http://www.essvote.com/HTML/products/m100.html Vogue Election Systems AutoScan; http://www.vogueelection.com/products_autoscan.html
Report on Election Systems Reform. July, 2001. George Washington University, Institute For Communitarian Policy Studies. http://www.gwu.edu/~icps/ExecutiveSummary.pdf "In conclusion, our analysis points to paper ballot, fill-in-the-dot, precinct-count optical mark reader systems as the clear technological option of choice on cost-benefit grounds."
Caltech-MIT Voting Technology Project: Voting - What Is, What Could Be. July 2001. http://www.vote.caltech.edu/Reports/2001report.html
45
Optical scan technology is a cost-effective and accountable system that meets the needs of both absentee and polling-place voters. Increasing capacity simply means adding more low-cost booths, and booths can easily be redistributed on election to decrease lines at the polls. Many believe the limited funds available under HAVA would be better spent on optical scan technology, known to be one of the most reliable voting technologies currently in use. After years of wrangling and protests, Secretary of State Ken Blackwell announced Wednesday that he will limit Ohio's uncompleted voting-machine conversion to a single device: the precinct-count optical-scan machine.137
137
Ohio pulls plug on electronic voting. Blackwell opts for people filling out ballots by hand. Cleveland Plain Dealer Bureau. January 13, 2005. By Julie Carr Smyth. http://www.cleveland.com/news/plaindealer/index.ssf?/base/news/1105612399208372.xml Examples: ES&S http://www.AutomarkTS.com; IVS http://www.ivsllc.com/ Avante http://www.aitechnology.com/votetrakker2/accessible_optical_voting.html http://www.vote-pad.us/
138
139
46
www.VotersUnite.Org
allows voters with visual impairments to review their selections. As they replay the audio tape, or re-read the Braille instructions, they point the wand at each candidate location to receive audible feedback indicating whether or not the location is marked. An opaque, sliding privacy shield sits in a pocket inside the front cover and slides part-way out to conceal the ballot as it is being deposited in a ballot box or precinct scanner.
140
"The Open Voting Consortium (OVC) is a non-profit organization dedicated to the development, maintenance, and delivery of open voting systems for use in public elections." http://www.openvotingconsortium.org/ Open system might plug up holes in the e-voting process. Baltimore Sun; April 1, 2004; By Mike Himowitz. http://www.baltimoresun.com/technology/custom/pluggedin/balpl.himowitz01apr01,0,4595991.column?coll=bal-pe-pluggedin http://gyaku.pair.com/~vote/ballot.html
141
142
47
143
144
Wrong Time for an E-Vote Glitch. Wired News. August 12, 2004. By Kim Zetter. http://www.wired.com/news/evote/0,2645,64569,00.html?tw=wn_tophead_2 The Ballot Integrity Project: Proposal for the Creation of a National Task Force. March 29,2004; By Citizens for Election Integrity http://www.wesavedemocracy.org/docs/BIP_TF-proposal_reviewed_032904.doc
145
48
www.VotersUnite.Org
[Note: Citizen groups like CASE Ohio forced this March 2004 contract to be reconsidered and managed to delay implementation of DREs. In January 2005, Blackwell chose optical scan for Ohio, and limited counties' choices to two vendors, ES&S and Diebold.148]
146
Three companies sign contracts to sell machines. San Jose Mercury News; Feb. 09, 2004; Associated Press http://www.mercurynews.com/mld/mercurynews/news/politics/7913465.htm
147
County to choose voting system; Three companies selected as eligible vendors Newark Ohio Advocate; February 11, 2004; By Lachelle Seymour, Advocate Reporter http://www.newarkadvocate.com/news/stories/20040211/localnews/394868.html Ohio pulls plug on electronic voting. Blackwell opts for people filling out ballots by hand. Cleveland Plain Dealer Bureau. January 13, 2005. By Julie Carr Smyth. http://www.cleveland.com/news/plaindealer/index.ssf?/base/news/1105612399208372.xml
148
49
Thus, while the use of DREs may reduce the total cost of printing paper ballots, the cost per ballot for printing absentee ballots and extra ballots is actually higher. In addition, cost breaks for large orders are reduced, further increasing the cost per paper ballot.
149
County priming public on new voting machines. San Diego Union Tribune; February 29, 2004; By Luis Monteagudo Jr. staff writer http://www.signonsandiego.com/news/metro/20040229-9999-news_1m29voting.html County proceeds on touch-screen voting machines. San Diego Union Tribune; November 26, 2003 ; By Helen Gao, staff writer http://www.signonsandiego.com/news/metro/20031126-9999_7m26vote.html
150
151
Hi-tech voting machines 'threaten' US polls. Scientist warns that electronic votes cannot be safeguarded. The Guardian; Monday February 16, 2004; Tim Radford and Dan Glaister http://www.guardian.co.uk/uselections2004/story/0,13918,1149135,00.html Campaign for Verifiable Voting in Maryland to Karl S. Aro, Department of Legislative Services, Annapolis, MD, Page 2; http://www.truevotemd.org/2003-12-22_Karl_Aro_Letter.pdf
50
www.VotersUnite.Org
deployment of touch screen voting machines to meet the needs of able-bodied voters appears to be an extravagant use of public funds. The hidden costs could easily negate (and overwhelm) any savings achieved from printing fewer paper ballots. Among these hidden costs are: 1. 2. 3. 4. 5. 6. Increased costs for secure and environmentally-controlled storage for these machines when they are not in use. Increased energy costs for keeping the backup batteries charged between elections. Increased labor costs for security when these machines are stored overnight at the polling place before an election. Increased costs for hardware maintenance and software upgrades for each of the thousands of such machines for a typical large county. Increased costs for expendable parts, including the backup batteries and smart cards used by these machines. Increased labor costs for verifying that each machine has the correct version of the software and firmware installed immediately before the start of every election and again immediately after each election is concluded. Increased labor costs for individually performing logic and accuracy tests on every one of thousands of machines prior to the start of every election and again immediately following each election. Increased labor costs for hiring additional poll workers (San Diego doubled the number of poll workers when it switched to DREs). Increased costs for poll worker training, both for longer training sessions and larger number of poll workers to train on using a much more complicated system.
7.
8. 9.
10. Massive costs for replacing these machines when they age and the technology they employ is no longer maintainable or supported by the vendor.
Example: Vogue Election Systems SpaceSaver 2000, http://www.vogueelection.com/products_spacesaver.html Example: Vogue Election Systems Vogue-I voting booth http://www.vogueelection.com/products_votingbooth.html
153
51
Enhancing each polling place's optical scan system to include an accessible voting method for the disabled would only require the addition of one ballot-marking device to accommodate the needs of blind, visually-impaired, and language-impaired voters. The devices print directly onto the same ballots completed manually by other voters and fed into the optical scanner. The AutoMARK, a ballot-marking device marketed by ES&S, was selling for about $5400 in the second quarter of 2005.154 So, figuring approximately $5,400 for the addition of the ballot-marking device, the following table shows the cost of three sizes of polling places: Polling Place Size 1-voting booth 5-voting booths 10-voting booths Optical Scanner $5,000 $5,000 $5,000 Voting Booths $250 $1,250 $2,500 Marking Device $5,400 $5,400 $5,400 Total $10,650 $11,650 $12,900
Precinct-count optical scan systems can provide a significant savings over DRE systems, even considering the additional costs of printing ballots, especially when you compare them with DRE systems that have voter-verifiable printers attached. The CalTech/MIT Voting Project report, published in July 2001, estimates that the cost of purchasing DREs without a voter-verifiable printer is over three times the cost of purchasing optical scanners. While the cost of operating DREs is about half the operating cost of optical scanners, it would take about 20 years of operation before the overall costs would be equivalent and this doesn't account for the other hidden costs of DREs outlined on page 50. The CalTech/MIT report says: "Even though optical scanning systems have much higher operating cost, the difference in the acquisition cost is sufficiently large that the total cost of the optical scanning system is somewhat lower over the fifteen-year operating life of the machinery. If we assume a twenty year lifespan, the costs are identical." 155 The CalTech/MIT study determined the acquisition and operating costs per voter as follows: Machine type DRE (Touch screen) Optical Scanning (in Precinct) Acquisition $18-25/voter $6-8/voter Operating $0.5-1/voter $1-2/voter
The Vote-PAD
This non-electronic assistive device (see page 46) can be used in conjunction with an optical scan system or hand-counted paper ballots.156 The cost is a minimal addition to the cost of printing ballots and could provide a reasonable solution, particularly for small paper ballot jurisdictions that do not want to, or cannot afford to, computerize their elections. The initial package costs approximately $1800 to $2200 for one polling place. This provides all necessary equipment and materials including twenty Vote-PAD booklets for use with twenty different ballots. At one Vote-PAD per polling place this would represent 20 elections or, on average, five years of elections.
154 155
The price of the AutoMARK has increased from $4,500 since the first publication of of "Myth Breakers." CalTech/MIT Voting Project, pages 23, 24. http://www.vote.caltech.edu/Reports/ At the time of this report, ballot-marking devices were not yet available, and voter-verifiable printers were not included in the capital costs of the DREs. http://www.vote-pad.us
156
52
www.VotersUnite.Org
157 158
53
159
Analysis of an Electronic Voting System. July 23, 2003; By Tadayoshi Kohno, Adam Stubblefield, Aviel D. Rubin, and Dan S. Wallach http://avirubin.com/vote/ http://www.dbm.maryland.gov/dbm_search/technology/toc_voting_system_report/votingsystemreportfinal.pdf Risk Assessment Report: Diebold AccuVote-TS Voting System and Processes. September 2, 2003; by SAIC. http://www.dbm.maryland.gov/dbm_search/technology/toc_voting_system_report/ votingsystemreportfinal.pdf
160 161
54
www.VotersUnite.Org
http://www.sos.state.oh.us/sos/hava/files/compuware.pdf http://www.raba.com/press/TA_Report_AccuVote.pdf E-Vote Still Flawed, Experts Say. Wired News; January 29, 2004; by Kim Zetter http://www.wired.com/news/business/0,1367,62109,00.html Ms. Jekot has been a computer programmer and consultant since 1984. She owns two copyrights on an accounting package. She has taught introductory programming at Lanier Technical Institute in Gainesville, Georgia and is currently working on an medical application, due for release in 2005. Interview with Roxanne Jekot of Georgia (CountTheVote.org). January 2004. By Pokey Anderson. http://www.countthevote.org/av/VernaMP3.mp3
165
166
55
The other thing was the division of a group of people working on a set of problems, and then a second group of only three people who worked on really critical programs, programs that would be required to manipulate the vote. Those were the two things that I think stood out, almost immediately. Again, the unprofessionalism. As an instructor, someone who's taught, I was surprised that the majority of the code looked like my first year college students, you know, doing a project. Very amateurish, very poorly put together. Very badly laid out, very poorly planned.
Some Officials' Claims about Electronic Elections Don't Match the Facts
Georgia Secretary of State Cathy Cox
The claim: "Though Georgia Secretary of State Cathy Cox said the states 26,000 elections voting machines performed without any problems on Super Tuesday earlier this week, some lawmakers Thursday said the machines may nonetheless be vulnerable to fraud and wanted printed receipts to serve as proof of the computer tabulation."167 The facts: Walker County, Georgia. March 2004 Super Tuesday. Walker County election officials worked until after midnight, following Tuesdays election, to rectify problems tallying results. Problems became apparent with Walkers first returns about 9 p.m. when neighboring counties were wrapping up their tallies. A Diebold computer technician began providing incorrect numbers to news organizations. The botched returns were fed to the media for more than two hours after the polls closed before the problem was corrected. The voting machines have been used for six elections, three of which were for the same State House District 1 race. Problems have cropped up at every election.168
Printers Wanted for Voting Machines. WXIA-TV Atlanta; March 5, 2004; Reported By: Keith Whitney http://www.11alive.com/news/news_article.aspx?storyid=43824 Ballot card problems delayed election returns (Georgia). Walker County Messenger; March 4, 2004; Eric Beavers. Reproduced at: http://www.votersunite.org/article.asp?id=1417 Lou Dobbs Tonight. CNN; March 8, 2004; http://www.cnn.com/TRANSCRIPTS/0403/08/ldt.00.html Buyers' remorse. Broward Daily Business Review; February 3, 2003; by Matthew Haggman http://verify.stanford.edu/EVOTE/buyers-remorse.html Election glitch missed 103,000 votes in Florida county. CNN; November 8, 2002 http://www.cnn.com/2002/ALLPOLITICS/11/07/elec02.florida.votes.missing/ Human goofs, not machines, drag vote tally into next day. The Palm Beach Post, 14 March 2002; reported in "Black Box Voting" Chapter 2 by Bev Harris
168
169 170
171
172
56
www.VotersUnite.Org
7% presidential undervote county wide. 2 precincts had 25% presidential undervotes; 2% rate county wide Polls opened late. All 6 punch card machines damaged in transit.
One precinct recorded 3893 phantom votes for Bush. Overcharged e-voting systems failed to start up.
The claim: January, 2005. "Spokesman Carlo LoParo said these [precinct-based optical scan] machines - long Blackwell's favored technology - produce the required paper record and are more flexible and affordable than electronic machines."175 The facts: March 2004. "Mr. Blackwell hopes to convert 14 counties using punch-card ballots to electronic voting by August special elections and 13 more for the Nov. 2 election." 176
173
Votes from 134 residents were not counted. Miami Herald; January 07, 2004; By Erika Bolstad; http://www.miami.com/mld/ miamiherald/7655457.htm
174
How Ohio pulled it off. The Washington Times. November 17, 2004. By J. Kenneth Blackwell. http://washingtontimes.com/op-ed/20041116-085742-1497r.htm Ohio pulls plug on electronic voting. Blackwell opts for people filling out ballots by hand. Cleveland Plain Dealer Bureau. January 13, 2005. By Julie Carr Smyth. http://www.cleveland.com/news/plaindealer/index.ssf?/base/news/1105612399208372.xml
175
176
Expert warns of balloting devices. Toledo Blade. March 18, 2004. By Jim Provance, Blade Columbus Bureau. Reproduced at http://www.votersunite.org/article.asp?id=1620
57
177
178
179 180
181
182 183
Blind voters rip e-machines: They say defects thwart goal of enfranchising sight-impaired Mercury News; May 15, 2004; By Elise Ackerman http://www.mercurynews.com/mld/mercurynews/news/breaking_news/8673336.htm
58
www.VotersUnite.Org
[Noel Runyan, a blind voter and computer scientist who is an expert in designing accessible systems,] said the voting companies appeared to have ignored feedback they solicited from groups of blind voters as they were developing their systems. "I personally want them to be decertified for this election,'' Runyan said. "We need to make a strong statement that all these machines need to be redesigned on the user interface side. We've got a mistake here." In a conversation with VotersUnite, Noel Runyan said that these criticisms of the machines accessibility did not constitute a formal position of the Silicon Valley Council of the Blind. He did, however, emphasize that, many of the visually impaired voters found that it was very frustrating and difficult or impossible to vote with the Sequoia voting systems. He also pointed out, One of the most glaring problems with the Sequoia system is the fact that it did not permit simultaneous speech output and large print display, as an option. Many low vision voters need to use both their hearing and whatever eyesight they may have to manage to navigate successfully through an electronic ballot. Forcing these folks to use either only speech or only large print will not permit them to vote independently. Mr. Runyan also discussed the accessibility features of two other systems he has used: Diebold and Avante. He said that the Diebold systems provide audio and large print simultaneously, but the Diebold speech quality and response time were so poor that elderly voters and others with hearing problems would have serious difficulties understanding the speech of the systems. Avante, he pointed out, uses a low-quality synthetic TTS (Text To Speech) system that is difficult for most older folks to understand. Use of the synthetic voice was the primary reason that the Avante scored at the bottom in the Access World review.
No System Provides a VVPAT for the Voter to Remove from the Polls
The term "receipt" has been used by many who advocate a voter-verified paper audit trail. They point out that you get a receipt for deposits made at the ATM, and you should get a receipt for your electronic vote as well. Some proponents of paperless voting claim that VVPAT systems would allow voters to remove the paper records from the polling place, thus allowing for voteselling and coercing of voters.185 However, in a VVPAT system the "receipt" is preserved like the ballot itself, and no system provides a ballot that the voter removes from the polls. All VVPAT systems require the paper to be retained by election officials and securely stored at the polling site.
184
Questions and Answers on Direct Recording Electronic (DRE) Voting Machines, page 5. The League of Women Voters of the United States. 2003. Questions and Answers on Direct Recording Electronic (DRE) Voting Machines, page 8. The League of Women Voters of the United States. 2003.
185
59
Memorandum Opinion For The Principal Deputy Assistant Attorney General, Civil Rights Division. October 10, 2003; http://www.usdoj.gov/olc/drevotingsystems.htm http://www.electioncenter.org/. Click on Membership Information. Electronic Voting Machines Blasted by Scientists, Hacked by Author. Scoop. July 25, 2003. http://www.scoop.co.nz/mason/stories/HL0307/S00198.htm http://verify.stanford.edu/EVOTE/ECresponse.html
187 188
189
60
www.VotersUnite.Org
... Its executive director, R. Doug Lewis, confirmed this week that the center had taken donations from makers of electronic voting machines - Sequoia Voting Systems Inc. of Oakland, Calif., and Electronic Systems & Software Inc. of Omaha, Neb. In addition, donations came from "probably Diebold" Inc. of North Canton, Ohio, Lewis said. ... Lewis issued a report last year saying that "well-intentioned people, some of them even highly educated and respected, scare voters and public officials with claims that the voting equipment and/or its software can be manipulated to change the outcome of elections." The report went on to say: "Do not be misled into believing that elections are reliant upon technology which can be manipulated..." Lewis said he did not think accepting donations from the manufacturers presented any conflict of interest or breach of ethics. 190 Activities sponsored by the Election Center have been questioned by election officials as well as concerned citizens. For example, in August 2004 the annual Election Center conference for election officials included a dinner cruise down the Potomac, sponsored by Sequoia Voting Systems. Freddie Oakley, County Clerk/Recorder in Yolo County, California, commented: I have been distressed for a long time that the Election Center, which holds itself out as the impartial association for election officials, engages in and appears to encourage this kind of industry subsidy of get-togethers of election officials.191
190
Excerpt from: Group that called electronic vote secure got makers' aid. Philadelphia Inquirer; March 15, 2004; By Linda K. Harris. http://www.mercurynews.com/mld/mercurynews/news/world/8273865.htm
191
Press Release: Voting Machine Manufacturers Wine and Dine Election Officials. Institute for Public Accuracy. September 1, 2004. http://www.commondreams.org/news2004/0901-08.htm Excerpt from: Lobbyist made money from touch screen sales. St. Petersburg Times; October 6, 2002. Associated Press. http://www.google.com/search?q=cache:WJT4jZ8WH0AJ:www.sptimes.com/2002/10 /06/news_pf/State/Lobbyist_made_money_f.shtml+Florida+%2B+ES%26S+problems&hl=en&ie=UTF-8
192
193
Winners and Losers of 2003: Worst Technology. Fortune Magazine; December 2003; By Peter Lewis http://www.fortune.com/fortune/technology/articles/0,15114,558787,00.html
61
Afterword
One of the difficulties that faces a software company is the disconnect between the engineering department and the sales department. In order to please customers and make a sale, salespeople promise features that the engineering department is then obligated to develop. Sometimes, there is a serious discrepancy between the promises and the deliverables that are even possible often because the sales people are not tech-savvy enough to know that what they are promising just won't work, sometimes because they are just too excited about the potential sale. I was discussing this recently with a client who used to be a software developer and is now a software salesman. I pointed out that he had an advantage since he knew what would be possible for his engineering department to develop. He responded by telling me a joke. Q. What's the difference between a used car salesman and a software salesman? A. The used car salesman knows when he's lying. In an attempt to "bring elections into the 21st century," election system manufacturers promised Congress and the elections community that their computer-based systems provide the following features: Records and tallies votes accurately Ensures the secrecy of the ballot Operates reliably on election day Provides security from tampering Allows the disabled to vote independently Ensures a satisfactory voting experience for voters Prevents over-voting and reduces inadvertent under-voting Simplifies the administration of elections Reduces the cost of elections Provides a paper audit trail Meets federal standards for voting systems
The facts presented in this document show that no such electronic election system presently exists. Is such a system even possible? My experience in the software development industry gives me grave doubts, but even if it were possible, would it provide the transparency required for democracy? No. When the founding fathers established a democratic republic in the Constitution, they did not promise that keeping it would be convenient.
62
www.VotersUnite.Org