News Reporst of E-Voting in the 2004 U.S. Election

Following the messy 2000 U.S. presidential election in Florida, the federal government passed the “Help America Vote Act”. It encourages each state to adopt a “direct recording electronic voting system” (U.S. HAVA 2002 Sec. 295 a.3) . In the 2004 election electronic voting machines were widely used across the United States, despite serious concerns about their potential for error and fraud. This analysis examines news coverage of this issue.

The Technology

There are two main kinds of computerized voting machines in the United States: touch screen and optical scan (older electronic technologies are beyond the scope of this paper). Touchscreens offer the voter on-screen choices, which he or she must touch in order to cast a ballot. Optical scan machines instead accept a paper ballot completed by the voter. In both cases, the machine is controlled by software and the results are recorded in computer memory.

Compared to paper ballots, electronic voting offers one great advantage: in tabulating votes, there is no potential for human error. It is also flexible and fast in a country where ballots are varied and complex. Critics of voting machines warn of several dangers:

1. As with all software, there is a potential for bugs or for fraudulent code.

2. Computer security is never perfect – it may be possible for outsiders to break in and alter the voting record.

3. If there is doubt about the result, recounts are impossible without a paper record.

4. Voting software is not transparent. Any problems or tampering will likely be invisible to observers, possibly even if there is a paper record.

This fourth point is critical. An electronic voting machine is general purpose computer capable of performing virtually any calculation at all. For example, it would be possible to write software which would record the opposite of what the user selects – but only on November 2 between 8 AM and 8 PM. Such fraud could be automated on a large scale, and neither testing nor monitoring would reveal it to observers.

It is axiomatic of computer security that software can never be completely secure. The only way to be know that the software is operating correctly is for a trusted technical expert to examine the source code, or to independently verify the result. But software is written and controlled by people – if people can be corrupted, so can code. Both the software and the voting data are vulnerable to security breaches.

Why then do we trust machines for other tasks, such as bank transactions? The main reason is that those whose interests are at stake are able to audit the results, e.g. through paper bank statements. With voting, on the other hand, there is no monthly statement. Even if the machine issues a paper receipt, as voters we can't look for our vote on the list and check that it is correct. Voter interests do not coincide with the interests of others: for elections officials and manufacturers a smooth vote may be more important than accuracy, and political interests conflict with the interests of voters.

Electronic voting can never be as transparent as paper ballots. But if it is used, critics of recommend following Australia's lead and use open source software, whose operation can be examined by interesting parties (Economist 40). More importantly, groups such as Verified Voting demand a “voter verifiable audit trail” and frequent manual recounts – even when things appear to go smoothly – so that errors or fraud can be detected. As a computer scientist, I concur.

Meanwhile, there have been questions about the manufacturers. The CEO of Diebold, one of the main companies, famously said in 2003 that he was “committed to helping Ohio deliver its electoral votes to the president next year” (quoted in Economist 39). And testing and certification of the software, while inadequate protection, has not yet taken place (Economist 40). In an analysis of the Diebold source code in 2003, Kohno, Stubblefield, Rubin, and Wallach (1) determine that their system is so insecure that “voters, without any insider privileges, can cast unlimited votes” – the system “is unsuitable for use in a general election”.

Framing & Ideology

This is not an exhaustive survey of news reports. I have focused on five stories during the period of early voting leading up to the election, when reports of e-voting were most common. The risks of electronic voting are not an news “event” so the media rarely report on it except in the context of another story, such as Florida early voting.

The first story is a brief report from CNN Headline News, covering the first day of early voting (see Appendix II). It begins by stating “already there are reports of problems” – implicitly framing the vote as a potential repeat of 2000. Its conclusion belies this, stating that “overall voters say they're happy with the system”. By referring to the opinions of voters, this disclaims any knowledge of fairness and accuracy with the system, and relegates problems to irrelevance.

The second is a New York Times front-page story by a staff writer, “In Florida, Early Voting Means an Early Return to Problems”. The framing here also the 2000 Florida election and its problems. The comparison is confirmed in the first two sentences, which states that “like persistent ghosts, some of these problems immediately resurfaced”. The use of the word “some” this immediately limits the scope of criticism.

The Globe and Mail also ran a report on the 19^th from Associated Press and Reuters: “Glitches Few in Florida's Early Voting”. The dominant framing is again the the Florida election of 2000, although the article suggests that the process is much smoother this time. The first sentence confirms that “only a few glitches were reported”.

USA Today's front-page staff report on the 22^nd also frames the story in terms of 2000 as the title, “Florida is Focus of Furor Again” makes clear. It details a number of problems, lawsuits, and concerns about the first four days of early voting.

Finally, the ABC News ran a report on the 21^st, titled “Legal Logjam” (see Appendix III). Both Democrats and Republicans had teams of lawyers ready in case of problems; the frame of the story is that lawsuits are the greatest danger to the vote – it ends stating, “the only thing that will save the voters from the lawyers is a clear victory for one side or the other”.

Florida 2000

Florida in 2000 is a common thread in all five stories. USA Today and the Times reference the previous election explicitly in their headlines. The Globe makes the connection in the first sentence, which calls Florida “the U.S. state synonymous with the election fiasco of 2000”. All the reports are explicitly about Florida, with the exception of the ABC report, which is nevertheless reported from Miami.

The reason for choosing Florida is clear: conflict. The furor in the previous election may promise more of the same for journalists seeking a story. Furthermore, polls had indicated the result might again be close.

USA Today only mentions incidents in the previous election: the “six-week recount”, and “bitter disputes” in Duval county, including discarded ballots. ABC makes little reference to 2000, other than to say that “the Democrats are so determined to avoid a repeat of election 2000”. The Globe only mentions electronic voting machines in reference to 2000, machines which were “introduced since the troubled 2000 election”. In the context of the dominant frame of the story, which is that “glitches” are few, this implies that e-voting is the solution to problems.

The Globe's spin is directly contradicted by the Times and CNN stories. The first sentence in the CNN story compares problems this year with problems then: “already there are reports of problems”. The Times is equally blunt about trouble with its headline, and the first paragraph confirms this framing, calling the problems “persistent ghosts”.

Regardless of different reporting about the success of voting this time around, framing the election in terms of Florida in 2000 sets up a dangerous contrast: the Florida vote in 2000 was a mess, so improvements – or apparent improvements – would make this election a relative success. Similarly, with 2000 as a baseline, problems this time around will appear simply as more of the same. This bounds any debate about electronic voting machines: they could be better, or no worse.

The other problem with Florida is that it ex-nominates the other states. None of the types of problems mentioned were unique to Florida (none mentioned one problem that was: chaos caused by the recent hurricane). By focusing on Florida, the reports imply that problems there are exceptions to the rule. The ABC story does look beyond Florida, showing a map of 25 other states using electronic voting machines. Mentions of the rest of the country are sparse in the other reports, even though on that day early voting started in three other states and was already underway in four more (see Appendix I).

Specific Coverage of E-Voting

While none of these stories even mentions the risk of e-voting fraud, a 15-minute piece on PBS on 19 October did reviewed problems with the election system as a whole, and raised the issue explicitly: “allegations persist that they [e-voting machines] can be programmed to rig elections”. But the story did not expand on these “allegations”, other than to mention the value of a paper trail in case of “problems”. There is no mention of the equipment manufacturers or political interests that might lead to fraud.

Several stories suggested the electronic machines were working well. The CNN story, despite its “reports of problems” and a statement that “the touch screen system briefly crashed . . . in Orlando and surrounding suburbs”, concludes that “voters say they're happy with the system”. The impression is that these are teething troubles (there are only “reports” of problems and the crash was “brief”) and there is no serious risk with the machines; images of orderly rows of touch-screen machines contradict the idea that anything could be wrong. The Globe claims there were “no early reports of problems with the ATM-like touch-screen voting machines” – implying the voting machines are as dependable as automated banking.

The Times, despite framing its story in terms of “an early return to problems”, only mentions electronic voting once, and then only in the context of nameless voters who have “suspicion” about the machines. Readers might be inclined to put this down to latter-day luddism: unfortunately there is a tendency to dismiss such fears on the basis that “computers don't make mistakes”. USA Today's list of problems places e-voting in the context of lawsuits demanding “paper documentation of votes cast on computerized machines”.

The range of problems or risks associated with e-voting, according to these stories, ranges from none (according the Globe) to a “brief” crash (CNN) and unexplained “suspicion” (the Times) and a similarly unexplained lack of “paper documentation”. The scope of any debate about the machines is effectively limited to technophobia, legal, and logistics issues. In fact, counting all five stories, there were over twice as many mentions of logistics problems (14) and other ballot technologies (paper, mail-in, provisional, and absentee: 14) as of electronic voting (6). Just as the focus on Florida ex-nominates other states, so the focus on other issues makes electronic voting seem like less a problem.

Voting Problems in General

Electronic voting is mentioned in all five stories, and with the exception of the Globe all of the stories at least imply it could be a problem. Is the risk of general voting problems and/or fraud – not limited to e-voting – really the message of these stories?

Two of the reports, in the Times and USA Today, are explicitly critical of voting in their headlines, and the CNN story immediately talks about problems. ABC's lawyers imply problems which are discussed in the text. The Globe admits to “glitches” - a word which itself suggests the problems are transient and not systemic; a positive contrast with the story on the opposite page about a referendum in Belarus which fell “well short of democratic standards”. The Globe story is the only one with an almost uniformly positive framing of the vote; the other stories are either explicitly critical (the Times, USA Today, CNN), or implicitly (ABC's lawyers are there to deal with problems). The USA Today story takes an objective stance: it catalogs problems as isolated and discreet with no attempt to build perspective or a pattern, or even any reference to voting in other states.

Television imagery on CNN and ABC is uniformly positive about voting procedures. On CNN, the on-screen graphics show a patriotic American flag with the positive symbolism of a check mark in a box clearly indicating that all is well. The more positive connotations of the check mark are suspect, given that voters traditionally mark an X. A shot of a sign with a flag on it and the world “Vote Here” reinforces the message. In both stories, images show lines of people voting and orderly rows of touch-screen voting machines. Whatever the announcers might say, the reports clearly show people voting without problems.

The fundamental attitude to problems with electronic voting or even voting in general is that they are temporary details (crashes are “brief”) to be worked out - “glitches”, details of paperwork – paper receipts - which don't bear explaining, of the attitude of voters with “paranoia” (USA Today) or “suspicions” (the Times). These stories present isolated difficulties as exceptions to the rule that everything is fine.

Citizenship and the Contest

Voters are participants – as are the media – and citizens. The issue of voting is not a neutral issue with two sides: there can be no balance between voting and disenfranchisement. It is the responsibility and right of the citizen to vote, and it is the responsibility of the media to support that action.

These stories keep their objective distance. They treat the election as a spectacle or a contest (all three newspaper reports are adjacent to stories about polls). They interpellate their audience – presumably citizen voters in the case of all but the Globe – as spectators.

The ABC story is the most clear on this subject. It quotes a Democratic election worker who says she is there “to ensure everyone's right to vote is protected.” Lawyers are part of a mechanism which could challenge fraud. But they are presented as enemies of voters: “the only thing that will save the voters from the lawyers is a clear victory”. The focus on the profession – which for many many represents greed and deceit – distracts from the potential for fraud itself. For ABC, the interest of the voters is not the protection of their right to vote: it is an uncomplicated election. They are merely consumers of news, observers who don't want to have to wait. There is a strange long shot of Jeffrey Kofman in front of a business jet explaining they are to be used by the Democrats to fly lawyers to trouble spots. These lawyers work for the parties, not the voters, but at least they would challenge fraud. The wealth and power symbolized by the jet distances them from the voters, and makes the election a game for elites.

The Globe takes a similar tack. The second paragraph introduces Florida as a “battleground” and talks about Kerry's accusations against Bush, reframing the focus from voting to the election as contest. Similar language is used three more times – “war of words”, “attacks”, the election as a “race”. The second half of the article is devoted to the conflict between the two presidential candidates. The surrounding stories, “Partisan Ads Flood Ohio TV Screens” on the left with its large picture of boisterous campaigners holding signs, and “U.S. Political Polls Called 'Numerical Tower of Babel'”, only reinforce this perspective. Voting is simply part of the game. The Globe is Canadian, but for the mouse, the elephant's election is never just a game.

The USA Today story, while it presents a list of facts, is infected with the same spirit of competition. There are six occurrences of words suggesting a battle or contest: “biggest guns”, “political maneuvering”, “tie”, “won”, “lost”. Beneath the article is a color graph of a Gallup Poll, above it a banner about baseball. The conclusion also detracts from the seriousness of the issue: a voter says she is “a little paranoid about this election”.

Some of these terms are artefacts of English usage, but the papers play up the contest. Beyond alienating the audience from the election, this also alienates them from their votes. Contest language redirects any conflict so that it is between the parties: if there is a problem with the vote, it appears to be a problem for them, not for the voters. We can see this in several of the stories, particularly the Globe, where there are 7 quotes from party members and none from voters, elections officials, or other experts; ABC follows a similar pattern (see Appendix I for details). Ideologically, the election is presented as a game in which citizens' main role is as observers.

The Outcome

Since the election there has been some media coverage of the issue, which unfortunately arrived too late for detailed analysis. Initial reports were brief articles, such as those in the New York Times (Times Nov. 5) and National Post (Agence-France). ABC News ran a story on 9 November titled “Conspiracy Theories Abound after Bush Victory”, which concluded by blaming the Internet and quoting Mark Twain: “a lie can travel halfway around the world before the truth can put on his shoes” (Tapper & Miller). The Globe did not mention the issue until a half-page report in the Focus section on 20 November, titled “No One Cheated (But They Could Have)” – a conclusion itself contradicted by the content of the article, which quoted a computer scientist who claimed that if he had wished to, he could have formed a voting machine company and stolen several elections already (Freeman). The New York Times addressed the issue on 12 November with two articles, “Vote Fraud Theories, Spread by Blogs, Are Quickly Buried” (Zeller), and “Mostly Good Reviews for Electronic Voting” (Schwartz). Salon ran an article on 10 November, “Was the election stolen?” (Manjoo), which claimed “The system is clearly broken. But there is no evidence Bush won because of voter fraud.”

The problem is that even if the election result was accurate, the system is broken. This is not a partisan issue. What has happened here is that challenges to the process have been conflated with political challenges to the outcome. The media, attempting to appear balanced, cannot be seen favoring the Democrats even in the pursuit of truth.

Despite media dismissals, numerous cases have come to light which indicate there were indeed instances of problems or fraud with the electronic voting machines – some of which they reported. For example, several machines were unable to count the number of ballots submitted – in Florida, for example, beyond 32,000 votes the machine actually started subtracting (Freeman)¹. Another, in Ohio, gave Bush an extra 3,893 votes – in a precinct with a turnout of 638 (Agence-France). According to the Globe, the reason for this is still not know (Freeman). In San Francisco, machines didn't tabulate correctly (Times 5 Nov.). In the case of Ohio, the error is obvious. Given the opacity of the process, if the number were incorrect but did not exceed the number of voters, would we ever know? The problems we are aware of may only be the tip of the iceberg: a recent study from Berkeley found at least 130,000 Bush votes in Florida which it claims with 99.9% accuracy cannot be attributed to chance (Hout, Mangels, Carlson and Best Summary).

Why are the media so reluctant to take the issue of fraud seriously? From the perspective of a disinterested observer, representative democracy has three main functional advantages: 1) leadership transitions are smooth; 2) voting provides an outlet for dissent and conflict; 3) universal participation confers legitimacy. In other words, one of the main achievements of this system of elections is stability – something highly desirable for entrenched interests, like the media conglomerates and their advertising customers (including Democrats, who likely have more to lose from instability than from an election loss; this may explain their acceptance of the outcome).

This amoral stability is one of the driving ideologies of the election coverage. These articles only examine illegitimacy in terms of its appearance and its threat to stability – legal disputes, suspicious voters, unstable computer systems – and not its threat to the rights of voters. They frame the system as explicitly stable – degrees of difference from Florida 2000, election machines like ATMs, brief problems and glitches. Otherwise, problems with voting are chiefly a threat to the audience as consumers: they may lead to confusion and delay. The system itself is never truly threatened.

Conclusion

The electronic voting machines used in the 2004 U.S. presidential election present a significant opportunity for fraud on a massive scale. This was known long before the election, but coverage of the issue in the mainstream media was extremely weak: in no case examined here were electronic machines presented as fraud-prone. The nature of e-voting fraud can make make it difficult or impossible to prove, so the duty rests with knowledgeable parties – particularly the media – to publicize the danger so that voters can insist that systems are in place to guarantee that their votes are counted. The media failed in this duty; instead they presented the election as a contest, the voting process as stable, and their audience as neutral observers. The protections in place in 2004 were entirely inadequate. It is entirely possible, even with the 3% margin between candidates, that this election was stolen. The safeguards were so weak that we cannot prove it either way. Given the sparse media attention to the serious problems which have come to light, and the dismissal of even the possibility, most citizens may never even know that fraud was possible, or that no electronic election can be trusted until the system is fixed.

Works Cited

further references available at <http://del.icio.us/geof/e-voting>

Appendix I: Content Analysis

Mentions of...	CNN	NYT	Globe	USA	ABC	Total
words suggesting the election is a contest or war (“battle”, “battleground”, “race”, “war of words”, “attacks”, “strategy”, “biggest guns”, “maneuvering”, “win”, “lose”, “tie”, “army”, “jockeying”)	0	0	4	6	3	13
electronic voting	2	1	1	1	1	6
ballots (other than electronic: paper, provisional, absentee, mail-in)	1	6	1	3	3	14
logistics problems (actual – not hypothetical or predicted – in 2004): long lines, insufficient resources, computer trouble, voting hours	1	10	1	2	0	14
negative mentions of voting (in 2000)	0	3	2	4	1	9
Quotes by (or attributions to)...
- Democrats (not presented as voters)	0	1	4	1	3	9
- Republicans (not presented as voters)	0	1	3	0	1	5
- elections officials	0	10	0	0	11	11
- experts (scholars, lawyers, etc.)	0	2	0	0	1	3
- voters (even if party affiliation is given)	1	2	0	1	0	4
“Florida”	2	8	3	3	1	17
the name of a state other than Florida	3	3	92	13	0	13

¹ This individual was not identified, but he is presented as an elections official: he works at a polling station and explains provisional ballots, and no party affiliation or expertise is mentioned.

² Two of the other states listed are not in the context of voting: one simply states that Kerry is a Massachusettes senator, the other that Bush visited New Jersey.

³ Not mentioned in the context of voting: “In Ohio . . . the candidates are in a virtual tie.”

Appendix II: CNN Headline News Transcript, 18 October

Appendix III: ABC News Actual Transcript, 21 October