Purpose

Published by the National Ballot Integrity Project in October 2004.

Among the National Ballot Integrity Project's publications, the statement of purpose repays reading most directly. It is comparatively free of rhetoric, it states its premises before its conclusions, and it makes a claim about what an election process should do before it makes any claim about what was wrong. This 2004 document can be read alongside neutral context on how the same questions are handled now.

The premise, as published in 2004

The American system of government requires an election process that is freely accessible to all citizens of voting age, fully transparent to public oversight and accurate in recording, counting and reporting election results.

Three requirements, in that order: accessibility, transparency, accuracy. The rest of the statement argued that the second of the three — transparency to public oversight — had become harder to satisfy as vote counting moved onto computers, and that the third could not be independently checked without it. The project's position was not that any particular result was wrong. It was that the means of verification available to an ordinary member of the public had narrowed.

The 2004 argument, in summary

The project set out its concerns as a sequence. Presented in the past tense, and as the project's own contemporaneous claims, they were:

• Delegation to private vendors. Election administration functions that had previously been carried out by public officials — programming ballots, tabulating and aggregating totals — were increasingly performed with equipment and software supplied by a small number of private companies, under commercial confidentiality.
• Software that officials could not inspect. The project pointed to court decisions of the period holding that voting software was protected commercial property and could not be examined by the officials responsible for the election.
• No security baseline. It argued that no computerised election system then in use had been shown to meet minimum government standards for information security.
• A lightly regulated equipment market. It described the voting equipment industry as operating without mandatory, enforceable security standards.
• Undetectable error. It argued that where the only record of a vote is an electronic one, an error or manipulation in tabulation leaves nothing behind that an audit could find.
• Disappearing paper. It noted that many jurisdictions had eliminated marked paper ballots entirely in favour of direct-recording electronic equipment.
• Fewer external cross-checks. It observed that independently administered exit polling, historically one rough external check on reported results, had become less available.

The project proposed to assist and coordinate organisations working on the question at national, state and local level. That is the same coordinating role described on the About page, and the specific 2004 proposals that followed from it are set out on the manual parallel accounting page and in the optical-scan procedures.

Reading the argument two decades later

Two things are worth separating when reading this 2004 statement. The first is the underlying principle — that a vote-counting process should be observable and independently checkable — which is not contentious and is embedded in election law across the country. The second is the specific 2004 factual picture, which has changed substantially and in ways the project could not have anticipated.

On standards, that position has been formalised. The National Institute of Standards and Technology has a long-running voting technology programme that provides the technical basis for federal guidelines, and those guidelines are published openly as the Voluntary Voting System Guidelines. They cover functional requirements, security and accessibility, and voting systems are tested against them by laboratories that are themselves accredited. The guidelines are voluntary at federal level in the sense that states decide whether and how to adopt them; in practice most states now require federal testing, full certification, or their own state certification, and the National Conference of State Legislatures maintains a current comparison of state testing and certification requirements.

On the paper question, the direction of travel reversed after 2004. After 2004, systems moved broadly towards producing a voter-verifiable paper record — either hand-marked paper ballots read by a scanner, or a machine-printed record the voter can check. That shift happened through state legislation and procurement rather than through the federal courts.

On verification, the mechanism that emerged is more demanding than anything proposed in 2004. A risk-limiting audit is a statistical procedure that examines a sample of paper records large enough to give a stated level of confidence in the reported outcome, and to require a full hand count if the evidence is not strong enough. That is a genuinely different instrument from either a fixed-percentage recount or the parallel manual count the project proposed in 2004, and it exists in law in a growing number of states.

What has not been settled

It would be inaccurate to present every question raised in 2004 as closed. The economics of the equipment market, the depth of technical expertise available to small election offices, the trade-off between accessibility and paper-based counting, and the cost of thorough auditing are all live subjects of published research — much of it collected by the MIT Election Data and Science Lab, which publishes data and peer-reviewed work on voting technology and administration. Readers who want current evidence rather than a 2004 position should start there and with the federal sources above.

The statement records how one group of citizens framed a technical policy question before it had a settled institutional answer. The timeline of the period shows what was happening around it.