Report on Denver electronic pollbook problems

Here’s a link to a copy of the report on the Denver electronic pollbook problems. We wrote about these problems as reported in the media in the 2006 midterm election in Denver.

Here’s the report’s executive summary:

The general election of November 7, 2006 in Denver was marred by significant technical and operational errors, as well as a seeming lack of needed oversight in some key areas. These errors and omissions led to unacceptably long waiting times for voters and an abandonment rate estimated at 18,000-20,000 voters (approximately 20% of the anticipated physical turnout on Election Day). In addition, seemingly preventable problems with the tabulation of absentee ballots led to significant operational stresses within the DEC and delayed reporting on key races and measures for several days.

The most direct cause of voter inconvenience on Election Day was the repeated failure of the “electronic poll book” (“ePollBook”) software, which hampered the efforts of election judges staffing voting centers to search for voters as they arrived, indicate that they had arrived to vote, and forward them to a machine to cast their votes. The ePollBook, developed exclusively for DEC use by Sequoia Voting Systems, is of decidedly sub-professional architecture and construction and appears never to have been tested in any meaningful manner by either the vendor or by the DEC. This software’s failure to accommodate Election Day traffic led to lengthy lines developing at the registration desks of voting centers while voting machines stood idle. Well-publicized media reports concerning line lengths were broadcast throughout the day and likely contributed to dampening turnout among voters without the time or determination to devote multiple hours to casting their votes.

While the ePollBook’s considerable shortcomings represent the most direct cause of Election Day angst in Denver, we must caution readers against assuming that merely repairing or replacing it will ensure the smooth conduct of future elections. That the ePollBook was deployed at all in such an unready state is symbolic of a consistent pattern of substandard information technology management within the DEC. Given the increasing dependence of election processes on technology, the state of technology management within the DEC must be recognized as an operational risk to the City and County as it looks toward future elections.

In addition to technology concerns, the DEC’s conduct of the 2006 elections suffered from inadequate contingency planning (some technical, some purely operational) and errors in logistical operations and assumptions, especially given the number of significant environmental changes with which the DEC was wrestling in preparing for this election. In 2006, the DEC was coping with new voting machines, new scanning equipment, software upgrades, vacant staff and leadership positions, new leadership, and a fundamental shift from traditional precinct-based polling places to voting centers, at which a voter from any part of the County may vote. These environmental changes, in addition to several others, represent an extremely complex problem set, and one might expect a cautious, if not ultra-cautious attitude to prevail among those responsible for the election’s conduct. Instead, planning and due diligence activities were less thorough than needed.

In analyzing the causes underlying the difficulties of 2006, it is tempting to search for a single factor, act, or error on which to place all blame. The purpose of this assessment, however, is not merely to diagnose what went wrong in 2006 but also to surface information of use to Denver in conducting future elections. In that light, it is critical that the failures of 2006 be viewed in an appropriately broad context that takes into account disparate factors such as planning, management, technology, interagency politics, and the degree of environmental change surrounding the conduct of the 2006 election cycle.

I’ve not seen an official response to this report from Sequoia Voting Systems, but I suspect at some point there will be a response and when it is available, I’ll post it here.

UPDATE (1-16-2007): Here is Sequoia’s response to the report.