This is a section of the Cops Canvass Report
. See also San Diego County Canvass Procedure
Executive Summary & Introduction
This report is the result of a project initiated by Citizens' Oversight Projects (COPs)
in 2008 to document and review the procedures used by San Diego County Registrar of Voters
("RoV") to process elections and create their certified result. The report addresses chain-of-custody issues, whether adequate in-process records are maintained, and describes serious issues of concern, including the loss of ballots, unexplained central tabulator crashes, and ballot box tampering.
A key aspect of this review was an attempt to follow the paper-trail of 5% of the precincts in the county, and reconstruct the results of the election. Unfortunately, the paper trail is insufficient to reconstruct the entirety of the results for the sample, but the attempt is useful to point out that such a reconstruction is impossible without enhancing the paper trail and computer logs, as well as numerous other problems.
Despite a return to paper ballots throughout California, systems used to count and report the vote -- both computerized and paper-based -- remain subject to election fraud vulnerabilities. San Diego County is no exception. There are numerous opportunities to improve citizens’ oversight and the quality and confidence level of the result, as well as to reduce the cost to process each election. Our review exposed a number of procedures that are technically incorrect, do not follow the law, or provide an opportunity for compromised Ro V
officials to manipulate the outcome of the election. The top issues are summarized below.
- The most obvious deficiency of the Ro V is the lack of comprehensive policy and procedures documents. Such documentation is always the first step in any quality assurance program. It is essential for public oversight of elections processing so we can check that the procedures are correct and that they are being followed. In addition, there is no procedure to systematically deal with errors and omissions to improve those procedures and eliminate errors in the future, which is key to an effective quality assurance program. There is little confidence that, without any written plan, that the certified results of the election represent the will of the voters.
- A large number of voters remain disenfranchised due to the careless procedures of the Ro V. Although the Ro V is aware of these issues, they continue to maintain that there are no deficiencies. For example, the Post-Election Manual Tally ("PEMT") reviewed 1% of the precincts in the election and detected problems or lost ballots in 29% of those 17 precincts tested in the base sample. The Ro V ran the ballots for those precincts back through the scanners and confirmed the errors, but then stated that there were no errors at all because the rescan cleared them up. However, this perfunctory gesture merely confirmed that the election would likely have 29% of the precincts with errors (478 precincts) and about 500 ballots lost, or indeed many more. With errors of this magnitude, it may be necessary to completely reprocess the election or resort to a 100% manual tally. In addition, the Ro V prepared an inadequate report, producing only a five-page summary of the results of the PEMT instead of the recommended 800+ page (electronic) report specified by the Secretary of State ("SoS"), and they refused to enhance the report or explain why the ballots were missing.
- Our attempt to reconcile the number of signatures on sign-in rosters with the number of scanned ballots revealed the loss of about 9,000 ballots. It may be the case that these ballots are actively removed by the Ro V by an undocumented procedure. These ballots were not included in the PEMT, and their disposition never formally analyzed and remains a mystery. Somehow, most of these ballots reappear in the final canvass report, perhaps by adjusting the final figures.
- The audit trail is insufficient to allow an oversight group, such as COPs, to independently review that trail to check on the outcome of the election short of ordering a full recount of the ballots. Many of the ballots, such as the Vote-by-Mail (VBM) ballots are not reported in the central tabulator audit log, and there is no paper trail, such as scanner reports, for the VBM or provisional ballots. Access to this audit trail is difficult and delayed. Suggestions to correct these deficiency are enumerated.
- Tamper-evident seals are used by the Ro V to seal ballot containers. However, there were 99 cases of "tampering," i.e. broken or wrong seals, without any followup investigation, and the serial numbers on the seals are never recorded in companion documentation. Without such an investigation and report, the use of these seals is for show only. The careless handling ballots should be considered negligent, if not criminal.
- When each polling place closes, a "Ballot Statement" is prepared by poll workers to reconcile the number of blank ballots received with the number of voted ballots, provisional ballots, spoiled ballots, and blank ballots returned. In 83% of the sampled precincts, poll workers were unable to reconcile these counts, and in only one case (1.17%) did the poll workers reconcile the ballot counts and the count of ballots also matched the number of ballots later scanned by the Ro V.
- Review of the Diebold "GEMS" central tabulator audit log revealed a number of discrepancies that have not been fully explained, including system crashes, unrecognized error messages, undocumented starting and stopping, lack of an initial clearing of the tabulation, dozens of cases of capturing votes directly from DRE (touchscreen) machines contrary to the requirements set forth by the So S, and a general inadequacy of the audit log report.
- Although the Ro V has referred to a final "reconciliation procedure" to clear up the 9,000 missing ballots, 500 mis-scanned ballots, 83% of the sign-in rosters not reconciled with the count of ballots returned, and numerous central tabulator audit log discrepancies, there is no documented reconciliation procedure, no reconciliation report, no operator notes, and no way for anyone to know with confidence that the 133 precincts that were rescanned caught all the problems. There is no wonder that the Ro V is pressed for time during the canvass procedure as they operate on an ad-hoc basis and re-scan precincts up until the very last minute of the 29-day certification period and complain that they have inadequate time to complete the canvass.
This document enumerates the extent of this review, summarizes the results, and provides recommendations for improvement in procedures and law. Our data collection effort and the entirety of our interaction with the Ro V
has been open to the public and can be reviewed on the CitizensOversight.org
website. This report may be reviewed on the Internet
This report may be reviewed on the Internet - }} for the most recent revision. We invite comments, corrections, and enhancements to our review.
The lead investigator in this work is Raymond Lutz, founder and coordinator of COPs. Mr. Lutz has extensive experience in the document processing field, in quality assurance procedures, has worked with standards organizations in the printer/scanner/fax area and has a Master's of Science degree in Electronics Engineering. Mr. Lutz was a candidate in the November 2008 election for the 77th State Assembly District seat in the California State Legislature and has proposed the Open Canvass method
to virtually eliminate the possibility of electronic election fraud. Although related to this report, changes in procedure and law suggested by this report are not directly related to the Open Canvass method, and most are necessary regardless of the method of counting the ballots.
Status of the Investigation
Although the Ro V
was generally cooperative and courteous in complying with our requests, in several cases they refused to answer questions, limited our ability to scan some documents, and did not provide any contact with Premier Election Solutions (formerly Diebold Systems) to follow up on central tabulator crashes and to get relevant central tabulator documentation, per their suggestion. Therefore, we do not view this report as "final" since these and other questions are still outstanding. However, the egregious nature of the deficiencies is such that publishing the report was justified now rather than waiting for the full extent of the research to be completed. This report can also serve as a starting point for future reviews and should be of value as a template for review of other counties' procedures.
Since our review project was not sanctioned by any county, state, or federal department, we had no ability to demand compliance with our requests for timely access to public documents and information. Generally speaking, the Ro V
only reluctantly returned documents, and then only if it seemed they had no other choice. Therefore, our conclusions may be incomplete in inaccurate to some degree. However, any difficulty we had in reviewing the election system only underscores that the process is not observable and subject to documentation by citizens.
Methods of the Investigation
Our investigation included the following steps:
- Determine the procedures intended to be used by the Ro V.
- Check that these procedures were actually used.
- Point out deficiencies in following the intended procedures.
- Comment on the adequacy of those procedures and suggests improvements.
Unfortunately, because the Ro V
does not maintain comprehensive policy and procedures documentation, determining the intent of the Ro V
was much more difficult than we originally expected and required numerous document request, question and answer interactions with the Ro V
, interview sessions, and video recording their operation.
Therefore, the following methods were used in our interaction with the Ro V
- Video taping the procedures used by the Ro V
- See: Election-night processing on the February 5, 2008, presidential primary election. This video demonstrates the process for administering the election and the Post-Election Manual Tally. This one-hour video is a good starting place for anyone curious about the procedures used by the Ro V to process the canvass on election night. The procedure is summarized for purposes of this report in a later section. The video does not, however, include every procedure. Mail and provisional ballot processing as well as all procedures prior to when the ballots reached the central "Tally Center" are not included in the video.
- Conversations with Ro V officials in various meetings
- See: Nov. 21, 2008 Meeting with Deborah Seiler (video). After this initial meeting, the Registrar refused to allow us to record our meetings, and so we discontinued that approach and instead used written requests for information only. We also found that video recording was less than optimal because of the additional work required to turn the verbal comments of the Ro V into written documentation and the misperception that we were trying to catch the Ro V off-guard.
- Request/Response letters to the Registrar asking for information available under the California Public Records Act (CPRA)
- Responses to these requests were not always provided within the 10-day time period provided by the CPRA. At times, it was necessary to appear before the County Board of Supervisors to obtain compliance with the law. That said, the responses were generally thorough, and most of our questions have been answered. Ro V staff members were always courteous and respectful.
- Requests for information (Q&A) not available under the CPRA.
- These questions were sometimes skirted and avoided with the note that responses were not required under the CPRA, but cooperation improved as we continued our investigation. For example, we asked “Why were the 'lost' ballots not counted by the scanning equipment? What was the source of the error?” and the question was not answered because it is not a CPRA request. (This remains a serious issue covered by this report).
- 5% Sample of the November 2008 election.
- Central to our project was an independent sample of 5% of the precincts. The exact method of choosing the 5% sample (85 precincts) and the data collected are described in detail later in the report. This sample was used as a means to detect errors by the Ro V over and above the Post-Election Manual Tally (PEMT), which is the first 1% of our 5% sample (comprising 20% of our sample). Thus, the 1% selected by the Ro V was 17 precincts and we added an additional 68 precincts. However, we were unable to access the actual ballots, so it was impossible for us to perform a manual tally on the additional 68 precincts or repeat a manual tally on the 17 precincts already tallied by the Ro V.
Structure of this Document
There are three main components of this document:
- Procedure Summary - provides an overview of the entire procedure used by the San Diego County Ro V to process the election
- Five Percent Review - A description of our review methodology.
- Issues of Concern Categories - The results of our review are "Issues of Concern" and are grouped into categories to provide context and further detail. Although some issues apply to more than one category, we discuss each issue only once, but may refer to it in other categories to acknowledge that it is relevant.
Each Issue of Concern has several components:
- The first section of an Issue includes a number, a title, a brief description, and a longer description.
- The second section describes any proposals to change for the Ro V as a follow-up to this report.
- The third section details actions requested for the state legislature so as to improve state law.
There are two types of issues, the first is numbered starting with 'A,' which means it is an actual issue. The second type are numbered starting with 'C,' which indicate areas where we need to do more work to collect information in our project.
It is important to list a number of desired outcomes from the publication of this report. As mentioned, COPs has no legal authority to obligate the Ro V
, the County of San Diego, nor the State of California to take any steps as a result of this document. However, the following steps are requested:
Response by the Registrar of Voters
We have made quite a number of assertions throughout this document regarding the activities of the Ro V
. Some of these assertions may be inaccurate We invite the Ro V
to refute errors, suggest corrections and comment profusely.
Nearly all of the issues, with only a few exceptions, have specific questions or proposals for the Ro V
. Instead of repeating nearly all of the issues separately, we will simply suggest that the Ro V
attempt to address each and every issue enumerated in the Issues of Concern
, particularly with respect to the section "Questions to the Ro V
Steps by the Secretary of State
We encourage the Secretary Of State to review the Issues of Concern and comment on each issue, particularly where they can play a role in improving the performance of the Ro V
Steps by the State Legislature
For each of the Issue of Concern, we have explicitly detailed any changes in Elections Code that may be appropriate. We hereby request that the California State Legislature consider our recommendations carefully, and sponsor legislation that eliminates the risks to a viable election process created by the practices of the San Diego County Ro V
and likely repeated in other counties across the state.
Our representatives in the Legislature should review each and every Issue of Concern, particularly with respect to the section of each issue "Legislative". In each case, we urge the those reviewers to place each issue in context by reading the descriptions of the larger issue in the Procedure Summary
and Issues of Concern
rather than reading only the text of the specific Issue of Concern.
The following people were most helpful in the generation of this report:
- Raymond Lutz - primary investigator and report author.
- Linda Poniktera - participated in meetings, election night oversight, manual tally oversight, document review
- Sal Magellenez - participated in meetings, election night oversight
- Brina Rae Schuchman - Save-R-Vote - participated in meetings, election night oversight
- Lisa Davis - Cops member, entered almost all the scanner tape data.
- Valerie Sanfilippo - Assisted with roster inspection
- Joe Sain - Assisted with Ballot Statement data entry
- Ken Karan - Legal advice and document review
Next section: Documents Referenced