Forum Navigation
Topics
Log In
Log Out
:
Forum Search
New Today
New This Week
Advanced Search
Tree View
Forum Account
Edit Profile
Register
Forgot Password
Forum Tools
Help/Instructions
Policies
CLICK STATE TO SEE:
"WATCH LIST"
Marked with: 
"OPEN & HONEST"
Marked with: 
...
|
| 2-1-07: Arizona computer logs show ch... |
|
| Author |
Message |
    
Bev Harris
Frequent Voting Rights Forum Participant
Username: Site_admin
Post Number: 535
Registered: 10-2006
Best of Black Box? 
Votes: 5 (A keeper?) | | Posted on Thursday, February 1, 2007 - 6:40 pm: | 
|
By Jim March and John Brakey
PRE-ELECTION VOTE TOTAL ACCESS IN PIMA COUNTY ARIZ. 2002-2006: AN ANALYSIS
With the rise in popular mail-in voting, election administrators have a new all-knowing power to sneak a look up to two weeks before Election Day at how the vote is going, based on early returns. This kind of cheat trumps any Zogby Poll in financial value. In Pima County, we have found a pattern that indicates apparent improper access to early results records in 2004 and 2006.
EXECUTIVE SUMMARY:
In recent years approximately 50 percent of Arizona's voting is by mail. Any voter may request a mail-in ballot, and until 2006 the scanning of these votes could begin a week prior to election day. In 2006 this was extended to two weeks.
It is illegal to release this data prior to 8pm on election day.
Effective for the general election which took place on Nov. 7, 2006 a new post-election hand-count process was instituted per state law SB1557. In order to hand-audit mail-in votes, a complete "snapshot" of the vote totals (called a "Summary Report" by Diebold) is printed, then a stack of ballots are scanned, and then another Summary Report is printed. These documents were not supposed to be looked at, but rather sealed up with the ballot stack in question and marked as an "auditable ballot stack" due to the existence of the reports in the box.
Approximately four of these "audit boxes" were created each day, witnessed by party observers who noted the serial numbers of the tamper-evident seals put on the boxes.
UNEXPLAINED EXTRA COPIES
After the election, a review of the Diebold central tabulator audit logs revealed that on at least two occasions, there appear to be extra copies of the summary reports printed – copies that do not appear to closely match the usual pattern of audit box printouts, and do not match party eyewitness observer logs.
Concerned about what was going on with the extra copies of this sensitive data, our project asked for an audit and inventory of ALL summary reports printed before the general election.
LAWSUIT FILED TO GET ACCESS TO RECORDS
These records clearly fall under public records freedom of information regulations, but they have been locked in a vault with the actual ballots, which are now sealed, under Arizona law. Pima County has thus far refused access to these documents, even with a court-appointed overseer making sure the public records are all that are accessed. This issue is addressed in a lawsuit filed Feb. 1 on behalf of the Pima County Democratic Party by attorney Bill Risner.
Here is a copy of the lawsuit:
http://www.bbvdocs.org/AZ/pima-complaint.pdf
TABULATOR RECORDS SHOW ACCESS TO EARLY RESULTS
While waiting for litigation to free these important records, we asked for central tabulator audit logs going back to the year 2000.
"Audit boxes" and pre-election summary reports are a new thing, beginning with the Nov. 2006 general election. Therefore we assumed that no pre-election summary reports would be found in the tabulator logs.
Not so. Starting with the tenure of the latest elections director in Pima County (Brad Nelson, hired in 2003), parties unknown have either printed or peeked (via "Print Preview" to screen) at the vote totals before election day.
Under the previous elections director, based on the 2002 data, no such impropriety occurred.
Based on the findings of this study, citizens are encouraged to seek and analyze these records in other locations. The analysis below will provide a road map for how to do this.
This document reviews the evidence in detail, from the oldest data forward.
THE DATABASES
Anyone reading this should have access to a set of nine Adobe Acrobat PDF files. You will find links to them within the analysis below.
These were generated straight out of the original databases in front of our eyes, and are unedited. For each election discussed, we will cite the PDF file name and discuss it by page number, date and time stamp line.
2000 DATA: NO USABLE DATA
In 2000, the Diebold audit logs were not set up to record print events. No usable data on the issue of pre-election vote total printouts can be extracted. You can download those files here:
http://www.bbvdocs.org/AZ/091200-GEMS-AuditLog.pdf
http://www.bbvdocs.org/AZ/110700-GEMS-AuditLog.pdf
2002 SEPT. 10 PRIMARY ELECTION - NO EVIDENCE OF IMPROPER PROCEDURES
Files:
http://www.bbvdocs.org/AZ/091002-GEMS-AuditLog.pdf
On page 200 between the dates of Aug. 31 and Sept. 3 we see various "Logic And Accuracy" tests (L&A) and then we see "reset election" commands to clear out the test data. This is normal.
By Sept. 6 they were getting close to ready to scan the ballots. They reset to clear any data, print a summary report (usually used like a "zero report" to indicate there are no votes pre-loaded into the system):
09/06/02 10:59:11 User admin: Reset election
09/06/02 11:03:02 User admin: Printing Summary Report
09/06/02 11:24:12 User admin: Closing GEMS
The records do not show anyone logged in during the night. It is possible that ballots may have been scanned during the night in 2000 and 2002, because only the more recent versions of Diebold's mail-in ballot scanning firmware require GEMS to be operating while mail-ins are scanned. It should be noted that the Diebold AccuVote mail-in ballot scanners in 2000 and 2002 used memory cards and were capable of printing results tapes before going into the GEMS tabulator. This feature was removed around 2003, and the mail-in vote data is now captured only in GEMS.
In the 2002 election, they come in the next morning, log into GEMS and do another immediate check. There is no GEMS evidence that any ballots were scanned during the night, and this was probably also a "zero report" to show there are no votes pre-loaded in the system, although it's not possible to verify this without also examining the 2002 AccuVote memory cards or AccuVote audit log or results tape, items which are probably no longer available.
09/07/02 08:00:47 User admin: User Login
09/07/02 08:00:47 User admin: Open Election: Primary Election, September 10, 2002 (091002-pima primary) admin Host
09/07/02 08:01:27 User admin: Printing Summary Report
09/07/02 08:02:17 User admin: Printing Summary Report
(page 201)
Also on page 201, 44 seconds after it was legally permissible, we see the first of many summary reports showing the results:
09/10/02 20:00:44 User admin: Printing Summary Report
NOTE: you'll also see "cards cast reports" throughout the pre-election scanning period. These are simply an auditing count of how many ballots have passed through the system. The "cards cast" reports do not reveal who's winning. Only "Summary Reports" and "SOVC Reports" contain vote totals.
NOV. 5 2002 GENERAL ELECTION:
Filename:
http://www.bbvdocs.org/AZ/110502-GEMS-AuditLog.pdf
The pattern here is similar to the 2002 primaries. We can see the final "reset election" after testing and then an immediate pair of summaries with no time available to scan ballots:
10/29/02 12:37:16 User admin: Reset election
10/29/02 12:37:20 User admin: Backed up election to D:\Program Files\GEMS\Backup\110502-pima general.gbf
10/29/02 12:38:11 User admin: Printing Summary Report
10/29/02 12:39:33 User admin: Printing Summary Report
(page 278)
We see no more "summary reports" until election evening at the top of page 280.
2004 SEPT. 7 PRIMARY ELECTION:
Filename:
http://www.bbvdocs.org/AZ/090704-GEMS-AuditLog.pdf
This election starts normally at the top of page 57:
08/31/04 12:20:02 User admin: User Login
08/31/04 12:20:02 User admin: Open Election: pima primary 090704 (pima primary 090704) admin Host
08/31/04 12:20:07 User admin: Reset election
08/31/04 12:20:44 User admin: Printing Summary Report
Also on page 57 we see some previews of unusual reports:
08/31/04 16:08:40 User admin: Printing Report: Races With Candidates
08/31/04 16:09:16 User admin: Previewing Ballot Text Report
08/31/04 16:09:40 User admin: Previewing Report: AccuVote-OS Status By Upload Time Report
08/31/04 16:10:07 User admin: Previewing Report: Races With Candidates
Those reports do not contain vote totals. But then a few lines down:
09/03/04 08:39:07 User admin: Previewing Summary Report
"Previewing" means displaying the summary report to the screen instead of printing it – basically sneaking a peek. If Pima County had by then upgraded to the newer version of AccuVote firmware -- the current version, which requires GEMS to be operating while scanning ballots -- there was not enough time logged in to have put any votes into the system.
Now we see a printout made in the afternoon on election day (still page 57):
09/07/04 15:08:19 User admin: Printing Summary Report
On election afternoon, somebody was also curious enough to sneak a peek on the screen:
09/07/04 17:10:18 User admin: Previewing Summary Report
This was certainly improper.
NOV. 2, 2004 GENERAL ELECTION
Filename:
http://www.bbvdocs.org/AZ/110204-GEMS-AuditLog.pdf
The last "reset election"” after the tests is following by a significant delay before a summary report is printed:
10/26/04 11:00:38 User admin: Reset election
10/26/04 12:23:55 User admin: Changed in Voter Registration
10/26/04 12:36:24 User admin: Changed in Voter Registration
10/26/04 14:03:00 User admin: Changed in Voter Registration
10/26/04 14:20:53 User admin: Changed in Voter Registration
10/26/04 14:21:01 User admin: Backed up election to D:\Program Files\GEMS\Backup\pima general 110204.gbf
10/26/04 14:21:22 User admin: Closing GEMS
10/26/04 14:23:48 User admin: User Login
10/26/04 14:23:48 User admin: Open Election: Pima General 110204 (pima general 110204) admin Host
10/26/04 14:24:29 User admin: Printing Summary Report
(page 149)
"Changed in Voter Registration" may mean they were updating the very latest voter registration data so the turnout percentage would be accurate. They don't appear to have scanned any ballots before doing the post-reset summary report at 14:24:29.
The next day we see a peek:
10/27/04 06:54:45 User admin: Previewing Summary Report
By this time the database has been open plenty long enough to scan ballots. They might have started doing so, or not...it's impossible to tell but the record here does show bad procedures.
The next entry takes place after "Cards Cast" reports have been run. The cards cast report show how many ballots have been scanned. Then they preview the results:
10/28/04 11:46:09 User admin: Previewing Summary Report
(top of page 150)
This is a definite peek at live data. And by that evening they did more than peek:
10/28/04 18:03:06 User admin: Printing Summary Report
(page 150)
That document is now worth literally thousands of dollars. Go ask Zogby what a countywide poll would cost involving over 20,000 respondents – and then ask how much the price would go up if those were REAL VOTES already cast. Of course, Zogby can't provide such a poll because it's illegal.
Worse: The results report was run far enough in advance of the election to be of significant value to a political party, campaign or heavy-duty political activist.
Another premature results report was cut the night before the election:
11/01/04 17:44:41 User admin: Printing Summary Report
(bottom of page 150)
...and yet another illegal early look at results too place 15 minutes later (see top of page 151).
2006 MAY 16 SPECIAL ELECTION:
Filename:
http://www.bbvdocs.org/AZ/051606-GEMS-AuditLog.pdf
The usual post-test pattern starts on page 318:
05/10/06 08:21:41 User admin: Reset election
05/10/06 08:22:08 User admin: Printing Summary Report
05/10/06 08:38:47 User admin: Printing Summary Report
Our first problem is already evident: between 8:22am and 8:38am it's possible to scan a fair number of ballots and include those in the printout of 8:38. Unlike previous reports where two summaries come rapid-fire, the final summary of this pair could contain a small but statistically significant polling sample.
This shows either cheating or sloppy procedure. The next day we commit to going beyond sloppy:
05/11/06 09:56:49 User admin: Printing Summary Report
05/11/06 10:06:21 User admin: Printing Summary Report
By this time the database has been open for hours. It's hard to imagine a reason to print a new summary report unless there's new information in the system -- like scanned ballots.
SEPT 12 2006 PRIMARY:
Filename:
http://www.bbvdocs.org/AZ/GEMS-Primary-AuditLog-9-12-2006.pdf
This time the final post-test reset and summaries to obtain zero are again spaced out a bit:
08/30/06 16:45:46 User admin: Reset election
08/30/06 16:48:54 User admin: Changed in Voter Registration
08/30/06 16:54:28 User admin: Previewing Summary Report
08/30/06 16:55:11 User admin: Printing Summary Report
...
08/30/06 16:59:50 User admin: Printing Summary Report
(bottom of page 85, top of 86)
Procedural integrity erodes further with this:
09/07/06 16:18:39 User admin: Previewing Summary Report
09/07/06 16:19:00 User admin: Printing Summary Report
A couple of days later:
09/09/06 15:43:09 User admin: Printing Summary Report
09/09/06 16:01:03 User admin: Printing Summary Report
For those unused to military time, this would be 3:43pm and 4:01pm on the Saturday before the (Tuesday) election.
Pima County elections director Brad Nelson was questioned about this twice.
Before we obtained the primary 2006 audit log, we confirmed with Mr. Nelson that since SB1557 requiring hand-counting wasn't in effect yet, no summaries should have been printed.
He agreed.
Once we obtained the logs and it was clear there were printouts and a peek done, we questioned him again. With both of us (March and Brakey) present, Mr. Nelson said:
"Yes, while SB1557 wasn't yet in effect, it seemed possible the Pima County Board of Supervisors might ask for a voluntary hand-audit in order to test the SB1557 process prior to the November general election. Therefore, as we scanned mail-in ballots, I set aside two audit boxes with the appropriate printouts of summaries in the boxes so as to enable at least some hand-counting of mail-in ballots."
For any one audit box, two printouts must be made – one before and one after a stack is scanned.
The log shows three printouts and a peek ("Preview Summary Report").
The printouts of Saturday Sept. 9 could be an audit box. But it is troubling that on the evening of Sept. 9 a "robocall" (recorded political message by phone) went out from the Republican party to a subset of Pima County voters against Democratic state senate contender Ted Downing accusing him of a wild (and unsubstantiated) variety of sins against the electorate. Downing was particularly unpopular among supporters of the state's current voting systems by Diebold, Sequoia and ES&S. There was no Republican in the race: whichever Democrat won the primary would be unchallenged in the general election.
Did somebody learn from the summary reports that Downing was in a nail-biter of a race, and try to swing it against him at the last minute?
It's not impossible. Just the fact that we have to wonder about such things indicate that election integrity procedures are not what they should be in Pima County.
Nov. 7 2006 GENERAL ELECTION:
Filename:
http://www.bbvdocs.org/AZ/GEMS-General-AuditLog-11-7-2006.pdf
This one is the most complicated because legitimate sealed summaries had to be printed under party observation. We only noticed potential trouble (and went looking through older data) when we cross-referenced the audit log against our observer notes and spotted extra summary reports.
We were delayed access to this audit log until about 19 days after the most troublesome day (Nov. 2nd). By then the video of the room (a new feature) had auto-erased as the hard disk it was on filled up and collected new video. Thus our best possible method to figure out what happened evaporated.
We filed the lawsuit to get the rest of the records out of locked storage.
Jim March and John Brakey produced this report as Election Integrity Consultants for the Arizona Democratic Party
PERMISSION TO REPRINT GRANTED, WITH LINK TO http://www.blackboxvoting.org |
Jim March
Frequent Voting Rights Forum Participant
Username: Jimmarch
Post Number: 144
Registered: 05-2006
Best of Black Box?
Votes: 2 (A keeper?) | | Posted on Thursday, February 1, 2007 - 11:29 pm: |
|
Here's the kicker: we weren't looking for this when we began looking at Pima County elections. I don't think anybody has documented a pattern quite like it.
If these documents are going out the door, it IS election fraud, but of a type I don't think anybody has been looking for.
If it's not clear yet: if you're a county or state party coordinator and you get this data, you pull resources from the major winning AND losing campaigns and pile it all in the closer races.
Do this in enough jurisdictions, you can flip the partisan direction of congress.
See...we need to view all forms of election fraud as being similar to card counting in Vegas. Some people can play Blackjack and keep track of which cards are played and hence which are left. Doing so under the right circumstances flips the odds from about 6% against you to about 1.5% for you...which is enough to make good money if you're careful.
Election fraud can be similar...and if the perps are smart that IS what it's going to look like. They don't need to flip every race, just tilt the odds.
We have to look for anything that can do that. |
Joseph Hall
Frequent Voting Rights Forum Participant
Username: Joehall
Post Number: 115
Registered: 01-2005
Best of Black Box? N/A
Votes: 0 (A keeper?) | | Posted on Friday, February 2, 2007 - 6:21 pm: |
|
Fascinating. |
Gentry Lange
Voting Rights Forum Participant
Username: Gentrylange
Post Number: 16
Registered: 09-2005
Best of Black Box? 
Votes: 1 (A keeper?) | | Posted on Friday, March 16, 2007 - 8:43 am: |
|
This is a new element to vote-by mail systems I had not yet considered. Really interesting. Great work. |
|
|
