Dr. Shiva Ayyadurai's EchoMail 180-page presentation for the Arizona Senate Republicans' forensic audit on Sept. 24 shows a 25-percent surge in duplicate early voting ballot return envelopes (EVBREs) between Nov. 4 and Nov. 6 in the 2020 Maricopa County Presidential Election. His report also showed 34,448 EVBRE's "originating from 17,126 unique voters." The duplicates Shiva found were additional to those found in the Cyber Ninjas report. Shiva's 99-page summary report can be found here.
Maricopa County never showed how many ballot envelopes they received. The County only showed how many early votes they counted. Shiva did not compare the signatures they examined with the voter registration signatures because the voter registration rolls were never made available to the audit team. However, reportedly, if a signature can't be matched, the County contacts the voter.
Below is the summary of Shiva's findings as compared with the first Maricopa audit. Maricopa reported 6,545 more unique EVBs than Shiva's team while they reported no duplicate ballots.
Shiva Table/Shiva findings v Maricopa Canvass
Duplicates Surge After The Election
The increase in duplicate EVBs is shown in the graph below in the last six days of the counting, which ended on Nov. 9. No duplicates were reported in the Maricopa County canvass that Dr. Shiva used as a basis for his investigation. The blue line indicates the EVBRE's received throughout the election. The red line represents blank signatures, and the green line represents scribbles.
The second graph below shows all the EVBREs, and the duplicates received in the same graph. "For two days, on 11/7/2020 and 11/9/2020," said Shiva, "Over 96% of the daily EVBRE are duplicates. On the days of 11/5/2020, 11/7/2020, and 11/9/2020, the daily duplicates percentage is nine to ten times more than the highest daily duplicate percentage recorded on 10/19/2020." He also notes anomalies on 10/16/2020, 10/26/2020, and 11/2/2020.
Various Findings Concerning the Duplicates
Developed by Dr. Shiva Ayyadurai, EchoMail is an automatic classification and pattern recognition program to help identify and organize documents. He developed the program to "automatically analyze and classify President Clinton's emails." He used the same process to more confidently verify what counts as a voter's signature on a Maricopa County ballot.
Dr. Shiva analyzed data contained in the signature box on the EVBs at a very granular level, categorizing them by signature, blank, likely blank, and scribble. It is unknown how the County defined the signatures it examined.
Shiva explained EVBRE "is the protective vehicle by which the early voting ballot (EVB) is transported and processed. The authentication/verification of the signature on the EVB return envelope is critical to the reliability of the process." Shiva maintains that it is difficult to ascertain proper verification of a vote if one cannot reliably check signatures to ensure a person's ballot equates to the same person's vote.
Shiva classified the duplicates as 2, 3, and 4 copy duplicates. Oddly, many duplicate blanks were verified, stamped, and approved by the County. Some were stamped and approved in the signature box region of the EVBs. There were also EVBs with the same name, address, signature, and phone number with two different voter IDs. The County confirmed that voters should not share voter IDs.
Shiva/2,3,4 copy duplicates
There were discrepancies between Shiva's numbers and those shown by the Maricopa County canvass. Shiva stated the bad signature and the no signature EVBs were surprisingly low in number compared to those reported in 2016. Shiva found over four times more bad signatures (scribbles) than Maricopa County found.
Shiva Report/Arizona Audit
To put the bad signature anomaly into perspective, it means there was allegedly only one bad signature for every 3,268 EVBs in the County's election. The graph below shows the proportions of that perspective:
The inverse relationship between bad signatures and total EVBs, comparing the 2016 and 2020 elections, is shown in a table from the Maricopa County Recorder's office below. There are more ballots, fewer bad signatures, and blanks for 2020. Fewer ballots, more bad signatures, and blanks for 2016. One would expect that with more people voting, the errors would increase, for example.
Maricopa County Canvass/MC Recorder
Maricopa County's explanation for scribbles is that some voters cannot fully sign their names on the EVBRE, so the County visually approves the signatures.
Shiva also noted some odd patterns in the illegible signatures when he compared the signature submissions before the election with the ones after. There was a 3% to 97% increased illegibility rate of signatures four days after the election. Four weeks prior to the election, the inverse was true.
Dr. Shiva concluded his report with many unanswered questions, some of which are pictured in the screenshot below:
As Senate President Karen Fann explained in her press conference after the audit report presentations, many questions may have been in service of the final reports by Maricopa County had they cooperated with the auditors. They resisted the auditors at every turn.
"We were doing an audit to the best of our abilities, and can you imagine if you were an auditor and you went in to audit one of our agencies or a big corporation, and they said we're only gonna let you see these books over here, we're not gonna let you see them over here. So, an auditor can only do the audit on what they're available to see. And, of course, that audit is going to have a disclaimer on it that says, 'I did this to the best of my ability. It doesn't answer all the questions, but I can tell you what this leads me to, and I can tell you some suggestions about what it may or may not be. But it needs further investigation. That's the best we can do'."
This is the third in a series of reports by UncoverDC breaking down the forensic audit reports submitted by Cyber Ninjas, CyFIR, and EchoMail. A report from Jovan Hutton Pulitzer on paper and kinematics is forthcoming, as is a report on the findings from the routers, admin passwords, and hardware keys. No date has been given for the arrival of those reports.