Review of Jovan Pulitzer's "Paper Analysis Report" for Maricopa County

Raymond Lutz -- Citizens' Oversight -- 2022-07-08

Video is available at this url:

https://rumble.com/v1a992g-the-truth-behind-arizonas-paper-ballots-jovan-pulitzers-bombshell-paper-ana.html

Actual report starts at 1:14:00

Introduction

The "Paper Analysis Report for Maricopa County" by Jovan Pulitzer is filled with numerous false claims and baseless conclusions. There is very little evidence actually provided. In each section, Pulitzer simply makes the claim over and over and that is about it. Nevertheless, we felt it was important to fully review this video in case people would quote it as evidence that the election was stolen in 2020.

In the sections that follow, I will place the statements made by the video in a box, and then I will make comments about those statements.

OVERALL CLAIMS

The first slide provides overall claims to be covered later. No extensive comment yet on these. We must note though that making a lot of measurements does not mean anything is proven. We notice that he says the "deep forensic investigations down to the precinct level,"  but it is interesting that he does not go down to the ballot level, which would seem to be deeper.

PREDETERMINED BALLOTS CLAIM

In the next section Pulitzer makes the claim that ballots were "predetermined" using an algorithm. However, he does not say how this algorithm works, nor how they could discriminate a "predetermined" ballot from a regular ballot, just that they were "detected using digital forensics, forensic mathematics, forensic accounting, and rhythm analysis. This is further combined with computer vision, machine learning and artificial intelligence."

He does not answer:

• How is a predetermined ballot detected?
• How is a predetermined ballot "inserted" into the election?
• What is the algorithm used to insert these?
• Assuming these were predetermined in an effort to "steal" the election from Trump, why were any predetermined ballots inserted for TRUMP?
• Since the total number of ballots cast was 2,089,563^[1], if 406,972 of those were "predetermined" then nearly 20% of the ballots were predetermined.
• If 91.62% were cast for Joe Biden, then 406,972 * .9162 = 372,868 Predetermined ballots were cast for Joe Biden, and up to (the rest) 34,104 were cast for Trump.
• But Pulitzer also says that "33% of the ELECTION DAY BALLOTS were PREDETERMINED and assigned to DONALD J. TRUMP." Since 167,878 ballots were cast on election day, 167,878 * .33 = 55,400 predetermined ballots were cast for Trump.
• --> But 34,104 does not equal 55,400. So Pulitzer's numbers are not consistent.

PALLET PACKING CLAIM

Here, Pulitzer claims that boxes were opened and reopened and "pilfered."

First of all, there are no laws or regulations regarding how batches are placed into boxes and onto pallets. These are all jurisdiction specific.

He says that "141 individual batches took two days or more to process." In our review of the boxes and batch reports, there were no batches that took more than one day to process. However, the boxes of several (usually 7) batches may have been processed on separate days, and then placed in the same box. Thus, this claim that batches took more than one day is not correct, but it could be correct if the word "box" is used instead of "batch", because many times, a single box of several batches may include batches that were scanned on different days, depending on when they stopped the shift.  They also used several different high-speed scanners and as the batches were completed, they may have been interleaved within the box. For example Box 1 might contain a first batch from scanner 1, and a second batch from scanner 2, etc. This does not mean the batches were pilfered, but could be misinterpreted as pilfering.

With all this said, it would be better to have a consistent box packing methodology that would keep batches from each scanner together.

He says: "We have boxes and layers of boxes where 1 Candidate won all batches. Which is not consistent with the ending turnout."

But that is consistent with how batches might swing one way or the other at random. Every batch will show one of the two top candidates for President as the winner in that batch. In this close election, an entire box might indeed have one candidate the winner out of 10,341 batches in 1675 boxes.

He says 1394 of the batches were "compromised", but what does that mean? He then jumps to a completely unsupported conclusion that "Ballots were removed from and added to the batches and boxes over time. The ballot boxes were pilfered in order to make the paper match the digital outcome."

OUT-OF-CALIBRATION BALLOT PRINTING CLAIM

One of the procedures used in paper ballot analysis is the idea that the front and back of the page were not properly registered to each other in front and back. This can commonly occur in laser and other printing technologies because the image printed on the page may be offset on the left vs. right, and then when the page is printed on the back, the image may be offset in the other direction.

It is a standard printing quality check to have a cross-hair target on one side with a gap in the middle, and then a plus-sign on the other so that when held up to the light, it is possible to see the two at the same time. Here is an example of an extreme case where the plus sign should be in the middle but is moved up and to the left with respect to the front side of the ballot. But it is still within the circle, and thus not actually out of spec.

Pulitzer says that voters were treated differently because the two sides were not lined up exactly. But it only makes a difference IF their vote was not counted properly. No one really cares if the two sides are lined up if the machine scan still interpret the marks correctly.

Mis alignment of the two sides could be a problem if the registration of the two sides was drastically off, if thin paper were used, and felt-tip pens used, because then ink from one side might be included in the evaluation area of the target on the other side.

But marks bleeding through from the back have no impact on the functioning of the recognition of those marks, because they are not close to the target areas:

But these so-called "Ghost Votes" did not affect the outcome of the election, and in Pulitzer does not provide even one example of a ballot where the vote was not read correctly. Thus, the 120,867 ballots with "Ghost Votes" (which should be called only Ghost Marks) did not change the votes on those ballots as recognized by the voting machines. This was further checked by the hand-count which did not find any appreciable error in the counting of votes.

Ballot Design can be improved:

We can notice that in the Nov 2020 election, the ballots placed the active ovals on the right side of each column, instead of on the left as seen in the prior image for the 2012 election. If the ovals are placed on the right side of each column, then in the left column, they will be next to ovals from the back in the center column. In the image above the Trump oval is close to a mark from the back, whereas the vote for Biden or Jorgensen do not have an oval nearby from the back. At the same time, the ovals in the right-most column will have no chance of being near an oval from the back.

This brings up the important issue that placing ovals on the left is a superior design because there are no nearby ovals from the back for the most consequential contests at the top of the ballot, while there is a higher chance for those "down-ballot" ovals in the center and right columns to be corrupted by ink soaking through from the other side.

Further, it would be advantageous to design each side of the ballot to be offset with respect to the other side. If ovals are on the left, the entire ballot grid and image content should be offset slightly to the right on the paper on each side. This will move the ovals further away from each other.

Based on the analysis above, offsetting the two sides of the ballot may be a good design decision, and with targets on the right, it will be best to move the entire printed grid to the left, since the targets are on the right. In the image above, if the red dots do the right and below the center of the target are the location of the plus sign on the back with respect to the front, then

these would actually be PREFERRED because they move the targets away from each other and reduce the chance that ink soaking through from the other side can be misinterpreted.

Thus, Pulitzer's claim that if the back is offset as in this diagram, voters are not being treated equally. Yes, that is technically true, but the offset actually improves the situation for those voters!

WRONG PAPER TYPE CLAIM

Pultizer claims that the wrong paper type was used. "Roland vote secure" paper is supposed to be used to avoid bleed through, even though the targets are offset from one side to the other to avoid any misinterpretation of bleed through. Thus the wrong type of paper can result in an error only if the paper is thinner, allowing bleed-through, and combined with really poor registration from one side to the other, or ballto mis-design so that bleed-through marks could be interpreted as votes. However, there were no examples provided of any votes being misinterpreted due to the wrong paper type.

GHOST VOTES AND FORCED ADJUDICATION

In this section, Pulitzer finally combines the claim of misalignment between the sides of the ballot, with thin paper and bleed through, to claim that this caused many ballots to be adjudicated.

Pulitzer does not provide any cases when adjudication was performed incorrectly. Adjudication is generally viewed as a good thing, because it allows a human operator to review the evaluation by the machine, particularly in difficult cases, when the voter may have added a "hesitation mark" on the ballot (a very light mark in the target area of one option when the other option was clearly marked), or other complex marking (cross out one and write "NO" then vote the other option), or in the case of any write-ins.

It is typical in close contests for jurisdictions to review all overvotes and writeins to make sure the outcome is correct. Reviewing 6% of the ballots should mean that the interpretation of the vote was checked by humans, and does not (in itself) mean that any improper changes occurred. Thus, this claim that this changed the outcome of the election is unfounded.

NON-COLOR PRINTED BALLOTS CLAIM

In this section, the claim is that the ballots should have been printed in color rather than black and white. But ballots printed on-demand are B/W ballots, and they are legitimate ballots. There is no reason ballots can't be printed with lower-cost B/W printing technologies.

Here Pultizer is grasping for straws.

SUSPICIOUS MACHINE-MARKING CLAIM

Pulitzer claims that he can tell the difference between a machine marked ballot and a human-marked ballot, yet say that machine cast votes are undetectable from Sharpie votes. He even plays a clip of a Dominion sales pitch where they describe that when disabled voters use a ballot marking device (BMD) that they can use a library of human-looking marks, so that the vote of a single or just a few disabled voters could not be identified by looking at the ballots.

Pulitzer does not provide an example how machine marking is used nor any example of actual machine marking that could change the outcome.

SUSPICIOUS BALLOT FOLD-MARKS CLAIM

Pultizer claims that by reviewing fold marks, 142,091 ballots either had folds when they should not have or did not have folds when they should have. There is no correlation provided to any hacking scenario.

Even if such folds were not detected, this could have been due to insufficient ability to detect the folds. And the number of ballots affected, if evenly distributed, would not affect the outcome.

Pulitzer just says there was outright fraud without describing how the lack of or extra folds means that fraud exists, nor where the ballots came from if they were not folded, for example. Most of the cases cited, such as "ballot cracking" mentioned later do not have anomalous folds.

COUNTERFEIT BALLOTS CLAIM

Pultizer now claims that if the paper was incorrect, was not printed property side-to-side, did not have the right print density, then they automatically can be considered "counterfeit." But the term "counterfeit" normally requires an intention to create a fake ballot, rather than one that is legitimate but not quite perfect. This is completely without any merit. Plus, those counterfeit ballots were nearly evenly split between the candidates. Why would a malicious actor intent on "stealing" the election from Trump also insert counterfeit ballots for Trump?

IMPROPER ADJUDICATIONS CLAIM

This is yet another round on the idea that adjudicating ballots means the election workers reviewing the ballot images are biased and inserted improper votes for Biden over Trump. However, Pulitzer does not provide any examples of that. He makes a big deal about the idea that the adjudications were made by looking at ballot images rather than at the physical ballots, but this is obviously much easier to process logistically than working with paper, and the reason almost all commerce today is conducted by reviewing digital images of documents rather than documents on paper.

This is not a "time hack" and the fact that there were corrections to the machine count, due to strange marking by voters, invalid write-ins, and other factors does not mean the outcome can be questioned as a result.

MISSING RETURNED BALLOTS AND "BALLOT CRACKING" CLAIM

The first part of this claim does have some merit. Ballots mailed to votes that were subsequently returned as undeliverable, should have been kept with other election data as specified in 52 US CODE  20701 for 22 months after the election. This election data should not have been destroyed.

However, the fact that they were destroyed does not automatically mean that the reason was due to a widespread "ballot cracking" scheme. He mentions that such a scheme might leave the signature blank or use a bogus signature.

But black signatures are not accepted without curing, and non-matched signatures are not accepted. We are told that 100% of all signatures were checked by human-eye comparisons with references. Thus, if a ballot-cracking scheme were to be successful, they would have also needed the signatures.

But Runbeck, the mailing subcontractor, may have had access to ballot signatures as well, and if this were the case, then such a scheme would have been possible.

If deployed, then such a scheme would have likely been detectable when voters complained that they were recorded as voted when they did not vote, and those cases would be widespread.

We don't understand the idea that the number of ballots returned as undeliverable differs when reported by Maricopa County and the EAC.

PHANTOM VOTERS CLAIM

Various methods of detecting "fake or fraudulent" voters on the rolls are inherently error prone due to incomplete matching and imprecise information about moving dates. The list of issues:

> duplicated voters. Same first name, same last name in a different county.

> Married with maiden name still voting.

> College students that left town but then are all of a sudden at their parents home.

> Fraudulent, does not exist.

> Name makes no sense, just gibberish, and never votes typically.

These issues, when investigated carefully, exposed only a very few fraudulent voters. In the research by the Senate Contractor, "Cyber Ninjas," they found as well that even if all the claims of phantom voters were true, that the party split was nearly even but had more R voters than D voters, and removing those ballots from the result likely would have only further buttressed Biden's reported victory.

EXCESSIVE PROVISIONAL BALLOTS CLAIM

In the 2020 General Election, there were only 6,198 provisional ballots. If 66% percent were rejected, that would be 4091 cases.

But in 2016, there were 52,173 provisional ballots, and 15,250 were rejected.^[2]

Thus, Pulitzer should be complimenting Maricopa county for vastly reducing the number of provisional ballots, and further reducing the raw number of rejects from 15,250 to 4091. Instead, Pulitzer using a deceptive technique of comparing rejection rates of the total number of provisional ballots while not acknowledging that the number of provisional ballots was vastly reduced to less than 12% of those used in 2016.

If nothing else, this red flag should be indicative of how the entire report should be viewed.

The lower number of provisional ballots occurred because of the introduction of e-poll books and the use of ballot printing on demand, so more provisional ballots could immediately be treated as a normal ballot.

CONCLUSION

Pulitzer makes many over-the-top claims about fraud, ballot cracking and other schemes that are not supported by any evidence at all. There was one issue that was correctly raised, regarding the destruction of the returned as undeliverable ballots. This was a processing violation by Maricopa County, but does not automatically mean that a "ballot cracking" scheme was the result. Most of these issues were separated out into separate cases, but it came down to sometimes thinner paper than specified was used, and sometimes the two sides were not perfectly aligned, and the use of sharpies on thin paper could allow bleed-through of the marks. But since the design of the ballots places targets in different columns between the two sides there were no documented cases where all these conspired to cause even one vote to be altered.

The claims that alignment of the sides of the ballot is a violation of a voter's 14th amendment rights of equal treatment are baseless unless the votes of those voters were misinterpreted, and they were not.

Gathering the data for this analysis was extremely time consuming because workers had to hand-position ballots under cameras and microscopes. Assuming that each one took ten seconds to position, and all procedure were interleaved, the total time required was about 2.1 million * 10 seconds = 5,833 hours. Not including any other costs, just capturing the data incurred at least $600K, given the use of 4 workers to process the images in a production line. There were also costs for storing and processing the data, so this would be a lower estimate of the overall cost.

Given the lack of any demonstrable evidence of any legitimate hack, this was entirely a waste of time and resources.