OPINION: This article contains commentary which may reflect the author's opinion
There are a lot of questions surrounding the events in the State Farm Arena in Fulton County during Election 2020. All 148,318 absentee ballots were reportedly scanned in Fulton County’s ballot processing center at the Arena.
Those ballots processed in the Arena were supposed to be authenticated first. The voter logs were to be used to identify voters, and then the signatures would be verified. Separated from their envelopes, the ballots were then to be arranged in batches of 100. Each batch of ballots was then scanned using a Cannon DR-G2140 high-speed machine [as can be seen in the video from the Arena on Election night.] The image of every ballot was saved simultaneously.
The Fulton County clerk produced scanned images of absentee ballots as a result of a previous lawsuit. These images have been made public and are currently under review by a variety of individuals and groups.
An individual examining the data (RonC) spotted something he could not explain. As the ballot images are scanned, timestamps are recorded in the image files. The timestamps indicate that the image files were created faster than the capabilities of the scanners. Often several times faster.
On page 23 of the manual for the Cannon DR-G2140, the specifications are listed:
Document sheets can be scanned at a maximum speed of 140 per minute, or roughly 2.3 pages per second, according to the machine’s manual. The rate is based on scanning standard 8.5 inches by 11 inches paper. At 18 inches long, absentee ballots are approximately one-and-a-half times longer than this.
In analyzing the surveillance video from State Farm Arena, a group of professionals evaluated the actual scan times, which were on average 80 seconds per batch (with no paper jams or other complications). That’s about 1.25 ballots per second.
However, the image timestamps indicate something else. The spreadsheet below contains sample data with the timestamps of the images as they were created. The above information was obtained from Fulton County and was used in the above case.
Images of actual ballots scanned by unit 5162 on 11-05-2020 are shown below.
In one second, 8 ballots were scanned, and in the very next second, 6 ballots were scanned. How did fourteen ballots get scanned in two seconds?
Thus, it appears that the ballot images were created at a speed that is physically impossible.
Results like these are not possible.
In the image files, the same situation is repeated on thousands of ballots. These are not expected or “normal” results.
Several experts were consulted about these anomalies by the group of professionals. With regard to the specifics of the arrangement, such as the fact that the files are stored on Ethernet-connected, high-performance network drives, and other specific technical parameters, there remains no clear and justifiable explanation for the timestamps. The impossible scan times, even when averaging all the images within each batch, are not explained by lag-time, buffering, or some form of computer processor- or network-caused “ebb and flow”.
Plus, there’s more. Expanding this dataset to include the modified file timestamps reveals additional issues:
Only after an adjudication process has been completed should a ballot image file be modified. If the machine is unable to identify a voter’s selection for various reasons, the ballot is supposed to be sent to a bipartisan team of election judges who look at the markings and attempt to understand the voter’s intent. Once the team votes and decides, the original image is revised to reflect their decision.
All of the files have been modified according to the modified timestamps above.
Unless they are being processed for adjudication, ballot images should not be modified. The use of adjudication should be limited to rare circumstances. [We are not sure of the official percentage, but we think adjudication should only occur in less than 2% of all ballots processed.] Perhaps there is an explanation for all of them being adjudicated. Could this be due to a technicality that only affected these ballots? It appears, however, that all the images were altered at the same time and within one hour of being scanned.
We can see a much larger problem if we zoom out to the 30,000-foot level and look at all batches from the same scanner as shown below. In the first place, each batch has been modified as shown on the far right. Each batch’s scan time is in yellow, and the numbers in red show the amount of time between scanning and modifying.
As an example, batch 001 was completed at 1:51 pm on 10/22. 192 hours later, on 10/30, the same batch was modified at 1:38 pm.
Earlier that day, the batch was scanned and then modified eight days later.
Between 3 and 8 days after scanning, 44 batches on scanner 5150 were modified. A large period of time is not only unjustifiable, but also unacceptable for processing ballots. There are extreme variations in the process, and these findings cannot be ignored or ignored.
On the page, the last 4 batches show 0 hours between scan time and modification time. The last four batches scanned were in fact modified before the previous 44 batches. As a result, there is no possibility of some consistent procedure or a forced process causing batches to be modified on the same date.
In addition, it’s important to remember that the above findings only pertain to the one of five machines used to scan absentee ballots, a process that ran until November 5th.
RonC sought understanding and answers for his findings by contacting the logical government agency responsible for such matters-the Fulton County Board of Registration and Elections (BRE).
RonC detailed his findings in an email that discussed the scanning speed issue and image file modification issues. Rick Barron, the county’s election director, was also contacted by the BRE for answers. Mr. Barron’s email reply to the board is provided below:
Mr. Barron’s email provides four key takeaways:
Firstly, it is obvious that Mr. Barron feels the same way about the impossible scanning speed and described RonC’s finding as “bizarre”. However, this information was not simply a “claim,” as described by Mr. Barron, but rather a factual observation based on Fulton County’s data.
Second, Mr. Barron directly states: “Ballot images that are not adjudicated cannot be modified.” Again, he reaffirms our understanding and expectations.
Thirdly, he reports that the equipment is proprietary and that security manuals aren’t available. Mr. Barron was probably referring to the electoral process in a broader sense or as a whole; however, the scanners used were consumer off-the-shelf machines available through many retailers.
As a final note, we did follow Mr. Barron’s advice to watch the State Farm Arena surveillance video to see how the scanners worked. They are slow, as he stated. Indeed, they are so slow that the images of the ballots that are produced by the scanner are produced faster than the machine can scan the ballots. How extremely odd is that?