One month ago, Red Voice Media reported on New York’s “Voter Matrix.” The Voter Matrix is a hidden structure buried within New York’s voter rolls, algorithmically-produced, complex, and can allow tracking, retrieval, and interaction with fraudulent registrations if accessed by the right software. The Matrix was discovered by a team of researchers from New York Citizen’s Audit (NYCA), who made their findings public at an event known as “The Pit” produced by True the Votes’ Gregg Philips and Catherine Engelbrecht.
Now, NYCA has discovered that New York is not alone. New Jersey has something going on in their voter rolls also. In both cases, a fair analogy can be made with the WWII-era Enigma machine used by Germany to encrypt and decrypt messages. The Enigma machine had a keyboard to type a plain text message. Each key was electronically connected to between three and nine manually adjusted rotors, each of which could be rotated to any one of 26 positions. The rotors scrambled the input from the keyboard so that the word “CAT” might come out as “MYG.” This encoded text caused illuminated letters to flash, which were then transcribed and sent to the intended recipient.
To decode an Enigma message, three things are needed: 1) the message, 2) rotor calibration settings, 3) an Enigma machine. In New York, and now New Jersey, the voter rolls themselves are the encoded message. The algorithm used to create the Matrix is the equivalent of the rotor calibration settings. All that is missing is the equivalent of the Enigma machine to make sense of the rotor settings. The equivalent to this in the voter rolls is a software tool designed with the Matrix in mind, possibly the same tool that introduced the Matrix into the voter rolls.
The value of a tool like this follows from two lines of evidence. The first is that fraudulent ballots appear to have been introduced into the official count in the 2020 election (and likely others). Second, fraudulent registrations have been found in multiple states. Fraudulent ballots are more likely to be discovered if they cannot be reconciled with an equal number of registrations. Fraudulent registrations are needed to avoid conflict with real voters.
We know the fake ballots and fake registrations exist. However, another problem remains: how to access the fraudulent registrations so that votes can be assigned to them? Without a map of some kind to locate the fraudulent registrations, they are as good as lost forever in an ocean of millions of records. The algorithmically-driven Matrix can solve this problem.
The Voter Matrix is similar to a transformation matrix found in 3D animation software. A transformation matrix is used to move, rotate, or scale objects. The Voter Matrix uses an algorithm to shift voter records into a highly idiosyncratic, unlikely to be discovered, pattern. The pattern can be accessed by anyone with prior knowledge of the Matrix and the right software.
NYCA found five patterns in their rolls. One was used in 52 counties, a variation of it in five others, another variation in one, a totally different pattern in four, and one more in a segregated portion of the voter rolls shared by all 62 counties. In New Jersey, there appears to be one algorithm shared across all 21 counties. The New Jersey pattern is totally different from anything found in New York but it does the same thing: it shifts the position of voter ID numbers. The new locations are only accessible to keyholders with knowledge of how the numbers were transformed.
In New York, ID numbers were transformed by using County ID numbers (CID) to control the position of State ID (SBOEID) numbers. For instance, the CID numbers 000005, 000006, 000007, and 0000012 are connected to records with the SBOEID numbers 20,307,393, 20,308,504, 20,309,615 and 20,310,726. If records are sorted by CID number, the SBOEID numbers fall into the order shown below (Figure 1). Notice that the gaps between CID numbers do not match the gaps between SBOEID numbers, each of which is 1,111 units higher than the previous number. This disconnect proves that the pattern observed in the SBOEID numbers is not a reflection of the gaps in the CID numbers. To accomplish this, the two sets of numbers must be coordinated at the time they were created, each with a different purpose. The CID number establishes order, the SBOEID number establishes value.
New Jersey does not use a CID number as in New York. Instead, they use a “displayID” (DID) (Figure 2), and store a “Legacy_ID” (LID). The LID is an old (or original) ID number used before the new numbering system was introduced around the year 2018.
Analysis of New Jersey’s DID numbers revealed that they restructured the numbers to conceal hidden order. To see the New Jersey Matrix, one has to literally chop the DID number into three pieces and rearrange them in a different order. First, the leading alpha character must be separated and converted to a hexadecimal (hex) value. Then, the remaining 10 numbers must be divided into the first 5 numbers (Left ID) and the last five numbers (Right ID). These three segments must then be rearranged as: Right ID & Alpha & Left ID (Figure 3). It would be like reading the word “Razzmatazz” as “ErqueAlbuq,” or “Fabricated” as “Catedfabri.” This arrangement is unintuitive and therefore cumbersome for any legitimate use.
It is, however, an excellent method to conceal what has been done. There are no normal tools available in spreadsheet or database software that would allow the DID numbers to be sorted in their reconfigured form. To do that, one must either manually rebuild the numbers, or use custom software to read the last five digits of a number first, followed by the first six after converting one of them to a hex value.
As a programmer on NYCA’s team explained, the original form of New Jersey’s ID numbers do not naturally sort into any logical pattern. For instance, If the records are sorted by registration date, the legacy and display ID numbers do not follow the same pattern as the dates. Sorting either ID by itself also fails to create a coherent pattern. Reconfiguring the Display ID does create a coherent pattern. One of its more obvious characteristics is that all of the numbers are separated by even multiples of 100,000 and all of the alpha characters are grouped in alphabetical order, starting with “I” through the last character used, “P,” and then continuing “A” through “H.”
If New Jersey’s Display ID numbers are rearranged in the Right-Hex Alpha-Left (RHAL) configuration illustrated in Figure 3, a number of structural elements become evident. First, Legacy ID numbers now follow the same order as the RHAL Display ID. Second, the total number space allotted to the Right ID portion of each number becomes divided into three distinct partitions, just as in New York. Also like New York, the first and last partitions are similar and the one in the middle is different. NYCA has dubbed these, “in-range” and “out of range.” The pattern found in the out of range partitions (Figure 4), dubbed “Arcade” is designed to appear random and effectively disguises the presence of the in-range pattern.
The Right ID ranges are:
Low Out of Range: 00000-51,173
High Out of Range: 54,074-99,999
The in-range numbers account for only 2.9% of the available number space but 89.42% of the records are crammed into it. This suggests that the numbers are being used as tokens rather than as numbers. Otherwise, one would expect the full range of numbers to be used but it isn’t. Out of 6,459,433 records, only 68,392 of 100,000 numbers are used for the last five digits.
Another curiosity is that the numbers are definitely assigned to counties but in varying percentages. This is like having a zip code where 85% of the mail went to New Jersey, 10% went to Vermont, and the remaining five percent was split between 15 states. As a zip code, it is worthless. And yet, it clearly is designated primarily as a New Jersey zip code. Every county in New Jersey has a consecutive range of records with the same RightIDs that are assigned to it in very high percentages (above 80%). The remaining records are distributed to every other county in New Jersey based on a perfect normal distribution curve (Figure 5) in a pattern called “Icicle” by NYCA.
The Left ID is interesting because it never goes above 65,535. Instead, it increases from 0 through 65,535 and then cycles over and over again. This number is the highest number that can be represented in two bytes. To a programmer, this number is well-known because it represents an upper limit. This is similar to the way a non-programmer would take notice of an even number like 100,000. In hex, 65,535 is represented as “FFFF,” and the next number, 65,536, is “100,000” in hex. At the end of every cycle, the letter changes to the next letter in alphabetical order.
There are other interesting characteristics to New Jersey’s voter ID numbers. For the purpose of this article, the salient points are these: First, they are complicated. Second, the only way to extract meaning from the numbers requires reconstructing them in a specific way. Third, reconstructing the numbers creates an opportunity to utilize the new configuration to mine the records.
It is unlikely that any person working for the Board of Elections would have a legitimate reason to look for the patterns found by NYCA. It is also unlikely the patterns would have been found accidentally or intentionally during the normal course of business. Concealing structure by literally remaking each number following a simple set of rules is an excellent way to conceal what was done. This is different from appending a number with a checksum value, as is often done to ensure data integrity. In that case, one or more digits can be added to the end of a number but the number itself would not have to be inverted.
Restructuring the numbers does make some sorting tasks more efficient but only if the intention is to first invert the number, as was done. There is no advantage if the numbers are left as-is.
We know ballots were harvested by ballot “mules” and deposited in drop boxes across the country. We know that likely fraudulent registrations exist in many states, including New Jersey. We also know that a fraudulent registration is useless unless it can be found clandestinely, without anyone else discovering their presence. The way Display ID numbers are generated in New Jersey can be used in exactly that way.
And last, NYCA did a cursory check for suspicious registrations in New Jersey. They looked for people with multiple ID numbers (clones), people who registered before they were born, people who were too old to be alive, registrations on federal holidays, and other problems. They flagged 114,548 records. The Right ID numbers for all of the flagged records started at 51,174 and ended at 54,073. In other words, every single suspicious record they found fit precisely within the boundaries of the in-range partition. There were no strays. And the punch line is this: they generated the list before they discovered the in-range partition.