top of page


Updated: Apr 2



The information presented here is a top-level summary of the “Fingerprints of Fraud, Volume 1” report. Much additional supporting information can be found in the complete report, available at

The findings are taken from the 2020 General Election Cast Vote Records (CVR) of 202 counties from nine states, Arizona, California, Colorado, Idaho, Nevada, New Jersey, Ohio, Oregon, West Virginia. A CVR is a record produced by the election computer which shows each ballot processed and the votes which were counted on it, in most cases, in the order of processing. (If the order of the CVR was changed or sorted, this analysis cannot be performed.) They can also include the precinct, tabulator, batch, ballot style, and voting method. No voter-specific information is included.

Because mail-in ballots can be expected to be received and processed by the county in a random order, only mail-in ballots are used in the analysis. Since they show a random receipt pattern (based on metrics such as precinct distribution of the ballots), the votes upon them should generally follow the Law of Large Numbers, which dictates that the more samples you have of a random population, the closer their cumulative average should be to the final average. (Think of a series of coin flips or dice rolls).

  1. 168 of the 202 counties, or 80%, fall into a similar pattern where there is an initial high advantage for the Democrat candidate, and at some time before the midway point that average gradually turns toward the Republican candidate. Following is a sample that shows this pattern in Mesa County, Colorado.

The red lines on the first chart, which shows the percent of mail-in votes for Donald Trump during the counting (shown in blue), indicate the range which would be expected from the Law of Large Numbers. Instead of expressing the expected pattern, there is a large initial series of votes favorable to Biden, followed by the general gradual increase to the final average. The second chart shows the Trump percentage for each consecutive 100-ballot batch, and the gradual upward pattern is apparent. As I first noticed this phenomenon in this county, I refer to this as the “Mesa Pattern”.

2. The 168 counties referenced above also share a “predictive” aspect. Namely, the percentage which Donald Trump had at the end of the counting is 1.1 to 1.3x the percentage he had at the midpoint (with most hovering closely around 1.2).

3. Nearly all the remaining counties which did not show the Mesa Pattern were either very small (under 4,000 total ballots) or showed signs that the CVR was sorted by precinct or other index.

4. Statistically speaking, the chances of 168 counties in 9 states spread across the country demonstrating this same unnatural pattern are too low to be calculated.

5. The 168 counties showing the Mesa Pattern spanned five different election vendors: Dominion Voting Systems, ES&S, Hart Intercivic, Clear Ballot, and Smartmatic.

6. Numerous counties in other states also demonstrate this pattern and will be detailed in future volumes of the report.

These findings demonstrate proof that the mail-in votes in the 2020 General elections in those states were altered via a software algorithm. Please see the entire report (URL above) to view the complete details of each county analyzed.



Since the release of “The Fingerprints of Fraud Volume One”, numerous questions have been asked. The most common questions are answered in full below.

1. How do you know the “randomness assumption” cited in the report isn’t just the way people voted, with Democrats sending their mail-in ballots early and Republicans sending them in late?

Reports of the receipt dates of absentee ballots which are able to be tied to a specific voter and thus their party affiliation are difficult to obtain. I have evaluated several which are available, and while some show this phenomenon to some degree, that degree is not significant enough to explain the findings. This rebuttal theory has other problems.

a. While the party of the voter may be known, their actual vote cannot. Defections of registered Democrats in the 2020 election is a documented phenomenon¹.

b. Many states have many voters registered as non-affiliated, independent, or to a third party. This interferes greatly with making any assumptions about the actual voting patterns.

c. Reports of mail-in ballot receipts would need to have their chain of custody proven to the actual date of receipt.

d. The similarity of the mid to end Trump percentage between the 161 counties listed in the report would require that not only did the pattern (Democrats voted mail-in early, and Republicans also voted later) exist, it existed with the same statistical fingerprint in all of them.

e. Some counties processed their mail-in ballots as received, particularly states like Colorado in which the counting began on October 19ᵗʰ. Others stored the mail-in ballots until they could be all counted on election day. An example of the latter is Salem County, New Jersey, in which I was told by the clerks that the absentee ballots were added to bins as they came in. In this case, the ballots would be taken out in a random fashion, perhaps even the opposite order they were received. Nevertheless, both are examples of the Mesa Pattern.

f. Some counties do not exhibit the Mesa Pattern in their mail-in ballots. One would have to believe that the people in those counties behaved completely differently than most of the rest of the country.

2. Why are there differences in the different Mesa Pattern counties, for instance when the rise begins and places within the rise that shift up or down?

The computer algorithm required to produce the findings in the report is behaving as a controller and is constantly monitoring the real votes being processed with the goal of achieving as close to a desired result as possible. Unexpected organic votes require the controller to adjust in the same way the cruise controller in an automobile must periodically adjust the amount of fuel and brake action to achieve the desired speed.

3. Why would some audits/recounts match the original results if those results are manipulated?

This answer assumes that the original ballots were not altered, a case I discussed in the original report.

While the Mesa Pattern proves computer manipulation of the order of the votes, it does not necessarily imply computer manipulation of the votes themselves. My research, along with the research of others23, shows that significant number of fraudulent mail-in ballots were injected into counties at the beginning of the 2020 election. If this injection was sufficient to achieve the desired final outcome, the algorithm would need only monitor and smooth the results throughout the election.

4. So, this is how the election was stolen?

This is one of the ways. The algorithm was quite probably manipulating the other types of ballots as well as mail-in. But as those are nearly always counted or sorted by precinct, there can’t be any randomness assumption, and thus the same methods that detect the Mesa Pattern in mail-in votes doesn’t apply. In addition, there have been numerous other “attack vectors” identified by other researchers. It is my belief that a combination of methods was used in each county depending upon what would work best in that county.



A simulation verification of the report's concepts


A simulation depicting two examples of the expected voting pattern. 99.7% of the time, the data will be within 3 Signa Limits, as it is for these examples.


The voting pattern for Mesa County, Colorado. This represents the Mesa County pattern.


A simulation of the Mesa County Pattern, illustrating how the Mesa County Pattern is occurs.

67 views0 comments


bottom of page