Fact Checking the AP's Fact Check on 2000 Mules
The AP's hit piece on Dinesh D'Souza's documentary that ran in over 200 newspapers is full of gaping holes, flawed assumptions, and blatant speculation.
Dinesh D’Souza’s documentary, 2000 Mules, follows Catherine Engelbrecht1 and Gregg Phillips2 as they use cell phone data combined with video surveillance to identify 2,000 individuals who may have deposited between 380,000 and 810,000 illegal ballots in dropboxes in just five states. The two passionate entrepreneurs leveraged decades of software experience and millions in donations to conduct the study to determine if election fraud could have changed the outcome of the 2020 election.
Almost immediately after the release of the documentary, Ali Swenson from the Associated Press wrote a hit piece titled, “Fact check: Gaping holes in election fraud claims of ‘2000 Mules’ movie.” Swenson’s article appeared in over 200 papers across the United States and is being used by social media to censor the study and the documentary. Perhaps worse these so-called fact-checks are being used to undermine the credibility of the social media users exposing them to ridicule and shame. For example, a friend who streamed the 2000 Mules documentary last night shared a link to the movie on their Facebook timeline recommending it as a “must-see”. Facebook sent them a warning that they were alerting their friends, family, and business associates that they were sharing misinformation. Facebook takes the warning a step further threatening to begin suppressing their feed if they ever spread misinformation again - the goal is to get users to self-censor. Ironically, Facebook has no issue sharing Swenson’s article attempting to poke holes in Engelbrecht and Phillips’ study despite the fact that she makes numerous flawed assumptions and blatantly false assertions. Example of a Facebook warning:
There is no doubt that D’Souza, Engelbrecht, and Phillips believe their data proves widespread election fraud occurred in 2020. And while I will spend the remainder of this article challenging Swenson’s false claims about the documentary and the study, she is right about one thing: the study, even if true, does not prove that the fraud resulted in President Biden’s election. The truth is that ONLY a federal investigation will be able to determine who benefited from the fraud. Engelbrecht and Phillips have identified how the fraud was perpetrated and who the perpetrators are - now it is time for the DoJ, FBI, and or a Special Counsel to open a formal investigation into the 2,000 individuals who may have stolen the 2020 election. The investigation will not change the outcome of the 2020 election but it will allow us to take the steps necessary to ensure it never happens again.
Cellphone Location Accuracy
Ali Swenson with the AP reported that cellphone location data is not precise enough to confirm whether or not someone deposited a ballot into a dropbox. Swenson reached that conclusion after speaking with Dr. Aaron Striegel, a professor of computer science and engineering at the University of Notre Dame. Striegel conducted a two-year study of two hundred Notre Dame freshmen back in 2011 detailing the difficulty he experienced tracking cell phone location at scale in a paper titled, “Lessons Learned from the NetSense Smartphone Study.”3 Swenson got the quote she was looking for when Striegel explained, “You could use cellular evidence to say this person was in that area, but to say they were at the ballot box, you’re stretching it a lot.”4
Of course, Striegel had neither watched the documentary nor read the Engelbrecht-Phillips “Study” he was helping Swenson debunk. If he had he would have realized that the Study never claimed that the cell phone location data used showed whether or not someone deposited a ballot into a dropbox.
While the location data used in the Study achieved an average resolution between three and six feet it could not, as Dr. Striegel pointed out to the AP reporter, determine whether or not someone dropped a ballot into a nearby dropbox. In fact, it was only the first piece of the puzzle.
The researchers began their efforts by drawing imaginary geofences around each dropbox. They decided to flag every cell phone that entered at least ten different geofences between October 1st and November 3rd. Then they looked at these flagged cell phones and determined if there were any common areas visited by these cell phones. The researchers then drew geofences around these secondary locations and applied a second flag to any cell phone that visited at least five of these secondary locations between October 1st and November 3rd.
Next, the researchers took historical location data for these cell phones to filter out any device that regularly entered these geofences prior to the October 1st and November 3rd period - Uber, FedEx, USPS, etc. The researchers also used data enrichment services to conduct identity resolution to filter out people who carry multiple cellular devices - iPhone, iPad, Apple Watch, Hotspots, etc.5
Using these filters researchers were able to narrow trillions of location pings and millions of devices down to just 2,000 - a very manageable number. The researchers then compared the time-stamped location of each device with the corresponding time-stamped video from each location. If the researchers saw videos of the same individual dropping off ballots at multiple dropboxes they knew they had caught a so-called ballot mule.
When Dr. Striegel gave Swenson his expert opinion, having not watched the documentary or read the Study, he was unaware that the reported location of each cell phone was only one small piece of a much larger puzzle. We reached out to Dr. Striegel for comment but did not receive a response to messages we sent via email and voicemail.
Ali Swenson reported that the 2,000 people who were observed dropping off multiple ballots at ten or more different dropboxes may have been simply dropping off completed ballots for family or household members or people with disabilities.
For example, Swenson cited a Michigan voter named Larry Campbell who told the AP that he legally dropped off six ballots at a local dropbox for himself, his wife, and his four adult children. Larry was not featured in the film nor was he counted among the 2,000 so-called mules. Even if Larry had dropped each of the six ballots at six different dropboxes around the state of Michigan he would not have been included in the Study. To be included he would have needed to find four additional ballots and drop them off at four additional dropbox locations AND visit at least five of the secondary ballot distribution areas geofenced by the researchers.
Swenson suggests that there may have been legitimate reasons why someone might have visited five ballot distribution locations and more than ten dropbox locations. She notes that many of the secondary locations geofenced by the researchers were controlled by nonprofits or NGOs and delivery drivers, postal workers, cab drivers, poll workers, and elected officials may have had legitimate reasons to visit on a given day. As noted earlier the Study controlled for devices that historically had entered ten or more dropbox geofences and five secondary geofences filtering out devices belonging to people associated with Uber, FedEx, USPS, etc.
Swenson argues that “pattern of life” filtering would not filter out election workers who spend more time at dropbox locations during election season. While she is correct that the historical location data would not filter out election workers, they were able to do so very quickly after observing the timestamped video. When they matched a device to a video of an election worker picking up ballots they would flag that device as belonging to an election worker and not as a mule. Of course, once they identified the device IDs of election workers they were able to track their movements throughout the day - bringing an entirely different set of concerns to light.
In another example, Swenson claims the fact that cab drivers’ daily paths don’t follow a pattern would prevent the researchers from filtering them out of their results. The historical filters used in the Study didn’t rely on a “route of life” but on a “pattern of life”. If a cab driver picked up and dropped off people at different hotels, restaurants, shops, or offices each day they would no doubt pass through at least ten geofences - perhaps not the same geofences but at least ten and as a result, they would be omitted from the results. Finally, she suggests that someone whose routine recently changed might be included in the results. Of course, she is correct. If someone began visiting ten or more locations with dropboxes starting on October 1st as well as five or more secondary locations where ballots are distributed they would be flagged. But once timestamped videos of the ten dropboxes were reviewed and failed to show them dropping off ballots they would be excluded from the results.
What Swenson doesn’t seem to grasp is the fact that the Study was never meant to be used to overturn the results of the 2020 election. It was designed to expose a potential flaw in our dropbox voting and provide law enforcement with a roadmap to prosecute those who were involved in the fraud. Anyone who was identified but didn’t participate would no doubt be exonerated by law enforcement.
Philadelphia’s Failure to Provide Video of Dropboxes
Ali Swenson points out that election officials from Philadelphia failed to provide the researchers with timestamped video of the city’s dropboxes and as a result, their findings have not been corroborated by video. While it is true that election officials in Philadelphia did not provide the group with video the researchers did obtain extensive time-stamped video surveillance of the dropboxes taken by the Trump campaign.
In fact, both of the officials Swenson reached for comment attempted to stop the Trump campaign from collecting video surveillance of the city’s dropboxes. Swenson reported that Pennsylvania state Senator Sharif Street believed he counted as “several” of the so-called mules because he carries “a cellphone, a watch with a cellular connection, a tablet with a cellular connection, and a mobile hotspot — four devices whose locations can be tracked”. Senator Street had not watched the documentary or reviewed the Study or he would have known that the researchers conducted identity resolution to link multiple devices used by a single individual. In fact, Senator Street didn’t seem to realize that simply passing by the same dropbox seven or eight times a day would not cause his device to be flagged. Street would have had to visit ten different dropboxes AND five secondary locations to be flagged. We reached out to Senator Street for comment but he did not respond to our inquiry.
Swenson quoted Philadelphia’s city election commission spokesman, Nick Custodio as saying that the claims made in the documentary and Study matched other allegations that had been debunked or disproven despite the fact that Custodio had not seen the documentary or read the Study. Custodio explained, “The Trump campaign and others filed an unprecedented litany of cases challenging Philadelphia’s election with dubious and unsubstantiated allegations of fraud, all of which were quickly and resoundingly rejected by both state and federal courts.” But a review of the various cases in Pennsylvania reveals that none of them relied on the data or methods utilized by the researchers in this particular Study. We reached out to Custodio asking him to show us which challenge was similar and why it was rejected unfortunately Custodio did not respond to our inquiry.
Ali Swenson claimed that there was no way the researchers could have determined the political affiliations of the 242 people they flagged as potential mules in Atlanta. To start, the researchers never claimed they could determine the political affiliation of these flagged individuals. Instead, the researchers simply noted that 67 of the 242 flagged individuals participated in violent BLM and Antifa riots in Atlanta. They were able to do this by comparing their data with data collected by a company called Mobilewalla. Buzzfeed reported that the company had been tracking the cellphones of individuals who attended BLM and Antifa riots in Atlanta and appended it with demographic data.6 It turns out if you know where someone sleeps at night you can determine with great accuracy who they are by appending their address with data from companies like Experian.7 Mobilewalla’s append allowed the researchers to determine the gender, race, and criminal history of the 67 flagged individuals - most were male, Black, had a criminal history, and not registered to vote.
Ever the optimist, Swenson suggests that some of these 67 people could have been peaceful protesters, police, firefighters, or business owners in the area - and again she is right. Some of the 67 people who had visited at least ten dropboxes and five secondary locations AND attended a BLM/Antifa riot might have been peaceful posters, police, firefighters, or business owners - but it is also very likely they were involved in ballot fraud. Again, the objective of the study isn’t to overturn the outcome of the 2020 election but to provide a roadmap to law enforcement to unmask the 242 individuals flagged by the study to find out if laws were broken.
Gloves in Atlanta
The researchers began seeing videos of the 242 individuals they had flagged in Georgia wearing latex gloves starting on December 24th during the primary election. The researchers speculated that it might have something to do with the FBI’s indictment of two women in Arizona for alleged ballot harvesting. Ali Swenson discounts this in her article by pointing out that the Arizona indictment didn’t mention the fact that the FBI used fingerprint identification to catch the women. Once again Swenson is correct - the indictment didn’t mention fingerprint evidence but what she ignores is the fact that the media was reporting that ALL election officials, including elected officials, had been fingerprinted as part of the investigation. The media was reporting that fingerprints were being obtained from ballots by the FBI - true or not it is very likely that anyone participating in an illegal ballot scheme similar to the one foiled in Arizona would be aware of the reports.8
Ali Swenson with the AP suggests that it was more likely that the 242 individuals flagged by the researchers in Atlanta began wearing latex gloves on December 24th because of cold weather. There are just two problems with Ali’s theory. First, latex gloves provide almost no protection from the cold. Second, temperatures in Atlanta were in the mid-sixties - the day after Christmas it was in the 70s.9 Swenson then suggests that the flagged individuals who began wearing gloves on the 24th may have done so as a result of the risks associated with Covid-19. The fact that the individuals removed their gloves as soon as they deposited the ballots in the dropboxes suggests that they were either worried about fingerprints or about catching Covid from whomever they receive the ballots from - either way further investigation is warranted.
The researchers also noted that they began seeing the people they flagged dropping off ballots at various dropboxes and taking pictures as they did. Swenson argues that many voters take photos of their ballot envelopes before submitting them. Again, Swenson is right but how many American voters take pictures as they drop ballots off at ten or more dropbox locations? Would they post all of these on their Instagram or TikTok accounts? Sometimes I think Ali is gaslighting the reader by taking a ‘fact’ out of context and then pretending by doing so she has debunked the entire documentary. While it is true that normal voters take pictures of themselves voting ALL OF THE TIME it is also true that normal voters don’t take pictures of themselves voting MULTIPLE TIMES.
Ali Swenson asked Derek Muller, a law professor at the University of Iowa if there was any evidence that a massive ballot harvesting scheme dumped a large number of votes for one candidate into dropboxes. Muller, who had not seen the documentary or reviewed the Study, claimed there was NO evidence of such a scheme and speculated that if there were it would likely be caught quickly because people can’t keep secrets. I would argue that criminals involved in lucrative felonies to advance their ideological goals rarely admit their crimes.
Swenson interviewed Dr. Barry Burden, a University of Wisconsin-Madison political science professor who claimed that absentee ballots were verified by signature and tracked closely. The reality is that Wisconsin does NOT conduct any sort of signature verification of absentee ballots. Neither does Georgia, Pennsylvania, New Mexico, Virginia, and Maryland. Dr. Burden also noted that “It seems impossible in that system for a nefarious actor to dump lots of ballots that were never requested by voters and were never issued by election officials.” The professor should be excused for not realizing that the ballots in question were both requested by voters and issued by election officials because he didn’t watch the documentary or read the Study. We reached out to Dr. Burden for clarification and he explained that “I am extremely busy this week and won't be able to answer your questions.”
The Scope of the Scheme
Ali Swenson and her colleagues in the media have long held that any election fraud that took place in 2020 wasn’t large enough to impact the outcome of the election. She correctly points out that no one knows if the ballots collected and deposited by the 2,000 individuals identified in the Study were for Trump or for Biden. All we do know is that if the 2,000 individuals identified by this Study deposited ballots for Biden the outcome of the election was impacted. Either way, the American people deserve to know that their elections are secure.
On January 6th the FBI collected the cell phone data of every protester at the Capitol.10 Using that data they were able to unmask the identity of everyone who was at the Capitol that day resulting in FBI agents visiting the homes of more than 2,000 Americans who participated in the protests.11 The Engelbrecht-Phillips Study collected the cell phone data of 2,000 Americans who may have participated in an election fraud so large that it might have changed the outcome of the 2020 election. To date, the DoJ and FBI have refused to review the Study much less open an investigation. The goal of Dinesh D’Souza’s documentary was to put pressure on the Department of Justice to appoint a Special Counsel to open an investigation based on the findings and data collected in the Engelbrecht-Phillips Study. The FBI was able to interview and investigate more than 2,000 Americans from every corner of the country in less than a month based on cell phone data collected at the Capitol on January 6th - surely it would take less time to interview and investigate 2,000 Americans from just five cities. Perhaps if they did, another event like January 6th could be avoided.
Catherine Engelbrecht is a graduate of the University of Houston and a resident of Austin, Texas. She is the co-founder and CXO of CoverMe an AI-powered SaaS application for hospitals and long-term care providers. She is also the president of True the Vote Inc. a 501(c)(3) organization established to improve the integrity of elections in the United States. The organization is the country’s largest voters’ rights organization protecting election integrity since 2010. Her co-founder, Taylor Phillips, worked with Greg Phillips at AutoGov and previously served as Director of Information Technology for True the Vote. In 2011 CPAC presented Engelbrecht with the Ronald Reagan Freedom Award in recognition of her work at True the Vote. Politico named Engelbrecht to their “50 to Watch” list recognizing her as a rising talent on the national political scene.
Engelbrecht was making enough of an impact to cause the Lois Lerner Director Exempt Organizations for President Obama’s IRS to target her organization. In less than a year True the Vote received twenty-three audits, investigations, and inquiries by the IRS, DOJ, FBI, ATF, and OSHA. In response, Engelbrecht sued the IRS and the administration for targeting her organization in 2013. In 2014 Lois Lerner was held in Contempt of Congress for refusing to testify about Engelbrecht’s claims and when her emails were subpoenaed she claimed she had lost all of them as a result of a hard drive crash. In 2015 the IRS admitted to the court that Lerner had a second, secret, email account that she used to conduct IRS business using the name of her dog “Toby Miles”. Finally, in 2019 the court found that the IRS and Lois Lerner unconstitutionally discriminated against True the Vote awarding the organization the maximum attorney fees allowed under the law as a result of the agency’s unethical behavior in the case.
Shortly after the 2020 election Catherine Engelbrecht’s True the Vote accepted a $2.5 million donation from Frederic Eshelman a North Carolina billionaire. The billionaire was disappointed when True the Vote’s efforts did not overturn the results of the election and filed suit in federal and state court to get a refund. Both lawsuits were dismissed.
Gregg Phillips is a 62-year-old resident of Austin, Texas. Phillips has extensive experience using powerful data analytic tools to solve problems faced by state agencies. For example, in the mid-1990s Phillips ran the Mississippi Department of Human Services ultimately leaving the department to join a firm called Synesis Corporation that won various contracts from the state during and after his tenure. Then in 2003, he was hired to serve as Deputy Commissioner for the Texas Health and Human Services Commission to help draft legislation to help privatize the state’s billion-dollar human services system. After completing his work on the bill in 2004 Phillips left HHS and started AutoGov to build a software platform to aid caseworkers in state health and human services department in determining patient eligibility. The company’s software, called CaseVue, uses machine learning tools to fuse applicant data with public and private sourced data to determine the likelihood of eligibility in real-time to caseworkers all over the country. The company employed 24 people and generated about $5 million per year in revenue.
After Phillips’ election fraud investigations became public and cited by then-President Trump various news outlets began running hit pieces designed to undermine his credibility. The Associated Press accused him of various conflicts of interest as his company, AutoGov, did business with his former employers in the Mississippi and Texas Department of Health & Human Services. Additionally, the Guardian reported that in 2014 Phillips and his ex-wife underpaid their taxes by $100,961. Phillips explained that there was a disagreement over which party owed the taxes and how much was actually owed - the amount was determined to be less than $50,000 and that a settlement with the IRS was reached without incident. The Guardian reported that almost 30 years ago Phillips lied on a job application when he claimed he majored in finance at the University of Alabama while Phillips actually received a bachelor of science degree in commerce and business administration. The Guardian also reported that his ex-wife’s new husband once made an unsubstantiated claim that Phillips may have failed to pay child support at some point after his divorce.