The Washington PostDemocracy Dies in Darkness

‘2000 Mules’ offers the least convincing election-fraud theory yet

Dinesh D'Souza at a May 6 campaign rally for Republican Senate candidate Mehmet Oz in Greensburg, Pa. (Jeff Swensen/Getty Images)
14 min

There’s one scene in particular that I think summarizes the irredeemable flaws of Dinesh D’Souza’s new movie “2000 Mules,” in which he purports to demonstrate rampant illegality surrounding the 2020 presidential election. The film has become a central part of Donald Trump’s assertions about the election, with the former president hosting a screening last week at his Mar-a-Lago resort. But, interestingly, the most revealing scene doesn’t have anything to do with the election at all.

In it, D’Souza is hearing from a man named Gregg Phillips about how cellphone geotracking works. In short, your phone has various tools that allow it to know roughly where it is at any given moment, data that is often collected through apps and shared with companies that aggregate data for marketers. Phillips uses that data, which also includes time stamps, to show that only a few phones were in the vicinity of a fatal shooting in Atlanta — an incident that Phillips’s colleague Catherine Engelbrecht describes as “ebbing on cold-case status.”

“You could see, visually, that there were only a handful of unique devices that could possibly have pulled the trigger,” Phillips says. He shows a circle overlaid on a map, within which five dots of different colors are visible — dots indicating “the only potential legitimate shooters,” he says. He explains that, having done this analysis, his team turned information about those devices over to the FBI.

“Now, I read, they've arrested two suspects,” D'Souza says.

“They have,” Phillips says, somberly.

There’s a reason for this scene. Phillips and Engelbrecht’s analysis of geotracking data is the crux of D’Souza’s claims about there being an army of people who were dispatched to collect ballots before the presidential election. If data can be used to identify and arrest criminals in one case, the movie would have us believe, it can be similarly used in the case of all this alleged election fraud.

But looking at the case more closely, you see how the impression you’re meant to have is wildly misleading. The shooting led to the death of Secoriea Turner on July 4, 2020. It was far from a “cold case” — police arrested a suspect about two weeks later after he turned himself in. A second suspect was arrested in early August 2021 — not by federal law enforcement but by state officials. There is no indication that geolocation data played a role in either arrest, much less data provided by Phillips’s team. Update: NPR later confirmed that the data did not play a role.

In other words, D’Souza is elevating shaky, misrepresented, incomplete claims to bolster his rhetoric — as I said, an apt summary of the movie overall.

D’Souza declined to comment for this article.

Sign up for How To Read This Chart, a weekly data newsletter from Philip Bump

“2000 Mules” can be broken out into three basic components. There’s the geolocation-based material that’s the heart of D’Souza’s assertions about the election. The second half of the movie is a broader effort to undergird the geolocation claims, an attempt to build a foundation of how and why a rampant ballot collection scheme might have been undertaken. And then there’s the third part, a sort of cable-news-style panel conversation with D’Souza and several other conservative and right-wing pundits. (All of those pundits, incidentally, have shows with Salem Media Group, which served as executive producer of the film.) By the end, the pundits have been convinced that rampant fraud occurred, with former Trump administration official Sebastian Gorka outlining all of the evidence that had been presented “empirically” in support of the claim.

There is no such empirical evidence, by a long shot. That geolocation data from Phillips and Engelbrecht’s group, True the Vote, which also has executive-producer credits on the film, is used as a purportedly data-driven latticework on which everything else hangs. But beyond lots of harrumphing about how revealing this data is, we see very little of it.

The theory that Phillips and Engelbrecht present is that nonprofit organizations employed people to collect ballots and then drop them into drop boxes in various cities. They call this “ballot trafficking,” a term meant to connote illegality akin to the transport of narcotics. The people carrying the ballots, then, become “mules” and the nonprofit groups “stash houses.” To test this theory, they obtained a large amount of anonymized cellphone geolocation data and tried to figure out how often individual phones appeared near drop box sites or near those nonprofit groups.

By itself, this is a dubious approach. As the Associated Press pointed out in a fact check of the film, there’s no way by just using cellphone data to know whether someone visited a ballot drop box, particularly since those boxes were installed in high-traffic areas. Last month, I spoke with an expert on geolocation who made clear that the imprecision of phone geolocation would make specifying that a phone was actually near a drop box (and not, say, 10 feet from it) nearly impossible. The film makes repeated comparisons to federal law enforcement’s ability to identify people who entered the Capitol on Jan. 6, but even if the phone’s location is off by 20 feet, it’s still obvious when you’re inside a large building. (In one shot, the film shows geolocated data inside the Capitol — with positions surrounded by large circles of uncertainty that make this point clearly.)

In essence, we're just asked to trust that True the Vote found what it says it found. That by itself is probably not wise.

Phillips first rose to national attention in 2016 when he claimed, without any evidence, that millions of people had voted illegally in that year’s presidential election. Trump jumped on the claim, but Phillips never presented any evidence it had occurred. There was no reason to assume it had.

So we get sweeping claims about how many “mules” True the Vote identified in each city and the average number of drop boxes each visited. We’re shown one map of the travels of one “mule” throughout one city on one day, but even that is simply offered by Phillips as representing “a smoothed-out pattern of life” that we’re asked to assume is accurate. Everything else is just offered in the aggregate.

To bolster the claim, though, the movie spends a lot of time showing video from ballot drop boxes, obtained with public records requests. Phillips and Engelbrecht narrate what we're seeing, framing all of it as evidence in support of their theory.

In one bit of footage, we see a woman come and use a drop box. She puts in “a small stack” of ballots, Phillips says, “maybe three, maybe four,” and then removes latex gloves that she had been wearing and throws them away. This happened at 1 a.m., the True the Vote team says, making it more suspicious.


  • It’s not at all clear that the woman is putting more than one ballot in the box. There’s just one thin white rectangle that gets slipped into the box.
  • This was on Jan. 5, 2021, during Georgia’s runoff election, so it had nothing to do with the presidential race.
  • The woman is wearing gloves and a mask — suggesting that she is taking precautions against the coronavirus.

Is it hard to believe someone might wear latex gloves to access a publicly used drop box if one is worried about infection?

We are later told that the True the Vote team noticed the “mules” wearing gloves only after a late-December election-fraud indictment in Arizona — an indictment that stemmed, we are told, from fingerprints being used to catch culprits. That indictment involved the collection and submission of four ballots, and it’s not clear that fingerprints were an important part of tracking down the alleged culprits. But this additional — and hard to verify in the moment — piece of information makes the woman’s actions seem much more nefarious.

The True the Vote people claim that they identified the same woman “in a number of locations” and that she went to “dozens and dozens over the course of these two elections.” But we don’t see the map of her geolocated activity. We don’t even see a map of her cellphone going to that drop box on that night in January 2021. We don’t see video footage of her at another drop box. We’re just asked to believe that all of this occurred, without evidence.

Then there's the guy with a bike. He arrives at a drop box, removes a ballot from his backpack and puts it in the drop box.

“You also see him get sort of frustrated as he starts to leave,” Phillips claims, though there’s no obvious evidence of frustration. “Because, guess what? At this point, they had started requiring the mules, apparently, to take pictures of the stuffing of the ballots. It appears that that’s how they get paid.”

You then see the guy park his bike by the drop box and take a picture of both.

“If you’re just casting your own ballot,” Engelbrecht says, “what reason in the world would you have to come back and take a picture of the box?”

The answer is obvious. The particular drop box in this case can be easily tracked to the Ponce de Leon Library in Atlanta. If you look on Instagram, there are a number of people posting photos of themselves voting at this drop box. (For example.) There are any number of other photos people posted on social media showing themselves, say, riding a bike to drop off a ballot. Elections officials encouraged sharing voting experiences on social media in a bid to drive up turnout.

But that’s the sum total of the evidence offered against “bike guy” — that he took a picture. Again, he doesn’t appear to have multiple ballots. Again, there’s no geolocated data putting him there. It’s just this one snippet of purported sketchiness that comes down to a guy taking a picture of himself voting. Notice, too, that Phillips doesn’t purport to have any information from anywhere else that taking a photo was a requirement to get paid — he just says this is “apparently” a new requirement. (The lady with the gloves didn’t take a picture.) Phillips just layers that story on top of the video to make it seem as if the guy with the bike was part of a conspiracy and True the Vote had unpacked this complicated scheme.

At no point is there evidence presented of people getting ballots from a nonprofit group and dropping them in drop boxes. There’s one anonymized interview with a woman in Arizona who claims this happened, but there’s no geolocated or video evidence of ballot drops being made. Despite a confused scene with the pundit panel in which it’s alleged that maybe some of these ballots were submitted on behalf of dead people or people who moved, there’s no evidence of that happening. There aren’t even very many snippets of people casting more than one ballot — a practice that wasn’t itself necessarily illegal.

One segment of the movie shows a man depositing multiple ballots in a county in Georgia. But there’s a trivial — and legal — answer, as the Atlanta Journal-Constitution’s Mark Niesse reported: He was dropping off ballots for his family, which he’s allowed by law to do.

That the film does not even try to show that man making any geolocated visits to other drop boxes should by now not need to be mentioned. What appears to have happened is not that True the Vote found “mules” and then placed them at drop boxes using governmental video surveillance. Instead, the group appears to be trying to imply overlap between two large, distinct data sets — one with dots on a map and one with unusual-looking activity at drop boxes.

If there’s no proof of “mules” running around, the film’s second-half effort to explain how the system works is unimportant. (It’s mostly a mishmash of old claims about fraud and the au courant effort to cast increasing voter turnout as devious.)

But none of this is really the point. The point, instead, is that the viewers come away from the movie believing that they were right all along about the election being stolen. And so D’Souza scratches that itch by adding up the number of “mules,” the number of drop box visits and the average number of ballots deposited to determine how this network of criminality swung the election to Joe Biden.

It's useful here to be blunt: Every part of the calculus that D'Souza uses to show that Trump really won is nonsense, as he himself inadvertently makes clear.

First, it depends on True the Vote’s “mule” estimates being accurate, which for the reasons stated above should not be assumed. Second, it weirdly relies on the average number of drop box visits per “mule” instead of just a total number of visits, which one would think True the Vote could provide. Third, it assumes that those votes are invalid or would not otherwise have been cast, which is not a defensible assumption. (In fact, speaking at a legislative hearing in Wisconsin in March, Engelbrecht noted that “we’re not suggesting that the ballots that were cast were illegal ballots.”) And fourth, it relies on True the Vote’s estimate that each drop box visit included the drop of five ballots on average.

Consider that for a moment. What on Earth could that be based on? There is one scene in which True the Vote notes that a drop box in Georgia had more ballots than would have been expected based on the number of visits observed in video footage during the 24 hours prior, but there’s no attempt to understand why that might be the case. Did they see someone roll up with a giant stack of ballots? If so, why isn’t that in the movie? Even if all of the rest of this were true, there’s simply no way to know how many ballots were dropped in a drop box by a “mule.”

After declaring that Trump would have won the election based on the math above, D’Souza does something even odder: He just assumes that there were more mules than True the Vote counted and, for no explained reason, that they were averaging three instead of five ballots per drop. Suddenly, Trump wins every contested state!

And if there was a secret effort to dress chipmunks up as people and cast in-person ballots for Trump to the tune of, oh, 40 percent of turnout in each state, removing that criminal activity gives Biden a massive electoral victory! I have video of a chipmunk in my yard that I believe is carrying a ballot, so who’s to say my theory isn’t accurate? If we make up whatever numbers we want, we can do all sorts of interesting things.

At its heart, “2000 Mules” is a triumph of capitalism. There’s huge demand for proving that Trump didn’t lose in 2020, and this film provides just enough of a veneer of authority to let people collapse comfortably into that belief. That it doesn’t survive even mild external scrutiny is as irrelevant as pointing out contradictions in a religious text is to recent converts: They want to believe what they want to believe.

“Their ability to keep their side ignorant is total,” radio host Dennis Prager said during the pundit panel portion of the film. It’s an interesting commentary on how partisan belief works, certainly.


An earlier version of this story misspelled the name of Atlanta Journal-Constitution reporter Mark Niesse.