The Washington PostDemocracy Dies in Darkness

‘Ballot trafficking’ is the next front in the unending fight over 2020

A voter delivers her ballot to an official ballot drop box in Milwaukee on Nov. 3, 2020. (Sara Stathas for The Washington Post)
11 min

On Tuesday night, the adoring crowd that Donald Trump addressed at Mar-a-Lago included a few more familiar faces than normal. The former president convened a number of his allies and former aides for a special event: the premiere of a film centered on claims that the 2020 election was rigged. His favorite people, his favorite topic, a special night.

The Washington Post’s Josh Dawsey was there and noted that Trump made reference to something else coming down the pike on the same subject.

“Trump promised that next week, the crowd would see levels of ‘ballot harvesting they never thought possible’ as part of a report from an outside group,” Dawsey reported. “It was unclear what he meant, but he said his advisers had been working with the project.”

One might understandably be skeptical that a weeks-away promise of information from Trump might not materialize, given how often such pledges have gone unfulfilled. But there’s good reason to think that, this time, some information is coming.

It’s just that we’ve already seen what’s coming, and it’s no more evidence of rampant electoral fraud than anything else.

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

In recent months, Wisconsin has become the center of the doomed fight to prove that the 2020 election was stolen from Trump. A report compiled by a Republican who was once elected to the state’s Supreme Court elevated familiar unfounded claims that the integrity of the election was compromised, with Michael Gableman, that former justice, suggesting that somehow the results could be rescinded, which they can’t. (Gableman, Dawsey writes, was in attendance at Mar-a-Lago on Tuesday and was praised by Trump.)

Gableman’s report was followed with a hearing by the state assembly’s Campaigns and Elections Committee last month, in which a group called True the Vote presented what it claimed was evidence of individuals collecting ballots and putting them into drop boxes in the state. The speakers, Catherine Engelbrecht and Gregg Phillips, presented sophisticated-looking analysis of cellphone location data that, they claimed, showed more than 100 people shuttling ballots around Milwaukee in advance of the 2020 election.

You might have seen this same data alluded to in the trailer for conservative commentator Dinesh D’Souza’s coming-at-some-point movie “2000 Mules.” (D’Souza, you’ll recall, was pardoned by Trump after pleading guilty to campaign finance charges.) That trailer uses the data collected by True the Vote to suggest that thousands of people in multiple states were involved in moving ballots around. Nefariously.

Before we go further, it’s worth evaluating this claim by itself. We know that companies can buy anonymized cellphone data from which information about people’s daily habits can be gleaned. The New York Times did a lengthy report on the marketplace for this information, a report that was included in the True the Vote presentation in Wisconsin as a way to validate its findings.

But there is a big difference between, say, identifying someone who worked at Microsoft and visited the Amazon headquarters for a job interview — as the Times report did — and figuring out that someone is visiting a specific drop box in a location. In 2020, Milwaukee placed ballot drop boxes in more than a dozen locations, including city libraries. Can cellphone geolocation data actually differentiate between visiting a library and visiting a drop box at that library?

Larry Daniel is an expert on the subject. He’s written multiple books as part of a two-decade career and holds certifications in Global Positioning System (GPS) forensics and cellphone location forensics. I spoke with him Wednesday morning to determine how precise geolocation on your phone can be when collected through apps.

“If you allow it to use precise location, which is your GPS, it can pinpoint you to within probably 30 feet,” Daniel told me. That’s a bit broader than what might be expected under ideal conditions for the satellite-based system (open sky, no obstructions) and a reflection of constraints placed on consumer devices. (Surveyors, Daniel noted, use much more accurate positioning devices.) He used the example of Waze, the navigation app. It can tell when you’re driving on the road, but not what lane you’re in.

And that’s using GPS. Your phone also can generally locate you using nearby WiFi access points and cellphone towers, but that’s a lot less accurate, Daniel said. What’s more, cellphones are regularly but not constantly recording your position. In other words, a ping (recorded location) directly next to a drop box outside a library might be you dropping off a ballot — or it could be you walking by on your way into the building. Or it could be you driving by on the street.

The True the Vote team worked around this by setting constraints on what it was looking for.

“The whole idea behind all of this is that we go in and we geofence, if you will, draw a geocoded polygon around a particular location,” Phillips said. Locations they looked at included drop box sites as well as nonprofit organizations that the group suspected were involved in an effort to harvest ballots. “Then we compile the cellphone signal data and build the pattern of life in such a way that we’re able to then determine when certain phones pass within a certain area.”

To describe how geofencing worked, he used the example of his sitting in that room at that time: Someone looking for pings in that vicinity would return a hit on his phone. He then said it would work if you were to geofence the specific desk at which he sat — a level of accuracy that Daniel would tell us is too precise to be able to isolate.

It’s important to note that True the Vote began with an assumption about what activity would look like. It was trying to find a specific pattern of movement and then found it, raising the distinct prospect of cherry-picking. They specifically looked only at movement between private organizations and drop boxes to limit how much processing power they needed to use. But perhaps they might have gotten the same patterns by looking at people who visited a particular Chipotle and several of the libraries where there were drop boxes. We can’t say.

Phillips claimed that their data showed 138 people making 3,568 visits to drop boxes — though of course they could say with certainty only that those individuals were tracked within the polygon drawn around a drop box location. (Here, too, a question: How loosely were those boundaries drawn?) This was an indication of what he called “ballot trafficking,” an intentionally more egregious-sounding descriptor than the usual “ballot harvesting.” And then Phillips went further.

“We believe 7 percent of the mail-in ballots, or approximately 1.9 million, I think, here in Wisconsin,” he said, “were cast as a result of trafficking.”

Where this figure came from is entirely unclear — and patently ludicrous. He’s arguing that 1.9 million ballots were illegally collected and submitted … but that it takes fancy geotracking data analysis to uncover that? The more important tell came a few sentences later. Phillips said that 7 percent “is a number that is held in every single community across the country that we’ve looked in.” In every community where they did their analysis, they found that 1 in 14 ballots was harvested and submitted? That consistency reminds me of Douglas Frank’s mathematical analyses of fraud — analyses that yield a consistent result because of how he massages the numbers, not because there was actual fraud.

Like Frank, True the Vote is not approaching this dispassionately. “Ballot harvesting,” Phillips said, is part of a multitiered system of collection, distribution, mailed ballots and various other things. He claimed that 14 of the devices they had identified as suspicious “participated in one or more of the violent riots in Wisconsin during 2020” — a claim that’s impossible to take seriously without clarification about how what constitutes “participation” and what constitutes “violent riot.”

At this point, we should note that you might have heard of Phillips before. A few weeks after the 2016 election, he claimed on Twitter that millions of illegal votes had been cast — a claim that Trump quickly elevated. It was indefensible nonsense at the time, given that the vote still hadn’t been certified in a number of states, and his promises that he would eventually prove his claim went predictably unfulfilled. This is the guy that the state of Wisconsin now asks that we take seriously in his analysis of complicated and hazy digital data about phone usage.

“It’s quite evident to those of us that have spent the last 15 months of our lives developing these numbers,” Phillips said in concluding his testimony, “that this was indeed an organized crime that was perpetrated on Americans.”

His colleague Engelbrecht then took over, quickly modifying Phillips’s claim.

“I want to make very clear that we’re not suggesting that the ballots that were cast were illegal ballots,” she said. “What we’re saying is that the process was abused.”

This, of course, is a massive qualifier, given the intended outcome of the Republican-led hearing and the enthusiasm from Trumpworld about its “revelations.” Here, again, the claim wasn’t that voters committed fraud. It is, instead, that Republicans don’t like how those votes got to the place to be counted. Engelbrecht offered all sorts of warnings about how this system might introduce potential fraud but offered no evidence that such fraud had occurred. In fact, one of the legislators in the hearing asked if she had seen any evidence in Wisconsin of the sorts of purported problems she claimed occurred with ballot harvesting; Engelbrecht talked around it by way of saying that she hadn’t.

And then there’s the kicker: ballot collection and drop boxes weren’t illegal in Wisconsin in 2020. The state Supreme Court has since ruled that they should not be allowed, but even if nonprofit organizations were collecting ballots and dropping them off, it was not illegal to do so.

In other states where D’Souza’s trailer suggests True the Vote collected data, the rules are different. Phillips pointedly mentioned the former mayor of San Luis, Ariz., who had been arrested for harvesting ballots. He mentioned that arrest at the outset of his comments, setting the tone for what he would later describe as “an organized crime” related to collecting ballots.

That mayor was accused of collecting five ballots. That’s presumably a bit less than 7 percent of the total in a city of more than 35,000 people.

The important point, though, is the one that Engelbrecht ceded — that Georgia Secretary of State Brad Raffensperger (R) made when similar allegations about ballot harvesting were made in his state. A voter casting a legal vote that is then submitted illegally (in states where collecting ballots is illegal) is still a legal vote. This doesn’t prove that Trump was the real winner in 2020, as Trump is eager to have you believe.

Since night fell Nov. 3, 2020, there have been scores of efforts to insinuate wrongdoing in the election based on activity that’s characterized as suspicious. Boxes under tables. Cardboard over windows. That sort of thing. Here, there isn’t even an allegation that illegal votes were cast. Instead, it’s just an effort to imply some sweeping, vile effort on the part of the left to hand the election to Joe Biden — to reinforce Trump’s general assertions about the deviousness of their shared opponents.

Trump, of course, is happy for the help.