A Look at Specific Fraud Claims in the 2020 U.S. Election

The run-up to and aftermath from the 2020 election has been rough, and it hasn’t been made any easier by the flooding of social media with dubious claims of fraud. One source of many such claims is in this article from The Federalist.

I’m almost certain that the originators of most of those claims are often suggesting them in bad faith, but for the purpose of this article, I’ll assume the people repeating them do actually care about the integrity of our elections. Thus, instead of insulting them or posting memes (well maybe just one, but not until later), which would be an unproductive exercise, I’m going to take them at their word and discuss the realities of some specific claims I see repeated all over Facebook, Twitter, NextDoor, Reddit, and other web-based hell holes where our humanity goes to die.

Some of these are more like categories of claims (e.g. more votes than registered voters), and you’ll likely run into many permutations. I’ll try to include the most common ones here, but I may add others if they are sufficiently unique to warrant addressing them individually.

Claim 1: Michigan “found” 140,000 votes overnight, which all went to Biden

The tweet that launched a thousand screeds

It turns out the source of this jump was a mere clerical error. Abigail Bowen, Shiawassee County elections clerk, reported that an extra “0” was added to a Biden vote tally, recording it as 153,710 rather than 15,371. She has been interviewed regarding this by various outlets, including the New York Times and AFP Fact Check, discussing how the error was caught and corrected. So on its face, this is not an instance of fraud.

However, it’s worth looking at this example a little further to tease out a common theme that exacerbates the spread of claims of fraud: ignorance about the way votes are tallied, checked, recorded, and reported. We’ll see versions of this problem in the other claims I address, but the specific complaint I’ve seen here is that “the fact that a typo could cause votes to jump like this is proof that we can’t trust the tallies”. Let’s dig a little deeper to understand what happened.

In our Federalist system every state runs its own elections, and every county runs theirs (within the bounds of those rules and procedures defined by the state). There are lots of ways things are done, and unofficial totals are often manually reported. In this case, per the aforementioned reporting from the New York Times, Ms. Bowen and her team sent the county’s unofficial vote totals to Michigan’s state-level elections officials early Wednesday morning, after finishing their tally.

These unofficial numbers were checked for accuracy, and within 20 minutes the state team called the county to verify the correctness of the tally, noting that Shiawassee County doesn’t even have the population to produce that many votes. The figure was immediately corrected, and the totals were updated. As you can see below, 15,371 votes were recorded, not 153,710.

per ElectionReporting.com

What happened here is that the system worked as intended. There is an inevitable human element in elections, and in fact there always must be to ensure that they are free and fair. Checks and balances are built in to the system, and those audit measures finding faults and correcting them is not front page news, or it least shouldn’t be, if one is trying to make a good faith effort at safeguarding our elections. If you’re worried about fraud this incident with the Michigan overnight count should be something to celebrate, not turn into a Twitterstorm of innuendo.

Finally, to be fair to Matt Mackowiak, he eventually retracted his tweet, although he still didn’t admit to bad faith.

Claim 2: A jump on the Wisconsin graph for Biden without any increase for Trump means fraudulent votes were “injected” to catch him up to Trump

Flattening the curve, indeed

This has become a very popular one, and it’s understandable. If you’re not used to reading graphs, and if you think about vote counting in “horse race” terms from a layperson’s perspective, it does seem unusual that you’d see such a dramatic jump. It’s now taken on a life of its own in context-less proto-meme form:

Well there’s your problem…

Now there are several things we can discuss regarding this, but I’ve already given you everything you need to know to solve the mystery. Notice anything?

It’s not like the lines mean anything, right?

That’s right, the meme that’s being passed around doesn’t even represent the actual graph in question! It’s a visual lie designed to create outrage. It shows Biden suddenly and dramatically jumping vertically up while Trump gains no votes, while in reality both campaigns jumped up, but Biden jumped up more. On that left side graph there is a red line for Trump’s new votes that lies “behind” the blue line.

There was no “all for Biden, none for Trump” vote dump. The reality is that counties in this case released large batches of results all at once, and the lion’s share of them were from Milwaukee absentee ballots. If you’ve been paying attention you’ll know that Milwaukee (county), population 950,000, is what passes for an urban area in Wisconsin, and urban areas tend to vote Democratic. Furthermore, the combined factors of Donald Trump calling mail-in ballots a scam and Biden supporters being more likely to take social distancing seriously mean that absentee ballots are generally more Democratic in 2020.

Hillary Clinton won Milwaukee 77% to 18% back in 2016, and polls leading up to this year’s election suggested Biden would do even better. So when Milwaukee reported all of its 170,000 absentee ballots all at once, it naturally caused a large spike in the graph for Wisconsin, with Biden receiving the majority of the new votes. But remember, it was not a spike just for Biden, so there’s no reason to suspect fraud. I agree that it would be suspicious if they reported that every absentee vote in Milwaukee was for Biden. But they didn’t, that’s just a lie intended to take advantage of people misunderstanding how election results are reported. There were Trump votes too, just far fewer than for Biden.

Like the last claim we examined, this one relies on people’s (understandably) incorrect intuition about how elections work. Because counting the votes take time, and news agencies want to keep our eyeballs on them with breathless updates on the tallies, many people look at the election like a horse race. First one candidate is ahead, then another…it’s very exciting! And indeed, it would be suspicious if halfway through the race one horse suddenly started running 75 miles per hour for a little while while the rest were stuck at normal speed.

But that’s not reality. When the polls closed, all the votes were already determined. They all existed already. The outcome was decided, and the candidates were not actually racing. We can actually count the votes in any order we want, and it is the order we count them in that determines the shape of the graph. Milwaukee could’ve chosen instead to count absentee votes in groups of 1000 every 15 minutes over the next 42 hours and the graph wouldn’t have had a big jump.

Thus, the order of counting doesn’t tell us anything about fraud. The whole moving-graphs “horse race” aspect of it is a human fiction. It’s not real; Biden had won Wisconsin when the polls closed, we just didn’t know it yet. If we could count the votes instantaneously none of these fraud claims would even exist, but the outcome would be the same.

Normally there’s not as much opportunity for the horse race aspect to confuse people as much because vote counts have happened faster. But 2020 has conspired to take us back to an earlier, slower time for elections (albeit with the instant media access we have today). Which brings me to the next claim…

Claim 3: The delay in vote counts is evidence of fraud

Apologies for the quality of this first one, it is a screenshot of someone else’s camera shot of a screen of the tweet. That’s how you can tell it’s a reliable primary source.

Was thinking of printing this out and then taking another picture of it and scanning that in and using it instead

I’ve seen this screenshot passed around over and over on Facebook the last few days. This person is supposedly an election official in Michigan. Of course, when I looked at Twitter I found a protected profile for someone supposedly from Indiana:


But I’ll ignore that for the purposes of this discussion. Sand(ra/i) says that any votes cast after 8:00 on Election Day are “not valid”, and that “any votes counted after midnight are fraud”. She also alludes to Claim 1, the Great Shiawassee County Extra-Zero Caper.

She seems to be implying that things should’ve been all wrapped up before midnight in order to be valid. Now, I can imagine a world in which that’s possible, but there are several things working against Michigan election officials.

First, and most obviously, the pandemic has massively expanded the use of mail-in ballots. I suppose this wouldn’t be an issue of a state required that all mail-in ballots be received by say, the day of the election at 6AM, and didn’t have that many of them. Michigan does in fact require that mailed ballots be received by the day of the election, albeit by 8:00 PM, making a complete count by midnight a tall order.

But surely they could get a head start on people mailing them in early? You’d think so, except that the GOP passed legislation preventing Michigan from counting any mail-in ballots until the day before the election. Incidentally, similar GOP-backed laws in Pennsylvania and Wisconsin, both critical swing states, prevented any mail-in ballot counting until Election Day itself.

So it turns out it’s a bit disingenuous for Republicans to complain about mail-in vote counting delays given that their own party has been a key factor in tying the hands of local election officials.

Before we move on from Sand(ra/i), I would be remiss if I didn’t link to my new favorite YouTube video about how she’s “a self-educated watercolourist from Holland, Michigan who has made momentous progress in a brief timeframe”. Every day I understand the Internet a little less, but I feel like that video must be scamming me somehow when I watch it.

Another claim I see is that bigger states like California have managed to count things right away, so why can’t Michigan et al?

Damn Californians and their [checks notes] functional elections!

As in the prior example, it seems a little disingenuous to complain when part of what California has going for it is that their state government didn’t prevent local elections officials from counting mail-in ballots before Election Day. But leaving that out, this one goes back to the issue of laypeople not understanding how elections work in the US.

It turns out that when this commenter made this complaint California was actually sitting at 77% of precincts reporting, versus 95% for Pennsylvania, the swing state in question in this particular discussion. So in fact, big ol’ California was actually behind, and had not “post[ed] their results almost instantaneously”.

The thing is, the media had already called the state for Biden, regardless of the lag in vote counting, and they did do it quite early. For many Americans the fancy maps on TV and news sites are the election, even though they are totally disconnected from any kind of official count. This commenter had the very natural impression that California completed their election tally within a few hours on Tuesday night, when in fact they were only at 77% several days later; it’s just that those remaining votes were totally unlikely to change the victor.

Finally, I’ve heard in several conversations the assertion that “if it takes a week to count the votes there’s no way to ensure it’s legitimate”. Now, I’m not sure what one has to do with the other, but the reality is the election won’t be truly decided until the electors vote on December 14th, 2020. California won’t even complete its certification of its results until December 11th. So this kind of “one week” deadline is essentially meaningless. One can imagine just how long it took to tabulate the vote in the 1800s; it’s just that no one was tweeting about it.

This all stems, again, from a deficit of electoral knowledge. I’m pretty sure I was never even taught how presidential elections actually work in school, beyond the general concept of the electoral college. We need to increase democratic literacy if we have a hope of combating the misinformation that feeds on this kind of ignorance.

Claim 4: Faulty software gives extra votes to Biden

Explaining what happened here requires us to dive a little bit into how vote tallying works in Antrim County (and many places). Per interviews with county officials, voting machines scan and retain voter ballots. When a machine is ready to tally its ballots, it prints a final tally of votes for each candidate in each race. Officials then use a piece of election management software to add up these totals and produce provisional counts.

In this case, Antrim County had made a human mistake: they hadn’t updated the software that calculates the unofficial results to match the software on the individual machines, causing a miscalculation when it added up the totals.

It’s important to note here that no individual votes were affected whatsoever. There were not “6,000 votes changed from Trump to Biden”. The only issue was that a wrong unofficial total was calculated, and it was quickly caught, reported, and fixed. In addition, they hand-reviewed the printed tallies from each precinct, just to make sure. Some people have claimed that individual ballots were being hand-reviewed, which Wisconsin Secretary of State Jocelyn Benson says is “completely false”.

It’s also clear that this was an isolated, human error by one county clerk. The claim that the election itself is in question because this software is widely-used has no logical basis. Furthermore, there’s nothing to suggest that such a bug would act in a partisan manner (were it to actually exist) so claims of conspiracy are totally unfounded.

This is another case of the checks and balances in the system working correctly. A human error occurred and it was rapidly caught, fixed, and transparently reported. Anyone who is truthfully concerned about election integrity should applaud the actions taken by election officials, rather than demanding their heads.

Claim 5: More voters voted than were registered

I’m pretty sure we’re about 64 re-compressions deep on this image, given the blur factor

This image purports to show that voter turnout in states like Pennsylvania is impossibly high. Of course we have no source for the image, just that its data was supposedly pulled from realclearpolitics.com and worldpopulationreview.com. The first one is a well-known right-of-center political news site. The second, which I’d never visited before today, appears to be a straightforward world geographic and demographic info site run by a guy named Shane.

I didn’t get a chance to ask Shane about these numbers, but something seemed immediately off to me when I first saw them. And sure enough, I was right: as of October 21st, there were already 9 million people registered to vote in Pennsylvania. So on its face, the chart contains factually incorrect information: 9,000,000 ≠ 6,469,000 and 9,000,000 > 7,041,572 (I knew three semesters of collegiate discrete math would come in handy one day). In addition, Michigan, Minnesota, Nevada, and Wisconsin all allow same-day voter registration, so it’s perfectly possible that they would have vote counts higher than the number of voters registered prior to Election Day.

I have refrained thus far from showing any social media comments other than the sources of these claims, but I just have to demonstrate the sort of ingrained misinformation we’re seeing now. When confronted with the fact that 9 million is greater than 6.5 million, and thus the chart they posted was wrong, this was one Facebook denizen’s reaction:


I don’t really have to say anything else about that. It’s too depressing.

Claim 6: Sharpie-Gate

I mean, is it really surprising someone voting for marijuana legalization might mark a few extra dots? ;)

Beyond making me extra-angry by insulting belabored poll-workers who give their time and talents to ensure that our democracy functions, this claim is pretty easy to prove false. Arizona officials have themselves debunked it on several fronts.

First, Maricopa County elections officials noted that their tabulation machines will read blue or black ink, as well as Sharpie marks. Second, it turns out the machines only read the ovals, and the ballot uses off-center columns, both measures ensuring that bleed-through is not a problem for counting votes.

Finally, the state’s election procedures manual notes that there are processes in place to count every vote, even if the machine has an issue reading it. These include ballot duplication, in which an election worker fills out a new ballot using the voter’s choices to ensure it will be read properly by a machine. Per the manual and Pima County election officials, this process is overseen by a bipartisan Ballot Duplication Board. It’s meant for ballots that using the wrong ink or are physically crumpled or damaged, and it would apply if bleed-through were an issue for some reason as well.

Claim 7: Pennsylvania Postmarks

…ballots received within this period that lack a postmark or other proof of mailing, or for which the postmark or other poof of mailing is illegible, will be presumed to have been mailed by Election Day unless a preponderance of the evidence demonstrates that it was mailed after Election Day.

“Preponderance of the evidence” is a legal standard that essentially means “more likely than not”. Basically if there is more evidence for one conclusion than the other, that’s the one you go with (this is the typical standard in civil cases, versus the criminal conviction standard of “beyond a reasonable doubt”).

In the interest of fairness, it is true that this ruling was made. It is in fact the case that as long as a Pennsylvania mail-in ballot is received by 3 days after Election Day it will be counted, unless a postmark or other evidence shows it was mailed after Election Day.

One claim related to this ruling is that this Pennsylvania Supreme Court decision is in violation of Article 2 of the U.S. Constitution, specifically Section 1:

Each State shall appoint, in such Manner as the Legislature thereof may direct, a Number of Electors, equal to the whole Number of Senators and Representatives to which the State may be entitled in the Congress

The idea is that in this case, the state courts, not the legislature, has performed election policy-making. This (like most things to do with the Constitution) is a complex area of jurisprudence. The Pennsylvania courts would contend that they were merely interpreting the state constitution, specifically Article 1, §5:

Elections shall be free and equal; and no power, civil or military, shall at any time interfere to prevent the free exercise of the right of suffrage.

They believe that based on previous jurisprudence surrounding the Election Code, it should be “liberally construed so as not to deprive, inter alia, electors of their right to elect a candidate of their choice”. Their specific findings are in the complete ruling.

The U.S. Supreme Court declined to grant Republicans’ request to stay the effects of this ruling, so it stands on Election Day. I’m not qualified to adjudicate this claim, but I’m sure we’ll see more from the courts on this. As of 11:00PM CST on November 9th, 2020, 45,103 late-arriving Biden ballots would have to be thrown out in order to swing the election. That seems like a lot, around 2% of the total mail-in ballot count. I’d be very surprised if 1 out of every 50 ballots both arrived late and were cast for Biden.

The other related claim following from that is the idea that allowing ballots received by three days after Election Day the presumption of having been mailed on time inherently leads to fraud. This, in my opinion, is not well-supported. Some people have used it as evidence that Claim 1 was indicative a bunch of new post-Election-Day fraudulent votes, allowed in even though they were not postmarked, but I’ve already debunked that claim.

Really what Claim 7 shows in isolation is the fundamental difference in philosophies about voting: presumption of eligibility vs presumption of ineligibility. The parties tend to fall on opposite sides. Any claim that this policy has lead to fraud has yet to be documented, so there’s really nothing to “debunk” here per se.

Edited to add: We now have an actual accusation of ballot fraud in Pennsylvania, although it doesn’t rely on the presumption of innocence, but rather actual accusation of fraudulent postmarking. See my analysis here.


I hope this has been a helpful, somewhat-quick overview of the most common claims of fraud in the 2020 U.S. election. I strongly believe that we in the United States have free and fair elections; if anything, it’s not the mechanics of the election that are the problem, but exactly these claims that are undermining confidence in our democratic systems.

There are a few additional claims I’d like to address, specifically regarding partisan election observers and outside presence at counting locations. I’ll likely add those in the future. Feel free to reach out to me on Twitter or leave a response on Medium.

Edited to add: I’ve posted a follow up piece addressing the allegations of postmark back-dating in Pennsylvania:

Site Reliability Engineer, former infantryman. Always open to discussion opportunities and ideas.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store