Welcome, friend! I’m so glad you’re here! Here is the fifth in a series on resisting autocrats from the inside out—by staying seated in our own center, in the knowing of our own heart. Here are the previous four:
39. The Knowing Inside
56. Thirsty for the Waters of Life
57. Holding Firm to an Open Heart
58. How Do We Know What Is True?
Today, on being committed to truthfulness—living by the heart, the place of truth. This means aligning with reality as closely as we can—finding and considering solid evidence when it is available. Then following the evidence where it leads, no matter our preconceptions.
Posts and podcasts at Nature :: Spirit are all free, always free. But if you are able, your paid subscription helps me put in the time for this research. It’s a tremendous boost! However you participate, though, I’m grateful for your presence. Thank you for being here! May you receive something here to help and support you.
So I’m wading into deep waters today—into something that until a few weeks ago I didn’t know much about: election integrity.
These are not waters I ever wanted to enter. They’ve been thoroughly poisoned by Trump. Insurrection and death at the Capitol, election systems breached, trust in the voting process wrecked, hatred on full display—the poison has infected millions across the country. And countering all those lies has meant putting up such a righteous defense of our voting system that now anyone who voices any doubts at all about elections sounds like the party of crazy.
Poison Can Work Twice
Poison can work twice. The original lies deceive, then the defense—if it’s rigid or partisan—can obscure the truth all over again. The lies have done double duty on us.
The way out of lies is following the evidence. Using critical thinking. Being committed to the truth, no matter our preconceptions. Being clear-eyed and openhearted, ready to follow the evidence wherever it leads.
So today I’m wading into these troubled waters. I’ve spent weeks digging deep into evidence and claims about the 2024 election. Learning about a whole new field of study. Tracking down references. Downloading research papers. Assessing methods.
Because I’ve always had doubts about that vote.
We Saw What We Saw
I mean, we saw what we saw, right? On one side, tens of thousands of people pumped up and joyful, crowd sizes growing every time. And on the other side, limp rallies, same old hate, and people walking out before it was even over, bored with it all. Such a striking contrast! I thought it would hold true on election night too.
But the numbers told a different story. And numbers don’t lie, right? So I’ve been working very hard to come to terms with a bitter reality: that we the American people are not who I thought we were.
But what if there is another possibility? What if numbers do lie? What if bad actors can make them lie? And what if there are ways to find out—prying those lies out of the numbers and bringing them to light?
Election Forensics
As it turns out, thoughtful experts in statistics have been working on these exact questions for decades. So I’ve been diving deep into the websites of the Election Truth Alliance and SMART Elections, both of them nonpartisan, nonprofit groups working for election accuracy. And through their work I’m learning about a new field called “election forensics.”
Election forensics means analyzing elections data using statistical methods. One of the pioneers in the field, Dr. Walter Mebane, is a professor of political science and statistics at the University of Michigan, and he defines it this way: election forensics is “using statistical methods to determine whether the results of an election accurately reflect the intentions of the electors.” Researchers begin with the public voting data posted after an election and apply statistical analysis to them. The process can reveal places where numbers are skewed in ways that wouldn’t likely happen in a fair election. Places of “weirdness.”
International Origins
The credit for some of these methods goes back to statisticians in Russia a couple decades ago who got suspicious of the elections results in their own country. So they started applying their own tools to the numbers. And they began finding weirdness.
One strange thing they found was that certain voting areas were reporting unusually high turnout—80 or 90 percent—and in all those places the winning candidate received an unusually high percentage of the votes.
Now, in a fair election there should be no correlation at all between voter turnout and winning candidate; the winner’s share should look the same no matter how many people are voting at this or that location. The researchers say that a pattern like this is consistent with fraud—in this case ballot stuffing.
And if this is getting complicated, don’t worry—it did for me too! I didn’t begin to understand it until I saw the pretty graphs, which is why I want to show them to you too.
The Russian Tail
Here’s how it works. In a fair election, when you look at how many votes the winner got relative to voter turnout, the scatterplot should look like a nice, neat circle—no long tails running off in any direction. You can see this in the graphic below, in each of the squares in the third column: there’s a circle clustering in the middle of the square near the voter turnout of about 50 percent. And if you look at the bottom of the second column you can see too that Switzerland’s actual results match that pattern. A turnout of about 50 percent, with results clustered in a clean circle, no tails. A fair election.
But with ballot stuffing, the circle gets stretched into a comet with a tail. When extra votes are added at some locations for the winning candidate, it pushes up the voter turnout for those locations. So the circle starts leaning to the right, toward higher turnout. And as turnout increases—to 70, 80, 90 percent—the winning candidate gets a higher and higher percentage of the votes, so the tail also drifts upward toward higher vote share. And pretty soon you’ve got a nice long comet tail.
So if you look at Russia in 2011 and 2012 in the first column, you can see that pretty tail reaching upward to the right. And the tail actually hits the top right corner of the graph, because some areas apparently reported 100 percent turnout, with the winning candidate receiving 100 percent of the vote in those areas!
So the statisticians have given a name to this pretty tail. It shows up so often in Russian elections that they call it the “Russian tail” or “Putin’s tail.”
And—here’s the part that concerns us—when statisticians use those same methods to look at our 2024 presidential election, they find similar weirdness.
The Election Truth Alliance, or ETA, has analyzed the election data for a number of states, and I want to talk about two of their reports here: Pennsylvania and Minnesota. You can find all their reports at their website.
Pennsylvania
The ETA report on Pennsylvania, a swing state, raises several red flags.
The ETA took a close look at three separate Pennsylvania counties, and they found that Trump’s votes in those counties rose as voter turnout went up.
Uh-oh, this shouldn’t happen in a fair election.
Here is the ETA’s scatterplot for one of those counties, Allegheny, the home of Pittsburgh. In this scatterplot a fair election shows up as horizontal trendlines—vote shares remaining consistent no matter how many votes are cast in a particular place. And in fact when ETA analyzed the mail-in votes for Allegheny County they found these horizontal trendlines. All good!
When they looked at ballots cast on election day, they saw something sharply different. As vote count went up, Trump’s share of the vote increased dramatically while Harris’s decreased. The trendlines are so far from horizontal that they cross. The ETA says this pattern is consistent with “a trend that has been observed in other elections internationally where vote manipulation is suspected.”
And what I hear them say? We’re looking at an American version of the Russian tail.
But why would this weirdness show up in election-day votes but not mail-in votes? This too is concerning and can be a sign of manipulation. It can happen when bad actors rig only one part of an election and don’t bother interfering in all segments of the election equally.
There’s another concerning fact about Pennsylvania: This weirdness shows up only in places with higher numbers of votes. It’s not real clear in that graph, but the ETA says that they did not find the weirdness until after, say, 200 or 250 votes had been counted. Beyond that threshold, Trump’s percentage of the vote rose higher and higher. We’ll come back to this anomaly in a moment.
Then the ETA performed a second kind of analysis for all of Pennsylvania: dropoff votes. Dropoff is when the candidate at the top of the ticket gets a different number of votes from the candidate of the same party in a different race. Some dropoff is always expected in a presidential race—people might vote for only the president and no one else, or people might split their ticket between a president of one party and other candidates of a different party. But without tampering, dropoff rates should be pretty consistent over all kinds of votes.
In Pennsylvania’s mail-in ballots, dropoff rates between Harris and Trump were quite similar: 1.48 percent and 1.96 percent. These are well within expected ranges. You can see on the graph that there is a nice randomness to the various counties—no distinct differences overall. This is how fair elections look.
But election-day votes show a sharply different picture. Trump’s dropoff rate shrinks to almost nothing, .87 percent, while Harris’s grows to an an unusually high 4.51 percent. And ETA says that in one county her dropoff rate reached almost 11 percent. It would mean that a whopping 11 percent of voters in that county voted for Democrats for Congress but did not vote for Harris for president. That’s very hard to believe—just plain weird.
So notice in this graph how the randomness between red and blue bars almost disappears as the colors separate out into distinct patterns. This is a red flag for vote manipulation.
The folks at ETA were so concerned by what they found that they submitted their analysis to Dr. Walter Mebane, the election forensics expert. He has developed statistical models for actually estimating how many ballots in a given election might have been distorted.
Dr. Mebane analyzed Pennsylvania’s voting data using two of his models. One of his models estimated 25,000 distorted votes, and the other estimated 225,000.
Would that have been enough to change the outcome of the election? In Pennsylvania the margin of victory was a little over 120,000 votes, so yes, his higher estimate of 225,000 would have clearly changed the outcome.
Proving Vote Manipulation
These are serious concerns about the accuracy of the vote. And they are arrived at by careful, methodical scrutiny. By following the evidence. Analyzing the numbers.
But the numbers alone can’t give us the full truth. So it is important to say here: Weird statistical patterns are not proof of vote manipulation. Weirdness only raises red flags. It shows us where more scrutiny is needed.
Let me repeat that: Anomalies in the voting statistics do not prove that fraud or manipulation took place. But they do show where we need to look more closely.
The only way to prove vote manipulation is by performing a hand audit of the paper voting records, to see what the actual physical records say. Audits like these will only be done if voters demand them—so there is plenty you can do right now, and more on that in a moment.
But now let’s turn to Minnesota, where some hand counting of ballots did take place.
Minnesota
Minnesota provides a great comparison with Pennsylvania. For one thing, it’s not a swing state, so if there are bad actors operating in swing states, Minnesota might fall under their radar. But if weirdness shows up outside of swing states, it could mean that there are election security problems nationwide, not just in the purple states. Second, Minnesota is great for comparison because some areas in Minnesota actually count their ballots by hand instead of by machine. Even better, some of them use both counting systems—certain precincts by hand, others by machine within the same area—which provides an instant check on the accuracy of the machines.
You can find the ETA’s Minnesota report on their website as well as in a video that the ETA put together to explain it.
The Election Truth Alliance looked at District 8, a rural area in Minnesota that is primarily Republican. District 8 uses both hand counts and machine counts, so it’s perfect for checking up on the machines.
The results of the hand-counted ballots show a pattern that’s expected in a fair election.
You can see this in the graph on the right, where the trendlines are horizontal—roughly the same 60 percent for Trump and 40 percent for Harris through all size precincts. This shows a fair and honest win for Trump in those precincts.
But the precincts that counted their ballots by machine—the graph on the left—show something very different. Trump’s share of the vote increases almost in lockstep with the rate of turnout, and Harris’s share decreases at the same rate. So in precincts with a turnout of 70 or 80 percent, Trump receives 70 or 80 percent of the vote while Harris’s vote share shrinks to 30 or 20 percent.
These results are consistent with fraud, like the pretty Russian tail.
Senate and House Races Too
What’s more, when the ETA looked at the Senate and House races in Minnesota, they found the exact same disparity between hand and machine counts. In votes counted by machines, the Democratic candidate lost vote share in areas of higher turnout and the Republican candidate gained vote share at the same rate.
It Could Be the Tabulators
These results are alarming. Again, they don’t prove that vote manipulation took place, but they do wave a serious red flag. More investigation is needed! That’s where you come in, and we’ll return to this in a moment.
The fact that the weirdness shows up in the machine-counted ballots but not in the hand-counted ones points a finger at the machines. Maybe they were compromised.
So the ETA is raising questions about the voting tabulators. And they’re raising concerns about these tabulators because they’re finding the exact same weirdness in all the states they examine—not just the swing states of Pennsylvania, North Carolina, and Nevada, but also low-stakes states like Minnesota. It raises the possibility that votes were manipulated by the tabulating machines all over the country.
It Can’t Happen Here!
But!—I hear you say—it can’t happen here!
. . . because our machines are secure!
And I say, How do we know? Should we take someone’s word for it?
No, we should follow the evidence. And the evidence is raising red flags. Maybe the machines are not as secure as we thought they were. Recall that the Washington Post reported in 2022 that bad actors had gained access to the software of the election systems in Georgia and other states and shared it with Trump allies. As far as I can tell, no one ever made them give it back.
But how, you ask, can the machines be compromised when there are so many of them? Surely the votes are secure . . .
. . . because the voting system is so decentralized you’d have to hack into thousands of different machines!
As the ETA explains, the administration of our elections is very decentralized, but the tabulating machines are not. They’re actually quite centralized. Every vote-counting system in this country comes from one of only three vendors, and you can see the breakdown in this graphic.
One vendor alone, ES&S, supplies almost two-thirds of the tabulators in this country. So programming malware into the machines of that one supplier could potentially affect two-thirds of the voting precincts in the country.
And just this week, as I was writing this, news broke that Dominion, the company that supplies a quarter of the voting systems in the country, has just been sold to a former Republican election official—someone known for his election-denier language.
I think that the era of believing that our voting machines are secure is pretty much over.
Or, you say, it can’t happen here . . .
. . . because post-election audits would catch it!
Most states do a careful statistical sampling of their votes after an election to ensure that the machines did their job and the votes are accurate. It’s called a risk-limiting audit, and it’s a best practice for election accuracy.
But, as the Election Truth Alliance explains, software can be designed to evade audits. For example, machines can be programmed for different windows of time: to turn on a program for a period of time and turn it off for another. Do you remember the Volkswagen emissions scandal from some years back? The company intentionally programmed certain diesel engines to show lower-than-actual emissions—but only during emissions testing. When the engines were actually being driven on the road, they produced up to 40 times the emissions that had been recorded during testing.
Or a program can be written to kick in only under certain conditions, such as after a certain number of votes have been counted.
Remember that Pennsylvania data? The weirdness appeared only above a threshold of 200 or more votes.
So if an audit samples only a small number of votes—2 percent is common—the manipulation might not show up at all. And it would affect primarily larger precincts with higher turnout—the exact places where interference could change the outcome of an election.
The point is, hackers are sophisticated. They can easily program software to evade audits.
Election Forensics Is Not a Cure-All
Election forensics won’t solve all our problems. We are still a severely divided country. Even if the people did not elect this president—and we don’t yet have proof—tens of millions of people in this country still voted for his brand of hate and corruption. If the 2024 election was able to be manipulated, it was because the contest was already close. When the margin of victory is thin, bad actors can hide their work a lot better.
Knowing if votes were manipulated won’t restore civic trust. It won’t help people think critically about what they read in social media. It won’t reduce the massive inequality in this country so that billionaires exist who can buy elections from the presidency on down. It won’t overturn Citizens United. And it certainly won’t do anything to heal the hearts of people who are hating and fearing neighbors who are different from them—Brown and Black people, immigrants, trans children. So much healing and restoring needs to happen!
And even if we were to find out that no, the American people did not choose this president, what would happen then? Discovering it would not automatically restore the rightful government. We have no processes in place for righting such a terrible wrong. We have never faced such treason, such catastrophe.
But what election forensics does do is give us an up-to-date tool for finding cyber crime. It gives us another way to secure our elections—and this is where you come in. Because if election forensics is pointing to fraud in the 2024 election, then our midterms coming up next year are in alarming danger. And we have to act fast!
What You Can Do Right Now
What you can do right now is call for “hand audits of paper voting records” in the 2024 election. (This is the ETA’s language, and it’s important to use it.) The people who have the power to order these audits are local and state elections officials, and they will need to be convinced to do it. Public pressure can help. If you want to take action, you can use the ETA’s Audit Advocacy Toolkit, where you’ll find templates for emailing government officials.
You can also sign up with the group SMART Elections for online training in election verification. That group is running a series of online workshops to train local people in nonpartisan, data-based election verification techniques so you can help document and verify the accuracy of your own local elections.
And you can talk with your friends and family about the very serious problems that very serious people are finding in our elections data. Spread the word now so that people get used to the idea that election data can hide weirdness and election forensics can uncover it. Tell your friends and family about the weirdness that shows up in the 2024 election.
Stay clear-eyed and openhearted, willing to follow the evidence wherever it leads. Encourage your circles of people to do the same.
There are many other ways to help our elections—like working to overturn Citizens United or abolish the electoral college—but honestly, all of it is useless if we don’t have free and fair elections. To keep a democracy, we have to ensure that every vote is counted accurately. If we can’t do that, we may not have another fair election, ever. So the best thing to do now is to work for hand audits of the previous election or become trained in accurate verification of the next one. Time is short; the midterms are upon us.
Did the People Choose This President?
So, did the American people choose this president? Elections forensics is raising serious red flags. If we did not choose this nightmare of hatred and cruelty and grift, it means that this country in fact has the people power and the collective will to turn things around. We are larger and stronger than we think.
And if, by uncovering vote manipulation, we could prevent such corruption from ever again taking power or holding onto it, that would make all the difference—for democracy now and for the country our children will inherit.
It’s on us to find the truth.
I’m closing comments for this post—sorry!—because on this topic it’s actually more helpful for you to talk with the in-person people in your life. Please spread this information! Help it travel as far as possible, as fast as possible. Share this post, or go to the ETA website and share their graphics. Write your own post. Get election forensics in front of major media figures. Help this info go viral! Time until the midterms is so short!
For digging deeper
Dr. Walter Mebane’s definition of election forensics comes from his talk for the National Press Foundation, “Detecting Voter Fraud,” August 1, 2024. His report on Pennsylvania is “eforensics Analysis of the 2024 President Election in Pennsylvania,” with summary and PDF available at “Pennsylvania Working Paper—Dr. Walter Mebane,” Election Truth Alliance.
Two Russian statisticians, Peter Klimek and Sergei Shpilkin, get a lot of credit for pioneering the field of election forensics. Here are some of their papers:
Peter Klimek et al., “Statistical Detection of Systematic Election Irregularities,” Proceedings of the National Academy of Sciences 109, no. 4 (Oct. 9, 2012):16469–73. Available for download at this link.
Sergei Shpilkin, “Statistical Analysis of Elections,” Election Statistics Roundtable, 2017. Available for download at this link.
Dmitry Kobak, Sergey Shpilkin, and Maxim S. Pshenichkov, “Putin’s Peaks: Russian Election Data Revisited,” Significance 15, no. 3 (June 2018). Available for download at the Oxford Academic site at the same link.
For a French news story about applying Shpilkin’s method to a recent Russian election, see Sébastian Seibt, “‘Shpilkin Method’: Statistical Tool Gauges Voter Fraud in Putin Landslide,” France 24, March 20, 2024.
For more on machine security: Emma Brown and Jon Swaine, “Inside the Secretive Effort by Trump Allies to Access Voting Machines,” Washington Post, Oct. 28, 2022. Because of concerns related to machine security, seven experts in computer science and election security wrote a Letter to Vice President Harris right after the election asking her to request hand recounts in the swing states. They detailed the evidence from the trial of Sydney Powell for election interference in Coffee County, citing depositions from this trial plus additional WaPo sources to show that operatives had gained illegal access to voting machine data in multiple states and that they had copied it, shared it with Trump allies, and retained it. The affected software came from both ES&S and Dominion, whose equipment, as the letter states, “counts nearly 70% of all votes nationwide.”
Learn more about voting and vote-counting methods in other countries. Wikipedia’s “Vote counting” entry is a place to start, and the section titled “When manual counts happen” references media stories that explain vote counting in quite a number of countries.
The websites of the Election Truth Alliance and SMART Elections both hold a wealth of sources and resources for election security, election tech, and election best practices. Support these groups if you can. Check out the ETA’s Audit Advocacy Toolkit. Poke around on their Promotional Material Resource Center to see a wonderful array of infographics—visual explainers that you can post anywhere.
