Electronic Election Theft in America

Address to the Brownstone Supper Club, Philadelphia, 8/8/24

Last Saturday, I was on a train to Eugene Oregon to spend the weekend with my friend who goes by Martin Truther for his internet presence. I first knew him from an election integrity conferences in 2006. As I introduced Martin to a woman I met on the train, she asked us if we were “election deniers”.

Where do I begin?

Election theft has been an unmentionable topic since at least the 1980s, when James and Kenneth Collier wrote about their experience in a book called VoteScam. They witnessed ballot box fraud first-hand, but as they brought their story to one newspaper after another, they repeatedly found that the editors refused to get involved.

The story I bring to you this evening starts with the introduction of electronic voting in 2002. After that, Republicans started manipulating elections from a distance, using software, and the Democrats didn’t do anything to stop or even to expose them. It’s an ultimate irony that Trump is the one who tried to bring charges of election theft to the American public, and the election integrity field carries the stigma of Trump’s reputation. Just as anyone who asks for evidence of vaccine safety is “antivax” and anyone who questions the CO₂ narrative is a “climate denier”, and anyone who questions the narrow interpretation of Darwin is a “science denier” — we have now become “election deniers”.

---

My introduction to this field came in 2004. You may remember the “hanging chad” debacle in Florida in 2000, and as a response, the Help America Vote Act was passed in 2001. This provided subsidies for local communities to go digital with electronic voting machines. 

In the spring of 2004, Paul Krugman was writing a series of op-eds in the NYTimes describing the vulnerabilities of the voting machines, noting that the companies that made these machines had corrupt pasts, and links to the Republican party. In August, a feature article by Ronnie Dugger came out in The Nation detailing what looked like electronic theft in Georgia, 2002. Georgia had just installed new Diebold voting machines, and Republicans won upsets in a Governor’s race and a Senate race against very popular Democratic incumbents, both of whom were heavily favored in polls. Dugger told the story which would later be corroborated when Bev Harris of Black Box Voting revealed a leaked computer folder that was not supposed to be publicly available on the Diebold website. The folder was named “Rob Georgia”. 

So I was primed, watching the results on November 2, 2004, Bush vs Kerry, and the results were so close, that by midnight it was clear that Ohio was the last swing state. Whichever way Ohio went, so went the Electoral College. All eyes were on Ohio, when the Secretary of State’s computer went down, with Kerry leading Bush by 100,000 votes. For two hours, Ohio results went black. When the computer came back on line, Bush was ahead of Kerry by 100,000 votes, and he kept his lead through the night. Bush “won” the election.

Stories from Ohio the next day were quite damning. In college towns and big city Democratic strongholds, there was a shortage of voting machines — even as thousands of voting machines remained warehoused by the Secretary of State on the theory that they were not needed. Republicans in the suburbs breezed in and out of the polling place, but Democrats in the cities were waiting on line outside, sometimes for many hours in the freezing rain.

There were more details of this story, unknown to any of us at the time, but they gradually came out over the ensuing years. Ken Blackwell, the Secretary of State in charge of running the election, was also chair of the Bush Campaign in Ohio. Blackwell had contracted with a private firm called Gov-Tech Solutions to mirror the Ohio SoS website so there would be a backup, “just in case something went wrong”. Gov-Tech Solutions was headed by a man named Mike Connell, who also happened to be the “IT Guru” of the Bush White House. He had previously taken the blame when hundreds of thousands of White House emails, subpoenaed by Congress, were “accidentally” erased, so the subpoena couldn’t be honored. Connell was a born-again Christian and a private pilot. A lawsuit over the election and the role of Gov-Tech Solutions gradually worked its way through the courts at a lawyerly pace, and it was revealed that Connell’s backup computer included a back door that allowed White House staff to alter election results in the computer’s memory. Well, it was 2008 before Connell was scheduled for a deposition before the Court. Connell was unlikely to lie under oath because of his religion. He reportedly received warnings from Karl Rove that his life was in danger if he spilled the beans. In the first run of the deposition, the lawyers for the State stood up and objected to every question, and Connell didn’t have to say anything. After the judge ruled that of course, Connell had to answer these questions, the deposition was rescheduled for January. But in December, Connell’s single-engine plane fell out of the sky and he was killed. Federal investigators were immediately on the scene and told the local authorities to go home. The accident was ruled to be caused by weather, even though skies were clear at the time.

Back to 2004, Jon Simon was way ahead of me. He was not only watching the results but screen-saving the exit poll results the moment polls closed in each state. Here’s what Jon had figured out: There are “exit polls” and there are “adjusted exit polls”. The exit poll results reported on line and on the radio were surveys taken at thousands of statistically-representative polling stations throughout election day. As long as voting continued, the stations were barred from projecting winners. But as soon as voting ended, they posted poll results. Then as the “real” results came in, the exit poll projections were gradually merged with the official count, until by morning the original exit polls were completely invisible, and the official counts replaced the results that were still labeled “exit polls”. 

So, by capturing the exit polls from each state in those first minutes when they were still unadulterated, Jon was able to compare exit polls with official outcomes. What he found was that the differences all favored Republicans. In states where the election was close, the differences were larger compared to uncontested states. Jon invented the clever term “Red Shift” — borrowed from astronomy — to describe the fact that official results seemed to be unaccountably skewed toward Republicans. The Red Shift for Ohio was 11%. The distribution of red shifts was a bell-shaped curve as we expected but instead of being centered on zero, it was centered on 2.7%. It looked like the entire distribution had been shifted 2.7% toward Republicans. The odds of this being a chance occurrence were astronomically small

Steve Freeman, a local professor at Wharton school, wrote up Jon’s results in a document that was circulated on the internet just a few days after the election. I was in touch with Jon and Steve and a handful of other activists around the country. I was teaching a Temple University statistics course at the time, and volunteered to write up the statistical analysis.

Here’s a key moment in my personal radicalization. We wrote up our preliminary results, signed by myself, other college professors, lawyers and engineers. We sent a note to the NYTimes news desk. We didn’t receive a response. Well, that’s the NYTimes. Of course they’re very busy and it’s hard to reach them. But then a couple of days later was a front page article under the headline, “Vote fraud theories spread by blogs are quickly buried.” They ridiculed us, declared us debunked, without ever talking to any of our credible analysts. The only person quoted was Kathy Dopp, a Junior High math teacher from Utah, a woman with a good heart and good instincts, who was simply in over her head. It was easy to make fun of her. 

This was a first for me…the first time in my life that I had personal evidence that the Times had falsified a news report. And falsified on behalf of Republicans. WtF? Why would the Times be going against common sense, against statistics, against some if its own columnists to declare that the election was honest? In a piece that could only favor Republicans and one of the most reviled presidents in memory?

Footnote: Kathy Dopp is living in a Boston suburb. In 2022, she was fighting the good fight for COVID truth, when she was hospitalized and medicated against her will. I don’t know the full story.   

Our experience with the NYTimes proved to be a harbinger of things to come. We took our story of Republican malfeasance to Democratic leadership in the House. We took it to various Democratic think tanks and policy institutes. We went to the left-leaning press. No one followed through. 

And Paul Krugman? He had predicted electronic vote theft before the 2004 election. His worst fears came to pass. What did he write about that? Krugman was “on leave” from the Times for three months. When he returned, he never wrote another word about vote theft. I have a friend who went to MIT with Krugman many years ago, and asked him to pass on our query — what happened? My friend reported that Paul looked uncomfortable and said, “I can’t talk about that now.”

I want to transition now from personal story-telling to the big picture

There are many ways to manipulate election results even before the first vote is cast —

• Gerrymandering makes Democratic votes count less than Republican votes. “Natural” gerrymandering statistically disenfranchises city-dwellers.

• And of course the Senate and the Electoral College are a kind of elitist bias built into the Constitution. 

• The two-party system makes sure that we must choose between two candidates we don’t like.

• Campaign finance by big corporations is well-known. We all know the story of Citizens United and corporate personhood.

• Mainstream media manipulate people by spinning the news and not reporting embarrassments. They dig up Bobby’s “brain worm” testimony from a divorce hearing 12 years ago, but bury stories like Hunter Biden’s laptop or Ashley Biden’s diary. Most egregiously, they tell us that anyone who questions the official vote count is guilty of “insurrection”.

• Voter registration processes make it difficult for poor people and people of color. This story has been covered doggedly by Greg Palast.

• Working people in the cities often wait in long lines to vote, sometimes miss an hour of pay, while voting is a breeze in the suburbs.

All these biases pull in the same direction, toward corporate-friendly candidates, against working people, against populist leaders, against peace and freedom and environmental stewardship. But these biases are not the subject I came to talk about. This evening I’m only reporting on direct manipulation of the vote count, after all this other dirty work is completed. And this is a further pull in the same direction, concentrating power in fewer hands, disenfranchising the majority.

True majority rule is an idea that the wealthy and the powerful have sought to undermine at every opportunity, going back to ancient Athens, continuing through the compromises built into the US Constitution by the Founding Fathers.

Election manipulation, too, is as old as elections, and the US is no exception. The history of America’s democracy is rife with stories of corruption from the very beginning. In 1792, the first scandal rocked a Congressional vote in Georgia, and as a result, no one was sent to Congress from that district. Andrew Jackson won the popular vote in 1824, but he was cheated out of the electoral college vote by a back-room deal between John Quincy Adams and Henry Clay. Democrat Samuel Tilden won the popular vote in 1876, and disputed vote counts led to a commission being formed to resolve the issue. They voted along party lines to resolve all disputed counts, leading to a one-vote electoral college victory for Republican Rutherford B Hayes.

There has been a lot of labor-intensive stuffing of ballot boxes. Mayor Daley of Chicago was famous for getting out the vote of dead people. Lyndon Johnson’s election to Congress in 1948 deserves special mention. In the Democratic primary, he was down 20,000 votes until a group of supporters “found” a box full of ballots, all alphabetized and in the same handwriting using the same ink, and every one of them voted for LBJ.

Both parties participated in election theft, and local political machines did the dirty work. 

All this was before 2002, when everything changed.

Retail-level vote theft by both parties was swamped by wholesale vote theft, almost all by Republicans. 

In 2001, Congress passed the Help America Vote Act, which promoted computerized voting. Computers may make it easier to vote, especially for people with some kinds of disabilities. But what we gain in convenience, we lose in transparency. In touchscreen machines, there is no record, no evidence. It is easiest and to hack them, and risk of being caught is very low. But scanners like the ones that count our votes in Pennsylvania are also easy to hack. In theory, the hack can be detected if the votes are re-counted by hand. But this has almost never happened. In the few cases where recounts were obtained, the same stack of papers were run over again through the same corrupted voting machines. 

The software that counts votes — incidentally — has been ruled by our Supreme Court to be a trade secret of the companies that manufactures the machines. The counties that buy election machines have no right to inspect the software that runs them.

Computerized voting led to the blatant, in-your-face theft of a governor’s race and a senate race in Georgia in 2002, as documented by Bev Harris of Black Box Voting.

Computerized voting is an election manipulator’s wet dream. You don’t have to have a network of local operatives stuffing ballot boxes. You don’t have an extensive infrastructure that must be kept secret. You can centralize the process, and steal votes wholesale by modifying the software that counts votes. Better than “behind closed doors in a smoke-filled room” — vote counting takes place in the pitch black of cyber space. And with our Court Decision about “trade secrets”, you can be sure there ain’t nobody looking over your shoulder. 

We have a variety of evidence that vote counts between 2002 and 2018 were consistently biased toward Republicans. That includes 2016. It appears that Trump was not legitimately elected, and that Hillary should have won the electoral college, in addition to the popular vote. Hillary never contested the 2016 results. My guess is that it’s because she had stolen the primaries from Bernie Sanders, and could not afford to have the press peeking inside that Pandora’s box. But Jill Stein, bless her heart, was a Green Party candidate in that election and so she had standing to challenge the vote. In a single Thanksgiving weekend, she raised $7 million in a crowd-funding campaign to challenge the vote count in Wisconsin, Michigan, and Pennsylvania. There was a lot of resistance to hand counts, even after money was raised to pay for the labor-intensive effort. In key Wisconsin counties, they just re-counted the votes using the same suspect op-scan machines. In Pennsylvania, there were legal challenges preventing a full recount. In Michigan, a Republican judge halted the recount. 

Anecdotal evidence

There are lots of juicy stories. 

I’ve already told you the story of the Columbus, OH computer of the Secretary of State, rerouted to a basement in Chattanooga that was controlled by George Bush’s IT guru, Mike Connell, whose plane was shot down over Ohio before he could give testimony.

The only other murder I’m aware of was Athan Gibbs, president of the TruVote Collaborative in Tennessee. Gibbs created an open-source voting system, counting votes with the same efficiency as a corporate voting machine, but with complete transparency. He was run off the road by a truck, and died in 2004. 

Many commissions and academics and computer scientists have studied the vulnerability of our election machinery and concluded that stealing votes is laughably easy. There was an academic study out of Princeton by computer scientist Ariel Feldman. Other studies were done by UC San Diego, and by the Brennan Center for Justice in New York. There have been articles in Wired Magazine, the NY Times Magazine, and Politico. Alex Halderman and Andrew Appel have held press conferences where they demonstrated how to hack a voting machine. 

Clint Curtis was a computer programmer in Florida. In 2003, he was hired by Rep. Tom Feeney to program a voting machine with a back door so that anyone who knew the trick and had the right password could change vote counts undetected. Curtis did the work he was hired to do, thinking that this was a demonstration to help secure voting systems. But when he learned that the system he designed was actually used to shift votes for Feeney, he became a whistle-blower. He wrote about his story. He testified before Congress. He even ran against Feeney in a Congressional election that led to a disputed vote count. The case was investigated by Raymond Lemme of Florida’s Inspector General’s office until Lemme was found dead in a motel room, ruled a “suicide”. Maybe that’s a third murder.

Congressman Vic Rawl, a popular and formidable candidate for Senate was challenging an incumbent Tea Party Republican, Sen Jim DeMint in South Carolina. But Rawl was defeated in the 2010 Democratic primary by a complete unknown named Alvin Greene. Very convenient for DeMint, who was dragged down by a sex scandal and might well have lost to Rawl. Greene “won” the primary in a landslide, despite the fact that he did not campaign, had no funding, no ads or signs, no web site, and no political background.

This one was covered in the New York Times.

“...everyone wants to know how Mr. Greene, an unemployed Army veteran who had been completely unknown until Tuesday, inexplicably defeated a heavily favored former legislator and judge to become the state's Democratic nominee for the Senate — and the state's latest political circus act. Even in Manning, a town of 4,000 where everybody knows everybody, nobody seems to know Alvin Greene.” 

Greene continued his non-campaign, and DeMint coasted to victory,  essentially unopposed in November.

Victoria Collier is the daughter of one of the brothers who authored the book VoteScam in 1993. In 2012, she wrote an article for Harpers Magazine from which I take the following quote:

Election Day is now dominated by a handful of secretive corporations with interlocking ownership, strong partisan ties to the far right, and executives who revolve among them like beans in a shell game

...As it happens, many of the key staffers behind our major voting-machine companies have been accused or convicted of a dizzying array of white-collar crimes, including conspiracy, bribery, bid rigging, computer fraud, tax fraud, stock fraud, mail fraud, extortion, and drug trafficking.

Statistical evidence

This is where I was most involved in research and analysis. Exit polls were our best tool during this time, and we would drill down on details of the exit polls, where they agreed with the official count and where they were different, to try to figure out what happened. Remember, we defined the redshift as a difference between the official count and the exit polls favoring Republicans. The RedShift has continued in every Federal election through 2020, with some qualifications that I’ll talk about.

The national exit poll was conducted by Edison-Mitofsky, which had a contract with the news networks over many election cycles. The pollsters saw their job as predicting the official outcome before the official count came in, so they gave the talking heads something to talk about during the long hours on election night. They did not want to see their work used to second-guess the official count. They saw this as a threat to their business model. 

As I said, 2004 Bush vs Kerry was the election where I got my start, and we created enough of a stink that Edison-Mitofsky prepared a report, defending their methodology, and offering us key details about the results that would not have been available otherwise. We went through that report with a fine-tooth comb, looking for any evidence of bias. Bias would mean that more Kerry voters than Bush voters responded to the poll, or that they included too many Democratic precincts and not enough Republican precincts in the places they chose to sample. Neither of these turned out to be the case.

In January, 2005, just a few miles from here on the Penn campus, there was a public debate between Warren Mitofsky, head of the National Exit Poll consortium, and my friend Steve Freeman, who was a professor at Wharton School at the time. They were debating whether Mitofsky’s poll was flawed.

The funny thing was that Freeman argued that the poll was reliable and accurate. Mitofsky was arguing that his own poll was fatally flawed. We were arguing that the exit polls were accurate and they constituted evidence that the vote had been stolen. Mitofsky was arguing that the official vote count was just fine, and it was his own poll that was mistaken.

Over the years, the situation became more and more absurd, more Orwellian or perhaps I should say Kafka-esque. As the redshift became more pronounced and more consistent with each new election, the pollsters started to adjust their methodology. Their goal was not to “get the right answer”. Their goal was to accurately predict who would be officially announced as winner a few hours later.

In the academic literature of polling, the question was raised: Why do we keep underestimating the Republicans? How can we reform our methodology to give better predictions? 

So they started to skew their polls, polling more Republicans and fewer Democrats. As the thumb on the scale of America’s elections got heavier, the goalpost moved, and their methodology moved to accommodate it. Not just exit polls, but pre-election polls were affected. Jon Simon documents all this in gory detail in his book, called Code Red. 

To summarize a decade of exit poll studies: All national elections were affected, not just presidential elections. One of the telltale signs was that polls tended to be accurate in most Congressional elections, which  were lopsided 60-40 wins for the incumbent, Democrat or Republican. In general, the exit polls were accurate in these races. But in tight contests where a few points could make the difference, we found consistent red shifts. The official results were more Republican than the polls predicted. 

As the exit polls became less useful to us, we sought clever ways to investigate what was really going on. We conducted our own exit polls, on a much smaller scale than Edison-Mitofsky. My friend Jon Simon came up with two ideas, and we collaborated on studies in 2006 and 2010.

In 2006, we commissioned a cheap, automated telephone poll, a computerized voice and a very low response rate. Normally, when the response rate is just 2 or 3%, the uncertainty is ridiculously high, and you can’t conclude anything. But Jon’s plan was to choose areas of the country where there was one lopsided race and one tight race in the same Congressional district. Maybe the Congressman was favored to win in a landslide, but the Governor was in a tight race. Sure enough, we found that our poll accurately tracked the landslide race, whereas the same answers from the same set of people indicated a redshift in the competitive race.

In 2010, we studied the special election in Massachusetts to replace Senator Ted Kennedy after he died. The race was crucial for a filibuster-proof Democratic majority in the Senate, which Obama needed to pass a public option as part of Obamacare. In an upset, Kennedy was succeeded by Republican Scott Brown, in one of the most Democratic states in the country. Of course, there were no exit polls. But what Jon did was to compare the percentages in those few counties that used hand-counted paper ballots with the great many counties that used voting machines. Martha Coakley, the Democrat, won easily in the hand-counted regions and Brown won in the machine-counted regions. Of course, we also looked at demographics and voter registrations. These should have led to the opposite result. The hand-counted regions were more Republican, both in their demographics and their party registration percentages. So we took this as solid evidence that the vote had been stolen.

We showed the Democrats this evidence and they brushed us off. We began to suspect that the Democrats wanted to lose this race. If they had a filibuster-proof majority, they would have to pass single payer health care — they would have no excuse. But as it was, they claimed that the Republicans had threatened to filibuster and forced them to accept Obamacare based on private insurers, guaranteeing the private sector even bigger profits than before.

What happened in 2020?

But 2020 was special. Because of COVID, 45% of voters voted by mail. More Democrats than Republicans bought into the pandemic scare, so the in-person ballots were skewed Republican. 60% of mail-ins were for Biden, while 60% of election-day votes were for Trump. Obviously, you can’t do exit polls for mail-in voters. As a result, the 2020 election was the least transparent in US history. 

For 18 years, there was clear evidence that votes were being stolen by Republicans, and the Democrats didn’t do anything to stop it. But with much weaker evidence, Trump threw a hissy-fit about a stolen election.

I had mixed feelings. Overwhelmingly, I was glad that someone was at last demanding that we look at how our votes are being counted. This was an opportunity for reform. On the other hand, Trump’s case was weak, and because of his reputation and his personality, he was easily discredited.

As a statistician, I have to say we have no way to know if Trump’s claim that the election was stolen is valid. But he had some compelling stories behind him. Here in Pennsylvania, the law says that mail-in votes must be received by election day in order to be counted. By order of governor Shapiro, counting of mail-in ballots continued through Friday, and postmarks were not checked to see if they were even mailed by election day. Trump was winning on Tuesday, and Biden won on Friday.

TrueTheVote.org collected evidence on the ground that Democrats had harvested unused ballots, filled them out, and paid “mules” to deposit them in small batches into drop boxes. The extent to which Democrats worked to discredit their work and force a retraction of the film suggests to me that there is both basic truth and a lot of shoddy research in Dinesh D’Souza’s film on the subject.

There were anecdotes about irregularities in Georgia and in Arizona especially. Trump’s lawyers, Rudy Giuliani, Sidney Powell and Jenna Ellis made a lot of noise, claiming that they had compelling evidence in their back pocket and they were about to reveal it. Then they folded their hand, packed up and went silent. Why? Did they really have no evidence at all, or did somebody get to them?

We had no exit poll results in 2020, but we had pre-election polls and other predictors that are less reliable. According to the information that we had, Trump did better than we should have expected, not worse. There is tentative evidence for a redshift in 2020, as in previous years. For example, before the election, the NYTimes called 27 House races as toss-ups they couldn’t predict. Come election day, 27 out of 27 “toss-ups” were won by Republicans. Simon speculates that Trump’s postal director stalled the delivery of mail-in ballots in Democratic strongholds.

My best guess is that there was corruption on both sides of the 2020 election, with Democrats dominating the old fashioned retail fraud and Republicans dominating the electronic vote theft. But evidence is too weak to take a stand — except to say that it is unacceptable that our election machinery is so opaque as to leave us guessing.

In the end, 2020 put the last nail in the coffin of our election integrity movement. Questioning the election was equated with an “insurrection”. Even Fox News rolled over, firing Tucker Carlson who gave Giuliani and Powell a platform.

Last year, Dominion Voting Machine sued Fox for “defamation” because they allowed Giuliani and Powell to speak on their network. The standards of proof for Dominion to prevail in such a lawsuit are ridiculously high. For one thing, Dominion would have had to prove that the election was not stolen — a claim for which, as I have said, there is no evidence. Inexplicably, Fox settled the suit without a defense. They settled for ¾ of a billion dollars — an amount far larger than the total revenue of Dominion in its entire history. A ridiculously large settlement for a case which Dominion almost certainly would have lost.

What was the reason? All I know is that the USA has the least accountable, least transparent system for counting votes of any Western democracy. And it is now dangerous to ask questions about our system of counting votes. You could be thrown in jail like January 6ers. You could be sued for hundreds of millions, like Fox News. The Democrats — who have been overwhelmingly the losers in systematic election theft over two decades — the Democrats have been the ones to discredit any questions about the way our votes are counted. 

It’s almost enough to make me think — maybe neither the Republicans nor the Democrats want votes to be counted honestly.