Transcription
Today I'm going to be speaking with Nathan Taylor, executive director of public engagement for the Election Truth Alliance. This is quite, quite honestly, Nathan, this is one of the most requested interviews over the last couple of months with folks saying, "You've got to hear Nathan out and hear what he has to say." So, can maybe what I'll do is I'll set up my understanding of some of the claims that have been made about the election, and then you'll explain to me what your claim is. Is that fair?
Yeah, let's do it.
Okay, so Greg Palast put out a piece some time ago. I looked at the piece. The piece made a number of claims that I found to be a little too circumstantial for me to make a definitive statement about 2024. They included extrapolations of prior, for example, black voter behavior in terms of the split between Democrats and Republicans, assumptions about changes to the number of registered voters, extrapolating from that in some cases with eight or 10-year-old data to make claims about 2024. It didn't feel super solid to me. And so I came away from those claims sort of saying, "Listen, I'm open to hearing anything. This I don't find convincing." So, let's now give, with that as the context, give us, at the most basic level, the claim you're making, and then we will talk about the evidence.
Yeah, sure. And if I may, I'd love to throw a little bit of background at you, um, just for my work history and the ETA. So, my personal background, um, I have a degree in cybersecurity and information systems. I am also, I was a 25 Bravo computer tech specialist in the army. um, actually commissioned as a signal officer as well. So, I do know a little bit about computers, computer systems, um, some data analysis. And on the civilian side, I worked, um, with a company that provided real-time network monitoring for multiple entities like banks. So, I've done a lot of data analysis. I've, I've looked at network security.
And what happened was, right after the election, um, there were multiple red flags that brought the ETA together. So, the ETA at this point, there's three executives: me, Lily, and Jive. Lily and Jive have different backgrounds in either statistics and finance or political science. So, we all got together, and that's when we sat down and we said this. We said, "There are some interesting things with the election. Let's do a little bit of investigation. Let's see what we can, um, start looking at to determine if our elections are being run correctly, free, fair, safe, and secure." That was the concern. We wanted to verify the vote, to validate. And so that is what we did. We started, um, we started bringing in volunteers with various backgrounds in statistics. Some of them have PhDs in statistics. Some of them are, um, you know, auditors or computer scientists. And that's when we sat down and we said, "Okay, how can we look at elections and how could we perhaps apply statistics to look at elections to flag any concerns?" And at the time, um, we, we knew that there was audits in the elections. We reviewed those. And if you're familiar, audits, that's a standard practice in finance, engineering, public safety. So, auditing elections shouldn't be that much of a concern. But immediately, when we did start, you're talking about this right after the election, the concern was, as you said, um, people wanted very credible, very well-established, um, concerns. And the common pushback was, "Election denial was on the rise." So, just want to make that clear. This is not election denial. This is election integrity and auditing. This is a standard practice. And because of that mindset, we said, "Okay, let's do this. If you can apply statistics to detect fraud in banking, if you can apply statistics to do safety and quality testing in engineering, why can't we use statistics to flag non-human or abnormal voting behavior? That could mean vote manipulation." That's what we said. And then we discovered this is already a thing. Um, it's happening outside the US already. There are already election fraud experts such as Roman Udat, Sergey Spilin, Peter Cleick, at all, who've been applying election, um, statistics to Russian elections, to elections in Afghanistan, Kenya, and Bolivia. And they have all begun to find credible concerns of election fraud. So, we took their techniques, um, and the ETA has ran with those methods, and we've begun to analyze swing states and non-swing states for the '24 election, and then we begin to publish those reports. So, that's the background. Um, as you said, I'd love to kind of walk you through some of the things we do, some of the things we find, and then the big picture. So, any questions?
Um, yeah. So, let's, in order to pick one thing, because we could spend hours on this, let's go to, is there, and then you walk me through the example if you have it. Is there a state that was reported as won by one candidate that your analysis leads you to believe that was wrongly reported and, in fact, the will of the voters was that the other candidate should have been awarded that state?
Yes, we've published two different reports now, both for swing states, three in total, but two of them meet that criteria: Pennsylvania and North Carolina. Um, if you're familiar with Pennsylvania, there's a lot of controversy around Pennsylvania, especially on election night. I think they had over 30 bomb threats, um, and multiple machine failures across the state. So, um, what we've done for Pennsylvania, and the report is on our website right now, elections.org, and we've analyzed the entire state of Pennsylvania, every county in the state. We've also analyzed three main, very strong Democratic counties and put those up, uh, in the report. So, what our statistics is finding and what our statistics does. So, for example, we like to use the work of Dr. uh, or the work of Dr. Walter Mebban and Peter Clem as examples. So, I do have a visual for you. Now is probably a great time to to reference it. So, what we've done, um, on the first slide for these visuals, this is a heat map, and as I said, this is an example of how we do statistical analysis of elections. This is a method developed by Peter Peter Kick, um, at all. So, this is a group of election fraud experts and statisticians, and they've said, "Let's look at Russian elections, let's look at Ugandan elections." And they put out a paper, and on the right side of the screen, we reference their statistical analysis of those elections. So, we see Russia in 2011, Russia in 2012, and we see a, um, a Uganda, which I don't know the exact date, but it's, it's in their paper as one of their, their foundations. And so, what a heat map does is we look at every precinct in a county. Um, this data is public. You can go get it from the county's websites. You can get it from the state. So, we took publicly available data and we developed this heat map on the left for you. In California, is an example of what we argue is more of a normal and expected election. Per the work of Peter Clemet, they say a, a non-manipulated election should be more of a circle. There should be no noticeable shifts in precincts of high versus low turnout. And that's what we're, we're measuring. So, the x-axis is turnout. Turnout is just how many votes were cast compared to how many people could vote. And that's it. So, if you're 50% turnout, half of your voters showed up and voted. And then, uh, the, the y-axis is the amount of votes each candidate got. We're just looking at one candidate for these heat maps, and that's Donald Trump, win candidate in a majority of the swing state. All the, all the swing states.
So, Nathan, if I may here, just to make sure we're not missing anything, the idea here is, by looking at voter turnout versus candidate performance, you're accounting for the size of the precinct. In other words, we're already adjusting for population density in so far as we're not looking at the number of people that voted. We're looking at the percentage of eligible voters in a precinct. So, we, this is not about, "Oh, well, it makes sense because denser urban areas tend to lean one way." We've accounted for that by looking at voter turnout rather than number of votes.
Yes. And this is, as I said, based exactly on the work of, um, Peter Clem and other experts that have looked at other elections. So, we're, we're actually going to compare US elections to some European and other elections.
Got it.
And I'll run you through, just very quickly, what is concerning and what would be potential vote manipulation on a heat map.
If you look at Russia on the right side of this, a heat map that begins to stretch up and to the right could be an indication of vote manipulation. Um, specifically, it may be an effect of ballot stuffing, whether it's physical ballot stuffing or digital ballot stuffing. We don't know. That's where you would go and investigate on the ground. But as we see in Russia, in both 2011, 2012, they have key indicators that they are potentially manipulating the elections in Russia by stuffing. And you actually see some places in Russia in 2012 that are at 100% turnout, and that's in the top right of the little, uh, grid there for the heat map. So, as you can see, California, we're going to use California as a baseline here. This is, um, Shasta County. This was won by Trump. So, this is what's important. That county was won by Trump, but it looks statistically normal. Looks like a circle.
Trump genuinely won that county, is your belief?
Yes. Yes. Statistically speaking, there are no, no red flags here. It looks very normal.
And then, so we'll go to the third slide for you. And this is, as you said, are we concerned that there was multiple states, a single state that could have had vote manipulation change the outcome of the election at the presidential level? So, Pennsylvania '24. This is every precinct in every county in Pennsylvania. And this took a lot of work to pull all that data together, but we visualized it here for you on the left. And immediately, we see a serious concern that instead of what we would expect in a free and fair election, which is a more circular distribution, we see a strong shift where in places of higher turnout, in precincts of higher turnout, independent of county. So, we're looking at Philadelphia, we're looking at Allegheny County, we're looking at Erie County. These are some of the more blue counties in the state. Even counties in precincts over 60% Republican candidate for president in those precincts, they begin to get more votes consistently. And what's important here is, as you, as most people would say, maybe there's, you know, some really strong pockets of Republican voters in some places. This effect is consistent across a majority of the precincts. This isn't just a handful. This is almost algorithmic in nature, and that's what our analysis is showing here. And the other half of it is, as well, there's almost an artificial wall. And you can see it for Pennsylvania, where instead of that more circular, round effect, we see a stretching around 85% turnout. So, it almost seems like these precincts could have been, in this case, this could be vote manipulation, stuffing. They could have been stuffed up to around 85%, 80, 85%, and then any votes after that point deleted. So, the, if we are to assume the worst here, and then we'll get to the differential diagnosis in a moment, and I have some other questions. If we assume the worst, the technique would be, maybe the voter turnout didn't really go as high as it looks like in some of these areas. It's like over 90%. I'm looking at the chart. Is it like 92, 93% in some of these counties? Is that right?
Yeah, in some of these counties and pre. So, maybe instead of really having that level of turnout, the real level of turnout was lower, but there was manipulation, either physical or digital, as you say, ballot stuffing or through technological systems, which credited votes above and beyond the real votes to candidate Trump. Is that essentially the sort of, like, most dystopian interpretation of this?
I would say this. We don't truly know how much of this could be stuffing versus switching, but we do know the scale of impact. So, we don't know exactly what the real turnout could have been, um, which was still expected to be very high. But we do know that switching, deleting votes, and stuffing votes could create this effect. And we are able to estimate the amount of votes that could have been impacted at the presidential race in Pennsylvania. Um, and per our analysis and per the work of Dr. Walter Mebban, he's a PhD statistician professor at the University of, um, I'll have to double-check which exact university he is. Yeah.
I think it's Michigan.
Um, and so he's put up a paper as well. He actually looked at Pennsylvania. So,
As you said, do we think this is enough to change the outcome of Pennsylvania for the '24 election? 120,000 or so, 121 or 22,000 votes is what Trump was winning by in Pennsylvania for the, or for the presidential race. Our analysis says potentially 190,000 votes could be manipulated, stuffing, switching. Um, and Dr. Meban even supports and his model, his method even says up to maybe 210,000. So, yes, if this is vote manipulation, if these effects from high turnout bending fitting one candidate is because of some type of digital vote manipulation, this would be enough to change Pennsylvania. And we're seeing the same effect in North Carolina, which it would be enough as well. The margin is 184,000. We, we have over 190,000 flagged anomalous votes.
And it could be enough to change the outcome in Nevada as well. And these are just the states we've published. We've been looking at other states, other swing states that raise the same concerns statistically. We just haven't yet estimated the potential impact.
So, as of right now, if indeed, let's assume it's true, Harris really won Pennsylvania, Harris really won North Carolina, Harris really won Nevada, Trump actually still has 270 in that scenario. But what you're saying is this is only what you've published and you believe there is more.
Yes. So, the common pushback we get is, "Everyone says every state runs their own election. Every state uses different systems." What we've started to find as we do this analysis is, what is the common trend between all of these swing states and all of these other states that we've not published yet? We are finding these concerns statistically. And it's very simple and very concerning. This is where my background with cybersecurity starts to jump in. Is we did find that 70, at least 70% of the US uses the top two voting systems, and these include the tabulation machines. And so, this is a fact I want you to understand. If you are able to compromise these systems, these tabulation systems, in 2024, nearly 100% of votes were counted by tabulation machines at some step in the process, even the hand-marked paper ballots. And so, that's very important. Is because we're seeing in these swing states, they're all using these same systems. And we're starting to find in, uh, states that aren't swing states that do use these systems, the same effects. And as of this moment, that leans more towards the technological manipulation rather than the physical ballot stuffing. Right? I mean, because the physical ballot stuffing would not be as relevantly mediated by the same tabulators. The tabulators seem to point more to the technological manipulation.
Yes. The consistency, how aggressive and consistent this is across multiple precincts and counties, um, points to it being a system instead of people going and doing things because it's too, um, algorithmic. We can actually, we've begun to now analyze, is there a consistent pattern? Is there an algorithm? Is there, you know, a formula? And we're able to start seeing that there may be a formula in how these votes are being shifted around, and that'll be coming out forward.
Yeah, go ahead.
One of the questions I know my audience will have is, does the scenario Nathan is drawing up require individuals at either every county, precinct, etc., or would this, and again, this is all, I'm not telling my audience this is what happened. And I'm asking questions to try to figure out exactly what the claims are. Would this be something that theoretically could be pre-installed in some of these machines such that the machines already know what to do, and it doesn't require someone in Pennsylvania, in Philadelphia, precinct 39, is pressing a button?
Yes. The concern is that whether through a compromise of the vendors, through a common, uh, point, like, for example, some of these systems use hard drives. If you were able to influence them before those hard drives ever get sent out, or putting code on the machines. Yes, you could do this before the election. Um, cybersecurity-wise, an example of this is, this is just malware that would have a time to window to execute. Um, it wouldn't be manipulating votes or changing things before the election or after the election, just during that window. And we actually found, and this is something people know about, but maybe not as readily, is we, there's an example of this have already happened, um, with technology. Have you heard of Dieselgate?
Dieselgate. Oh, this is the, um, Volkswagen, uh, gas mileage scandal, I guess we would call it.
Yes. So, Dieselgate was where Volkswagen had a software program to cheat emissions. And when they were auditing the machines, um, in a lab or when testing them, the, the cars ran normally and their emissions was normal.
Right.
But the data from emissions, um, reports wasn't matching up. There was way higher emissions than what should have been accounted for. So, some independent nonprofits actually got together and they took a Volkswagen system and they put a, um, tester on the back of the vehicle in the trunk and on the tailpipe, and they drove it on the highway. And in use, the Volkswagen began to output higher emissions. So, they had designed a way to cheat audits so that when they're being tested,
They look normal.
They look normal, but when they're in use, they begin to act, um, in this way. If it was voting systems, you could do the same thing, whether by design or through malware. During testing, you wouldn't catch any problems, but when they're being used, you could manipulate votes. And the other half of the battle is, we're finding the largest voting system brand in the US. They do seem to have cellular modems in their tabulators that counties are using to send early reports to the counties from the precincts. We've begun to investigate this as well, because the common pushback is, "Our systems aren't connected to the internet." Your tabulators could be, and if I have the means to compromise them, um, digitally, I don't ever have to be anywhere near these systems when they're operating. I can compromise them before, during testing, put malware on them at any point in the process, and then manipulate votes. And it would be very difficult to catch that without sufficient audits and without statistical investigation to go and figure out where to look.
All right. I want to ask you about one more thing, and then I'll sort of, uh, give my sense to my audience of what I think makes sense to do based on on what you're asserting. I want to just propose a differential diagnosis. Right. What we're pointing to here is an association between counties. Sorry, is it counties or precincts with higher turnout?
In this case, precincts.
What you've identified is an association, uh, between higher turnout at the precinct level and a larger share of the vote going to Donald Trump. That's the association that you've identified that we're focusing in on here. Is it possible? Is it possible, or from a statistical perspective, how would we contradict the idea that there are certain parts of states where, due to what's happening socio-culturally, it makes sense that the higher turnout correlates with more support for Trump? For example, I'll propose one possibility. In some areas, just based on activism and who's there, new voters said, "Let's get together and vote Trump." It was specifically a Trump-up supporting effort. And so that explains why you've got an anomalous larger turnout share, but it is because of some state-level activism or precinct-level activism that it was mostly Trump supporters that said, "Hey, we're going to turn out more than we did in previous times." Is that a plausible differential diagnosis here?
That, that is a good concern. But two things. The first, we're seeing this effect even in places where the population wouldn't match the results, um, in some of the most heavily Democratic registered, uh, places in the US.
Nevada, Las Vegas. We're still seeing this effect. But this effect isn't happening, um, right at the beginning. It happens almost after a certain threshold of votes or a certain threshold of turnout. And that's what's weird. Is if we were seeing high, uh, votes for Trump from the beginning, or from low turnout to high turnout places, then yes, this would be normal. But it almost seems to begin to happen after a threshold. And here's what's interesting. This is the last little piece. We haven't revealed this yet. So, this is the first time anyone's ever going to see this. We asked the question, which is, if the concerns we're finding is a compromise of the systems, what if we look at the places that did hand-count some of their precincts and some of their counties in the '24 election?
And that is the second slide we have for you. And this is very simple. This is '24. This is in Minnesota. So, St. Louis County, they hand-count a few of their precincts. Um, and they do have almost like a split where some of the, some of the, like, half the county will do machine and half the county will do hand counting. And in this case, we compared the two. This is the same county, and this is '24. And we found that in the precincts that hand-counted their results, Trump won this county overall. By the way, in, in this sense, he got more votes. I think he got a little closer to 55% of the votes in this county. We found that in precincts that hand-counted their results, Trump was receiving on average around 40% of the vote. In precincts that they machine-counted using these tabulation systems that we're flagging as a concern, Trump got 7% more of the vote consistently across all of those precincts. That put him above the 50% margin. And statistically speaking, this does not look like a free and fair, um, result. It actually exhibits effects of potential switching of votes for the Michigan.
Now, again, differential diagnosis, playing devil's advocate. Yeah. Is it possible that there are reasons that correlate with Trump voting why Trump voting precincts would be more likely to machine count than hand count? Right? Because we've got to back that out in order to be able to make the claim.
Yeah, that's a good point. That's something you would be able to identify as you dive a little bit deeper into historical, uh, representations, which we will be releasing as we go forward. Is we actually have the same data multiple elections going in the past. And, um, we asked the question then of when did these places actually start using voting systems? And we'll be publishing that as we go forward. But no, these are good concerns. I will highlight one other thing then, which is less about the data but more about the intent. So, as we said, if you were to pull this off, right, it would take access to these systems or compromises of these systems, um, and honestly, like, that would be hard to go unnoticed, right? But we do have examples of these systems being targeted. Back in, um, 2024, and November 13th, Free Speech for People had a group of cybersecurity experts say, "Hey, we know about known attempts to compromise the election systems across the US." Examples of this was in Mesa County, Colorado in 2021, when actors breached voting systems in, uh, potential benefit of, you know, Trump's administration or Trump's campaign at the time. U, we see the same examples in Coffee County and Georgia. So, what's interesting to us is if this is a compromise, um, as we said, we're going to keep putting these reports out there. We're going to keep asking these questions. And I love the devil's advocate because that, that's what you have to do to test things.
Yeah.
But we, we've moved forward. The ETA has moved forward from statistical reporting to actual on-the-ground investigation. So, that, that is the other half of what we do now. Is we will continue to put out these statistical analysis, but we ask the question, if this is vote manipulation, let's go to these places and let's work with the people on the ground, and let's see if we can find evidence that this is a real concern. And we are moving forward with two cases, uh, litigation where we have found sufficient concerns of the vote manipulation. We've not published these yet. We were hoping to file our first lawsuit this month, where we're, you know, we've actually hit the opposite of what you'd expect, which is we have so much that we're struggling to cut out the things that we think are less effective and and put in. Um, and I can maybe tell you just a few examples of what we are finding. Well, you know what, I would rather do is just get there were a couple things I wanted to ask about and then I think maybe what you're, that that might be best left for a second interview, but so, so my audience knows, right now you've talked about Pennsylvania, North Carolina, and Nevada in this interview has published stuff that would still give Trump 271. Is it your belief right now, as you speak to me, based on everything you have published and not yet published, that Donald Trump did not genuinely win the 2024 election?
As of right now, every set of data and additional evidence that we have found points to a serious concern that our election system could be compromised. It could have changed the outcome of not only the presidential election of the '24, but it could have also impacted state, senate, and house, um, positions that would go on to to impact Congress.
Okay. So, here's what I'm going to say to my audience and encourage them to do, which is what I'm going to do. Nathan has laid out this case. He's provided some visuals. The first thing I would do is now I'm going to go and first try to verify that the data Nathan's working with is accurate. That's the first thing. Is the input data accurate? I'm then going to evaluate his conclusions, delve more deeply into differential diagnosis, and that's the process I will start to use to make a decision. And then, what I would love to do, Nathan, is after you, you have more to reveal, and after I've had a chance to do that, we could have you back and have another conversation.
Yeah, that would be great. Our data is actually, our, our methods and our data is public at data.electiontruthalliance.org. We built a dashboard, digital dashboard, so you can actually go download the data for yourself, compare it to the county and state reports, and then you can do your own method or analysis or apply our method analysis yourself on the dashboard.
All right. We've been speaking with Nathan Taylor, executive director of public engagement for the Election Truth Alliance. Nathan, I appreciate you being here.
Of course. This is opening up a very serious potential can of worms, which we're going to investigate. I look forward to speaking to you again.
Yeah, thank you so much for having me, and I, I have a lot more to share.