Trevor Sullivan
@trevsully.bsky.social
This is a parody of the actual (much worse) charts used in the announcement presentation.
August 8, 2025 at 3:13 AM
This is a parody of the actual (much worse) charts used in the announcement presentation.
What part of "that's not what a Russian Tail looks like" don't you understand?
Elections aren't normally distributed. The original Udot analysis that invented the term "russian tail" is predicated on the fact that elections aren't normally distributed.
Elections aren't normally distributed. The original Udot analysis that invented the term "russian tail" is predicated on the fact that elections aren't normally distributed.
July 31, 2025 at 5:53 PM
What part of "that's not what a Russian Tail looks like" don't you understand?
Elections aren't normally distributed. The original Udot analysis that invented the term "russian tail" is predicated on the fact that elections aren't normally distributed.
Elections aren't normally distributed. The original Udot analysis that invented the term "russian tail" is predicated on the fact that elections aren't normally distributed.
Really this is just telling me that you didn't read it, because on one reply you say "they're right because they used Udot's method" and on the other "they're right because they fitted a bell curve", when the ENTIRE POINT of Udot's method is that elections DON'T adhere to single bell curves.
July 31, 2025 at 2:47 AM
Really this is just telling me that you didn't read it, because on one reply you say "they're right because they used Udot's method" and on the other "they're right because they fitted a bell curve", when the ENTIRE POINT of Udot's method is that elections DON'T adhere to single bell curves.
The more interesting topic is whether or not there are cross-dropoff votes, like you said.
This would be another "smoking gun" that's easy to find but ETA chose not to.
Sure enough, it doesn't support the idea. Early votes, the ones that were supposedly manipulated, have the LEAST cross-dropoff!
This would be another "smoking gun" that's easy to find but ETA chose not to.
Sure enough, it doesn't support the idea. Early votes, the ones that were supposedly manipulated, have the LEAST cross-dropoff!
July 29, 2025 at 11:46 PM
The more interesting topic is whether or not there are cross-dropoff votes, like you said.
This would be another "smoking gun" that's easy to find but ETA chose not to.
Sure enough, it doesn't support the idea. Early votes, the ones that were supposedly manipulated, have the LEAST cross-dropoff!
This would be another "smoking gun" that's easy to find but ETA chose not to.
Sure enough, it doesn't support the idea. Early votes, the ones that were supposedly manipulated, have the LEAST cross-dropoff!
I was gonna pass on this, but I realized I have a chance to put it to bed, so, congrats you made me do more work lol.
So, that isn't alleged by ETA specifically, though they do imply it.
What ETA points out is that Harris had a lower dropoff rate, and they wiggle their eyebrows suggestively abt it
So, that isn't alleged by ETA specifically, though they do imply it.
What ETA points out is that Harris had a lower dropoff rate, and they wiggle their eyebrows suggestively abt it
July 29, 2025 at 11:46 PM
I was gonna pass on this, but I realized I have a chance to put it to bed, so, congrats you made me do more work lol.
So, that isn't alleged by ETA specifically, though they do imply it.
What ETA points out is that Harris had a lower dropoff rate, and they wiggle their eyebrows suggestively abt it
So, that isn't alleged by ETA specifically, though they do imply it.
What ETA points out is that Harris had a lower dropoff rate, and they wiggle their eyebrows suggestively abt it
The simulation produces the EXACT same "statistically Impossible" pattern ETA claims is proof of fraud.
No manipulation. No special tuning. Just normal geographic differences in voting habits.
You can mess with the parameters and rerun the simulation yourself on the website.
No manipulation. No special tuning. Just normal geographic differences in voting habits.
You can mess with the parameters and rerun the simulation yourself on the website.
July 29, 2025 at 1:26 PM
The simulation produces the EXACT same "statistically Impossible" pattern ETA claims is proof of fraud.
No manipulation. No special tuning. Just normal geographic differences in voting habits.
You can mess with the parameters and rerun the simulation yourself on the website.
No manipulation. No special tuning. Just normal geographic differences in voting habits.
You can mess with the parameters and rerun the simulation yourself on the website.
Because when you actually make that graph, as I have, it completely disproves their theory. Every machine shows a normal random walk, no sudden changes at 250 votes.
Again this is the easiest graph of all to make. They must have tried it, saw that it disproves them, and decided not to publish it.
Again this is the easiest graph of all to make. They must have tried it, saw that it disproves them, and decided not to publish it.
July 29, 2025 at 1:26 PM
Because when you actually make that graph, as I have, it completely disproves their theory. Every machine shows a normal random walk, no sudden changes at 250 votes.
Again this is the easiest graph of all to make. They must have tried it, saw that it disproves them, and decided not to publish it.
Again this is the easiest graph of all to make. They must have tried it, saw that it disproves them, and decided not to publish it.
Their second argument is to show a histogram with a "Russian Tail".
Two problems:
1. That's not what a Russian Tail looks like
2. When you properly separate urban/rural data (liek the original Georgia analysis that coined the term) it disappears.
See for yourself:
Two problems:
1. That's not what a Russian Tail looks like
2. When you properly separate urban/rural data (liek the original Georgia analysis that coined the term) it disappears.
See for yourself:
July 29, 2025 at 1:26 PM
Their second argument is to show a histogram with a "Russian Tail".
Two problems:
1. That's not what a Russian Tail looks like
2. When you properly separate urban/rural data (liek the original Georgia analysis that coined the term) it disappears.
See for yourself:
Two problems:
1. That's not what a Russian Tail looks like
2. When you properly separate urban/rural data (liek the original Georgia analysis that coined the term) it disappears.
See for yourself:
The real issue is this: they are seeing the Law of Large Numbers and calling it fraud.
Flip a coin 10 times - getting 7 heads isn't weird.
Flip a coin 1000 times - you'll get ~50/50 every time.
Same with votes. More counted, closer to the true preference.
I made a game to help show this:
Flip a coin 10 times - getting 7 heads isn't weird.
Flip a coin 1000 times - you'll get ~50/50 every time.
Same with votes. More counted, closer to the true preference.
I made a game to help show this:
July 29, 2025 at 1:26 PM
The real issue is this: they are seeing the Law of Large Numbers and calling it fraud.
Flip a coin 10 times - getting 7 heads isn't weird.
Flip a coin 1000 times - you'll get ~50/50 every time.
Same with votes. More counted, closer to the true preference.
I made a game to help show this:
Flip a coin 10 times - getting 7 heads isn't weird.
Flip a coin 1000 times - you'll get ~50/50 every time.
Same with votes. More counted, closer to the true preference.
I made a game to help show this:
But this pattern appears in EVERY election. Even in their own examples of "normal" elections.
They hid this by using different X-axis scales for different years. Here's what it looks like when you compare the data at the same X scale: the difference vanishes.
They hid this by using different X-axis scales for different years. Here's what it looks like when you compare the data at the same X scale: the difference vanishes.
July 29, 2025 at 1:26 PM
But this pattern appears in EVERY election. Even in their own examples of "normal" elections.
They hid this by using different X-axis scales for different years. Here's what it looks like when you compare the data at the same X scale: the difference vanishes.
They hid this by using different X-axis scales for different years. Here's what it looks like when you compare the data at the same X scale: the difference vanishes.
ETA's big claim: Voting machines started "flipping votes" after counting ~250 ballots. They say this pattern is "inexplicable", and that the patterns in the scatterplots they produced from Clark County data are "statistically impossible"
July 29, 2025 at 1:26 PM
ETA's big claim: Voting machines started "flipping votes" after counting ~250 ballots. They say this pattern is "inexplicable", and that the patterns in the scatterplots they produced from Clark County data are "statistically impossible"
Irresponsible reporting. There's over 100 reported incidents of guards passing drugs to inmates, but only 4 cases of drugs in mail out of 17,000 letters.
The company's own CEO calls this effort a "trojan horse" so they can extort inmates with $0.50/email, $0.25/min of calls, SMS limited to 2/week.
The company's own CEO calls this effort a "trojan horse" so they can extort inmates with $0.50/email, $0.25/min of calls, SMS limited to 2/week.
July 25, 2025 at 2:54 PM
Irresponsible reporting. There's over 100 reported incidents of guards passing drugs to inmates, but only 4 cases of drugs in mail out of 17,000 letters.
The company's own CEO calls this effort a "trojan horse" so they can extort inmates with $0.50/email, $0.25/min of calls, SMS limited to 2/week.
The company's own CEO calls this effort a "trojan horse" so they can extort inmates with $0.50/email, $0.25/min of calls, SMS limited to 2/week.
@hankgreen.bsky.social's new video makes a point that's been my internet bugbear for a while now.
So many bad ideas on the internet (cough superintelligence cough) are based on the erroneous belief that exponential curves exist. They don't.
Every exponential is a sigmoid that hasn't inflected yet.
So many bad ideas on the internet (cough superintelligence cough) are based on the erroneous belief that exponential curves exist. They don't.
Every exponential is a sigmoid that hasn't inflected yet.
July 7, 2025 at 1:46 PM
@hankgreen.bsky.social's new video makes a point that's been my internet bugbear for a while now.
So many bad ideas on the internet (cough superintelligence cough) are based on the erroneous belief that exponential curves exist. They don't.
Every exponential is a sigmoid that hasn't inflected yet.
So many bad ideas on the internet (cough superintelligence cough) are based on the erroneous belief that exponential curves exist. They don't.
Every exponential is a sigmoid that hasn't inflected yet.
Election Truth Alliance are somewhere between incompetent and liars. The "suspicious pattern" they're seeing is what we in the stats world call "bigger samples are better".
Took me about an hour to make this chart with NV data. There's no big step up at 250 like they claim.
Took me about an hour to make this chart with NV data. There's no big step up at 250 like they claim.
July 1, 2025 at 3:33 AM
Election Truth Alliance are somewhere between incompetent and liars. The "suspicious pattern" they're seeing is what we in the stats world call "bigger samples are better".
Took me about an hour to make this chart with NV data. There's no big step up at 250 like they claim.
Took me about an hour to make this chart with NV data. There's no big step up at 250 like they claim.
This is turning into a thing. I've made like 5 simulations and over 12 visualizations, I'm going to start putting together some kind of interactive blog post.
Here's the latest complete-lack-of-a-smoking-gun. If their theory was right, there would be a huge step up at 250 here.
Here's the latest complete-lack-of-a-smoking-gun. If their theory was right, there would be a huge step up at 250 here.
July 1, 2025 at 3:28 AM
This is turning into a thing. I've made like 5 simulations and over 12 visualizations, I'm going to start putting together some kind of interactive blog post.
Here's the latest complete-lack-of-a-smoking-gun. If their theory was right, there would be a huge step up at 250 here.
Here's the latest complete-lack-of-a-smoking-gun. If their theory was right, there would be a huge step up at 250 here.
This got so stuck in my craw, I went and did the simulations. Check it. This set of parameters produces results that look EXACTLY like the Clark County data.
I'll play with it some more and try to get my hands on the original data to see if some of the demographic assumptions I made are correct.
I'll play with it some more and try to get my hands on the original data to see if some of the demographic assumptions I made are correct.
June 27, 2025 at 4:20 AM
This got so stuck in my craw, I went and did the simulations. Check it. This set of parameters produces results that look EXACTLY like the Clark County data.
I'll play with it some more and try to get my hands on the original data to see if some of the demographic assumptions I made are correct.
I'll play with it some more and try to get my hands on the original data to see if some of the demographic assumptions I made are correct.
In an election with a stark urban/rural divide, we expect these graphs to be not normal but bimodal, which they clearly are.
And since there were less machines in rural districts with higher early turnout, we expect a right-centered rural group with low deviation, which is exactly what we find.
And since there were less machines in rural districts with higher early turnout, we expect a right-centered rural group with low deviation, which is exactly what we find.
June 26, 2025 at 9:46 PM
In an election with a stark urban/rural divide, we expect these graphs to be not normal but bimodal, which they clearly are.
And since there were less machines in rural districts with higher early turnout, we expect a right-centered rural group with low deviation, which is exactly what we find.
And since there were less machines in rural districts with higher early turnout, we expect a right-centered rural group with low deviation, which is exactly what we find.
For completeness, here's what a "Russian tail" actually looks like, an elbow in the distribution that is flat all the way to the edge, which means that a weirdly large number of districts were unanimous.
If anything there's an anomalous Pro-Kamala tail.
If anything there's an anomalous Pro-Kamala tail.
June 26, 2025 at 9:46 PM
For completeness, here's what a "Russian tail" actually looks like, an elbow in the distribution that is flat all the way to the edge, which means that a weirdly large number of districts were unanimous.
If anything there's an anomalous Pro-Kamala tail.
If anything there's an anomalous Pro-Kamala tail.
"Little precedent"? How about "in direct violation of the unambiguous letter of 150-year-old law"
June 9, 2025 at 10:55 PM
"Little precedent"? How about "in direct violation of the unambiguous letter of 150-year-old law"
Couldn't fit an "illegally" in that headline, huh?
June 9, 2025 at 10:52 PM
Couldn't fit an "illegally" in that headline, huh?
It varies from state to state. Some mimic the federal system with executive appointments, others have at large elections, some have more complex systems with recommendation councils, executive appointment, AND popular confirmation all at once.
April 2, 2025 at 2:23 PM
It varies from state to state. Some mimic the federal system with executive appointments, others have at large elections, some have more complex systems with recommendation councils, executive appointment, AND popular confirmation all at once.
It's hard to stop that happening when the people in the drivers seat are doing it on purpose.
This quote from Vance is not repeated enough, maybe because people don't catch the reference. He compares himself and Trump favorably to Caesar, ie, the guy that ended democracy and republicanism in Rome
This quote from Vance is not repeated enough, maybe because people don't catch the reference. He compares himself and Trump favorably to Caesar, ie, the guy that ended democracy and republicanism in Rome
February 8, 2025 at 6:59 PM
It's hard to stop that happening when the people in the drivers seat are doing it on purpose.
This quote from Vance is not repeated enough, maybe because people don't catch the reference. He compares himself and Trump favorably to Caesar, ie, the guy that ended democracy and republicanism in Rome
This quote from Vance is not repeated enough, maybe because people don't catch the reference. He compares himself and Trump favorably to Caesar, ie, the guy that ended democracy and republicanism in Rome