Looking at election statistics
[Hat tip: commenter “I am Sparticus.”]
I found this video fascinating. I know enough about math and graphs (although nothing about programming) to more or less understand what they’re saying, but I can’t evaluate it independently. Also, I don’t know if it would have any clout in a court of law. In addition, why on earth would any voting software need a mode in which votes could be weighted like this? Why would that ever be okay rather than fraudulent?
The main speaker, Dr. Shiva Ayyadurai, is characterized by Wiki as a conspiracy theorist of dubious reputation. However, his scientific credentials seem on the up and up, and some of his assertions that Wiki characterizes as “misinformation” don’t seem all that suspect to me. Just to take one example: “In March 2020 he published an open letter to president Donald Trump where he wrote that a national lockdown was unnecessary and advocated that large doses of vitamins could prevent and cure the disease.” As time has gone on, there has been quite a bit of evidence that lockdowns haven’t helped and that Vitamin D can really improve outcomes, so I don’t see how such assertions can be dismissed out of hand. The jury is still out on so many things about COVID.
Ayyadurai also ran for the Senate in the Republican primary in Massachusetts in 2018, and ended up running as an Independent and getting 3.4% of the vote. This campaign slogan of his against Elizabeth Warren makes me smile: “Only a real Indian can defeat a fake Indian.” Guy’s got a sense of humor. Lastly, he and actress Fran Drescher were an item from 2014 to 2016. Make of that what you will.
At any rate, as with any other science, the facts and the conclusions should stand or fall on their own. If anyone more conversant with the statistical and programming issues than I would like to weigh in, feel free:
Lastly, he and actress Fran Drescher were an item from 2014 to 2016. Make of that what you will.
A 57 year old divorcee (whose husband had decided that after 21 years of marriage he’d prefer sodomitical associations)? Keeping company with the author of Enter Whining? Dr. Shiva enjoys a lark…
I haven’t watched this video yet, but I was put off immediately by the claim Ayyadurai invented email as a high school student in the late seventies.
Email, as we know it, came out of ARPANET in the last sixties/early seventies. ARPANET (Advanced Research Projects Agency Network) was the legendary DoD project which laid the foundations for most of the tech which became the internet. We definitely got our bang-for-the-buck from ARPANET funding.
Which doesn’t invalidate whatever arguments Ayyadurai makes in the video. However, I will watch with the mindset that this guy has been a cheap hustler in the past.
huxley:
Yes, there’s a big discussion on his Wiki page about the email claim. He does seem rather fishy to me, but you and I agree that his arguments on the election must be evaluated on their own merits or lack thereof. I included the biographical material on him as a caveat, however.
neo: To be sure.
I’ll just add that the early days of computing, of which ARPANET was very much a part, were an astonishing story of ingenuity and determination. I can’t say those days were exactly sacred — they weren’t — but a new era in human history was being born.
I don’t take kindly to those stealing valor from the giants of that time.
I have a degree in theoretical mathematics and have been a software developer for more than 30 years. They are absolutely correct about unverifiable inputs and outputs. I’ve been screaming about that part for years.
And I have imagined numerous ways a software developer could game the system. The data they show is exactly what I would expect of a rather poor software developer doing just that. If you’re a corrupt developer of party A, your algorithm will switch more votes in districts that are heavily populated by party B. Why? Because there are more votes to switch.
Having said that, I don’t think they are totally convincing. I’m not sure I would expect their “normal” distribution. You could argue that the straight party ticket is not indicative of the individual ballots. The biggest factor that argues in their favor is the steepness of the slope in their examples. I might expect a downward curve, but it wouldn’t be as clean as that, and it would be a smaller slope.
@ Chris: The steepness is such a clean line that it, as Bennie puts it, does not seem germane to being natural; and Dr. Shiva ponders may be an algorithm at work.
algorerhythm may be fixed by SCOTUS
Related:
https://noqreport.com/2020/11/11/data-deep-dive-on-dominion-voting-systems-offers-incontrovertible-proof-of-election-hack/
They assume that “individual candidate” voters should vote in proportions similar to those of “straight party” voters. However, I’m not aware of any reason why that assumption must be true. Maybe it’s true, but maybe it’s not. Given that the two groups deliberately acted differently (by definition), why would we assume they’d vote for Trump in the same proportions? If their assumption isn’t true, then their graphs are meaningless. If they present any evidence for their assumption, I must have missed it.
See for yourself. At 22:06 he starts showing a hypothetical example (to explain their method, etc.), and at 23:28 he starts showing an actual example (to present their suspicions). They make the assumption without showing evidence why it must be true.
NOTE 1: I only watched ~33 minutes of the video. I have no comment on whatever else was presented beyond that.
NOTE 2: I’m biased in their favor, because I already suspect some serious shenanigans in this election. So I’d actually like to be proven wrong here. Until I am, though, I find their presentation to be thoroughly unconvincing.
However, I will watch with the mindset that this guy has been a cheap hustler in the past.
You have lived too long on Earth, Hux. There’s a lot of things you need to “unlearn” first.
I don’t take kindly to those stealing valor from the giants of that time.
An emotional trigger, and not a logical argument or thinking.
The human mind associates certain things as “facts” even though it came from fake news, books, or other experiences that may or may not be accurate.
When this foundation is challenged, a trigger is pulled, a flag raised, and the defense of this identity must be given by the ego for the ego.
This resistance prevents people from looking at whether the news they grew up believing, is true or false. Same reason why Leftists think stuff about Trump.
It’s a psychological flaw of humans.
So, given the options:
1: Alien reverse engineered tech allowed DARPA humans to steal technology and present it as original Or
2. A huckster is claiming to have made email.
Which one looks more appealing?
To continue on my earlier thought.
The biggest problems with their analysis, is that there’s not enough of it. They showed us one heavily Democrat county. We need to see more. We need to see more Republican counties. We need to see what some of these same counties looked like when doing the same calculations for Biden. And we should look at at least a few other states too, some heavily Republican ones like Idaho, heavily Democrat ones like New York, more centrist states like Ohio and Florida, and even Pennsylvania.
Then we can compare/contrast, look for aberrations and similarities.
Only then is a real analysis possible.
What they have so far is an apparent red flag. Red flags are only possible indicators, not proof. That comes from either actual evidence or a great many more red flags.
They assume that “individual candidate” voters should vote in proportions similar to those of “straight party” voters.
That’s relatively easy to check and they may have merely skipped over it, as it would be basic.
The number of people who voted only for one candidate per election, should be in the data banks historically and statistically. Unless Americans just cared not to keep those records, which is fine.
Indians are rather meticulous about keeping records, given Vedic astrology cataloged data for 4000 plus years.
Red flags are only possible indicators, not proof. That comes from either actual evidence or a great many more red flags.
Good for Trump that there are a great many more red flags then…
In addition, why on earth would any voting software need a mode in which votes could be weighted like this? Why would that ever be okay rather than fraudulent?
When testing software it could be handy to have a weighted function to add X number of votes for Y number of touch-screen inputs to save time with particularly large test conditions. Your second question gets to the heart of the matter, though.
For a mission critical system like a medical device, avionics system, missile control, or electronic voting, I’d argue that it’s never OK to put test/debug code into the application; full stop. There are specific quality assurance and testing protocols developed over decades to deal with these types of systems. The two most common reasons for not using them are laziness or malice. Given the crap systems that electronic voting software companies have been known to release, and their absolute denial to subject their products to external audit and testing, I don’t think we can rule either of those two possibilities out.
Liebling Schatz:
They said that it could be a natural phenomenon except for the regularity of the line.
You have lived too long on Earth, Hux
Ymarsakar, DNW, etc: I’m done with being called “Hux.” I don’t consider if a pleasant diminutive. I don’t have the impression those who use it are doing so in a friendly manner. More of a frenemy thing, I’d say.
huxley. Six letters. Not much trouble, really.
They said that it could be a natural phenomenon except for the regularity of the line.
neo: Yes, that bothers me.
As Chris of Rights notes it would be good to see more examples, which would work better in a paper than a YouTube. (I guess they are trying to reach the broader public.)
Still, it’s a clever approach for cooking the vote, if that was done, in a manner standing up to superficial scrutiny.
huxley; Chris of Rights:
I also think much more info is needed.
I thing the best data would involve other states in which (a) it is suspected that the votes were cooked in some way; and (b) it has a straight party voting option. Problem is, Michigan is the only such state in which both (a) and (b) are present in 2020.
However, there are 5 other states that were not disputed but which also have straight ticket options: Alabama, Kentucky, Oklahoma, South Carolina, and Indiana. The problem is that there’s no suspicion that such an algorithm would be used in those states, if indeed it is ever used.
What could be done is to compare the voting patterns, though. If these states exhibit a similar pattern to that in Michigan, it could mean that it’s a normal pattern OR it could mean the same algorithm was used in all the states. However, if they exhibit a very different pattern from Michigan’s, that would indicate that perhaps there was something fishy that happened in Michigan and not in the other states.
I take back the statement of him inventing email. I forgot to delete that statement before posting as it was close to bedtime. You can denigrate that statement and the man, but the core HYPOTHOSIS remains the same.
“The PROGRAMMING CODE in the tabulator appears to skew Trump results downward in REPUBLICAN PRECINCTS where the harvesting could be better hidden.”
It is a small sample size for sure but it does open the door to larger investigation as the downward drift is so pronounced especially since Wayne County (or testing other states results as suggested by Neo) does not show the same pattern. In the earlier post, I did not include information that I sent to a high Trump Campaign official I casually know but the hypothesis can easily be tested.
Message to the Trump Staffer:
This proves that big data can prove anomalies. I was always puzzled why Trump loss in MI was so large. The PROGRAMMING CODE in the tabulator appears to skew Trump results downward in REPUBLICAN PRECINCTS where the harvesting could be better hidden.
If the same machines are used to CANVAS the votes it would return the same results. It would be easy to test by taking a Wayne (Detroit) county machine and run the Oakland/Macomb ballets through it and vice versa provided there was no software refreshes in the meantime. If the test shows a problem then a lawsuit can definitely be won on this.
I have implemented several Enterprise Systems in my day.
It appears we have to get to the source code and compare Wayne county code vs. other counties.
We need to see the updates to the programming.
If there is a different version on the Republican machines then that is a red flag or if there is an integer value that changed
There should be a log of updates.
Was there updates right before the election? (several affidavits state there was)
Was there an update during election night? (the three hour pause)
Was there an update after the election? (unknown at this time)
We needed an absolutely trusted source to examine the code. Changes in code never quite disappear.
Someone like Binney who proved that no way was the DNC computers hacked but instead the information was downloaded on a thumb drive could be a resource.
Ps. For all the people criticizing the team, since the Doctor and the other main presenter are people of color you are displaying your white privilege and systematic racism by criticizing them….not.
Neo: They said that it could be a natural phenomenon except for the regularity of the line.
Yes, it is a fairly straight line, but it’s also a fat line. They didn’t say what the line’s correlation coefficient is (unless they mentioned it after the 33 minute mark), so we don’t know how good the fit is. We need that information to draw natural/unnatural conclusions.
Meanwhile, here’s a possible explanation for a fairly straight line. I’m just suggesting a hypothesis, not claiming it’s a solid gold explanation. — A voter casts an independent candidate vote, because she wants to buck the trend. In a Dem-majority precinct, she likes most Dems but specifically wants to vote for Trump. The more extreme the Dem majority, the more trend bucking she’s doing, and the more likely it is that an independent candidate vote will be a vote for Trump. And vice versa for a Rep-majority precinct. This single motivation has the effect of proportionally raising the left half of their plot and proportionally lowering the right side. And now there’s a fairly straight line running from upper-left to lower-right.
BTW, I love your blog! I’ve been a fan for many years!
Me: They assume that “individual candidate” voters should vote in proportions similar to those of “straight party” voters.
Ymarsakar: That’s relatively easy to check and they may have merely skipped over it, as it would be basic. The number of people who voted only for one candidate per election, should be in the data banks historically and statistically. Unless Americans just cared not to keep those records, which is fine.
Yes, that might justify their assumption. It would be nice to see them demonstrate that with data.
Libeling Schatz; Ymarsakar:
I don’t think that comparing those who voted in previous elections only for one candidate would answer the question. “Voting only for one candidate” is not what an “individual candidate” voter is, although it sounds that way.
“Individual candidate” voters in this case refers to those who vote, but not straight party vote. In a state that offers a straight party option, you can just check one box and that means you are voting for every single candidate that party offers. But if you don’t check that box, then you are choosing to vote “individual candidate” – which means that you are choosing to vote for each candidate in each class one by one by one rather than party line. You can split the ticket, voting for Republicans for some races and Democrats for others, or you can choose to skip some races, or you can vote for someone in every single race, or you can just vote for someone for one race and not vote for any candidate in any of the others.
To compare with the study discussed in the video, I believe you would have to take each county and plot the votes for the candidate that were cast under the straight party choice compared to the votes for the candidate that were cast under the individual choice option, and put the differences on a graph as in the video, and see if the patterns in the graphs match.
Liebling Scharz:
Glad you like the blog.
I had a similar thought to yours about the motivation to vote straight ticket versus to cross over just for the presidential vote. Problem is, I believe, the findings were only for Trump voters and not for the Biden voters at all. What’s more, the effect increased for Trump voters the more Republican the precinct was, but it did not increase for Biden voters or even show up at all. At least, that’s my understanding.
Neo asks if this evidence of exceptional patterned voting is legally useful.
On the day after the election, law blogger VivaFrei hosted a streamed reaction analysis on vote fraud and the election with brilliant federal criminal defense lawyer, Robert Barnes.
In the last third, Barnes mentions that he used to work as an election lawyer. And he helped a candidate win a mayoral contest. (And vote cheating was involved; his client later went to prison but for unrelated charges.)
Apparently, improbable statistics are used to show patterns to reveal disparate impacts. For voter suppression, involving the civil rights violation of racism, these involve suspicious under vote results.
But for ballot stuffing suggesting vote fraud, the reverse, over voting — or else undervote patterns of the targeted opponent. As in this video.
And yes, statistical evidence is used in Court as evidence. This game doesn’t prove the case, but merely as one part of a larger pattern of a case proving that a free and fair standard for election is highly improbable. And therefore that a remedy is required.
Classical liberal blogger in SW England, Carl Benjamin, outlines the half dozen or more issues that consistently suggest Biden vote fraud.
https://www.youtube.com/watch?v=kHHsHZ5fnfk&t=1629s
The first 12 minutes are essential, but up to 32 minutes delves more thoroughly. “I feel like I in a sea of red flags [suspicious warnings]!” he exclaims. Benjamin then (37m) explains how cheating is immoral to the right, but for the left, it is a heroic quest! And (at 40m) goes through the BBC 2026 piece on detecting vote fraud in developing countries “How to detect vote rigging” the US checks off five of their six.
Will no one rid me of these troublesome YouTubes?
Give me a paper or a decent set web pages I can print out, then read all about election fraud in one place, at my leisure, and scribble notes in the margins!
Googling “election fraud 2020” I find:
_____________________________________
The Times Called Officials in Every State: No Evidence of …
http://www.nytimes.com
1 day ago — The president and his allies have baselessly claimed that rampant voter fraud stole victory from him. Officials contacted by The Times said that …
Barr’s Voter Fraud Decision Inflames Tensions With …
http://www.nytimes.com
33 mins ago — The department’s handling of President Trump’s dealings with Ukraine and a potential case against the former interior secretary exacerbated …
Voter fraud used to be rampant. Now it’s an anomaly. http://www.nationalgeographic.com
8 hours ago — Gone are the days when bribes and voting by voice were commonplace at the polls. Today secret ballots and improved security measures have …
Victims of Voter Fraud: Initial Announcement of Deceased …
http://www.donaldjtrump.com
13 hours ago — Having confidence in our elections means knowing that votes are legally cast and that voters themselves are legally eligible to cast a ballot.
US election: Justice lawyer quits after attorney general orders …
http://www.bbc.com
2 days ago — Donald Trump refuses to accept Joe Biden’s projected victory, and has made unsubstantiated fraud claims. The president’s campaign is seeking …
_____________________________________
Move along folks. Nothing to see here. Trump’s just lying some more.
Neo: Problem is, I believe, the findings were only for Trump voters and not for the Biden voters at all.
I’ve since watched the entire video. I don’t believe they said that only Trump, not Biden, votes were affected. In fact, I’m not sure how the algorithm they described could affect just one candidate. If the algorithm really does give to Biden what it takes from Trump, then we should see a complementary, i.e. reversed, effect in any Biden graphs.
Still, maybe I’ve missed something obvious. If so, please correct me.
Regarding the “straight lines mean it’s not natural” idea — Their later plots each include TWO straight lines, one nearly horizontal and one very steep. Two lines could indicate either (a) some algorithm that’s only enabled when enough Rep voters are detected, or (b) some honest data that doesn’t form such a straight line after all.
Regarding straight lines in general — There’s always a best fit line for any real world data. However, that doesn’t mean the best fit is any good at all. For example, see their Wayne County plot at 38:00. They show the best fit line, but the fit looks awful. We need to know the correlation coefficients to judge how good the fits are.
All in all, I think their claims are interesting and worth pursuing. Maybe with more work they could make this into a “full” red flag. Right now, I say it’s worth “half” a red flag.
Regarding the “email” controversy: I used 10 minutes to follow the links and get some idea about the controversy.
He didn’t invent email, obviously. email had been already invented a few years before. But he coined the term and created the first program that managed email in the way they do nowadays. I’m talking about email clients like Outlook or Thunderbird.
In a nutshell, he didn’t invent email, but he created the first email client. The proper credit should be “contributor”.
His claim that he invented email is unfortunate. On the other hand, having created the first email client is a contribution that should be (at least) credited, for example, in the wikipedia ‘history of email’ page, which it’s not.
“Individual candidate” voters in this case refers to those who vote, but not straight party vote. In a state that offers a straight party option, you can just check one box and that means you are voting for every single candidate that party offers.
But those are not relevant to what people are analyzing. They only need the small segment that only voted for one candidate per election, and left the rest blank. These are outliers that should be easy to isolate, if officials had not deleted the records and ballot pictures.
Even if we can’t get Nixon on crime, we can get him on the cover up. IT’s the cover up, so to speak. Al Capone, taxes, not the other stuff.
Good overview of where the EV numbers are today:
________________________________________________
The facts are, as of Monday, a full week after election day, Biden is behind President Donald Trump in the undisputed electoral count — Biden’s 226 to Trump’s 232 (assuming Trump wins North Carolina) — with a full 80 outstanding electoral votes in seven states still in a legal fog and unlikely to be determined much before the December 14 state deadlines to report the count to Congress.
https://issuesinsights.com/2020/11/12/look-closely-trump-still-has-a-good-shot-at-winning-reelection/
________________________________________________
I remain unclear on how the audits/recounts, lawsuits and state legislatures work together and send electors to the Electoral College. Is it different in each state? Probably…
I could use a flowchart or a cute cartoon like this:
–“Schoolhouse Rock- How a Bill Becomes a Law”
https://www.youtube.com/watch?v=Otbml6WIQPo
Re: Ayyadurai …
Yann: You are more lenient than I am. Here’s a summary from SIGCIS (Special Interest Group for Computing, Information) on the matter. It’s pretty brutal.
__________________________________________________
V.A. Shiva Ayyadurai is not a member of the MIT faculty and did not invent email. In 1980 he created a small-scale electronic mail system used within University of Medicine and Dentistry of New Jersey, but this could not send messages outside the university and included no important features missing from earlier systems. The details of Ayyadurai’s program were never published, it was never commercialized, and it had no apparent influence on any further work in the field. He does not “hold the patent for email” or have a copyright on the word email, though in 1982 he did register a copyright claim covering the exact text of a program called “EMAIL.” The U.S. Government has not recognized him as the inventor of email and he did not win the Westinghouse Science Talent Search for his program. Electronic mail services were widely used in the 1960s and 1970s and were commercially available long before 1980. To substantiate his claim to be the “inventor of email” Ayyadurai would have to show that no electronic mail system was produced prior to 1980, and so he has recently created an absurdly specific and historically inaccurate definition of electronic mail designed to exclude earlier systems. Ayyadurai has not even been able to show that he was the first to contract “electronic mail” to “email” or “e-mail” – his first documented use is in 1981 whereas the Oxford English Dictionary shows a newspaper usage in 1979. Despite Ayyadurai’s energetic public relations campaign, which presents him as the victim of a racist conspiracy financed by corporate interests, he has not received support from any credible experts in email technology or the history of information technology. His claims have been widely debunked by technology bloggers and articles based on them have been retracted by the Washington Post and the Huffington Post.
–“Did V.A. Shiva Ayyadurai Invent Email? A Computer Historian Responds”
https://www.sigcis.org/Ayyadurai
__________________________________________________
SIGs (Special Interest Group) have been a primary vehicle for sharing information and discussion about computing since 1961 under the umbrella of the ACM (Association for Computing Machinery). Not fringe.
Sadly, Ayyadurai has been waging a campaign to justify his email claim which includes accusations of racism. From his own blog:
–“Yes, A “Darkie” Invented Email. Get Over It.”
https://vashiva.com/yes-a-darkie-invented-email-get-over-it/
Off the cuff responses to a couple of the comments above:
First, I believe you can get the corresponding graphic for Biden voters by suitably reflecting the X and Y axes of Dr. Shiva’s plots.
Second, looking at the plots, I’d say the correlation of Shiva’s X and Y variables has to be better than -0.9. This is just an impression based on a lot of time in data analysis.
Third, we did have a report of a Republican county where 6000 votes were shifted to Biden. That anecdote would be consistent with this analysis.
After wasting much scratch paper and more time, I have convinced myself that the situation described in the video can happen if votes are counted correctly. This can occur if the frequency of voters who vote straight ticket is large enough so that virtually all ticket splitting Republicans are anti-Trump voters.
Of course, this does not rule out chicanery, and, in the situation I have described, there are restrictions on the slope of the linear part.
It would help to know something about the proportion of ticket splitters in this dataset. If that proportion is low, then this result looks much more like the software problem described in the video.
H, that is the strongest emotional reaction i have detected from you in some time.
The name huxley belongs to the novels of ald huxley, as you seem to be a fan of. I wonder why you are so attached to this name, that is not yours. I use 3 letters because it is faster to type and not mispell on galaxy 10. Even the prez of usa, i shorted to 4 letter, trum.
I even short my own to 4 letters. 3 is certainly friendly enough, so where does the negative emotion come from?
Ymarsakar: Typing my full login name, as I chose it and now request it, should not be a hardship. It’s common courtesy. Perhaps you’ve heard of it.
My emotions are none of your business. I have no interest in your advice or analysis.
Nobody has responded to my comment on the video.
The only reason a Trump voter would request a “choice” ballot would be if they wanted to vote for a Democrat down-ticket. Otherwise they would take a straight-ticket ballot.
In a heavily-R district, wouldn’t it make sense that it would be more likely that a voter would want to vote a straight ticket? And therefore there would be fewer choice ballots in that district.
Also, the null hypothesis is that in the absence of fraud the choice ballots would be roughly equal to the straight ticket ballots, but he doesn’t show us any examples, so as far as I can tell that’s pure speculation.
I think there was fraud, but I don’t think this is it.