A “Transistor Function” algorithm has been used to alter voting patterns in Michigan. It has a “Weighted Race” feature.
Dr. Shiva Ayyadurai, an MIT engineer and Fulbright Scholar, with Phil Evans B.S.E.E. and Benny Smith (election commissioner and data analyst) have back-analyzed the pattern of voting in Michigan and figured out the electronic algorithm used to alter votes. Voting patterns are distorted in a way that can only be explained by a linear transformation (an algebraic equation) and he can approximate that equation and slope of the line. This means he can calculate how many votes were flipped and he does, and it’s massive.
This one form of fraud alone is enough to flip the election to Biden. That’s without all the postal vote backdating, the dead people voting, the out of state votes, the discarded ballots, the crooked media coverups, the pollsters fakery and the Pfizer hiding of the Covid Vaccine news. Trump won the election despite all the other trickery. If there had been real media coverage, honest vaccine headlines, exposure of Hunter Biden and honest polling the election probably could have been called by 8pm on Election day.
The action analysis starts at 13:30 minutes.
In disconcerting news, votes are stored as a decimal fraction. They are not even trying to record votes as whole numbers, as individual choices.
If the video goes missing, look here
What does it mean to be a citizen if the voting inputs and outputs are unverifiable.
It reads as rather melodramatic, but in a factual sense, if we accept rule by unverifiable and unaudited mechanisms, we are accepting that someone else has the power to chose who rules, not us. So when he says “It’s time people got educated or they will be enslaved.” It is merely an academic and obvious corollary.
The title of the presentation: A Mathematical Analysis of Trump-Biden 2020Election Vote Counts in Four Counties in the State of Michigan.
As I understand it, his analysis looks at the “down ticket” voting patterns in individual precincts of four counties in Michigan. The unnatural pattern is that the more Republican the precinct, the less likely they were to vote for Trump. Thousands of Trump votes were given to Biden in strongly Republican precincts. It’s possible that Republican voters could be fed up with Trump and choose to vote Biden for President, and the Engineering trio consider that. The red flag that is unarguable is that the ratio of Republican voters who chose Biden over Trump should be the same across precincts, and it is in precincts which are predominantly Democrat. But these engineers found a “transistor function”. After the percentage of Republicans rise in a precinct beyond a “cut off” point, this ratio changes, but it changes in a perfect and predictable linear way. The ratio depends on the percentage of Republican voters in a precinct. The more Republicans there are, the more likely they are to abandon Trump and vote a Biden-then-Republican ticket.
The dashed orange line should be a flat line, instead the downward slope “cuts in” and votes are increasingly distorted in precincts where more Republicans vote.
Adapted from the video presentation above.
The slope of that line is “too perfect” — it’s almost perfectly linear. The transistor effect kicks in and shows that the same algorithm was used in different counties across all precincts. The pattern is non-random. Even if you wanted to believe that Rep voters were tired of Trump it would not happen in a perfect line that depended on the number of Rep voters around you. (45 mins.)
So most of the votes were stolen from strongly Republican areas and flipped for Biden. Unless this is fixed, in future elections, they (whoever they are) will just change the algorithm to alter down ticket votes too.
I’ve only watched this once. If I have misunderstood any part of this, please correct me in comments and I’ll update the post.
There is a “too perfect” decline — that is linear. It suggests electronic manipulation — the only thing that can create the artificially, improbable pattern.
The same slope occurs in Early voting patters as well as Election Day voting patterns.
Trump in Macomb County is 12% to 22% more popular than the Republican party candidates in predominantly Democratic areas. In Republican dominant areas this effect reverses.
In WAyne County there was no cheating and there the pattern is totally different. Trump is universally more popular than his Rep colleagues, not less. Trump is leading people to vote Republican who otherwise chose non-Rep candidates.
Questions:
Are both parties are aware of these features? Are both Reps and Dems supporting the algorithm?
- Is a coup underway?
- Who is behind the coup?
- Did Biden win?
50 How do we get out of this (for future elections)?
To be a democracy, certain minimum standards must apply to elections:
- A Voter Registration card
- Open source software.
- Hand marked paper ballots records. The vote intent is obvious.
- The inputs must be verifiable.
- The ballot images must be stored for the long term in a legal format. These must be saved. (Some states are throwing the files away.)
- All elections should be audited automatically. Not just disputed elections.
The man is a hero who has been battling for fair elections. For asking where the ballot images were in previous elections he was taken off Twitter. No lawyers would defend him. He had to go to court and defend himself. The Federal statute was created saying that States must store the images.
Listening to trunews.com, assuming they were correct, it appears the voting machines may have had security issues.
If what was being said was correct, the manual for the machines apparently was also on the internet and it indicated how to remove batches of votes if duplicates occurred…..
Um…hang on….
Presumably if its also a centralized system, it would run a database. Databases by thier nature allow for inserts, updates and deletes.
How good was the system tamper protection and auditing? It should have a detailed audit log.
Forensic software can recover the last 20 deletes off a single computer disk. Memory can also be recovered.
And apparently the software is chnese made.
Interesting times….
281
Both sides of the house and the middle are corrupt. Donald Trump is an outsider. They have been trying for four years to get rid of him.
The swamp is deep. They believed they could control Trump because he is not a perfect individual. Hence ALL the investigations to find dirt. After five years of trying,….. nothing sticks.
They MUST have a compromised President. Otherwise blackmail is not possible.
Biden is the PERFECT US President.
502
Thank you for this insight. Of course the vote database has to have ability to do corrections and audit log everything.
Question now is whether these changes are legitimate. Bipartisan approval is required.
20
“Apparently” the software is Canadian/US made, if you are thinking of DOMINION.
WIKIPEDIA:
Dominion Voting Systems Corporation is a company that sells electronic voting hardware and software, including voting machines and tabulators, in the United States and Canada.[1] The company’s international headquarters are in Toronto, Canada, and its U.S. headquarters are in Denver, Colorado. The company carries out in-house software development for its customers in the United States, Canada, and Serbia.[2]
Dominion was the subject of a hoax originated by followers of the far-right QAnon conspiracy theory, later spread by One America News Network, President Donald Trump, and Trump surrogates and supporters. They alleged that the company’s voting machines had been compromised, resulting in millions of votes for Trump being deleted or switched to votes for Joe Biden in the 2020 presidential election. Some made false claims that Dominion had close ties to the Clinton family or other Democrats. There is no evidence supporting these claims, which have been debunked by various groups including election technology experts, government and voting industry officials, and the Cybersecurity and Infrastructure Security Agency (CISA).
Also apparently, an MIT engineer, an election commissioner and a data analyst are spreading QAnon’s conspiracy theory /sarc/.
11
Slightly OT but related….
The swamp are also panicking…i saw a clip whereby ex-CIA Brennan [snip] appeared to be asking Pence to invoke the 25th to get rid of Trump.
What are they worried about coming out?
Trump has just purged the military of what appear to be swamp critters…..
Put all the bits together…will we see an internal round up and crackdown with the apparent vite fraud as thr last straw?
301
Donald Trump Jr. is allegd to have said that his Dad should declassify EVERYTHING!
That is what Brennen and Clapper and Comey and teh Clintons are so afraid of.
391
No wonder they are wetting themselves..
DJT is a one man truth enema, specifically useful when deployed against swamp critters…
261
No alleged about it Karabar, he said it in a Tweet.
He also said very soon after election day that when the American public saw what they had found they would be disgusted.
41
If he does that, it couldn’t have happened to a nicer bunch of people.
60
Is a coup underway?
No.
Who is behind the coup?
That is a leading question your honour, the deep state is inscrutable and unknowable.
Did Biden win?
Yes.
359
Assuming it was a fair election, but it now appears it wasnt..ergo…no.
581
It now appears it was fair. Federal election commission called it the fairest and most scrutinised in history.
Ergo…yes. This is easy.
110
Sure. 100%. No bureaucrat has ever tried to cover their own failings or said something self evidently false with a straight face.
What does “most scrutinized” mean when there is video footage of Republicans being denied access to do scrutiny?
131
El Gordo
After that statement I will have to re-assess anything of yours with which I may have agreed in the past
410
Like something to do with climate change?
Are are discussing American politics, a poor foundation for exceptionality?
119
0 chance.
367
He hasn’t won. He’s been declared a winner by the media. They have no power. There are states in play that haven’t officially been called and are in dispute.
60
More and more,you seem a fan of opaque state entities having central control.
150
Is a coup underway?
Absolutely and it started 4 years ago when the left weaponized hate as their primary political lever.
Who’s behind the coup?
The Democrat leaders and their financiers who are afraid that Trump is their existential threat and will expose them for what they are. In fact, he already has.
Did Biden Win?
Not yet. It remains to be seen if enough of the corruption that distorted the election can be uncovered to flip the results and allow truth to prevail over hate.
242
The vagaries of the US system leave it open to abuse.
Trump should have won, there was an increased groundswell for him, but the other side had gained the support of the MSM and demonised Donald. Even with all the extra effort to encourage people to vote, a third of the electorate still refused to vote.
11
I have stepped away from some of the more unlikely and extreme conspiracy theories circulating but give credence to the notion that Pfizer deliberately withheld their announcement until after the Election. This is explored in the article here
“Pfizer vaccine completely unworkable”
https://lockdownsceptics.org/
The timeline is suspicious, but the very notion this vaccine will do the job also seems unlikely and Pfizer are accused of trying to drive up their share price by some and derailing Trump by others.
The NHS has many faults but it does have the structure to carry out mass vaccinations, but it is being realised that the prime method, through GP’s or hospitals, is impossible, as they do not have the facilities to move, store or deliver a vaccine that needs keeping at ultra low temperatures.
91
Maybe they should talk to the people who do artificial insemination of cattle.
The semen straws are stored in flasks of liquid nitrogen at a temperature of -320 degrees F (-196 C) and only thawed out when they are about to be used.
50
If just 10% of the claims of vote tampering are true we are in any outcome is invalid territory.
200
Looks like Biden got My Mann to write the algorithm and in return he gave him the science advisor gig in the WH
230
Yes, a reverse obverse hockey stick!
20
Gee! Eh?
80
Obviously a very saleable “feature”
“The Algorithm is a weighted race allocation method where the % is a weighted decimal vale and Dr. Shiva predicts he will be able to calculate the % based upon more data
Weighted Race is a documented feature in elections as early as 2001
ALL Major Vendors of election counting ( not just Dominion ) have this feature
Diebold had the original feature with the subsequent vendors using the Diebold source code @ 43 minutes”
(My bold there)
http://www.smalldeadanimals.com/index.php/2020/11/11/november-11-2020-reader-tips/#comment-1373223
110
Dominion bought out Diebold after the last election !!
30
Bloody Hell!!!
“Frank Gaffney is on speaking about his new website https://everylegalvote.com/country
The website allows you to visually change the fraud vote vs non-fraud vote counts by state down to the county level.
He also explains the cyber-manipulation of votes.
28 states are outsource their voting tabulation to Barcelona, Spain. The servers are tied to George Soros….”
http://www.smalldeadanimals.com/index.php/2020/11/12/november-12-2020-reader-tips/#comment-1373377
110
When it gets to SCOTUS perhaps the logic may go as follows:
1. The constitution and most State laws require that votes be scrutinised by both sides if the vote count is to be legal.
2. In those States where scrutisation by both sides was not allowed, the counting of votes from the time of such disallowal was therefore illegal
unless the votes were separated and re counted later with full scrutiny,
3. Given that late votes were not kept separate, as required by at least one SC Justice, and were deliberately mixed with earlier votes,
it is not possible to check whether these votes were legally counted.
4. Whether such votes were legally cast is moot, they may or may not have been. There is now no way of knowing. however it is certain that such votes were illegally counted.
5.The above situation places the outcome in so many States that it is impossible to to properly declare the outcome.
6.In view of the above, the determination of the outcome passes to Congress, as foreseen in the US Constitution. In this situation the Congress votes as fifty States, each State having one vote as determined by the majority in the Congress from that State
190
The problem is newly elected members of the house have been elected in the same fraudulent vote.
71
As I understand it the old house had 26 republican States as so determined
100
Fake democracy!?!? Ha ha. That sounds like fake news. Has the fake President left yet?
In the world of Trumpists, all is fake. Nothing is real. MAGA.
Count the votes again, Jo. I’m fine with that. But respect the will of the people. Not the will of Donald.
Fair? Or can Trump simply NEVER lose?
24
If the Dems were interested in healing they’d deal openly with the accusations of fraud instead of burying them, censoring them and bullying the whistleblowers.
If they wanted Republicans to respect them, they’d fix the voting rules in Democrat states. Indeed, they could run the whole election again with clean voter rolls, and only paper ballots. Would you accept that result?
It’ll never happen. The Dems need the dead voters, the multiple mail-ins and the electronic machines.
141
Sadly, so true. However, spreading fundamentally flawed election fraud theories, such as Dr Shiva’s theory, will only damage the ongoing genuine effort to fix the US election system.
11
What genuine effort?
20
1. Trump Legal team is actively conducting data analysis to identify fraud in the 2020 election.
2. Experienced Data analysts and pollsters (e.g. Richard Baris from People’s Pundit) spend a lot of time analyzing the election results with this in mind.
10
If the Dem’s were truly interested in transparency it would be trivially easy to publish all of the ballot images with names redacted. Each ballot image captured by machine will contain the timestamp of when it was entered so that the batches could be recreated, along with all of the votes. Additionally one could easily see if ballots were filled in by machine, or if consecutive ballots were marked by the same hand. Similarly all the ballot images could be published with signature attached but with votes redacted. Then all of these signatures could be compared to voter registration cards. This entire process could be done in the open and would be a great exercise. The most troubling thing about this whole charade is that not a single high profile dem has come forward to say sure we won and we can easily prove it and here is all the data. Why not ????
20
Some Dems were actively involved in the fraud. See the latest news about the role of the Dems in this project: https://www.youtube.com/watch?v=sq7TeUJwQD4&t=1175s
Sadly, it looks like some Republicans (e.g.. in Georgia) are also involved.
10
FYI, just in case you missed it, the latest update on the genuine effort: https://www.youtube.com/watch?v=sq7TeUJwQD4&t=1175s
00
Which of the “big tech” crowd provided the algorithm & who are the shareholders of the electronic vote counting tabulators used in the 5 states in question.
Heard a RUMOUR that Pelosi & 2 other Democrat supporters are major shareholders. I hope someone is investigating!
Obviously this was a planned fraud on the American people and probably took over 1year to organise & test. There will be evidence somewhere!
170
This is amateur hour stuff, can’t decide if it was ill planned by idjots or they simple don’t care if they get caught because they think they will get away with
181
Ignore red, see green.
21
“ Obviously this was a planned fraud on the American people…”
Yup. ‘cept that 1-year planning period you mention is off by, let’s say, ‘bout 3 1/2 years.
Just as soon as it became highly probable that DJT was gonna win the R primary *thats* when it began.
When he won that primary thats when it kicked into high gear, and Hitlerery got going in earnest.
I now, again, refer you all to watch the film “The Plot Against the President”. I know you guys in OZ can get it. It’s on many different platforms. If you’re an Amazon Prime member it’s $3.99 (US) to rent, $14.99 to “buy” (unlimited access).
It’ moves *very* quickly with tons of behind-the-scenes people whom you may have no idea about. So I’d consider buying the film so you can do your ID and background checks on the prime movers in the film.
It is really well done.
100
I remember early in 2016 the media certainly treated Trump as a joke candidate until it became obvious he wasn’t kidding. Very quickly he went from being just another wacky celebrity candidate to “The New Hitler”.
20
Hoh ho…. well now, if they could alter figures like that so easily and quickly ( don’t think it was an ‘over a cup of Coffee job’ on the back of a ciggy pakt) this has all been planned: Begs the OTHER question:- where was it previouusly tried out / manipulated ? Vaccine development, Viral infection models, Hey – CLIMATE Models…………?.
You CANNOT trust what you See, hear, or read …. What to do…….. Hmmmmm and the smell is getting worse, MSMs are flying around biting, winter precipitation is coming and the swamp will likely grow. GET THE DIGGERS OUT and DRAIN THE SWAMP, kill the MSMs.
120
Drain
The
Swamp.
140
There is a theory out there that it may have been tried out on the selection of Biden against Bernie.
130
Watching the vid up to floating point numbers…..you need to convert ASCII etc to FP so as to apply coefficients which some people call weightings
40
I’ve only watched this once. If I have misunderstood any part of this, please correct me in comments and I’ll update the post.
Are you on the legal team
017
Did they do the same analysis with Democrat data?
023
Did you watch the video incompetent fool?
191
Is this actually Crakar or are you channeling the ghost of Andy?
03
I watched the video and they did not replicate analysis with Democrat data. Someone else has though and it shows the same pattern except it is Republicans stealing votes from democrats. The analysis Dr Shiva did was a scam.
28
Where has someone else? Please, we’d like to see that. It’s easy to say it’s been done, and declare it a scam but where is that analysis?
[Note to readers that Crackar25 is NOT Crackar24. Crackar25 needs to pick a more unique name that does not appear to be a deliberate attempt to mislead. – Jo]
30
It seems that taking the difference of percentages is the cause of the odd-shaped plots.
https://www.youtube.com/watch?v=aokNwKx7gM8
00
well it wasn’t me… though I’m sure you know that but letting the other punters know.
Anyway… statistical distributions and their analysis are not trivial and this one has failed.
The big failure though was the lack of skepticism, though I note that this blogs skeptical credentials are not exactly well advertised any more.
01
Dr Shiva’s theory is formally debunked (numerically and mathematically) in this video
https://www.youtube.com/watch?v=aokNwKx7gM8
The theory is based on a fundamental flaw. Inded, it can be used by Biden to show that votes were taken from him.
20
Here’s a long response, but it’s length is necessary for clarity and completeness:
“Someone else” is Matt Parker of stand-up Maths on youtube—but I doubt one critical aspect of his evaluation: his x axis. I *think* that *if* Dr. Shiva, Phil Evans, and Bennie Smith (a democrat) are correct, then Matt Parker made a mistake when changing his x axis when he evaluated Biden/Democrats on the y axis. As I understand it (but perhaps I am wrong?) the x axis needs to be the same for both data plots, whether the y axis is Trump/republican or if the y axis is Biden/democrat. This is because Shiva and the others are comparing y to x. And x represents “how republican the precinct is.” So you also need to compare Biden/democrats (y axis) to the same metric of “how republican the precinct is” (x axis). Analogy: You can’t compare three to four, and then compare seven to eight; you have to also compare seven to four, then you have an evaluation of how both three and seven compare with four. Using the same x axis would then find the slope going up when putting Biden/democrat on the y axis.
Two data plots that illustrate: https://pasteboard.co/JAwiqVt.jpg
This image link is not mine, but was linked/posted by someone in the comments to Shiva’s video—there are so many comments I lost where it is, but pretty sure it was in a reply to one of the current top ten or so comments. The data plot creator simply said something like these charts explain the matter, and it was in response to comments/discussion about Matt Parker’s explanation. At the least, these images intend to point out the x axis needs to be the same, but might also intend to illustrate something that is obvious to someone better trained in math than I am.
Me being a fair textbook mathmetician, but not with experience formulating real life data to determine/prove a point, I could be wrong and am still trying to fully understand the two arguments (Shiva’s vs Matt Parker’s). But, this x axis point does seem correct. I think. Maybe. Let me check the back of the book for the answer.
10
EDIT TO MY PREVIOUS: Now realizing—my math brain comes around eventually—that because Matt Parker shows that you can produce a downward slope for Biden, then that would also “prove” (according to the point being made in Shiva’s video) that Biden is having votes taken from him the “more democrat a precinct is”. So, according to Matt Parker, that makes the Shiva presentation innaccurate—you can’t have it both ways, votes taken from from Trump in more republican precincts and also have votes taken from Biden the more democrat a precinct is. (At least, if Matt’s data plots are being created accurately. I have yet to evaluate that closely enought to know.) That’s now what seems to me to be what Matt Parker is showing. But the back of the book doesn’t have the answer for this one 🙂
NOTE: this is not to say there can’t be something amiss with the voting machines (maybe there is, maybe there is not) but only says that Shiva’s presentation would be innaccurate. There is of course, the issue with observers not being allowed to closely watch the vote counting, dead voters, etc, which is pretty clear has occurred in some places.
10
Electronic but O/T
Chiefio surveying what is around outside the ysm
“Beyond Fox”
https://chiefio.wordpress.com/2020/11/12/beyond-fox/
20
I really really disagree with “How do we get out of this (for future elections)?”. This election matters. ‘We’ need to get this sorted out now, so that we get this election right.
330
The fact that they convert to floating point says they are manipulating data. Essentially you have integer say 56432 when converted to floating point it becomes 56432.0 and allows to to get decimal point numbers.
It looks like it works based on % of votes perhaps over a time period. It’s called transistor because it acts like a switch
90
The statistical arguments give an indication of the scale of the election F…
But they do not count in a court room. Brave postal workers are risking their employment to give actual evidence.
There must be a lot more. An operation to achieve the scale of voting manipulation that has been suggested requires alot of people, maybe thousands of people.
That means there should be a lot of communication. The FBI likely already has it. Will anyone break ranks there?
131
Perhaps Trump (and the many presidents before him) have failed to impose secure and reliable voting systems across the USA because they have all realised that fraud breaks both ways although they often criticise nations who don’t even have superficial excuses for a democracy. Corruption exists on unknown global scales and money plays a huge role in the subterfuge of the popular vote. Sadly that corruption has now entered science, academia and our media, with free speech endangered.
The first priority for the next US President (whoever it is) is surely to clean up that democratic mess, insist on standard voting methods across all States and Counties, with proper policing and much stiffer penalties for anyone in politics or business found to be cheating in any way at all. If we voters cannot get politicians to secure our democratic processes then we really have lost the plot. Doesn’t matter who does it as long as it is done forensically and properly
100
US presidential elections are run by the legislative assemblies os each state, not by a Federal body. It’s written in their constitution to limit Federal powers from centrally manipulating elections so presidents are essentially powerless in this regard .
90
I agree that the President has little power to intervene with state electoral processes but I think Trump, if re-elected could and should establish a fixed set of electoral standards and procedures which the states could be encouraged to follow in the interests of future fair, honest and standardised elections.
30
It’s worth a thought but he will have to work hard to convince 50 competitive scatterbrains … and they’ll all probably want their own curlicues and friccasees …
10
I think this relates to what he was talking about. In the week before the election Paul Murray gave figures for the mail in and absentee votes by voter registration and the rough rule of thumb was that because of the anticipated number of on the day Trump voters Democrats would need to get a ratio of 2/ 1 of the mail in ballots to have a chance of winning. All the states ( ironically except for Pennsylvania ) were no where near 2/1 in favour of Biden registrations. Yet on the night it’s clear that the counting of the mail in ballots were not consistent with the voter registrations ( which were official figures) meaning either heaps of Trump supporters voted for Biden or 100% of independents voted for Biden ( these are more likely to split 50/50.
Based on the prepoll voter registration of almost all the swing states Trump should’ve won all of these comfortably unless there was fraudulent activity / or heaps of glitches.
I must say I have been suspicious about a fix being in for a long time such that a Biden win orchestrated the way it has been was not only feasible but considered likely. The polls for one seemed to lean to a certain victory. This was to act as a reinforcement of the result2. I read an article which predicted virtually the exact scenario and that Trump would lead on the night but would be overrun by Biden by the mail in ballots. 3. They kept asking Trump beforehand to accept the election result making it look dodgy if he didn’t even though Biden said that he wouldn’t claim victory until the result was certified by the electoral college.
It appears to me that what has made the fix in relation to the election a bit more difficult for the Biden camp has been the sheer size of the Trump vote. This has meant that the Democrats have had to coerce / manufacture many more dodgy votes to stay in front. Not just tens of thousands but hundreds of thousand and that’s where instances such as the 180,000 Biden only votes in Wisconsin and the 130000 99% Biden votes came through In Michigan in the middle of the night totally altering the dynamics of the results in those states. Even if these ballots had come from a democratic stronghold one would’ve reasonably expected about 30000-40000 Trump votes. With the results in those to states relatively close one would like to think that some sort of audit ( as well as count) should be undertaken.
The way this has gone it’s clear for the integrity of the election and Biden’s legitimacy there needs to be recounts and audits in those states where there is reasonable suspicions about the conduct of the vote . Ot
231
“How do we get out of this ?”
The simple answer is to use what works – pen + paper + manual counting + double checking.
Any sort of technology is open to errors/fraud.
170
(Long very sore sigh…..) Yes……. tell the younger generations ….. and the old trendy foaggies as well ( ultra liberal ? ). Technology for the sake of technology never did anything economically or environmentally worthwile – only aids in getting “lazy” folk to Play at working. as we farmers may say ‘spread the shite for others to collect’ !!
90
It looks suspicious – especially given that it is apparently a design specification feature of the software! But with software bugs and/or user error, you can get the illusion of order/design, it’s very easy to come up with a scenario whereby an error only occurs when a particular level is reached and then iterates proportionally to something.
90
Mistakes are random in their effect
20
Hi folks, don’t be fooled: The vertical axis of Shiva’s plot is a counterintuitive variable, and he fools himself as you can see at 45:10 in his video: The Mitt Romney example is simply wrong, in fact that example itself corresponds to a line with negative slope. To see this, assume a fraction alpha of all voters vote for the opposite president than the party they vote for, say alpha=0.2 for example. Assume we have N voters voting for democrat and republican parties N=n_d+n_r. The republican presidential candidate then gets p_r = n_r (1-alpha) + n_d alpha votes. The vertical axis is then (p_r-n_r)/N = alpha (1-2 n_r/N) For alpha = 0.2 is this for example: 0.2 – 0.4 n_r/N Voila! there you have your linear curve with negative slope.
73
“How do we get out of this (for future elections)?”
We need to get out of this NOW, because if we don’t, we may never get another chance in any future election.
100
“As I understand it, his analysis looks at the “down ticket” voting patterns in individual precincts of four counties in Michigan. The unnatural pattern is that the more Republican the precinct, the less likely they were to vote for Trump.”
This is not correct. They are comparing voting percentages on “straight-party” tickets versus “split-vote tickets”. This is not the same thing as comparing votes for President versus down ticket voting patterns. In fact, they do not look at how the “split-vote tickets” voted down ballot. They examined patterns in votes for Trump versus Biden in the two types of ballots.
More explicitly, they put on the horizontal axis the percentage of “straight party” ballots that were Republican. On the vertical axis they put the DIFFERENCE between the percentage of non-straight-vote tickets that chose Trump for President minus the percentage of straight-vote tickets that were Republican.
Fitting a line to the data across precincts, there might be some explanation for a downward sloping line in such data. The idea would be that as the percentage of voters choosing a straight-party Republican vote goes up, more of the people splitting tickets would be voters who favored Republicans but did not like Trump. Their dislike of Trump is why they split their ticket. Conversely, as the proportion of straight-party votes for Democrats goes up (you move back toward the origin on the x-axis), the ticket splitters would tend to be more people who generally vote Democrat but they wanted to split their ticket to vote Trump for President. This is shown in the video as a single downward-sloping line with precinct data scattered around that line.
The suspicious thing here is not that there is a downward-sloping line. It is first that there seems to be two straight lines with a kink. You would expect a more gentle curve than a sharp “knee bend.” Second, the downward sloping straight line seems to have the same slope across different counties. In a more natural outcome you would expect the downward sloping line to vary more in slope across across counties.
80
As a follow up to my comment:
Looking again at the Oakland County results in the head post, I have some doubt that the difference between a single straight line fit to the data versus the horizontal then bent straight line would be statistically significant. Whether or not I am right, if they want their analysis to be taken seriously by professional analysts they should fit a single line and test for statistical significance of that fit versus the “knee-bend” pair of lines. Similarly, with regard to the second point, they did not give the slopes of the lines in the video. They looked similar, but I would like to see the actual numbers. They should present them and again test whether they are statistically significantly different from each other.
I understand this is a video for a non-professional audience. However, to take it to the next step, formal statistical analyses need to be carried out.
40
Peter Hartley, this would seem to be fairly easy to test for if the original paper ballots have not been destroyed: randomly select several precincts and hand count the paper ballots. Compare to machine recorded counts and look for any patterns that develop. While at it cross check the number of voter signatures in the test precincts. (to attempt to rule out ballot stuffing) The only thing this would not be able to determine is fraudulently replacing (and destroying) Trump ballots with Biden ballots.
If the paper ballots no longer exist then it is a big problem for the election officials because Federal law insists they keep ALL records for a period of time.
30
Interesting theory, but I’d like to see the same analysis and graphs for Biden’s votes. They’d have to show Biden under performing in Dem heavy precincts and over performing in GOP areas. I.e. Slope of the linear trend in the opposite direction. A lot more could be done that might either support or debunk this claim. I’ll wait for more info before getting excited.
30
Good point Rick. In fact, this theory is fundamentally flawed. See a formal analysis at
https://www.youtube.com/watch?v=aokNwKx7gM8
The graph for Biden has a very similar (negative) slope!
10
Three modern Presidents have upset the Deep State.
1) Kennedy. It didn’t work out well for him.
2) Reagan. They tried to kill him.
3) Trump. Look what they’ve done to him.
60
You can add Dr Ron Paul to that list in the GOP primaries the year Romney lost
40
Dr Shiva has tweeted out to both the Biden and Trump camps to say the are happy for anyone to check their methodology and data. So they are being very open about it.
50
The theory is fundamentally flawed. See my reply to Rick (#28)
00
As I type this, the numbers for Michigan are as follows.
President Imposter Biden 2,790,648
Real President Trump 2,644,525
Difference 146,123
Less 138,000 fraudulent votes, 8,123
So if there is 138,000 detected fraudulent votes in just four counties and there is then only 8,123 votes in favour of Biden, once fraudulent votes in other counties are detected, it will be a win for Trump.
51
Michigan gop senators request full audit before results are certified….the list of items to be investigated is the same as in PA, fraud everywhere you look.
The Georgia recount will be a good test I think 17k votes there were flipped
41
Gotta say, it’s kind of smart- switch as many Republican votes as you can get away with to Democrat without changing the outcome in a county, so each county “should” still come up as Republican but the extra stolen votes will tilt the entire state to Democrat.
Might have got away with it if Antrim county didn’t get flagged. The glitch was that they switched too many votes in too small of a county.
50
Good to see this getting some coverage.
But there’s still a long way to go… And “… There’s many a slip between cup and lip”
10
CISA official (election security) just fired after CISA apparently release statement that there was no previous electronic voter fraud and the 2020 election was the most secure EVAH!
40
If I understand this correctly, the assumption they make is that the %R/D split represented in the Straight Party Voting (SPV) can be accepted as representative of the precinct in general and therefore any significant deviation from that %R/D split in the Individual Voting (IV) is unexpected – or if there IS any deviation it should be a consistent deviation across all precincts ie. IV voters might be x% more or less likely to vote for Trump but importantly there is no reason that would change depending on %R/D split in the precinct.
It’s a compelling argument but, playing devil’s advocate, I have a few questions.
It would be nice to know % mix voting SPV vs IV? Is that split reasonably consistent across the precincts or does it vary? If it varies, is there any correlation between %R/D support and %SPV/IV voting. (eg. if IV voting is more prevalent in precincts with greater Republican support, could this potentially explain the pattern of increasing deviation as Republican support increases). I think this is unlikely but it would be good to eliminate that possibility.
I would also like to see more examples of ‘normal’ (no cheating) patterns to be fully convinced. The only example given (Wayne County) was an extremely Democrat county with hardly any precincts with Republican vote above 25%. The algorithm they describe supposedly doesn’t kick in until Republican support is above 20-25% so it is not necessarily inconsistent with the previous patterns. Comparison with patterns from a couple of counties in a safe Republican state (with both SPV and IV voting; and no evidence of tampering) would be very convincing if they produce the horizontal trend line predicted.
Also for those who would like to see a working demonstration of what can be done here is a video with Benny Smith (from the video above) manipulating results from a voting database using a program he wrote.
30
Not sure what happened to the link. I’ll try again …
https://www.youtube.com/watch?v=Fob-AGgZn44
10
The same types of plots have to be made for Democrat Straight Ticket and Biden. If they look the same, then the plots shown in the video have no particular significance, since the new plots would suggest that votes were shifted from Biden to Trump. This would mean that the interpretation from the plots in the video is wrong. If the Democrats/Biden plots look very different, then you would have shown something worth looking into further. Please supply the new plots for the same counties, and then we can have a more useful discussion.
30
Thomas, I would be very interested to see those plots. have you got a link?
10
See at https://www.youtube.com/watch?v=aokNwKx7gM8
00
well duh. I made that comment above. Yesterday.
Besides the 138000 vote output has been explained and is not the on the ground account so who really cares?
14
So, what you’re saying, G.a., is that criminals wouldn’t think of committing a crime?
This is criminal activity, approved by criminals, for the purpose of committing a crime.
https://www.thegatewaypundit.com/2020/11/breaking-outsiders-usbs-vcards-allowed-pennsylvania-counting-areas-no-observers-present/
What part of “that’s NOT legal” don’t you understand?
31
the part where nothing happened.
22
I wrote a message to Dr. Shiva Ayyadurai on his Web site. I do not have the data, so I cannot make any plots. Perhaps the original source was Bennie Smith. I do not have a contact method for him (yet, but perhaps will find one). It seems that it would be difficult to find Phil Evans, based on a brief search online.
By the way, I happen to have three engineering degrees from MIT, and I was very interested when I saw that an engineer from MIT had a video on the subject. I am skeptical for the moment, because I have dealt with data for many years, and I know that it is easy to misinterpret a trend if there is no further investigation to determine significance.
20
Dr A. C. Grayling, Professor of Philosophy and Master of the New College of the Humanities, London, in his 2017 book “Democracy and Its Crisis” concludes with some
This extract forms less than 0.5% of the material in the book (less than one page from 215 pages excluding the Table of Contents, Index and Preface.).
Grayling, A. C :”Democracy and It’s Crisis.” © 2017, One World Publications, London
ISBN 978-1-78607-298-4
The book has an interesting bibliography but you can acquire your own copy to access that.
20
Hmm, he’s author of a stack of books but one title jumped out of the stack at me:
The Art of Always Being Right.(Schopenhauer) 2004.
Interesting 😀
00
AS far as I’m concerned, in some way the USA was asking for this kind of situation. If a republic is going to have 50 state elections where there is no common thread with voting rules and procedures, then it deserves what it gets. We now have the ridiculous situation where, on December 14, the 50 Electoral College states get together and confirm who they voted for. It is clear to me who rules the USA….the States, not the Senate or Congress. As for fraud, let Trump prove it. I am sick and tired of the whispering conspiracy theories floating around…DONALD, PROVE IT !!
14
Congratulations Jo, that cool $1,000,000.00 must surely now be coming your way:
https://www.foxnews.com/politics/texas-lt-gov-dan-patrick-1-million-reward-evidence-arrest-conviction-voter
If not and you’ve simply been fooled by the Red Haze of election day and #LoserTrump’s own brand Lügenpresse
https://www.commondreams.org/views/2019/08/09/leading-civil-rights-lawyer-shows-20-ways-trump-copying-hitlers-early-rhetoric-andIts time to grow up and move on – for Democracy to work it has to work BOTH ways!
https://thenewdaily.com.au/news/world/us-news/trump-news/2020/11/13/former-leaders-urge-trump-to-concede-defeat/
Though the phrases “Donald Trump” and “act responsibly” appear mutually exclusive! ;-D
24
Got your message loud and clear.
Stuff the plebs; All power to the political Elites.
20
OK, I wanted to take encouragement from this video, but I can’t. Basically, I think the analysis is garbage. Here’s the train of comments I left on this video at youtube to explain why. Note that I went into it hopeful, and finished bitterly disappointed:
1) Isn’t the downward slope in the graph around the 25 minute mark exactly what we would expect to see, if Trump managed to move many non-traditional Republicans to vote for him personally? The story I’m seeing is that he moved previously rusted on Democrats, particularly blacks and Hispanics, to vote for him discussed at unheard of levels for a recent Republican e.g. he scored around 20% of the black male vote, when McCain, Romney etc got 5 – 12%. This was a personal “We like Trump” effect in these communities. Similarly, the Republican “Establishment” arguably hated as much as the Democrats, so his personal share of traditional Republican voters was likely to be lower than normal. We even saw the very public phenomenon of “Never Trump” advocates among establishment republican pundits. Presumably these people would be inclined to vote for the other Republicans at a higher rate than they voted for Trump, and they will be concentrated in Republican heavy areas.
Maybe I’m missing something here? If so, what exactly? (Note that I want Trump to win, thin he should definitely not concede yet, and want him to try every possible constitutionally valid route to achieve victory).
2) OK, I’m at the 32 minute mark now. How do you justify putting a level trend line through the first bit of the data, then the descending trend through the rest of it. Anyone eyeballing that data would see one slightly less rapidly descending trend line fitting all of the data. Similarly, if you randomly hacked out the data from 20 – 40% on the x-axis and fitted a trend, that would probably be comparably flat to the 0 -20% segment. What exactly is your justification for separating two separate domains here?
3) OK, I’m at the 34:25 mark now, dealing with Macomb County. There is no justification for drawing a break in trend in that data. This is ridiculous, and extremely disappointing. This is virtually giving the other side ammo.
4)38 minutes in. Finally something thing that looks weird, and they don’t even touch on it. What is going between the 0 and about 7% marks (x-axis), with the descending straight line of points immediately below the y-axis? Maybe simply some artefact of some very small precincts combined with integer counts, but at least it looks unnatural.
5) 44 minute mark. No, the question is not whether this pattern (the break in trend) is natural. The question is whether it exists. Seriously, what criteria or algorithm have you used to objectively identify this break in trend? It is certainly not evident just looking at the scatter of points.
6) 45 minute mark. Yes, you would expect to see a flat trend, UNLESS there was a simultaneous second factor in action, where the most Democrat inclined constituencies were MORE likely to vote for Trump that for standard Republicans. Gosh, and isn’t that exactly what happened in this election, and exactly what we see in the Democrat dominated counties chart you showed earlier? Under those circumstances, we’d expect to see exactly what we do see here.
Seriously, we have Biden’s son’s email regarding kickbacks for the big fella, Biden’s whole campaign based on the “Fine People” hoax with the full support of the media, near total suppression of Trump friendly facts by Big Tech, and this is what we go after?
10
Jo, good find. The line seems to be too perfect but the more important thing is that it went down, indicating that the more Republican you are, the more you hated Trump.
13
Not quite. It is measuring the difference between how much an area likes Trump, and how much that same area likes Republicans in general. So what it really indicates is that the more Republican an area is (and presumably has been historically), the more the residents dislike Trump compared to standard / establishment Republicans. More importantly, it shows the more Democrat and area is (e.g. majority African American), the residents are to vote for Trump than a standard Republican. Isn’t exactly what we have just witnessed with Trumps record levels (for a Republican) of black and Hispanic support, and the long campaign waged by Establishment Republicans against Trump, the decrease in white college educated supporters etc?
10
I found this video very difficult to understand (have watched it several times). Have asked the folks over at Catallaxyfiles.com for explanations but nobody has yet come up with anything that makes sense.
It seems that everybody’s eyes are glazing over for the mathematical and graphical bits, but people are then taking on board the conclusions Dr Shiva presents.
To me, the effect he identifies is exactly the effect we would expect for precincts with no fraud. That being: the more R leaning the precinct the more they will vote via the “straight party” method and the less via the “individual candidates” method.
Note that by voting “straight party” they are still voting for Trump.
Does the maths and explanations for the graphical data make sense to anybody here? Thanks.
00
Tom, if they voted the President and Party ticket 1 to 1 the dot would be on 0%. There would be no excess or deficiency.
Above the line, people voted for Trump but not the Rep Party. Below the line they voted for the Rep ticket but then someone else for President.
Why were there so many votes for “Biden Plus the Rep Party” in the most Rep suburbs? Perhaps there was a real movement for Reps who didn’t like Trump. But what makes this bizarre is that the line is perfect. People in 80% Rep suburbs voted for Biden at exactly the right ratio to put them exactly in between the 90% Rep suburbs and the 70% Rep Suburbs. In the real world thousands of humans don’t act with this sort of linear perfection on a complicated question. We get bell curves, gaussian curves, bends in the line. Phase shifts…
Their point is that only a linear algorithm (that works like a transistor on a signal) takes a flat line and adds one “elbow” bend and keeps it linear but in a new direction. It’s the mark of an equation that starts by ignoring Democrat suburbs and then applies a “change” to all suburbs with more Rep voters in a “weighted” way that makes the most Rep suburbs the most affected.
Does that help?
61
In thinking about the video that was posted I believe that it is just a reflection of crossover vote and is not indicative of fraud. Which is not to say that fraud did not occur just that this pattern is not necessarily reflective of it.
If you just assume that some fixed percentage of voters defected then you would generate the pattern as presented. For example. suppose you had 5 precincts of 100 voters that contained 90R. 70R, 50R, 30R, 10R registered voters. Say now that voting is 50% straight party, 50% split. In the split voting assuming that there is a crossover rate of 10% so that 10% of all split ballot voters vote against their registration for president.
registered voters
prec regR regD
A 90 10
B 70 30
C. 50. 50
D 30. 70
E. 10. 90
Votes assuming 50% straight party and no defections. All straight party voters vote ther registration.
prec R D
A 45 5
B 35 15
C. 25. 25
D 15. 35
E. 5. 45
In the non straight party voting assume a scenario with a 10% crossover rate with no net defection.
prec Rreg DVotes Rvotes Dreg Dvotes. Rvotes. Total R Votes. Total D votes
A 45 4.5. 40.5 5. 4.5 .5 41 9
B 35 3.5 31.5. 15. 13.5. 1.5. 33 17
C. 25. 2.5 22.5. 25. 22.5. 2.5. 25 25
D 15 1.5 13.5. 35. 31.5. 3.5 17 33
E. 5 . 5. 4.5. 45. 40.5. 4.5. 9. 41
The calculated points would be straight party, split ballot – straight party
A. 45, -4
B. 35, -2
C. 25, 0
D. 15, 2
E. 5, 4
All these points are on a straight line with negative slope just as seen. To the extent that the line is not straight is perhaps more interesting.
There is however strong indications of fraud in Michigan voting in a separate data set. Post to follow.
20
DAvid, if that is true, we should see the same pattern in most places not just in Swing states. Shiva et al say that in Wayne County they don’t find the manipulated pattern, they find the flat pattern. I presume they checked other places too.
The MIT team suggest that if there was a straight percentage who split the line would be flat across the graph. They are graphing a percentage above and below expected.
30
Hi Jo,
In order for the line to be straight horizontal there would have to be no defection from either side. Or the number of natural R’s and D’s would have to be the same and the defection rates equal.
As long as a county has a difference in natural D and R (so not 50-50) then if there is a fixed percent of defection on both sides then in absolute terms the side that has more natural voters will lose more absolute votes. The greater the natural disequilibrium the more the absolute difference. Since the plot is no switch, switch – no switch the greater the absolute number of no switch the farther below the 0 line this will end up. The example I posted above is exactly this.
The statement in the video that says if the preference remains constant the line will be horizontal is just wrong. The example above shows that clearly.
The people that vote straight ticket are highly likely not to have switched so it comes down to the split ticket folks who by definition have switched something (forget about 3rd party for the moment).
I think the interesting part of the data is that the straight line bends to flat where there are more D voters in precinct. That may be indicative of an issue cause it would suggest that in large D precincts the percentage of defections is lower than in precincts with a smaller D percentage. While one can make up some story to fit that it is somewhat dubious and we can probably test it out geographically.
If there was fraud it was almost certainly committed by stuffing extra D votes for D who were registered but did not vote. The stuffing would have to take place in heavy D counties where they controlled the counting places. The data from Shiva actual supports this. Stealing R votes electronically is too easy to catch since the vote tabs will be wrong right off the bat and will be caught by any machine recount and also an in person count.
The only way to catch this is to get access to the ballot images in the machines and look for duplicates against real ballots because the stuffing could not have known everything that would dribble in. Additionally the timestamps on the ballot image would tell a tale. Watch when they start deleting ballot images.
Unfortunately, unless they were sloppy with their fake ballots a hand recount won’t find anything either. If I were running the fraud the fake ballots would have been machine generated and would reasonably match the signatures on the voter registration cards. Maybe they weren’t quite so sophisticated and the hand recounts will catch something. If my supposition is correct, then they were already sloppy with the county vote targeting…..
I don’t hold out too much hope only because the public is so easily deceived on anything and everything that has any math in it.
10
” Watch when they start deleting ballot images.”
Aye there’s the rub. That’s one of Shiva’s key accusations: ballot images were never stored at all and the default setting had to be overriden in order to do so.
“In order for the line to be straight horizontal there would have to be no defection from either side.”
Not quite: it would indeed be flat horizontal if there is no defection on either side. That’s true. But you can’t go from “there will be some defection” straight to ” a gradient of -0.6 or stronger is also fine”. A gradient of -0.1, possibly even -0.2, might be fine, but there are other artefacts that you would expect to see with it. These straight lines at -0.6 are just not credible. Nowhere close.
I have now sourced the complete data set that Shiva used and have successfully replicated – as has Matt Parker who pointed me to it – the charts he presented. So I’ve got a lot of number crunching to do – I’ll post a proper analysis hopefully in the next couple of days.
The sharply downward sloping below the axis is super super super worrying. It does not fit with the real world, not with the rest of the data. Needs a proper debunk of Matt Parker’s pres to do so because the thing he’s missed is subtle.
20
Thanks David and TPG for the work you are putting into this.
10
Hi TPG,
Can you share the Shiva datasets ?
thanks
David
00
See links in Matt Parker’s vid – he also, good scientist that he is, shares a link to his actual spreadsheet with his analysis and charts so you can check the data is still good (it is, with exception of one precinct) and replicate the exact numbers (you can.)
A chunk of counties (including Kent) here: https://electionreporting.com/
I found Oakland here: https://results.enr.clarityelections.com/MI/Oakland/105840/web.264614/#/summary
Haven’t had time to do the rest – I’m 20 slides into showing exactly where the flaw is in Matt Parker’s analysis. It’s so subtle that blink and you miss it – that’s why so many are convinced by it. I don’t know whether to hope that Matt didn’t see the flaw (which would be poor for a mathematician) or whether it’s deliberate (which would be terrible), but he’s in a bind now. He will NOT want to be associated with anything that casts doubt on the integrity of the election, but his (flawed) video has gone viral. The question is how to get the takedown out…
00
Hi All
Please note that according to Dr Shiva’s new theory (see The Pedant-General), the “normal” curve is not a flat line, but rather a …… parabola (increasing, then flat, then decreasing).
So his new claim seems to be that a missing (increasing, flat) section implies fraud.
I also draw your attention to the fact (discussed by Matt Parker) that since typically in each precinct the Party and Individual parts are not of the same size, the difference between the two percentages is a meaningless metric and cannot be used as a proxy measure of defection.
00
Sorry for the cut and paste mishap. The URL of Dr Shiva’s news video is https://www.youtube.com/watch?v=R8xb6qJKJqU
Warning:
If you are a mathematician, you might be offended by Dr Shiva’s treatment of mathematicians in general and Matt Parker in particular.
00
The following is analysis on datasets that can be scraped from the NY Times website. The file michpres.raw was taken from
https://static01.nyt.com/elections-assets/2020/data/api/2020-11-03/race-page/michigan/president.json
I had previously done some work with the time series data in this set that shows the anomalous
Biden +140000 vote, and also shows votes being removed from Trump.
Working with the same Michigan dataset only this time focusing on the final vote totals on a county basis. Glancing at them it seemed to my eye that there were too many round numbers. So I thought if someone were to programmatically alter results it would seem natural to target a total number of votes for Biden and Trump within the margins of what would be reasonable within the constraints of the registration data and then to target a difference in the votes.
If this was the case and the programs targeted round number outcomes then sums and differences of votes should favor the round numbers and disfavor the numbers centered on 5. I wrote some code to do sums and differences in each county where all the votes were in (75 at the time of the analysis), and then extracted the 100’s digit and the 1000’s digit from the sums and differences. If the numbers were random you should see random totals 100’s and 1000’s digits. If the targets were round numbers of 1000’s or 100’s you would expect to see an excess of numbers 9 and 0 (corresponding to a bucket of 20% around 0, and deficits of 4,5 which is exactly what occurs.
Name 00 01 02 03 04 05 06 07 08 09
—————————————-
Sum100 08 05 07 13 05 05 06 07 07 12
Sum1000 13 08 05 06 04 06 06 10 11 06
Diff100 13 04 08 09 05 04 12 06 07 07
Diff1000 05 09 04 05 07 07 10 10 07 11
—————————————-
Total 39 26 24 33 21 22 34 33 32 36
The results are very, very far from random.
I subsequently ran monte carlo simulations to see what the frequency of this kind of distribution would look like and it came in about 1 in 1400 by chance.
Happy to supply all code and data
30
Thanks David! Interesting.
40
Votes were ‘removed’ from Trump?? Utter hogwash.
21
For the Michigan presidential race you can find time series data that contains all of their updates.
https://static01.nyt.com/elections-assets/2020/data/api/2020-11-03/race-page/michigan/president.json
Votes: 574417 Trump: 326269 Biden: 233213 Other: 14935 New: 560 Trump: -5420 Pct: 0.00 Biden: 3097 Pct: 5.53 Other: 2884 Pct: 5.15 Time:2020-11-04T01:51:52Z
In this time slice the number of new votes is 560. Trump loses 5420. Note that the actual number of new votes is included in the feed but only the percentage total for each candidate is included. The percentage is to three decimal places so like .488 trump .496 biden. The numbers for vote count for the candidates are backed out from the percentages so can be off by rounding .001. At the time of this slice the total vote wsa 574417 so the rounding error could only be 574. Losing 5420 votes is not possibly attributed to rounding.
Happy to supply you the code and the data.
20
Weighted race voting, as stated before is a feature of many voting systems. It’s also a feature required by the Michigan Democratic Party. How they hold their primary elections, internal party elections, etc – see Rules of the Michigan Democratic Party. Paragraph 2.6 & 4.0.
https://michigandems.com/wp-content/uploads/2019/05/Rules-for-Voting-and-Elections-converted.pdf
That said, Dr. Silva makes a very compelling argument, it is enough to believe their were irregularities in the voting in not just Michigan, but PA, WI, AZ, NV, and likely every state that uses a tabulator.
11
“have back-analyzed the pattern of voting in Michigan and figured out the electronic algorithm used to alter votes”
Gee sounds like what the CAGWarmistas do to temperature data and come up with CAGW.
I just cannot get over the gormlessness of this ‘Trump was robbed’ stuff on a site that is allegedly frequented by skeptics. The guy is the clown we all hoped he would not turn out to be, that he would have ‘grown into the job’ and actually drained the swamp. Instead he turned out to be some sort of petulant dinosaur who now lurks in the swamp himself.
22
‘No lawyer would defend him’.
Against who? Against what??
Utter nuttery.
The Feds have gone public to declare this election clean and fair.
As for your questions
Is this a coup?
Possibly.
Who is behind it?
Donald and Julie.
Who will be President?
Strange question. Biden IS the President elect. He’s juzt waiting on the current occupant to accept this reality.?
32
jo
If one does an internet search for Dr Shiva one immediately sees a welter of “conspiracy pillar” and the like.
The Left are following the well worn path of the Soviet Union where any dissenter is immediately labelled as a crackpot. The next thing that will happen is that they will start declaring people mentally ill and forcibly incarcerating them and drugging them.
The Left hate any who show them up, and being unable to engage just Resort to slander smearing and invoice to try to win the argument.
41
[Duplicate]
11
This post from StephenJ on Michael Smith News will hopefully assist all in understanding what took place and (IMHO) plausibly, why.
“If you look at the example on Antrim County Michigan there are some very exact movements.
1.Any final total reported by the computer system would have to approximately agree to the votes on hand at that time. Checked apparently by a process the yanks call canvassing.
The total votes that can be realistically recorded as being received has to have some relation to registered voters and past turnouts ie there is some upper limit.
If someone intends to fraudulently record votes they can not be sure of the numbers required until the vote has progressed to a substantial degree.
That carries with it another problem ; that the number of late votes recorded can not unrealistically skew towards one candidate.
How to overcome this? By making small adjustments to the totals for each candidate over a period.
How to minimise the number of false votes inserted? By destroying votes for the other candidate and replacing them with false votes.
While this is occurring the count should be proceeding so it will be an inexact operation.
The explanation for the cessation of counting on election night lies here. They needed to determine how much was required to achieve their results.
The shifting of relatively small amounts from Trumps total on numerous occasions laid the ground work for the destruction of the required number of votes and the insertion of the false votes without breaching the constraints referred to.
It was a fluke that someone noticed the movement on election night. “Here’s the votes , here’s the final tally; approximately equal; nothing to see here.”
The process reduced the number of votes for Biden that they had to show as being received from mail in votes.
I feel there is at least some element of truth in this.
Steve”
https://www.michaelsmithnews.com/2020/11/stephenjs-insightful-commentary-on-us-voteelection-fraud.html
21
I think this analysis is wrong.
at 19:56 the video says his graph plots y = c – x
c : trump vote by individual candidate method
x : republican vote by the straight-party method
when c is independent of x and its magnitude is not too big, you would see points scattered around a straight line with a slope of -1.
The appearance of a straight line is an inherent feature of the way these plots are made.
20
But that’s the point: you’re assuming c to be constant. That’s VERY unlikely: in fact, I would expect c to be very closely correlated with x.
i.e. c (the share of specific candidate votes) ought to be pretty much in line with the overall party affiliation (which is x).
if c is some reasonable approximation of x, then the equation becomes:
y = ~x – x = ~0
That’s the straight line on the x axis that Shiva is expecting. It’s what shows up in almost all the 2016 data too.
10
I don’t necessarily agree that this indicates fraud, it could just be voting patterns.
And the explanation given is poor.
The X-axis is how Republican the precinct was, based upon all-ticket voting.
The Y-axis is how Trump or Biden the selective voter was (who did not want the all-ticket choice).
One would have thought that selective voters in a precinct would be much the same as all-ticket voters. So a strongly Trump all-ticket precinct (X-axis), would be equally strong for Trump among its selective voters too. In which case the mass of blue precinct dots would be horizontal.
However, if the selective voters don’t like Trump so much, the precinct blue dots would be below the red line. And if they like Trump even more, the dots would be above the red line.
The data shows that the more strongly Republican the precinct was, the less the selective voters voted for Trump. And the more strongly Democrat the precinct was, the more the selectives voted for Trump.
This could indicate some kind of fraud, changing Trump votes to Biden. But it may simply be minority groups in Democrat precincts voting for Trump (plus the Democrat ‘down-vote-list’), while strong Republican areas were shying away from Trump (while still voting for the Republican ‘down-vote-list’).
RE
10
[Duplicate]
00
Hi Shiva made a math mistake, please watch https://www.youtube.com/watch?v=aokNwKx7gM8
20
It would appear that they are not very good mathematicians.
See Mat Parker’s (Stand-up maths)analysis at
https://youtu.be/aokNwKx7gM8
1. If you do the same thing usind Biden’s data and you get the same result.
2. It is a big maths no-no to add or subtract averages
3. Just keep it simple and compare the averages for each type of voting method and they are the same.
30
.
Here is the full explanation for this graph – it is merely a miscalculation.
See 12:00 – the golden rule, is you cannot subtract percentages.
Just plot the data, not the percentage, and the graph works.
https://youtu.be/aokNwKx7gM8
RE
20
Disclaimer: I’m pretty sure that fraud took place on a massive scale, and I really want Trump to prevail.
Unfortunately, Shiva’s “analysis” was even worse that I thought, and is likely to serve as ammo for Biden’s side. It turns out the downward trend is much easier to explain than I originally thought, to the point of being an almost theory free zone.
The guy at Stand-up Maths puts this theory out of its misery quickly and elegantly at:
https://www.youtube.com/watch?v=aokNwKx7gM8
30
Stop being a skeptic. Stop being scientific
01
See my comment below. Matt Parker is missing the point in some cases and misdirecting in others.
10
I saw your comments below. Matt is right and you are wrong. The position of the intercept does not suggest “abnormality”. Can you define what you mean by “normal”?
01
Plot of votes for Biden also slopes down in exactly the same way as Trump chart. Conclusion is that these charts are following the formula y-x = mx+b-x which is just a variant of the correlation chart of y= mx+b where m and b are contents for the correlation. In short, there is no fraud as, the Biden chart (not shown or apparently done by Dr Shiva) would have to show a positive gradient or at least a clear change to a positive gradient. It doesn’t, it follows exactly the same gradient as Trump’s.
31
Not in Wayne County.
So which county are you referring too?
Where are those graphs?
10
1. Why don’t you show your followers Matt Parker’s video?
2. Matt’s arguments are valid irrespective of the county.
01
I’m publishing those comments. No secrets here.
Where are the graphs showing the same slope (with elbow) happens in democrat counties?
Where are the graphs showign the transistor (elbow effect) also occurred in 2016. Pedant General shows they didn’t.
Thus I don’t see convincing evidence at all from Parker. But there’s nothing stopping you from persuading us.
10
The fact that there is variety of graphs with different shapes does not imply that there is a “fraud” out there. If you claim that any, or a certain, deviation from a “normal” curve implies “fraud”, then the burden of proof is on you. In his new videos Dr Shiva presents all sort of shapes, but fail to show that a diviation from his “norma” shape implies fraud.
There are ample of valid evidence that the 2020 election was indeed rigged. It is not necessary, in fact it is counter productive, to rely on bogus theories in an effort to fix the system.
Parker does not provide any evidence other than showing, formally and rigorously that the theory advanced in Dr Shiva’s first video is fundamentally flawed.
01
Dodged my question.
Bluster.
I’m not buying.
10
To show that Dr Shiva’s theory is flawed it is not necessary to investigate what happens in democrat counties. Matt’s argument is valid irrespective of what happens in democrat counties.
01
Matts argument cherry picked one county I’m told.
Your repeat assertions, and lack of examples that I request, suggest that Matt doesn’t have the goods.
Simply repeating things doesn’t make them true.
00
With regard to #59.1.1.1.2, you are either misinformed or you misunderstood what you were told. Why don’t watch Matt’s video for yourself?!
Matt did not pick a cherry. Indeed, he used the very same example that Dr Shiva use in his video to explain his theory.
Perhaps you are not aware of it, to destroy a theory it is sufficient to construct a counter-example, namely to show that it does not work in one of the cases that it is supposed to handle.
I keep repeating the same point because this is the very same point that you seem to miss. I suggest that you read the post that I repeat mentioning, where the author explains why he suspects that Dr Shiva is a data charlatan. My view on this matter is that either Dr Shiva is grossly incompetent or a charlatan. Given his scientific and math background, I have no reason to assume that he is incompetent.
01
I’ve read the back and forward between you and the Pedant General. PG explained why Parker picked that graph and not the other and why it was inconclusive with less data in the elbow region. I’m still waiting for you to provide 2016 or Democrat counties showing the same pattern as Shiva found.
You keep dodging the question.
00
I do not dodge the question. I merely argue that this question is senseless in the context of a discussion on the validity of Dr Shiva’s flawed theory, as presented in his first video.
Since you seem to be unfamiliar with the basic concept “counter example” in modeling, let us simplify the analysis.
Suppose that, for simplicity’s shake, none of the democrat counties exhibit the property that you attribute to the graph presented by Dr Shiva in his first video.
So what?
If his theory is valid, then it should be valid in the context of the specific graph used by Dr Shiva to illustrate his theory in action. But Matt Parker has shown that this is not the case.
If now Dr Shiva claims that his theory is valid in the context of other cases, then the burden of proof is on him: he will have to demonstrate this and explain in detail what these other cases are. He will also have to admit that the theory does not work in the context of the examples he used in the first video.
I’ll be more than happy to discuss this matter with you in private because I share your concerns about the situation in the USA regarding the 2020 election.
BTW, I also share your concerns regarding the climate change issue.
01
He refers to the counties examined by Parker. These counties were also used by Dr Shiva to test/illustrate his theory in his first vieo. I suggest again that you watch Parker’s video. It is very educational.
01
Congrats. This “doctor” has just shown you all how plots work. He just illustrated basic math: lines with negative slope descend. And he’s just plotting such data.
1. His data is EXPECTED to show a descending plot.
2. Biden plots are the same: then did republicans steal the votes from the dems?
Explanations:
* https://youtu.be/aokNwKx7gM8
* https://kabir-naim.medium.com/dr-shiva-ayyadurai-the-danger-of-data-charlatans-4f675ffe793c
Either this poor dr. is really bad at math. Or either he’s lying/laughing in your face with his made-up story
11
See my response below. It’s not just the slope (though a constant vote share across widely varying precincts is already odd). It’s the intercept as well.
Biden’s is well above the line, Trump’s below.
the 2016 data – and most of the 2020 data as well – show that this is NOT normal behaviour. Candidate vote share broadly tracks straight party vote share. This produces a flat line close to the x-axis, not the declining slope.
10
See my comment above.
01
Unfortunately, Dr. Shiva’s theory seems to be seriously flawed. See a formal explanation of the flaws at
https://www.youtube.com/watch?v=aokNwKx7gM8
PS. I am a Trump supporter and I am convinced that the election was rigged.
11
Careful with this one – apparently (I haven’t checked this as I don’t have the source data), the same pattern shows when you run the same analysis with Biden’s votes, which rules out fraud, at least in the form implied. There’s a fundamental flaw with the Y axis plot producing the negative gradient.
https://www.youtube.com/watch?v=aokNwKx7gM8
10
I’m sorry, but I have a PhD too, and I don’t think this analysis holds water. The total of votes for R.S.P and Trump non-R.S.P would add to the number of votes for Trump in that district. In other words, 100% of the votes for Trump in that district. An increase in percentage of one would mean a corresponding decrease in the other. This naturally leads to a linear effect in the plot independently of all other considerations.
10
You’ve misunderstood Peter. We’re not showing % of total Trump votes anywhere (which would of course always add to 100%).
Instead, we’re plotting Rep share of candidate votes vs Rep share of straight party votes.
Simple example:
A district with 1500 voters, 1000 of which vote straight party, 500 of which vote for a candidate
Straight Dem: 250
Straight Rep: 750
R share (x-value): 75%
Biden Candidate: 200
Trump candidate: 300
R share of candidate: 60%
y-value is 60 – 75 = -15
Trump has a total of 1050 votes in this example, but that fact is not relevant anywhere.
10
Correct. But, as explained by Matt Parker, this is exactly why subtracting the percentages is an indication that Dr Shiva is not aware of how bad is analysis actually is.
01
Fundamentally disagree – I’m really really frantically busy right now.
What Matt Parker has done is subtle so requires careful explanation to show. Once you see it, it is clear as f*cking day, but it will probably take a post – and a chance for Matt Parker to review and respond: this is a howler on his part so he will want a “right of reply”.
10
I’ll never trust an election again, that doesn’t involve paper ballots and picture ID/proof of citizenship. Thank you, CIA. #Hammer #Scorecard
10
Shiva has done a follow up video on this with more info. Can you do a post on the new info if you have time to watch it?
10
The redone is worse than the original flaw.
01
Can you please show Biden’s side to verify that it is opposite to trump’s. Would the rigging be obvious from the raw data sheets – precinct data records. These show that your analysis may be incorrect as the following shows –
Voters Republican Trump
Bowne Precinct 1 3359 1006 30% 1557 46%
Byron Precinct 1 3444 987 29% 1503 44%
Byron Precinct 4 3097 1003 32% 1463 47%
Gratton Precinct 1 1938 525 27% 855 44%
Solon Precinct 2 3584 1008 28% 1499 42%
10
https://www.youtube.com/watch?v=aokNwKx7gM8
CAN YOU REFUTE THIS ALTERNATIVE ANALYSIS.
01
Yes – see my comment below. Matt Parker has missed the point in some cases and is doing some quite serious misdirection in others.
I have repeated this analysis for all the data I could find for 2016 and you get the expected gentle curve starting at y= 10ish at x = 0, rising to 20ish at x = 50 then falling to y=0 at x = 100.
this is the generally expected behaviour – Trump slightly outperformed what would be expected straight ticket vote but the data tends to little or no under/over performance at either end.
This analysis showing massive, growing underperformance, in just a few counties is a massive red flag.
The “knee” is critical – Matt has chosen only to show the one chart from Shiva’s analysis where the knee is not clear. That’s misdirection.
10
Matt’s criticism is valid irrespective of the specific case. He deals with fundamental flaws in Dr Shiva’s theory that are present in all specific applications of the theory.
00
I created an Excel spreadsheet that can be used to “simulate” voting results based on some simple assumptions with adjustable parameters. I then created a plot like that by Dr. Shiva Ayyadurai and his colleagues in the video. The result is always a line with a slope of -1, and it works the same way for Trump and Biden. This will always happen if it is assumed that the non-straight-party voters split their votes between Trump and Biden with the same percentage, i.e., with a fixed percentage voting for Trump and independent of the percentage of Republican straight ticket voters. Although the results plotted by Dr. Shiva Ayyadurai and colleagues look impressive, the arithmetic of the problem shows that it they are the expected results, given their choice of variables to plot.
01
You are absolutely correct: if the Trump/Biden vote share is constant across precincts, you will indeed get a perfect slope of -1.
There are, however, two (big) problems with this approach:
1) the assumption that the Trump/Biden vote share would be constant across precincts.
2) the idea – this is Matt Parker’s big failing – that it is only the slope of -1 that matters.
To deal with these in turn….
1) Trump/Biden share being constant.
Would you assume that the Trump/Biden vote would be constant between counties? Or between demographics? Or between States? Obviously not. So why would it be constant between widely varying precincts?
If it varies by precinct within a county, the question then is HOW does it vary? A reasonable proxy would be that the vote share Trump/Biden might reasonably, give or take, be related to the predominance of Republican/Democrat supporters in a precinct. This is DEFINITELY NOT a constant.
This proxy is not going to be perfect:
– Some republicans are never Trumpers and would vote for a GOP senator, but for Biden -> Trump vote a bit lower than the prevailing party split
– On the other hand, Some democrats might have been horrified by some of the antics of the left and worried about their failure to condemn the looting -> Trump vote a bit higher than the prevailing party split.
The question is: what’s the balance of these – and myriad other – effects? My guess is that they should pretty much net out – the line should waver around 0. If a candidate is stronger than the party, it should be a little bit above, a little weaker it will be below.
All that’s happening with the subtraction is to make the line easier to see: instead of seeing a line which has a gradient of 0.9 or 1.1, you are just looking for above or below the x-axis. Shiva’s analysis is looking at exactly this. It’s a simple transform that, despite Matt Parker’s objection, he then shows to be completely mathematically valid later on.
2) It’s not JUST the slope – it’s the intercept.
Having a slope of -1 is already counter-intuitive: it says that the candidate vote share is completely unrelated to prevailing party leanings in a given precinct. But let’s say that this is what’s happening. It’s still not enough.
Two (ludicrous, extreme) examples, but also both using your constant share
a) line starts at 0% Rep vote, +100points to Trump, and declines to 100% Rep Vote, +0 to Trump. This is the case where Trump has 100% candidate share across the board.
b) line starts at 0% Rep vote, +0points to Trump, and declines to 100% Rep Vote, -100points to Trump. This is the case where Trump has 0% candidate share across the board.
Both of these have slope -1, but they are very different on the ground: it’s intercept – where the line crosses 0 that matters.
This is where the “knee” is important. Oakland county shows Trump polling around 15% BETTER than expected in 100% Dem precincts with that holding to around precincts that split 70% Dem/30% Rep when performance starts to suffer. So we’re being asked to believe that Trump does better than expected in Dem areas but massively worse in strongly Rep areas. Not only is it not credible, it flies in the face of most of the data – I have analysed as much data as I can get for 2016 (Jo – I’ve emailed this to you) and sure enough, candidates pretty much follow the straight party vote share.
We’re being asked to believe that Trump does a little bit better than might be expected in heavily D areas and HUGELY worse in strongly R areas. That’s just nonsense.
[Thanks! — Email coming — Jo]
10
If the Trump/Biden split varies among precincts, this will cause scatter around the line with slope -1, which is what is observed in the plots of actual (calculated) data. It would not cause the slope to be zero. It could cause the slope to vary from -1 as well, but not significantly, unless other factors are involved. The original analysis is basically plotting a variable A on the horizontal axis and a variable (Constant – A) on the vertical axis. It is not surprising if there is a negative slope.
Having a slope of -1 is counter-intuitive only in the sense that the combination of variables being plotted is counter-intuitive, while the arithmetic of the problem is straightforward (if lengthy). I did the calculations in Excel and then, by brute force, simplified the long equation (by combining formulas in related cells) to come to my conclusion that a slope of -1 is expected. I went through that process because I am both a supporter of fair elections and a supporter of Trump, so I am not at all biased against the idea of the possibility of vote fraud. I just wanted to understand what the original Dr. Shiva Ayyadurai et al. analysis meant.
I cannot find any information on voter registration by party on the Kent County, Michigan Web site, or how many voters of each party cast ballots, so I have no way to determine directly whether Trump did better or worse than expected in various precincts, unless I were to start making unfounded assumptions. Of course, it is obvious that some areas are more heavily R or D. I plotted % Trump (overall) vs. (% Straight R)/(% Straight R + % Straight D) and found that there is a strong correlation with a slope of 0.83 and intercept of 3.1. That slope value can be considered suspicious, and perhaps that is where we need to look at the mechanics (or electronics) of how votes were counted.
I also plotted % Trump (not straight party) vs. (% Straight R)/(% Straight R + % Straight D) and found that there is a weaker correlation with a slope of 0.60 and intercept of 9.0. In general, this supports the contention that we are being asked to believe that Trump did better than expected in heavily D areas and worse in strongly R areas. This may be suspicious, but since we do not know the party affiliations of the not-straight-party voters (in the available dataset), it requires further examination to see whether it is nonsense or has a possible rational explanation. Since the not-straight-party votes may be of a wide variety of kinds, they are far more complicated than straight party votes. We need to see the actual ballots.
30
Good comment.
10
Disagree Thomas.
“If the Trump/Biden split varies among precincts” – it definitely does. We see that it does for straight party votes. Why on earth would it NOT for the candidate votes? This is so staringly obvious that I’m struggling to see how anyone can seriously suggest that the Trump/Biden candidate vote split is not going to be at the very least related in some way to the straight party share.
If we plot straight share vs candidate share, you will get y= x + noise
Now instead of plotting y, you plot (y-x) and you have (y-x) = (x + noise) – x
That’s just noise around the x-axis. That’s what we should see. That’s what we do see in almost all the other precincts and in all the data I can find for 2016. So when you get a slope of -1, that suggests that the candidate share is NOT related to the straight party vote – that it’s a constant with respect to straight party share.
That’s so counter-intuitive and contrary to all the other data that the onus is on you to demonstrate how and why that happens.
“The original analysis is basically plotting a variable A on the horizontal axis and a variable (Constant – A) on the vertical axis. ”
That’s your problem again – it SHOULDN’T be a constant: it SHOULD be highly correlated with variable A. It’s only because there is something seriously weird going on – a constant vote share across precincts – that you get this -1 slope. THAT’S the red flag.
“I also plotted % Trump (not straight party) vs. (% Straight R)/(% Straight R + % Straight D)”
(% Straight R + % Straight D) = 1
Either I don’t see what you are doing here or this might be the clue to what you’ve done wrong.
Example. Precinct has 1500 voters in total.
1000 vote straight party:
R votes: 750
D Votes: 250
R % straight vote – x value: 75%
the remaining 500 vote for a candidate:
Trump (R): 300
Biden (D): 200
R % Candidate vote: 60%
we plot 60 – 75 = -15 on the y-axis.
Can you illustrate with these numbers what you are plotting?
“Since the not-straight-party votes may be of a wide variety of kinds, they are far more complicated than straight party votes. We need to see the actual ballots.”
But on this, I absolutely 100% agree.
20
There is a negative slope in the Dr. Shiva Ayyadurai plots. If you make such a plot for Trump, then the interpretation being put forward is that votes were taken from Trump and given to Biden, and the number of votes increases with the “redness” of the precinct.
If you use the exact same dataset and make the same type of plot for Biden, you will end up with a similar plot with essentially the same negative slope. The interpretation would presumably then be that votes were taken from Biden and given to Trump, and the number of votes increases with the “blueness” of the precinct.
I hope that it is obvious that both of these interpretations cannot be simultaneously true, since they are directly contradictory. Therefore, the source of the negative slopes is in how the variables for the axes are constructed, and not in any supposed vote shifting. This does NOT prove that there was no vote shifting, but the plots as given demonstrate nothing of the kind. Votes were not massively stolen from both candidates and given to the other, because there is a “conservation of votes” requirement.
My “simulation” in Excel, which is admittedly a simplification based on some assumptions, assumes that there is some fixed percentage of straight-party votes, some fixed percentage of Republicans among the non-straight-party votes, some fixed percentage of Trump votes for non-straight-party Republicans, and some fixed percentage of Biden votes for non-straight-party Democrats. The percentages can all be chosen independently. I calculate results for different straight-party Republican vote percentages from 10 to 90, and then make the same type of plot as Dr. Shiva Ayyadurai. The slope is always -1.
The key concept is that the straight-party Republican vote percentage is the horizontal axis variable in the eventual plot, and that same variable is included in the calculation for the vertical axis variable with a factor of -1 in front of it. If the other percentage values are constants, then the slope of the resulting plot will be -1 “by definition,” and it will be the same if Democrat and Biden votes are plotted. The negative slope is a necessary part of the arithmetic of the problem, and the fact that it is negative (instead of zero) indicates only that.
Unless there are radical shifts in the other percentages as a function of the straight-party Republican vote percentage (rather than being even roughly constant), then this “simulation” explains the general nature of the results. If there is an argument that the other percentages should not be assumed to be constant, then I would say that a different type of plot is needed that does not confound an expected result (negative slope) with a pattern that might lead someone to suspect fraud (Trump performance appears to get worse with increasing “redness” of precinct). That should not be difficult to do, so let’s do that. For me, the discussion of the negative slope in the present plots is over. It is not, in itself, meaningful at all.
01
True. As clearly indicated by Matt Parker, the negative slope does not imply “fraud”. Dr Shiva and his followers now argue that the issue is the difference (Biden vs Trump lines) in the slope and the value of the intercept. These arguments have no merit.
It is sad that experienced scientists support Dr Shiva’s fundamentally flawed theory. This may damage the on going effort to expose and fix the weakness in the US election system. There are plenty of reliable evidence that the 2020 election was indeed rigged and there is no need to rely on a flawed theory to convince the public that this is indeed so.
00
Thomas
“If you use the exact same dataset and make the same type of plot for Biden, you will end up with a similar plot with essentially the same negative slope. The interpretation would presumably then be that votes were taken from Biden and given to Trump, and the number of votes increases with the “blueness” of the precinct.”
No it wouldn’t – because the Biden line is still above the axis: he’s still a net positive, just not as strongly. This is the crucial interpretation error.
“I hope that it is obvious that both of these interpretations cannot be simultaneously true, since they are directly contradictory.”
I see what you are doing here but you’ve missed a step. Once you have your line, calculate the difference between the -1 slope and the x-axis – basically add the y-values across the range. This is the measure of over/under performance being claimed by Shiva. If the -1 slope is mostly above the x-axis, your net will be positive. If it’s mostly below, it will be negative. Crucially however, the corresponding totals for trump/Biden will net to zero – they aren’t contradictory.
This is also why the intercept is so important.
” I calculate results for different straight-party Republican vote percentages from 10 to 90, and then make the same type of plot as Dr. Shiva Ayyadurai. The slope is always -1.”
That’s because your starting assumption on the behaviour of Trump/Biden candidate voters is wrong. Your maths is completely correct: it’s the meaning of it that you are missing. If the -1 slope is so commonplace, we would see it often. We don’t.
“If the other percentage values are constants, then the slope of the resulting plot will be -1 “by definition,” ”
If the other percentage values are constants YOU ARE LOOKING AT FRAUD. That’s the point.
Send me your spreadsheet: thepedantgeneral – at – gmail dot com
In return, I’ll send you my analysis of the 2016 data that does not have this artefact.
Ruth,
“Dr Shiva and his followers now argue that the issue is the difference.”
I disagree – he hasn’t changed his story and he did talk to this in his first video: however, because this anomaly only seems to appear to Trump’s disadvantage and never to Biden’s, or at least never to anything like this degree – and it’s the degree that’s important – he didn’t labour the point.
This is not “changing his story”. Anything above the line is overperformance, anything below is underperformance vs a reasonable expectation.
The key here is this and it’s really simple: if I gave you any arbitrary grouping of voters and asked you to predict the approximate candidate share Trump/Biden, where would you start? Would you think that the general blueness/redness of the group would be an important – probably the most important – predictor? and if so, wouldd you think that straight ticket Rep/Dem share would be a good proxy for that?
That’s what this is about. All the previous data shows that the straight ticket share is about the best predictor for the candidate vote that there is. the -1 line shown here is evidence of a noticeable deviation from the expected – and historically observed – behaviour.
10
Disagree Thomas.
“If the Trump/Biden split varies among precincts” – it definitely does. We see that it does for straight party votes. Why on earth would it NOT for the candidate votes? This is so staringly obvious that I’m struggling to see how anyone can seriously suggest that the Trump/Biden candidate vote split is not going to be at the very least related in some way to the straight party share.
If we plot straight share vs candidate share, you will get y= x + noise
Now instead of plotting y, you plot (y-x) and you have (y-x) = (x + noise) – x
That’s just noise around the x-axis. That’s what we should see. That’s what we do see in almost all the other precincts and in all the data I can find for 2016. So when you get a slope of -1, that suggests that the candidate share is NOT related to the straight party vote – that it’s a constant with respect to straight party share.
That’s so counter-intuitive and contrary to all the other data that the onus is on you to demonstrate how and why that happens.
“The original analysis is basically plotting a variable A on the horizontal axis and a variable (Constant – A) on the vertical axis. ”
That’s your problem again – it SHOULDN’T be a constant: it SHOULD be highly correlated with variable A. It’s only because there is something seriously weird going on – a constant vote share across precincts – that you get this -1 slope. THAT’S the red flag.
“I also plotted % Trump (not straight party) vs. (% Straight R)/(% Straight R + % Straight D)”
(% Straight R + % Straight D) = 1
Either I don’t see what you are doing here or this might be the clue to what you’ve done wrong.
Example. Precinct has 1500 voters in total.
1000 vote straight party:
R votes: 750
D Votes: 250
R % straight vote – x value: 75%
the remaining 500 vote for a candidate:
Trump (R): 300
Biden (D): 200
R % Candidate vote: 60%
we plot 60 – 75 = -15 on the y-axis.
Can you illustrate with these numbers what you are plotting?
“Since the not-straight-party votes may be of a wide variety of kinds, they are far more complicated than straight party votes. We need to see the actual ballots.”
But on this, I absolutely 100% agree.
10
The horizontal axis variable is % Straight R, which equals (Straight R votes)/(All straight party votes).
The vertical axis variable is (% Not-Straight R) – (% Straight R). If % Not-Straight R is constant, then the expected slope is immediately -1 if a number of precincts are plotted. If % Not-Straight R is not constant, then something else will happen. The slope will only be zero if % Not-Straight R is generally EQUAL to % Straight R across precincts. But is that expected?
The “expected” result depends on your theory of who voted Straight-Party and your theory of what the reasons are for Not-Straight-Party votes.
According to my understanding, there is no requirement for someone who voted Straight Party Republican to be registered as a Republican. The same is true for a voter who is a Democrat or registered in any other party. A Straight R voter is not necessarily a Republican, nor is a Straight D voter necessarily a Democrat. Given that this election was quite polarized, there may have been more crossover votes than usual. I am not sure whether there is any way to determine that from the available data. A high ratio of (Straight R votes)/(All straight party votes) does not necessarily mean that a precinct is heavily Republican, although this would be expected.
Who are the Not-Straight-Party voters, and why are they not voting straight party? These voters must fall into some categories, such as Republicans who did not vote for Trump AND/OR at least one other Republican candidate for some office, Democrats who did not vote for Biden AND/OR at least one other Democrat candidate for some office, and those not registered as either Republican or Democrat who voted some type of split ticket. Remember that this represents about 40% of the voters (average percentage for precincts in Kent County Michigan), and that a split ticket does NOT necessarily mean Trump plus votes for Democrats other than Biden (or Biden plus votes for Republicans other than Trump).
I do not know of any a priori reason why it should be expected that the roughly 40% of non-straight party voters would have the same distribution between Trump and Biden as the roughly 60% of straight-party voters within each precinct. Let me use your example:
Example. Precinct has 1500 voters in total.
1000 vote straight party:
R votes: 750
D Votes: 250
R % straight vote – x value: 75%
the remaining 500 vote for a candidate:
Trump (R): 300
Biden (D): 200
R % Candidate vote: 60%
we plot 60 – 75 = -15 on the y-axis.
But is not -15 a HUGE discrepancy in what would seem to be a heavily Republican precinct (and even more so if there were 67% straight-party voters with 75% of them for Trump)? It is exactly results like these that lead to the Dr. Shiva Ayyadurai type of plot with a negative slope. For Kent County, Michigan, the slope is about -0.41 for Trump and -0.40 for Biden (and yes, the intercepts are different, but that is another topic). The slopes are less negative than -1, which is due to the fact that the non-straight-party voters have a fairly strong positive correlation with straight-party voters (in Kent County) in terms of how they voted (it has a least-squares line with slope near +0.6). Since that least-squares line has a slope less than +1, the ((% Not-Straight R) – (% Straight R)) vs. % Straight R plot has a negative slope. It is just the arithmetic of the method and available data.
All of this does NOT prove that there was no vote fraud, but neither does it show that there WAS vote fraud. It is simply the wrong way to try to identify whether vote fraud occurred, unless mountains of other evidence can be brought to bear to show that it can reliably detect fraud in other elections. The method does not appear to be very sensitive for this purpose, and a LOT more examples would be needed to even begin to convince me that it is a reliable way to look for fraud.
I wish that it were this easy to prove vote fraud, and I would be quite satisfied to see it in this case (being a Trump supporter), but I am not convinced at all. Has anyone looked at, say, hundreds of elections in different places to prove what are “normal” and “abnormal” results? If the datasets are available, this would not take all that long to do, and it would go a long way toward demonstrating whether such a method is useful. Even using only the existing dataset, it could be done for various down-ballot races to see what happens.
00
https://postimg.cc/143Hj2Qz
The above image is a plot of the ratio of absenteeMixedRepublican ballots / absenteeStraightRepublic ballots vs. Republic Percentage Vote
There does not appear to be any meaningful correlation
00
The above image is a plot of the absenteeMixedDemocrat ballots / absenteeStraightDem ballots vs. Republic Percentage Vote
Clearly there is strong positive correlation.
What this means is that in precincts where Republicans did very well the distribution of absentee ballots on the dem side shifted towards mixed ballots.
If this is an indication of ballot stuffing it would mean that single vote Biden ballots, or Biden plus some down ticket R’s were placed preferentially into republican precincts.
00
Matt Parker’s rebuttal is also flawed. Shiva may be a crackpot in many ways, but in this case he has almost certainly uncovered something really horrible.
Fundamentally Parker is ignoring the implication of the use of the delta on the y-axis. All this does is to show if the individual votes for Trump are over or under-performing relative to the low information/straight-ticket voter. That’s all it is.
Several elements:
– he makes a fuss that you “can’t subtract percentages” and that’s true to the extent that you can’t then use that delta to project back to an absolute number of voters but that’s not what’s being done. He subsequently then shows that this is mathematically completely fine – it merely takes a 1:1 straight line and projects it onto the x-axis.
– he studiously ignores the meaning of that over/under performance in the real world. We are being asked to believe that, as precincts become more Red (based on the straight ticket voters), the under performance gets noticeably – and predictably – worse.
– when he plots the lines for Biden, he also finds this downward line, but that’s also missing the point: the line for Biden does indeed show the (completely mathematically) complementary view, but the Biden line is shifted vertically: see at 8:08. We are being asked to believe that Biden polled that strongly in bright red precincts, and only in these precincts.
Crucially, however and to quote Steve McIntyre, “watch the pea”, specifically two of them:
1) there is a real question about the bend in the curve at ~30% – the question is whether this is flat from x = 0 to 30, or whether it’s a straight line. The one chart he uses to argue for a straight line, rather than one with a bend has very few points at low x values, so it’s hard to call either way. Matt uses Kent county where the knee is indeed unclear, but ignores e.g. Oakland county which is where Shiva makes his case. It’s almost as though Matt Parker has cherry picked the one county that allows him to ignore this point – it’s critical when you look at the rest of the maths.
2) when Matt shows how the transform creates a negative gradient, his starting line only goes up to 80% on the y-axis – i.e. Trump underperforming. That’s the very question at issue.
This is fraud – I have run these numbers for the 2016 election and you do not see this behaviour. I have 1800 precincts worth of data that shows very little net under/over performance – the data shown by Shiva is terrifying and Parker’s takedown misses the key points.
20
Hi,
Any chance you could post some links to the raw precinct data ?
Thanks
David
10
I’m afraid it’s a bit of a nightmare:
https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/LYWX3D
The Harvard Dataverse has a wide range of resources but with precinct level data for President, House and Senate races for 2016, House and Senate and the local dog catcher for 2018 and a bunch of other stuff.
Only a subset of precincts report straight tickets separately and they sometimes name them differently (“straight party”, “ticket” etc). I had to load to SQL to make sense of it as the files are large. Net result, I ended up with comparable data for the Presidential and House races at precinct level for 2016, around 1800 precincts in all. Almost nothing goes below -10 on the y-axis at x=100% – it’s noise around y =0: the various effects net out so that the candidate shares are closely correlated with the straight ticket share.
This sharp downward slope is undoubtedly an anomaly.
30
Sadly, you seem to have misunderstood Parker’s analysis and conclusions. Parker has show clearly and conclusively that Dr Shiva’s theory is fundamentally flawed. If you watch Dr Shiva’s second video and his response to Parker’s criticism you can clearly see that Dr Shiva is now trying to coverup the flaw in his theory( see https://www.youtube.com/watch?v=R8xb6qJKJqU&t=1510s). His attack on mathematicians is very telling.
01
Ruth,
Can you show me exactly where and how I have misunderstood Matt Parker’s analysis?
I’ve watched his video carefully and I have pointed to the key elements where I believe Parker’s analysis is flawed:
1) because Parker does not ever address the meaning of the regions above/below the line in terms of over/under performance, he does not see the significance of the intercept, nor that a gradient of -1 is really weird — wildly unlikely in the real world. Yes of course you can decompose the signal to a constant vote share and get a -1 line, but A CONSTANT VOTE SHARE IS EVIDENCE OF FRAUD IN THE REAL WORLD. Parker never addresses the real world meaning.
2) the knee in the chart is important. Parker uses the one chart in Shiva’s analysis that does not clearly exhibit it in order to say it doesn’t exist. That’s not good practice. He needs to look at the strongest case and knock that down.
3) when, towards the end, Parker shows how a 1:1 line maps to a negative slope and goes “tada!”, there are two further misdirections: firstly, his 1:1 isn’t: it actually only hits ~85%, so is only a gradient of 0.85. This _does_ map to a negative gradient but it’s nowhere near -1: it’s only -0.15. But even this apparently minor niggle is important in the real world: it suggests an underperformance – that Trump’s weakens in more strongly Republican areas.
Finally, if Matt Parker was correct and this was common, we would see it everywhere. We don’t. The examples shown by Shiva are the exceptions: they do not conform to expected real world behaviour and there is a clear and understood (and fraudulent) mechanism in the voting machines by which this result can happen – they need to be explained.
Re Shiva’s second vid – he’s a prickly character, not easily likeable and is clearly flawed in other ways, but his attack on mathematicians has some merit – he’s calling out my point 1) – that Parker never addresses the meaning of the data so misses the important artefacts. However, his analysis here is correct and he has identified a real and troubling anomaly. To that end, I don’t see the cover up in this vid. He’s clearly cross, but he’s just amplifying his original work – I can’t see any element of a cover up of any flaw.
Two things for you:
1) Can you explain – with time references to Matt Parker’s video – what I have missed or misinterpretted.
2) can you show – again with time references to Shiva’s second video – the cover up I’ve missed.
This is a genuine request: I have an Engineering and Maths background and am now a data scientist. I do this stuff for a living. If I’m wrong, I really want to know how and why. That does require you to show where my reasoning above is incorrect.
10
G’day General:
I”ll try my best to convince you that Matt Parker’s criticism is valid and that Dr Shiva original election fraud theory is fundamentally flawed.
However, I suspect that this will be a long iterative process. Therefore it would be much better to do this in private.
In a day or so, I’ll let you know how we can do it.
FYI, I am a retired academic. I taught college applied math subjects for over 30 years. I am convinced that the 2020 election was indeed rigged and that it would be possible to show it to unbiased persons.
I am looking forward to an interesting and productive interaction with you.
In the meantime, you may wish to read the relevant post “Dr Shiva Aaaudari & the danger of data charlatans” at https://kabir-naim.medium.com/dr-shiva-ayyadurai-the-danger-of-data-charlatans-4f675ffe793c
10
Ruth
That’s really kind. I would very much like to take you up on this: I too am busy so completely respect that this may take time.
If it helps, I am genuinely terrified by the implications of this. If it really is the case that the US election is unsafe and systematically fixed – in the way that Shiva alleges and which I also think is evidenced here – then we are doomed. Not just the USA: All of us. The represents the beginning of the end of the era of democracy. I would truly genuinely like to be shown to be in error.
If nothing else, this would be a significant learning exercise for me. I can – quite easily – see the flaws in Matt Parker’s rebuttal but only because I tried extremely hard to break Shiva’s analysis first and find a way to rationalise it so that we don’t have to confront the horrifying idea that Shiva might be right. If I’m wrong, then I’ve f*cked up big time and I want to know how I’ve missed it.
I can be reached at thepedantgeneral – at – gmail dot com
I’ll read your post in the meantime ahead of any discussion.
I am heartened that you have reached out and look forward to talking with you in due course.
10
Pedant-General
Sorry for the delay, we have just came back from our daily walk. I feel much younger.
I share your concerns. This is why I believe that it is very important not to use flawed theories in the effort to fix the system.
I’ll contact you within a day or two.
10
If you guys could do this online that would be great. There are a lot of other folks including myself that have suitable backgrounds to add to the conversation. FWIW I am on the side that there may well be fraud but the presentation by Shiva can be explained without having to resort to it. See for example the post 44 above. That being said there are some very strange anomalies.
I have been working with the precinct level data for oakland county michigan. There is a very odd relationship between the percentage of MixD ballots and republican vote percentage. Specifically as the republican percentage in a precinct increases so does the percentage of MixD / StrD ballots. This is much more odd than the Shiva data. There is no reason to expect that as you move into more republican counties that you would get higher percentages of ballots with Biden at the top and down ticket D’s missing. These percentages are not small numbers.
While Biden wins about .5545 of the straight line voting he wins fully .5945 of the mixed ballots. For comparison in 2016 Hillary won .5433 straight line votes and .5419 of the mixed votes. Going to see what this looks like on a precinct by precinct basis between 2016 and 2020. That relative swing in increase MixD share amounts to about 25,000 votes.
10
David:
I’ll discuss this possibility with The Pedant General and let you know of our decision. Personally, I believe that it will be more productive to do it first in private and then put our conclusions online for discussion. This can be done in stages by addressing and reporting on each of the key points in a sequential manner.
10
Ruth,
for ref, I looked at the link you included at it makes all the same errors that Matt Parker has made and that I’ve already pointed out above.
I’ve been in touch with Matt Parker just to check the data source – I’ve confirmed that I can replicate the numbers.
I’m working up a pack to explain the problems with Matt’s rebuttal – my points remain. Shiva – for all that he may be a nutjob – has found a serious problem here.
I’ll share with you as soon as I can.
10
The Pedant-General:
As agreed, I’ll try to explain to you, in private, the fundamental flaws in Dr Shiva’s (first) theory, and why his new theory is yet to be properly stated and validated.
10
I am watching this conversation and analysis and am impressed with genuine efforts to get to the truth of the situation. Looking forward to any progress in reconciling two different views.
10
Great, Jo!
It will help a lot if you could embed Matt Parker video in this page.
10
First, the Dominion voting system wasn’t used in Oakland Michigan, but Verity version 2.2.2 from Hart.
Second, the pattern, the slope Shiva mentions in his video also appears with other voting systems and also appears for a county where Trump won, by a lot.
So Shiva is wrong or lying, I would go with lying, he is a fraud.
See https://twitter.com/dpopa/status/1329587546897014786?s=21 for some revealing charts.
11
ok. those charts have the same problem as Parker’s analysis – the x-axes are mirrored for the two lines. Data points for precincts to the R on the trump line are on the l side of Biden chart. Plot on the same axis and you’ll see the divergence.
00
Some Allen is correct. Check again you claim, Pedant General.
In the graph for Biden in Matt Parker’s video, the x- axis represents, as it should (according to Shiva’s theory) the Democrat party vote. In both cases (Biden and Trump) the x-axis represents the respective party votes (republican for Trump and democrat for Biden. Watch here https://youtu.be/aokNwKx7gM8 and see that Matt emphasizes this by using different colors for the two axes. In short, when you draw this graph for Candidate A, the x-axis should (according to Dr Shiva’s theory), represent the party vote for that candidate, namely Candidate A.
Alternatively you can use the same axis for both candidates, and use different color to distinguish between points on the graph. Dr Shiva is doing this in his second video.
BTW I am still waiting for your reply to my email letter. If you have not received, please let me know here.
00
Your reply arrived in good health, PG.
Thanks again for the report.
00