The case for the occurence of algorithmic vote flipping

[dsw]

DSW- there are certainly lots of precincts where this trend will hold, BUT there are also precincts where the opposite is true- Romney loses a few votes. For me, we have to look at as many multiple- machine precincts with varied vote counts as possible. I really have no idea if we will see a trend or not.

I look forward to seeing multiple-machine precincts where the vote counts are similar, too. If those have significant differences, then either your assumption is wrong or there was tampering of a different sort. Or maybe just a probabilistic threshold. Or if there are no significant differences when the counts are similar, but there are when the counts are different, that would make your case stronger.

If fraud was at the tabulator, why release the EVM data at all, when it means having to construct fake EVM data in a way that is consistent with both the precinct total and the individual EVM totals? And if they were constructing fake EVM data, is there a plausible way of doing it that would result in a difference like that one data point, where the one machine looks pretty much like the other one except that Romney got a bunch of extra votes -- and they got the delegate count consistent in that case, but way off in other precincts?

 

[RonRules]

I don't assume that we're going to find anything here and RonRules feels confident that the central tabulation is the culprit.

I still am . That's because the nice PDF report they gave you came out of the Central Tabulator software.

I'll admit I'm wrong immediately, but right now I'm still sticking with the CT theory.

All I want is Romney in handcuffs. I don't care where the deed was done.

 

[dsw]

I need to "Precinct-Level data" in order to analyze this more conclusively. Please help me find this data.

Try:
http://www.commonwealthmagazine.org...ction-results/2002-gubernatorial-results.aspx
which links to a spreadsheet in the first paragraph.

 

[RonRules]

It reminds me of your probability calculation proving that the absentee vote for Mitt was *impossible*.

Give it up DSW. [personal attack removed by moderator]

The probability DSW is referring to, (for you new to this thread) is an absentee ballot count in the Village of Greendale, WI were Ron Paul got 3 votes vs. 133 for Mitt Romney. Comparing the statistics of those precincts with the overall performance of Ron Paul in Wisconsin, my statistical calculator came back with a probability result of ZERO.

In other words the calculator said those results were IMPOSSIBLE, considering the statistical population basis being compared to.

 

[dsw]

All I want is Romney in handcuffs. I don't care where the deed was done.

If you didn't misunderstand the math in your proof that Romney's absentee percentage was *impossible*, then it's a very straightforward proof. If the same pattern holds in more than one state, and you aren't wrong about the math, then you've caught him redhanded, and with a proof that anyone with just a little bit of background in statistics will understand.

If there was flipping in Iowa, and if you understand the math about flatlining graphs correctly, then a substantial percentage of the precincts there had to be flippers. The votes were counted manually, and in public, and Ron Paul supporters were encouraged to record the results of those manual counts. It's virtually certain (if you're right) that some of the flipper precincts had their manual counts reported independently, and (if you're right) they'll show tell-tale manipulation that moves votes from one candidate to another but keeps the total the same. One would raise suspicions and get big attention. Two would probably doom Romney's campaign. Three or more and he'll be shopping for mansions in a country without an extradition treaty. And, if you're right about the flipping, the chances of a few flipper counties being among those with independent reports of the manual counts are a virtual certainty. All that needs to be done is collect them, and compare against the central tabulator numbers, until you find the flippers.

Either of these, unless you've made a very basic mistake in your understanding of the math, would be the kind of easily-accessible proof that would knock Romney out of the race.

 

[dsw]

Give it up DSW. [personal attack removed by moderator]

The probability DSW is referring to, (for you new to this thread) is an absentee ballot count in the Village of Greendale, WI were Ron Paul got 3 votes vs. 133 for Mitt Romney. Comparing the statistics of those precincts with the overall performance of Ron Paul in Wisconsin, my statistical calculator came back with a probability result of ZERO.

In other words the calculator said those results were IMPOSSIBLE, considering the statistical population basis being compared to.

Um, I'm agreeing with you that (unless you are misunderstanding something very basic about the math) you've proven that the result is impossible. Wasn't there another outlier that turned out to be absentee votes, too, one where (if flipper math is correct) the last point on the graph took an impossible upward jump? I'll bet you can find others too, but impossible means impossible. If it's impossible without fraud, then it must be fraud (or else you don't understand the math).

I'm agreeing with you! You have proof that the result was impossible, unless there was fraud (or unless you have some basic misunderstandings about the math). You want to see Romney defeated. Proof that he benefitted from fraud would knock him out of the race. You have that proof (unless you are mistaken about the math), and in a very, very straightforward argument. Why let that proof languish in a comment on a forum like this? Put it out there, post it widely. It's such a simple, straightforward argument that anyone with just a little bit of background in statistics will quickly see it for what it is.

UPDATE! Mitt won Kalamazoo County in Michigan by just 663 votes ... 9,671 to 9,008. But the absentee vote went 2,479 to 1,052 for Romney. Plug that one into your calculator! Or Macomb county, Romney edges it out with 23,188 to 22,999 on election day but ... brace yourself ... 14,650 to 7,219 absentee! Holy cow. It's less than 1% difference at the polls but two-to-one absentee? What are the odds?? Either this is fraud, or ... well, you know.

 

[drummergirl]

Or maybe just a probabilistic threshold.

Please define this term.

If fraud was at the tabulator, why release the EVM data at all?

Because the election officials are ignorant of the theft in their own house.

 

[drummergirl]

Why let that proof languish in a comment on a forum like this? Put it out there, post it widely. It's such a simple, straightforward argument that anyone with just a little bit of background in statistics will quickly see it for what it is.

well, DSW, I've talked with lots of people about this IRL, online, everywhere I have influence. Of course, I'm not Rupert Murdoch. I happen to know that RonRules has gone to universities, presented this work to statistics professors and grad students, and is working with his local election officials. What exactly have you done? Besides make circular arguments, of course.

 

[drummergirl]

If you didn't misunderstand the math in your proof that Romney's absentee percentage was *impossible*, then it's a very straightforward proof. If the same pattern holds in more than one state, and you aren't wrong about the math, then you've caught him redhanded, and with a proof that anyone with just a little bit of background in statistics will understand.

If there was flipping in Iowa, and if you understand the math about flatlining graphs correctly, then a substantial percentage of the precincts there had to be flippers. The votes were counted manually, and in public, and Ron Paul supporters were encouraged to record the results of those manual counts. It's virtually certain (if you're right) that some of the flipper precincts had their manual counts reported independently, and (if you're right) they'll show tell-tale manipulation that moves votes from one candidate to another but keeps the total the same. One would raise suspicions and get big attention. Two would probably doom Romney's campaign. Three or more and he'll be shopping for mansions in a country without an extradition treaty. And, if you're right about the flipping, the chances of a few flipper counties being among those with independent reports of the manual counts are a virtual certainty. All that needs to be done is collect them, and compare against the central tabulator numbers, until you find the flippers.

Either of these, unless you've made a very basic mistake in your understanding of the math, would be the kind of easily-accessible proof that would knock Romney out of the race.

This has to be your most hideous post yet. DSW, go back under your bridge.

 

[dsw]

Please define this term.

(Probabilistic threshold) If there's a sharp threshold, with flipping above X votes total and not below, then pairs of EVMs that straddle the threshold point would show a sharp and repeatable contrast. A smarter algorithm would "throw the dice" to decide whether to flip, and maybe how much to flip, with the probability and maybe also magnitude of flipping increasing with larger vote totals. That wouldn't show up as dramatically but it should still show up if there's enough data to look at.

 

[drummergirl]

that's quite a bit different than the definitions of the two words would tend to indicate.

Probabilistic - having to do with the statistical possibility that an event will occur

Threshold - a minimum level needed for something to take place. e.g. A sustained nuclear reaction requires a mass of fissionable material above the critical mass threshold.

If there's a sharp threshold, with flipping above X votes total and not below, then pairs of EVMs that straddle the threshold point would show a sharp and repeatable contrast.

Yes, that's been discussed for a number of posts now. What do you see as "probabilistic" in such a number?

(Probabilistic threshold) If there's a sharp threshold, with flipping above X votes total and not below, then pairs of EVMs that straddle the threshold point would show a sharp and repeatable contrast. A smarter algorithm would "throw the dice" to decide whether to flip, and maybe how much to flip, with the probability and maybe also magnitude of flipping increasing with larger vote totals. That wouldn't show up as dramatically but it should still show up if there's enough data to look at.

 

[dsw]

This has to be your most hideous post yet. DSW, go back under your bridge.

I think the absentee argument is bogus, and is based on a misunderstanding similar to the one behind the "graph must flatline" argument. If it were valid then it would be a proof, and a very simple and straightforward one, of fraud. Enough to knock Romney out of the race. But it's not valid, and it's wrong in a very basic way.

If there were flipping in Iowa then the number of precincts where the manual counts were independently reported is large enough, and the number of precincts that would have had to have been flipped (if the "graph must flatline" argument were valid) would be large enough, that there would have had to have been some overlap and flipping would have been obvious. There were some "typo" kinds of errors, but no "flippers" with Romney +N and Paul -N or anything of the sort. So it's a debunk.

But I think you're right about one thing. Pointing out these things is clearly not welcome. And more and more I've descended into mocking what I regard as bogus arguments, and mocking people for pursuing something they feel passionately about, in whatever way they feel like pursuing it, is hideous.

Thanks for saying that. I apologize for the mocking tone.

 

[drummergirl]

If there were flipping in Iowa then the number of precincts where the manual counts were independently reported is large enough, and the number of precincts that would have had to have been flipped (if the "graph must flatline" argument were valid) would be large enough, that there would have had to have been some overlap and flipping would have been obvious. There were some "typo" kinds of errors, but no "flippers" with Romney +N and Paul -N or anything of the sort. So it's a debunk.

Unfortunately, you are wrong on this one. There has been a great deal of controversy in Iowa since election day about many issues. You keep saying everything is ok, but I've never seen anything from you but hand waving arguments to explain this:

and, now I must get some sleep; I have a gig tomorrow.

 

[Pauling]

I was just getting those typos out of the way, in response to the one he posted. He claims that there is evidence of over a hundred more discrepancies in Iowa, discrepancies between the manual count done in public and the result from the central tabulator. If there was flipping in Iowa -- correct me if I'm wrong on this estimate -- then based on your graph something like 15-30% of the precincts were flipped, starting with precincts as small as 70 total votes or so. If he's got 120+ verifiable discrepancies then it's a virtual certainty that some of those are in the flipped precincts.

Even just a handful that show Romney getting the same number of "extra" votes that other candidates are losing would *end* Romney's campaign. Imagine that happening before the Texas and California primaries.

There are no poll tapes in Iowa. But the count was done manually and in public, with observers allowed to record the results with video and photographs. A lot of them were recorded. Watchthevote2012 had an organized effort to get people to record and report the manual counts, and there were quite a few others posted to youtube, facebook, twitter, blogs, and in local papers.

If there was flipping in Iowa, then there is zero chance that the flipped precincts would turn out to have no overlap with the precincts that had the manual counts independently recorded. If there was flipping, then the proof that would end Romney's campaign is out there, and has been for more than three months. Better late than never but time is running out.

Yo dog...
http://www.cbsnews.com/8301-503544_162-57349486-503544/occupy-protests-send-gop-vote-counting-to-secret-location/

 

[melodicEntropy]

[system won't let me post an image]

For the sake of clarity, the X-axis is what, here?

 

[RonRules]

For the sake of clarity, the X-axis is what, here?

Cumulative Precinct Vote Tally expressed in %.

That's all the precincts vote counts added up with each other from left to right. The last point on each trace on the chart includes ALL precincts, therefore ALL demographics.

That's important to know because these types of charts tend to remove the demographics effects. If you plot pure demographics with these charts, they will flat line. (almost perfectly).

Here's an example of different types of Republican females, voter registration:

Also this:

This technique is so good that I suspect census fraud with Hispanic Democrats with the 2010 census. There are benefits to cheating the census: Electoral College votes, Delegates, Congressional Reps, Federal aid.

 

[affa]

Example: if every machine were programmed so that at 250 votes Romney begins receiving 5% extra of the vote from Gingrich, a precinct with a single EVM would be affected at a precinct vote of 250 votes; a 2- EVM precinct with equal votes in each machine would be affected at 500 votes- and so forth.

Interesting theory. And obviously, the actual number would show some variance. Assuming three machines in a line, and 900 voters, we might see the first machine get 400 voters (150 votes affected), the second machine get 300 voters (50 votes affected) and the third machine get 200 voters (0 votes affected), for a total of 200 affected votes. That's just an example, obviously - not just the voter totals, but the 'hinge point' as well.

And that's assuming it's done at the EVM and not the central tabulator, of course.

 

[affa]

But I think you're right about one thing. Pointing out these things is clearly not welcome. And more and more I've descended into mocking...

Did you ever stop to think you might be conflating or confusing the reasons some people dislike you? Just reread your own comment.

Perhaps you get a cold reception because of your mocking tone (and tendency to drive a conversation in circles by focusing on inconsequential minutia or by re-asking questions you've already asked a dozen times) and not because of the occasional salient point you may make.

You started out acting like a doddering professor slow on the uptake, asking over and over again for the same explanations. Then you shifted to repeating the same misleading statements over and over. Then you straight up became rude. And the whole time? You distract from the actual conversation being had, driving the conversation away from new points (like Romney vs. Romney delegates) into unimportant tangents. Yea, maybe that's why you don't get a warm welcome. Just maybe.

 

[Aden]

So if I understand things correctly, after California results come in, we can use the fantastic precinct-level demographic data to finally put to rest the notion that demographics are responsible for Romney doing better in larger precincts?

 

[drummergirl]

So if I understand things correctly, after California results come in, we can use the fantastic precinct-level demographic data to finally put to rest the notion that demographics are responsible for Romney doing better in larger precincts?

yes