If Vote-Counting in America Were Honest, Bernie Sanders Would Be President
A startling discovery has been made by election integrity activists across the country: US counties which employ vote-counting methods which are difficult to hack do not display the unusual data patterns seen in the 2016 general presidential election, the 2016 Democratic primary, and in at least one congressional primary, which are suggestive of electronic vote-padding. The statistical anomalies seem to have disproportionately benefited Hillary Clinton and one of her allies, Congresswoman Debbie Wasserman-Schultz.
Electronic vote-padding takes place when the vote count for candidates is manipulated by the internal software of optical-scan vote-counting machines, according to maliciously-inserted instructions. Optical-scan vote-counting machines count the votes marked on manually inserted paper ballots, and are in use in over 70% of US voting jurisdictions. The ease of performing such manipulations was demonstrated in the HBO documentary Hacking Democracy.
One pattern which greatly puzzles election analysts is smoothly increasing vote shares - percent of the vote - for a particular candidate, as the number of voters in a precinct increases. The odds of such a pattern happening repeatedly through random chance are statistically impossible. In a normal voting data distribution, data tends to level out and converge on an average.
The pattern is found in such electorally important states as New York (below.)
The pattern, also known as vote-flipping, was identified in 2012 by supporters of Republican presidential candidate Ron Paul. Examining election data for Ron Paul in the New Hampshire primary, mathematicians made the case that the Paul data was unnatural and indicative of fraud. The data showed what statisticians considered an oddity: as precincts were lined up according to the total number of votes cast, from smallest to largest, the percentage of votes garnered by Paul decreased while those of his main opponent, Mitt Romney, increased. Other candidates in the primary race were unaffected (below.)
Likewise, in the 2016 general presidential election, both at the state and county levels, smoothly increasing vote shares can be observed across the country, almost always to the benefit of Hillary Clinton. Almost always, but not quite. In Racine County, Wisconsin, Donald Trump shot ahead of Hillary Clinton in vote share as the number of precinct voters increased, prompting statistics experts to call the results "concerning."
For the sake of simplicity, only votes for the candidates of the two major parties are counted in some calculations. This does not affect the relevant patterns.
Why would one candidate or the other see a consistent, linear increasing vote share as the numbers of votes in precincts, or counties, increase? One hypothesis put forth to explain the 2016 general election anomalies is that more populous counties tend to be more urban, and therefore more Democrat-leaning and likely to vote for Clinton. But the Racine, Wisconsin Trump data wreaks havoc with this hypothesis. Racine County is considered part of the Milwaukee metropolitan area with a minority population of about 20%, a demographic which would seem favorable to Clinton.
A second objection is that such a hypothesis would not explain the steady increase in Clinton's vote share even within distinctly rural counties, whose demographics cannot be characterized as urban in any way. In Illinois and Michigan, two key swing states, vote share for Clinton increases linearly across precincts of all sizes, at a constant rate, and not just as the charts hit the large urban precincts of Chicago and Detroit. This can be seen in election analyst Richard Charnin's blog. If the urban-Democrat hypothesis were true, one would expect a flat-line distribution throughout perhaps two-thirds of the chart, consisting of rural precincts, then a sudden jump as larger city precincts come into play.
Further putting doubt on the urban-liberal hypothesis are the similar patterns shown during the 2016 Democratic primary between Hillary Clinton and Bernie Sanders. This time, the hypothesis ran that more populated urban centers tend to have more African-American voters, and Clinton did well with African-Americans. But this does not explain why Clinton's vote share steadily rises, in a nearly perfect, linear manner, in all small, rural counties, according to precinct size. The mathematical regularity of the pattern is uncanny.
Interestingly, when actual African-American population patterns are studied relative to county size, no smooth and regular increase can be seen which corresponds to vote share in urban precincts. Instead, African-American population spikes in very different sizes of precincts, which should register as bumps in Clinton vote share, but do not (below.)
Other evidence which casts a shadow on the liberal-urban-center hypothesis is the peculiarly normal behavior of the Jill Stein vote as counties get larger in population. If urban centers are more liberal, one would expect Green Party candidate Stein's vote share to increase somewhat along with Clinton's. This is especially so in college campus locations which were hotbeds of disaffected Sanders voters virulently opposed to Clinton.
Unusual Voting Patterns Absent from Counties with Vote-Counting Transparency
Perhaps the most damaging comparison to machine-counted votes which show increasing vote share behavior, is data from counties where vote-counting is performed by hand, or by optical scanners using open-source software. These counties are Columbia County, New York, and Humboldt County, California. Except for these two counties, and scattered exceptions at the town or precinct level, all election functions in the US are performed by machines which employ corporate-owned, proprietary software, which not even election authorities can examine.
Humboldt County is also the California County where Sanders received the highest percentage of votes in the state, beating Clinton 70% to 29%.
An examination of cumulative vote tally charts for recent elections in these counties shows no trace of any unusual patterns seen in machine-tallied districts, such as precinct-size-related vote shares. The chart shows typical early scattering of data points when the numbers are small, converging on a flat-line average as the sample size increases. This is normal statistical behavior (below.)
In open-source software, every line of programmed code can be examined for malicious instructions. If machines are to be relied on to count votes, such software is considered to be a vast improvement over proprietary software. But the international “gold standard” for integrity in elections is still the counting of paper ballots by hand, in public, immediately after an election. This is based on the logic that the shorter the time that ballots are left unattended before counting, the smaller the window for illegal behavior.
In Columbia County, paper ballots are hand-counted immediately after an election. Most election experts consider this system, properly executed, as the best, most hack-proof way of conducting elections, in districts or cities of any size. To the objection that employing a hand count system in someplace the size of New York City would be “impossible,” Columbia County Election Commissioner Virginia Martin counters that New York “has 100 times the voters we have, but also 100 times the resources and people.”
It is easy to express in mathematical terms the formulas which apply to increasing vote shares. In any straight-line graph, the general expression is y=ax +b, where “y” is the vertical axis variable, “x” is the horizontal axis variable, “a” is the slope of the line, a constant, and “b” is the “y”-axis intercept, also a constant. Therefore, at any point in the cumulative vote count along the “x” axis, the corresponding (perhaps desired?) vote share may be calculated.
For example, across counties in Illinois between Clinton and Trump, a nearly straight line slope in the Clinton shares of about 1/4 may be observed, with a corresponding negative -1/4 slope in the Trump line. At 3 million cumulative votes, for the slope 1/4 and “y” intercept 25% (which is merely the number of votes in the first, smallest precinct,) the formula yields 47%, almost precisely what is observed.
In the case of the 2016 New York Democratic Primary between Sanders and Clinton, pictured above, a simple parabolic function fits the curves with reasonable closeness. Hillary starts at a steep slope then tapers off in her accumulation of vote share as a comfortable margin of about ten points is assured, at about 120,000 cumulative votes. In fact, remarkably enough, one can identify not two smooth curvilinear lines, but three more or less straight lines of slightly different slopes, in which one can see what could be a "course correction" tweak at 120,000 votes, and another at about 170,000 votes.
An interesting discovery by election integrity analyst and computer expert Bennie Smith in 2016 was the capability, in many optical scan vote-counting machines, to register fractional votes. Since a single vote is by definition an integer, as is any multiple of single votes, Smith asked why vote-counting software would be enabled with a decimal point which could express fractional votes, and not just fractional percentages of the total vote. Oddly, this is the very capability a system would require for it to be programmed to deliver specific margins to predetermined winners. Activists labeled the discovery “fraction magic.”
If an election outcome were to be engineered by hackers wishing to deliver a certain percentage of the vote to a candidate, Smith reasoned, the variable which would remain unknown until near the end of the day would be voter turnout, i.e. the total number of votes. Therefore a decimal point would needed in the software's definition of a vote, to allow any degree of precision in manipulating the outcome.
In displaying the final public results, of course, the decimal point would be eliminated from the visible vote totals reported, resulting in what computer experts call a "round-off error" which often amounts to a single vote. Perhaps coincidentally, officials in Michigan in the 2016 contest between Clinton and Trump found a puzzling pattern of more votes present in precinct totals than voters who checked in, in 37% of Detroit's precincts. The size of the difference was almost always exactly one vote.
So unusual was the strange dangling vote in so many counties that Krista Haroutunian, chairwoman of the Wayne County Board of Canvassers, was prompted to tell the Detroit News:
“There’s always going to be small problems to some degree, but we didn’t expect the degree of problem we saw in Detroit. This isn’t normal,”
Is it Possible?
One of the principle objections which defenders of the current election system raise to suggestions of vote padding at the machine-count level is that with thousands of machines across the country, it would be impossible to program every one. However, this is not quite true. Precincts and counties are often connected by networks which transfer information such as vote totals at the end of the day. Malicious instructions and programming could theoretically propagate across networks in short periods of time.
As a quick test of this reasoning, a cumulative vote shares chart was constructed for a voting district, the City of Somerville, Massachusetts, where voting machines are not networked together. If this introduced a degree of "hack proofing," then the results in Somerville would show no unusual patterns. Sure enough, a cumulative vote shares chart of the city shows an almost textbook normal, flat-line distribution of data.
Just What is America Now? Do We Have Free Elections?
Is America a patchwork of electoral fiefdoms in which hacking is widespread, and winners are determined by a form of electronic warfare which rages behind the scenes?
If so, it is likely that party affiliation is less important than the prevailing power structure arrangements. Rather than the Jets and the Sharks, our political landscape might bear a greater resemblance to the Gangs of New York, with fluid and shifting alliances the norm. In the contest between Sanders and Clinton, many Republicans bore a far greater resemblance to Clinton than Sanders in terms of economic and foreign policy.
If no explanation of the mathematical formula hidden within some vote share charts is forthcoming, an initial speculation might be that, in 2016, Clinton forces were hard at work stealing votes both in swing states and padding the popular vote in non-swing states, with occasional exceptions where Trump forces were doing the same thing. In the end, the speculation might go, the Clinton hacking forces could not outpace an unforeseen groundswell for the perceived non-establishment Trump, without calling attention to extraordinarily unusual vote patterns in key states.
It is inescapable that in 2016, many Democrats and independents either stayed home, or voted for Trump. In 2008 just under 70 million voters turned out for Barack Obama. In 2016, Hillary Clinton garnered 65.8 million votes, or 4 million less than Obama in 2008. As heated as the election was, Clinton failed to inspire the numbers to come out that Obama did in 2008.
Hillary supporters who hear these arguments are quick to dismiss all such discussion as “conspiracy theories.” But if this is so, why are state and local election departments fighting tooth and nail to prevent citizens from viewing the only evidence which will prove whether these "conspiracies" are true or not? Meaning, the paper ballots? Numerous lawsuits have been filed across the country by citizens demanding that paper ballots be treated as public record. To date, election authorities have vigorously opposed all of these.
Not only are authorities reluctant to allow the supervised and orderly review by citizens of paper ballots; they are also opposing what one would think would be a compromise: the release of digital ballot images which are created by a significant portion of paper ballot vote-counting machines, which could easily be copied onto a DVD or posted online. (See WhoWhatWhy.org: "Will Wisconsin Allow Experts to Examine Digital Images of Recounted Ballots?")
Not only presidential races, but congressional races as well can show unusual vote patterns which may indicate election manipulation. In Florida, a Democratic primary challenger to incumbent congresswoman Debbie Wasserman-Schultz received the attention of noted statisticians when the results of the election showed a similar unusual vote pattern. In the blog HollerbackFilm.com, statisticians wrote:
"As more and more votes are added in, instead of the pattern approaching the average level of support for each candidate (a flat line) Wasserman Schultz' percentage climbs continuously in a mathematically precise pattern, and Canova's percentage decreases steadily...This is not the expected statistical pattern."
If suspicions of vote-padding are true, its impact on history is enormous. During the Democratic primaries, every head-to-head poll between Trump and Sanders indicated that Sanders, lacking Hillary's negative ratings, would have solidly trounced Trump in the general election. Proof of shenanigans among Hillary forces was rampant.
Most advanced, industrialized countries have returned to counting all ballots by hand as the most potent frustrater of vote-counting machine hacking. These countries include Germany, Canada, France, Ireland, Italy, Denmark, Finland, and 53 other countries. This year the Netherlands began hand-counting ballots in response to concerns over machine hacking. Numerous organizations have been pushing for a national policy of hand counting ballots.
Until America turns to a nationwide “gold standard,” like the rest of the advanced democracies, of hand-counted paper ballots, in public, Americans can never be confident that the will of the people is being truly reflected. They may continue to be saddled with a "lesser of two evils" choice between two unpalatable candidates, as many complained in 2016. Efforts to change the laws may be a good way to harness the energies unleashed by the current political climate, in a manner which takes aim at the root of the problem.
MORE INFORMATION #HandCountTheBallots
(Footnote: In the Democratic primary evidence of fraud on the part of election authorities on behalf of Clinton was rampant. It included millions of incorrectly purged voters in NY, AZ, and CA, for which official apologies were issued. The purges disproportionately affected voters leaning toward Sanders. Also witnessed but never investigated by the Obama DOJ was eyewitness testimony before the Chicago Board of Elections of fraud by Chicago election officials, who went as far as erasing paper vote tallies on a white board and altering the tally to match machine-counted vote totals. In San Diego, citizen observers of the vote count of provisional ballots took video of ballots on which votes for Sanders were visibly erased with white-out.)