Looking In All The Wrong Places For Evidence Of A Hacked Presidential Election


December 19, 2016

Much like the generals of the early 20th century who were accused of always preparing for the war already past but not the war upcoming, many of my Democrat, liberal, and progressive friends are still stuck in the backwash of the 2000 election between George W. Bush and Al Gore. That election, you may remember, hinged upon the results in Florida, which was initially called for Mr. Bush—along with the presidential election itself—until massive discrepancies discovered in the way the ballots had been interpreted in the first count caused a recount to be ordered. Despite promising signs—or, perhaps, because of promising signs—that the recount would reverse the Florida results, giving both Florida and the presidency to Mr. Gore, the United States Supreme Court stepped in, halted the counting, and handed the crown to Mr. Bush. Many on the left never forgave Mr. Gore for giving in at that point and conceding the election, and ever afterwards have held onto the pledge that if such a situation ever came up again, they would both call for recounts and fight to keep them going to the bitter end.

Thus, no-one should be surprised that following the surprise and shock of the Trump victory in the recent presidential election, many Democrats immediately called for and supported a recount in states where they thought a victory by Ms. Clinton might have been stolen.

But 2000 Florida, with its notorious hanging chads and interpretive counting, was a special case that could never be confused with what happened in 2016. And so even before this year’s recount began to go off the rails (“Michigan Recount Over; Pennsylvania Sets HearingWashington Post December 8, 2016; “Trump’s Lead In Wisconsin Barely Changes As Wisconsin’s Recount ContinuesMilwaukee Journal Sentinel December 7, 2016), there seemed little chance of it actually turning up enough deliberate fraud or inadvertent miscount to swing the election from Mr. Trump to Ms. Clinton.

The problem is, though there indeed may have been massive enough fraud in the 2016 election to give the victory to Mr. Trump, to paraphrase the old song, we were probably looking for that fraud in all the wrong places.

And I’m not talking about any possible interference by our friends across the way in Russia, but something closer to home and distinctly American. I’m talking about taking a look at ballots that by their very nature, can’t be recounted.

And if that sounds confusing, friends, stay a moment, and let me explain.

While there are some variations, balloting in American elections can basically be broken down into three current types: hand-counted paper ballots, paper ballots fed into electronic tabulators to be counted, and direct recording electronic (DRE) voting machines using no paper whatsoever, in which the voter casts a ballot by touching their choices onscreen and the machine provides a count of the totals at the end of the voting day.

Ballot-count fraud to some degree or another probably goes on all the time in local American elections using hand-counted and electronic-counted paper ballots, maybe enough to change the results in some local elections. That’s because such elections are less likely to be closely-monitored, and the amount of fraudulence needed to alter the result in such elections can often be small enough to go undetected. But to commit enough ballot-count fraud to change the results in statewide balloting—which is how each state’s electoral votes for a presidential candidate are decided—is a different thing entirely.

To achieve a fraudulent vote reversal enough to swing a statewide election, one would either have to spread the fraud around to thousands of precincts throughout the state—a logistical nightmare, since you’d have to have operatives located in all those precincts and each and every one of their operations would need to escape detection—or else you’d have to do such a massive amount of illegal vote-switching in a smaller number of precincts that it would be easily detectible just by looking at the results.

So while you can recount hand-counted and electronic-counted paper ballots all you want in a presidential election, there’s a minimal chance such a recount will find enough fraud in a single state in a presidential election to make the desired difference.

That leaves us to consider recounting in the third major type of American voting, touch-screen voting with no paper trail and an internal self-counting mechanism, what are commonly called DRE’s. Unfortunately, DRE recounting is not even a joke. What’s the point in doing it?

Because there are no voter-marked paper ballots to count in a DRE-based election, the only way to recount in such an election is to go back and ask each DRE to do the count again. But in all of your experiences with a computer, have you ever had one flash a message on the screen that said, “oh, wait, my bad, I screwed up the first time I did this”? That’s not the way computers work, and a touch-screen voting DRE is simply a form of a computer. Absent some actual internal electronic malfunction in a DRE that was either a one-time fluke or was since fixed by a technician, a recount is most likely to give the same result as the first count, even if the first count was a miscount, accidental or deliberate.

That brings us to three follow-up questions on figuring out if DRE’s altered the count in any particular election enough to change the results from one candidate to another.

First, is it possible to manipulate a DRE to switch votes from one candidate to another? Second, would it have been possible to switch enough votes on DRE's in the recent presidential election to both be undetectable from a cursory review of the results and also to swing the victory from Clinton to Trump? And finally, since recounting can’t detect such a vote-manipulation in a DRE, is there any other way to find out if such manipulation might have taken place?

The answers are yes, yes, and yes.

First, let’s deal with whether it’s possible to manipulate—or “hack,” to use the popular term—a DRE so that it switches votes from the candidate actually voted for to another candidate.

It is, theoretically, if we throw out what has come to be the common interpretation of “hacking.”

In most—though not all—instances when most people talk about “hacking” a computer, they’re talking about doing it in the way the Russians were accused of hacking the Democratic National Committee computers in the last election, that is, with no physical contact with the DRE at all, but by intercepting from a distance instructions going to the DRE’s or results coming back from the DRE’s, the "hacking" taking place through a closed, dedicated network or the wide-open internet.

The problem with hacking a DRE in this manner is that these machines—with few possible exceptions—aren’t connected to the internet or a network. Let’s repeat that, in case one missed the point. The DRE’s themselves can’t be hacked over the internet because DRE’s aren’t connected to the internet.

How, then, are vote totals from the DRE’s transmitted to the central county counting station?

At the end of the voting day, each machine counts its own vote totals in a self-contained counter, and, at the end of the voting day, these totals are called-in or typed in a separate email that is then sent to the central vote counting location by precinct officials. Presumably one could hack into the county email system and change the results as they are transmitted, but the miscount would be quickly detected once the actual DRE’s were physically transported to the county election headquarters and the counting mechanisms of each DRE manually checked against the electronically transmitted results.

The second standard way of hacking a DRE would be to physically enter the machines manually and manipulate the count in some way so as to change the result. A number of computer experts have said that such DRE physical hacking is possible (“How To Hack An Election In 7 MinutesPolitco.com August 5, 2016). This method of hands-on hacking has one major advantage, and one major disadvantage.

The advantage of hands-on hacking is that—similar to the way computer “viruses” often act—someone doing such hacking of a DRE could introduce a program into the DRE that both performs a certain act and then, once the act has been performed, erases any trace of both itself and the fact that such an act was performed. Relating that to vote-hacking, such a hacking virus would switch a set number of votes from candidate A’s column to candidate B’s column and then—upon erasing both itself and the original and correc vote count—leave only the altered vote count. There would be no electronic “fingerprint” to show that any alteration had ever been done.

The major disadvantage to hands-on hacking is that in order to physically hack a DRE, you’d have to have direct contact with the voting machines sometime before the final count is done. That could happen either in the central county storage spaces where the DRE’s are kept in bulk before delivery to the various precincts just before election day, or else in the precincts themselves either just before the polls open or while voting is going on. As we have noted in talking about paper ballots, it would take the manipulation of a large number of individual DRE vote counts in order to affect the statewide outcome in a presidential race. The problem is, the larger the number of physical contact hackings needed to affect that outcome, the more difficult it becomes to coordinate such actions, and the more likelihood that some minor screw-up by one of the hackers—like being seen while doing the hacking—would lead to detection.

Does that mean that it is difficult to impossible to hack enough to DRE’s to change a presidential election?

Difficult, but not impossible, as the bodyguard told Michael Corleone in The Godfather II.

The most effective way to hack a DRE system to change the result of a presidential election is not to hack the machines once they get into the hands of the various counties for use in elections, but would have been to program an altered count into the DRE’s before they ever leave the factory and first get shipped out.

How could this work, in theory?

As computers, DRE’s are run by programs—made up of millions of lines of embedded code—that tell it how to operate in an election. In its most basic form, the DRE’s computer program tells it that just before a new election, someone will feed it the names of, say, five candidates running for a particular office. The program will go on to instruct the DRE that whenever a new voter accesses the DRE during an election, the DRE will display the names of each of those candidates on the screen along with a touch-sensitive area on the screen next to each name and, when the voter touches the area next to the name of one of those voters, tally a vote in the column of the particular candidate voted for. The program then instructs the DRE to keep a running total of the votes for each candidate, which it will display when asked for it by a precinct administrator at the end of the voting day when the DRE's are being closed out.

People who know little about computers—and that includes me—generally expect that while a computer can and does make mistakes, one of the basic functions it generally doesn't make mistakes on is addition and counting, which is essentially what happens in totaling up votes in an election. But that leaves out the fact that computers do what their programmers tell them to do, and a programmer can just as easily instruct a computer to do a deliberate miscount as they can to instruct it to do an accurate count.

This means in practice that the computer in a single DRE—or a whole factory full of DRE’s—could be programmed to alter the vote total for any given candidate or series of candidates in an election. A programmer could simply write in a subroutine to the main program of these DRE’s that instructs which reads, essentially, that “in a presidential race, take away five votes from the Democratic Party candidate by counting five votes that should have gone for them as either misvotes or no notes.” Or, alternatively, the subroutine could instruct the DRE to move five votes from the Democratic candidate’s column to the Republican candidate’s column. Such a subroutine would have to be written in such a way that such vote-nullifying or vote-switching did not turn up in testing of the machines, but only in an actual election. Such a vote-nullifying or vote-switching would be undetectable by simply reviewing the vote count at the end of the day, since DRE’s have no independent voter-marked paper ballot to check the computer-reported totals against if some discrepancy is suspected. And, as we've seen, asking the DRE to do a recount in such a situation would only result in the same miscounted totals being given back.

There were, indeed, a number of reports around the country of DRE’s switching a vote onscreen while the voter was trying to vote. Voters reported casting a ballot for Mr. Trump, for example, but watching the machine immediately switch that vote to Ms. Clinton, for example, or vice versa.

This is not the type of DRE vote manipulation I'm talking about above, however. Any halfway sophisticated hacking and reprogramming of a DRE vote total would never give anyone looking at the DRE’s screen any inkling that a switch or mis-tally was taking place, but would only make the switch during its internal calculations, away from sight.

That brings us to the second question: could switching as few as five votes a DRE from Ms. Clinton to Ms. Trump have altered the results in the 2016 presidential election, stealing the presidency from Ms. Clinton and giving it to Mr. Trump?

The answer is yes, and it could have happened in as little as two states.

Speculate with me, for a moment.

Pennsylvania is one of the states won by Mr. Trump which also uses a large number of DRE voting machines. According to the nonpartisan VotePA website, which is trying to get rid of these machines, over 85% of Pennsylvania's voters” use DRE’s with no paper trail. To get a rough estimate—very rough—of how many DRE’s Pennsylvania operates totally in a typical statewide election, take, say, Dauphin County’s total number of precincts—134—as an average. If each of those precincts housed four DRE voting machines, a reasonable estimate, and maybe a little low, that would put the total number of DRE’s in that county at 536. If each one of those machines switched five votes from Ms. Clinton to Mr. Trump, that would have resulted in a change of 2,680 votes from Clinton to Trump in that one county alone. Multiply that by the 50 counties in Pennsylvania that voted with DRE’s in 2012, the latest broken-down totals I could find, and you end up with an illegal switch of 134,000 votes from Clinton to Trump.

How many votes did Mr. Trump win Pennsylvania by? 68,236.

Georgia, which Mr. Trump won by 233,616 votes, uses paperless DRE’s. Florida, which uses paperless DRE’s in some of its counties, was won by Mr. Trump by 119,770 votes. A change of five votes per DRE wouldn’t be enough to switch the election totals in either of those states, but up the switch total to, say, 25 a DRE, and either Georgia or Florida could have been stolen from Clinton by DRE program manipulation (the emphasis being on "could have").

Give any combination of two states switched from Mr. Trump’s electoral vote column to Ms. Clinton’s out of Florida’s 29 electoral votes, Pennsylvania’s 20 electoral votes, and Georgia’s 16, and Ms. Clinton would be the incoming President of the United States, not Mr. Trump.

Thus, while this is certainly not evidence that such DRE computer code manipulation actually did change the results of the 2016 presidential election, it suggests that such manipulation could have changed the results, at a low enough rate per DRE to make it difficult to detect simply by looking at the vote total results given back by those DRE’s.

So are we screwed and doomed to both a Trump presidency and a permanent Republican thumb-on-the-voting-scales because internal code vote-switching in DRE's is undetectable after the fact?

Actually, while DRE vote-switching is undetectable by normal recount, it is detectable by other means,. That’s because the one flaw in such a scheme or pre-programmed DRE vote-switching is that such embedded code miscounting would be meant to stay in the DRE’s permanently, not for just one presidential election, but for all of them while the particular DRE is in use. (You wouldn't want to program the miscount for just one election because it only helps in switching the results in an election if both the election overall and the vote in the particular state in question are close enough, and there's no way in advance to know when this might happen.) Anyway, since the vote-switching subroutine is intended to remain deep in the bowels of the DRE permanently, its discoverable by any person with sophisticated computer skills that has access to the internal code of the DRE’s.

But that’s where detecting such a subroutine gets a little difficult.

What it would take to discover any possible hidden vote-switching subroutines in a DRE code would be an internal code audit of those machines by a qualified computer technician. Unlike recounts, however, there appears to be no legal recourse in the various state election codes to trigger such a source code audit. That would take a court fight, a nasty one at that, since the companies which built and programmed these machines and sold them to the various counties and states jealously guard those source codes from the public and would use all their powers to keep those codes from being audited. But getting a court-ordered code audit is the only way to find out if such machines have not been pre-programmed to miscount presidential election totals—or other election totals, for that matter—in favor of one political party or another.

The question is, are there any organization or group of citizens in the country who believe enough in the integrity of the country’s voting system, and believe that these voting systems were compromised inside some DRE’s during the late election with an embedded code that remains to compromise and manipulate it again in the future, and are willing to take on this fight in court?


Contact Jesse Allen-Taylor at safero@earthlink.net
Writing Pages Of J. Douglas Allen-Taylor