Voting Machine Verifications Show Walker Lost Recall Election

The required verification of Wisconsin's electronic voting equipment after elections has shown that election fraud occurred in several counties during Scott Walker's recall election. The corrected election results show that Barrett won with 51% of the vote, matching the percentage of voters that voted for Democratic Senator John Lehman and favoring President Obama in the exit polls as well as more closely matching the 50/50 split exit polls for Walker and Barrett. It turns out that it wasn't the exit polls that needed to be adjusted, it was the election results.

As one of the first steps in the canvassing process after an election, each electronic voting machine and ballot counting machine is verified for its integrity. The post-election verification process is the same as the pre-election verification process. First, the integrity of the software on each machine is verified, and if the verification fails, it's assumed the machine was tampered with. If the machine passes, the function of the machine is tested. For optical scanners, election officials run a number of randomly selected completed paper ballots through the scanner and match the totals against a hand-count. For touchscreen voting machines, election officials repeat a number of votes from a random position in the paper log and match the totals against a hand-count. The machine has been compromised if the totals don't match.

The verification of the open software installed on the equipment uncovered nearly 100 machines across Wisconsin that had been compromised. It's believed that at least one person, most likely two or three, compromised the optical scanners after the pre-election verification by connecting a small device to the diagnostic port on the machine which automatically downloaded the modified software. The software only affected the gubernatorial election, which explains Lehman's win.

All of the votes counted with the compromised machines were recounted by verified machines or hand count, and the recounts show that 110,805 Barrett votes had been wrongfully switched to Walker by the compromised machines. It appears as though the modified software flipped Barrett votes at random short durations of time, so the modified software changed vote totals differently across machines. Outagamie County had the largest percentage of switched votes with 7,741 votes, 10% of the total votes in the county, but Walker still won the county nearly as much as he did in 2010. Milwaukee County had the largest number of flipped votes with 31,154, followed by Dane County with 10,682 flipped votes. The other counties with modified software were Brown, Calumet, Dodge, Door, Kewaunee, Kenosha, Marathon, Manitowac, Oconto, Racine, Shawano, and Winnebago. Interestingly, nearly every county is along Highway 41 or very close.

Since we control the open software on the machines, we can quickly plug the security hole at no cost to taxpayers for both detection and the fix. Several Wisconsin software engineers wrote a fix for the problem within a few hours of the announcement by the Government Accountability Board, and the new software will go through a number of tests before it's certified. Rest assured, our voting machines will be fixed and certified long before the next election thanks to it being open source software.

As a truly bipartisan measure, Scott Walker reportedly asked Tom Barrett to attend the brats and beers today. Walker conceded, "We have yet to determine who tampered with the equipment, but I'll relinquish my command to the rightful Governor of Wisconsin, Tom Barrett." The person(s) responsible for tampering with election equipment and swinging the election are still being investigated.

Unfortunately, someone may have tampered with voting equipment during the recall election, but we'll never be able to find out. Nearly everything above is just a dream. Voting machines are not open and are not verified after elections, and so there is no way to detect such election fraud. None of the precautions listed above are actually done in Wisconsin, but the possibility of someone swinging an election by tampering with voting machines so easily is real. Barrett could really have won with 51%, but we'll never know.

I'm very sorry if that was painful, but I hope it was effective in convincing you how important this issue is. It's a far more critical issue than voter fraud. We have no proof whether or not Walker really survived his recall, but we could. The next time a Democrat wins a big election, you can bet the Republicans will be shouting election fraud, knowing the problem exists but refusing to do anything about it. Lets get this fixed now.

I wrote a blog post following the Prosser/Kloppenburg election detailing the issues with our elections and the solutions available. By fixing these issues, we could detect and compensate for election fraud, and the story above could be a reality. The following is a minor update to the post, putting it into the context of this election.


Our most pressing issues with elections in Wisconsin continue to be:

1. Ballot security and integrity
2. Antiquated unsupported voting equipment
3. Insecure and inaccurate voting equipment
4. Voting equipment hardware design and software not open, owned and controlled by the people of Wisconsin
5. The integrity of every voting machine isn't verified before and after every election
6. Various vote-tallying processes are not open
7. Ease and likelihood of errors in reporting
8. Voting equipment and software is not uniform across the state

(1) Ballot security and integrity. There were many unsealed and ripped ballot bags during the Prosser/Kloppenburg election. Nothing was done to prevent this from occurring in future elections. The only reason we knew about these issues was because of the recount and the ability to watch some online. There could be far more issues than we are aware of from other elections.

(2) Antiquated unsupported voting equipment. The Prosser/Kloppenburg recount showed us that at least one approved model of our voting equipment, the Optech Eagle, is antiquated and must be removed from service. The original vote tallies from that election had to be deleted, against the law, in order to carry out the recount with the machines, because there were not enough memory packs and those memory packs are no longer manufactured. Although, I will say that I believe it's good that the issue forced a hand recount in parts of the state instead of simply re-feeding the ballots back through the same machines for a recount. This model should be replaced quickly, though the following issues may warrant some delay. A full review of all our voting equipment should occur first.

(3) Insecure and inaccurate voting equipment. Our electronic voting equipment is insecure and inaccurate, but you're not supposed to know that. The very equipment we use to cast and count votes can be manipulated without detection in seconds, swinging the results by any number of votes. There's not a single computer or security expert that would argue with that. However, even if tampering doesn't occur, our vote-tallying machines rarely count the exact number of votes. These are inevitable consequences of using electronics to cast and count our votes. Unfortunately, the design of our electronic voting machines and their margins of error are secrets kept tightly by their manufacturers. So we have no way of knowing just how insecure and inaccurate our machines are (more on this in (4)).

Quite near anyone with the knowledge to write moderately sophisticated computer programs can manipulate a voting machine and its vote tally in literally seconds. That's well into the tens of thousands if not hundreds of thousands of people in Wisconsin with that ability. Not at any other point in our country's history do so many people have such an ability to swing elections so dramatically. It really does take some time to stop and consider.

No one has any "bullet-proof" solutions to this problem, and it's likely we never will. If we want to count votes using electronic equipment, and I for one think it's a "good thing," we must have verifiable guarantees of voting machine security and accuracy. We don't have that now, and we almost certainly won't have that until we solve (4).

There are other issues regarding accuracy specific to the voting machines we use. A GAB memo from December 2009 shows that there were several issues with voting machines that we currently use in many parts of Wisconsin. An error message "error while printing" occurred 15% of the time during a test of the AutoMARK VAT. This was due to a malfunction in the system which required replacement, and the replacement had similar non-tallying related issues about 5% of the time. Such errors can cause inaccuracies, and they can also cause voter disenfranchisement as voters may be told to come back later or may have to wait for an excessive period of time.

The AutoMARK VAT is used as an option for voters with disabilities in many municipalities across Wisconsin. However, the GAB memo says that the testing by the GAB and testing by the Wisconsin Election Administration Council shows that

"The AutoMARK VAT does not provide full privacy and independence for voters with disabilities, especially voters with dexterity or motor disabilities, as voters may need assistance inserting the ballot, removing the ballot and placing the ballot in the ballot box or tabulator."

The Wisconsin Election Administration Council had even more to say. The memo notes several issues including vision-impaired voters won't be able to verify their vote, inadvertent steps that cause a cancellation of votes, the device doesn't meet 2005 US-EAC guidelines, "it takes longer to cast a ballot with the AutoMARK than manually marking the ballot with a marking device," and screen reading difficulties. Yet the AutoMARK VAT was approved by the GAB for Wisconsin voters with disabilities. The GAB said, "The ES&S voting system technically meets" the requirement of a voter to privately verify their votes. The GAB later notes in a bordered paragraph,

"The AutoMARK voting systems for which approval is being sought, do not change the degree of accessibility currently provided by previously approved AutoMARK systems."

I'm appalled that so many of our voting machines do "not provide full privacy and independence for voters with disabilities," and that the GAB would approve such a machine. Are we that desperate for voting equipment?

Another machine mentioned in the memo with issues regarding its ease of use is the intElect DS200. It may not be immediately clear that these issues are issues with accuracy. Any time a voter's intention doesn't get correctly included in the official results, the system is inaccurate. The more difficult a machine is to use, the less accurate it will be.

Similar touch-screen machines continue to have worrisome issues in other states as well. In the 2010 election, a touch-screen voting machine in Pennsylvania began casting votes for the opposite candidate from the one selected by the voter, and the machine required "recalibration" to resolve. We don't use the same machine here, but the same manufacturer, ES&S. So there's good reason to suspect the same issues can and may have happened here. Luckily for us, all of our touch-screens mark or print a paper ballot, but most people expect the machines won't make a mistake. So they may not properly inspect the results before casting their vote and walking away.

The GAB incorrectly states on their website "Adminstrative(sp) Code Chapter 5 Ballot and Electronic Voting Equipment Security insures all electronic voting systems used in Wisconsin are accurate and reliable." This is plain false, and it provides a false sense of security to those voters who aren't aware of the issues. I don't believe we will ever fully be able to insure electronic voting systems are accurate and reliable, but I believe we could publicly guarantee much higher security, accuracy and reliability if we solved issue (4).

Wisconsin does have some of the best electronic voting machine laws in the country, but they're far from perfect. What's worse is that at any time the GAB can exempt a machine from complying with Wisconsin law. The GAB can exempt a voting machine from Wisconsin law if they choose, or as they say, "for good cause" GAB 7.03(5). I don't see how exempting a voting machine from Wisconsin law is a good idea at any time for any reason.

(4) Voting equipment design not owned by the people of Wisconsin. When I say that the people of Wisconsin should own the designs for our voting equipment I mean that the hardware designs and the software source code should be open, i.e., in the public domain. I'll elaborate more on this in the solutions section later, but I'll briefly cover some highlights and comparisons now.

The manufacturers of our voting machines will never provide us with the information and control we need to conduct our elections in the most fair, open and transparent way. The people who run our elections, like your county clerk, have no control or idea of what's going on inside our voting machines. (They can, but I'm not sure who's gone through the trouble s.5.905(5).) It doesn't appear as though anyone from the state government or the Wisconsin public has inspected the code or designs of these machines. We're just expected to have blind faith in these systems, but we know the manufacturers can't and/or won't solve all of their problems.

Currently, certain portions of the software for every electronic voting machine model approved for use in Wisconsin is stored in an escrow s.5.905(2). It's unclear if every software version in use is stored in the escrow, and we have no way of knowing how much of any particular software is stored. The GAB most likely knows the exact components, but they don't make the information available on their website as they should. This escrow provision is meant to make us feel better about the insecure proprietary software, but it does little or nothing to increase the security of our voting equipment. We need all of the software components at the very least, and even that won't provide us with the security, accuracy and reliability that we could achieve with open voting equipment.

Even if the manufacturers gave public access to the design and software of their voting machines, we won't get the full value of an open system if the hardware designs and software source code aren't in the public domain. We wouldn't necessarily be able to ensure the integrity of a voting machine simply because we have that information, because the machine itself may have certain vulnerabilities we would have very little ability to control. If all we can do is look at the designs and code, and we're not be given the ability to implement modifications, what happens if we want to make a change but the manufacturer wouldn't agree to it? That's unacceptable and completely avoidable.

There are many advantages to creating an open election system, and I'll defer talking in detail about those for the solutions section. One of the most beneficial advantages of using an open election system is the amount of people who can inspect the design and code to ensure the utmost security and accuracy, anyone who would want to could. This doesn't make the machine less secure, because there will always be ways to "hack" a machine. Instead, vulnerabilities and issues can be spotted and resolved more quickly, including right on the spot by county clerks or other officials (through appropriate processes of course). This openness has been shown to produce highly successful software many times, e.g. Linux, Firefox and WordPress to name just a few.

We can do better than these voting machine companies, and we must, because there's no better solution to many of our issues than using an open election system.

(5) Integrity of voting machines not verified before and after every election. The Government Accountability board conducts periodic audits of a random selection of machines, but that won't detect a singular instance of an issue. Nor will the audits detect widespread issues that were created and manifested between audits. Even still, the GAB doesn't post the results of the audits, just the municipalities in which the audits took place. The GAB should be required to post the results of their audits. These audits are helpful, but they aren't sufficient for detecting all likely issues with our voting machines.

The integrity of a machine can only be inspected during a recount if a candidate requests permission from the GAB, and as long as they sign a Non-Disclosure Agreement s.5.905(4). This process should be automatic for every machine for every election, regardless of the closeness, because that's the only way we can have any hope to guarantee any amount of security, accuracy and reliability. Simply "matching numbers" during canvassing will not uncover many possible issues with our voting machines. So we need a process to ensure every machine is counting votes as accurately as possible during an election.

Verifying the integrity of a voting machine after an election includes making sure that the software currently on the machine is the software that actually counted the votes. Wisconsin Statue 5.905(3) states that "the verification procedure shall include a determination that the software components correspond to the instructions actually used by the system to count votes." However, the GAB may not have enough information or there just may not be any possible way to determine if the software instructions in a particular machine were the actual instructions used to count the votes in an election. I can't find any details as to how the GAB would make such a determination. If we solve (4) this issue becomes much easier to solve.

There are several ways to determine the accuracy of a voting machine without verifying its integrity, although the integrity is the ultimate test. Municipalities employ a simple pre-election test of running a predetermined set of votes through a machine and verifying that the counts match. It wouldn't be difficult to write software to pass the pre-election test but still manipulate the vote counts later. At the very least, this same test must be run on every machine after an election as well as before. However, without verifying the integrity of a machine or running a hand recount, there's no way to guarantee that the results from a machine match the actual votes.

As for recounts, there's really no point to a recount if each machine isn't inspected for its integrity, because the device should give back very near the same results a second time whether the device was manipulated or not. Fortunately, there are some hand recounts occurring throughout the state, but not nearly enough to provide information beyond most glaring types of discrepancies. Since recounts are meant to try to determine the actual vote count, why don't we at least inspect every machine automatically before proceeding with a recount? Otherwise, we're just another example of the classic definition of insanity, continuing to do the same thing but expecting different results.

(6) Various vote-tallying processes are not open. I already mentioned the issues with the closed vote-tallying processes within our voting equipment in (4). Here, I'm referring to vote-tallying process outside of our voting equipment. This GAB manual for county clerks says, "The counting of votes is always done publicly after the polls close at 8:00 p.m." (their emphasis) If the counting of votes is always done publicly, we would have found out about the Waukesha County error much sooner. So, this law is clearly not being enforced properly, and the GAB felt it necessary to emphasize "publicly" to officials who presumably should know that very well.

As evidenced by my reporting on the ballot bag issues, the availability of information during the Prosser/Kloppenburg recount was scarce. Yes, there was a live stream of the Waukesha County recount, but they're not the only county in Wisconsin. And even with the live stream we couldn't figure out exactly how many ballot bags had discrepancies and where they're from. There was no mention of the issues with the ballot bags on the GAB website, even though at the very least the Journal Sentinel, The CapTimes and WisPolitics had reported the issues.

Every county in Wisconsin should have a live stream of vote-tallying and machine verification. Though, even then, few of us have time to intently watch a single county. So, there should be a live stream of each county with the ability to look back at previously streamed events. There would be very little cost but a huge increase in election transparency. The cost could be further reduced by using third-party sites such as YouTube.

Any disputed ballots during a recount should be scanned and posted online for everyone in Wisconsin to see as they were in 2008 during the Minnesota recount for the U.S. Senate election between Al Franken and Norm Coleman. We should see what causes errors, so that we can learn from those ballots. Also, it makes the process much more transparent with very little extra effort. In fact, some voting machines take "photos" of ballots, and those photos could be quickly cropped appropriately and posted for all to see. We have online banking, why not "online" recounts?

(7) Ease and likelihood of errors in reporting. This was thought to only pertain to Waukesha County, but it also happened (on a slightly smaller scale) in Winnebago County during the Prosser/Kloppenburg recount. Therefore, it's probably just as likely in the rest of Wisconsin. This issue is not as serious if the previous six issues are resolved. However, until then, errors in reporting will continue to exacerbate the concerns we have.

(8) Voting equipment and software is not uniform across Wisconsin. We should have a standard set of strict vote-tallying processes and one standard set of voting machines, i.e., at most a few touch-screen machines and one paper ballot-tallying machine. Variation creates unnecessary complexity and cost, and complexity increases the chances of an error. We could reduce a lot of the learning curve involved in voting as well as the cost to train staff, volunteers and maintain the equipment. Obviously, I believe that we should standardize around a set of open voting machines.


We must seek real solutions to relevant issues by determining the best way to solve these issues. I for one won't be able to trust an election in Wisconsin until issues 1-6 are fixed, and I hope you feel just as worried.

Are There Real Solutions?

So what are some possible solutions? There's at least one very beneficial and realistic solution to many of these issues. Voting equipment whose hardware design and software is owned and controlled by the people of Wisconsin, i.e., open source, would be much more secure and instill much more voter confidence than any proprietary equipment ever could. We asked for nearly this back in 2005 via AB 627, but the bill was amended to remove the half-hearted provision before the bill was passed. Those who control our elections want desperately for it to stay that way, but it's not what's good for us. We will end up choosing this option at some point. So why not now?

Standardizing open voting equipment across the entire state of Wisconsin would provide greater benefits over those from standardizing over proprietary equipment. The learning curve for voters would be reduced, because everyone would use the same user-friendly and accessible system. Reporting of votes would be faster and less prone to error. The cost of maintaining and approving voting equipment would be drastically reduced. Many other costs at both the state and municipality level would be reduced. We would know exactly how secure and accurate our voting equipment is. There would be no doubt, because anyone could inspect the code. That also means that more people can help improve the system. There are valid reasons against complete homogeneous standardization, but that debate is for a later time.

I realize that I'm glossing over quite a few major details here. It's not guaranteed that the first, or even tenth..., version will be user-friendly and accessible. Over time it will no doubt improve, but by beginning with effective guidelines and true experts, the first version could easily be much more user-friendly and accessible than we've come to expect from our voting machines. This also applies to the costs of developing such a set of systems. Successful open source projects gain momentum quickly when they're driven by knowledgeable and thoughtful leaders. The more knowledgeable people who contribute to the project the less taxpayers will end up having to pay. I have no doubt that such a project would create more than enough enthusiasm from capable people in Wisconsin. We can do it, other less wealthy countries have successfully done this.

I also realize that converting to an open system can't happen overnight, but a deadline should have been set a long time ago and needs to be set now. I don't know how much it would cost the state to implement and convert to such a system, but it's clear we need to replace at the very least the unmaintainable equipment, and it's quite possible the conversion would cost less than the full cost of implementing and enforcing voter suppression with photo ID.

There's a very good chance we could partner with other states to share the burden of development. There would be no reason not to. These tough economic times add to the value of governments working together and sharing the software code they run on. Also, various other municipalities, states, organizations and countries have either begun or completed open source election systems that we could implement as is or modify to our desires. So, I believe converting to an open election system should be one of the top priorities in fixing our elections (if not the top priority).

I'd like to have a discussion about the relevant issues with our elections and possible real solutions. Any other suggestions for solving some of these issues? Any other issues I'm not listing? Is there anything I could clarify? Did I increase your concern?

Please, contact your state representatives and tell them why you're concerned with the security and accuracy of our voting equipment, not voter fraud. They won't do anything unless we demand it.