Judge: Wichita State statistician can’t have tapes to audit voting machines
[Judge] Lahey told Rathbun that he “won the battle” because the case could go to trial on the question of whether to have a recount. But he “lost the war,” because Clarkson won’t get access to the ballot tapes even if she wins at trial.
This is the final graph from my presentation at the Lawrence Symposium yesterday. My friend Eileen encouraged me to make a post about it as it’s one she feels quite illustrative of the reason for concern.
During the Q&A after my talk, I sorta kinda said I didn’t think the Governor legitimately won the popular vote in 2014. It occurred to me later that probably wasn’t a very ‘politically astute’ thing to say. I hope the Governor does not personal offense. Such thoughts didn’t occur to me at the time because I was talking about how I read the data shown in this chart. I get engrossed in the data analysis and don’t think about anything else.
This chart shows each of the four statewide races, Governor, Senator, Secretary of State, and Attorney General graphed in a cumulative summation analysis. In addition, the cumulative summation analysis of Wichita registered republican voters is provided to give a comparison.
In the cumulative summation models, as the models accumulates more data (think moving left to right across the graph) we expect it to converge to the average quickly and show decreasing differences randomly distributed around the mean. When we see a pattern of the model increasing or decreasing, it indicates that the sequencing of the summation is not random, but has a relationship with the share of the vote achieved.
In the graph above, the % registered republican voters shows a pattern that is not visually distinguishable from random. The four races all show a clearly non-random pattern with striking similarities. All show the pattern of increasing republican share of the vote with increasing votes cast.
These are not independent datasets. The four races all show a similar pattern because they are highly correlated. The pattern itself is the troubling thing.
Kris Kobach introduced HB 2543 last month. He supports auditing our voting system! I am glad of the moral support although I did have some concerns about the actual law he submitted.
Letter to Legislators re Kobach Audit Proposal
I testified about those concerns at the Elections committee meeting in Topeka today. I want to write a more detailed description of the experience later, but I did want to post a notification of this activity today. Legislators on both sides are listening. Everyone agreed that audits are doable, although perhaps costly in 2016, and a good idea in order for voters to have confidence that their votes are all counted correctly.
I count this as progress. Sadly, I am amazing ignorant of all that is required for a bill to become law in the state of Kansas. Fortunately, I have supporters in the legislator. I was pleased to meet some of them today.
I even met Mr. Kobach today. He was quite nice, not seeming to mind at all that I’d just testified that he shouldn’t be trusted to generate the sample list, it should be part of the law that the sample is randomly generated. No one can be trusted to generate the sample list. That’s why it’s required that it be randomly generated. After the election. Because no one can leak information before it’s been generated.
Some people have been know to take offense at that implicit insinuation, but he showed no signs of feeling that way, understanding that it’s not personal. Setting up non-random samples is the easiest and most obvious way to degrade the audit process into an empty expensive rite devoid of value. Even if Mr. Kobach can be trusted not to deliberately bias samples in that manner, can his successor, or the next after that? Not to mention the effect of unconscious bias on sample selection methodologies?
My understanding is that “Open Source” software is publicly available. That anyone interested can access the code.
This is no doubt a naive and fuzzy view of the concept, because Commissioner Jim Howell has responded as follows:
Please know, Henry Adkins equipment is based on an open source platform (linux) but all their software is proprietary. If the vendor represented their software as open source, then they were highly misleading. Their software for counting votes is exactly as proprietary as ES&S. If open source software were truly being used, then it could be argued that someone in Sedgwick County could potentially make changes to the software that could change the way the machines operate. I see this as a vulnerability as much as you and Ms. Clarkson see it as a preference. When the ability to change the software is totally out of our control, then we cannot be accused of making changes that might affect the outcome of the election and would also be difficult to defend or verify. Having this out of our control is a good thing in my opinion. Auditing the processes end-to-end (every election, randomly selected 1% of precincts) will give us confidence and assurance that everything is as it should be. Open source SW creates more questions and opens doors for distrust on the election systems. For the record, Sedgwick County does support HB2543 and we look forward to doing the audits as prescribed by law when this bill passes the legislative process.
If the vendor was misleading me and I fell for it, it wouldn’t be the first time. If I misunderstood the technical details and made a few false assumptions, it wouldn’t be the first time for that either. I’ve meant to go their website and check it out, but I’ve been rather busy at work this week and haven’t managed it yet. I’m just going to give my opinion without looking up definitions first and specifications first. It can be a very detail oriented task and it’s a bit late in the day for me to attempt it.
I disagree with Commissioner Howell that open source software implies that anyone can change it. Open source only allows people to see the code, check for backdoors, hardcoded passwords and other security breaches. It does not allow the software to be changed unless the evildoer also has physical or wireless access to the computing equipment. Access to the equipment is supposed to be rigorously controlled and without any wireless connections.
It’s arguable that open source grants more people the knowledge necessary to undermine the voting system if they can also get access. I find the additional transparency provided by open source software a worthwhile trade off for that increase in risk, but I can understand if others disagree on that point.
If the Henry Adkins software is actually proprietary, not open source, that fact would curb my enthusiasm for them quite a bit. I still consider their system superior to the ES&S for the other reasons I listed, but the difference is considerably less.
I want to give kudos to commissioner Howell for supporting routine audits of the electronic voting results. I think that’s great!
In an editorial last October, I discussed the three requirements I have for any new voting system. Slightly rephrased, those are:
1. A fully auditable paper trail
2. Open Source Software
3. Post-Election Audits
Kris Kobach introduced legislation in January regarding such audits and requiring them in Kansas prior to the results being officially certified. Although I have some concerns about his suggested methodology, an attempt has been made to address #3! This is great news!
By it’s very nature, electronic voting equipment introduces points of opacity into our voting system. In order to have faith in the official results presented by these machines, we must have public and transparent audits of these voting results. I am so grateful to discover that my discomfort at their lack is shared by many other Kansans that it is now being discussed by our elected officials charged with making sure that our voting system is accessible and accurate for all Kansans.
Two vendors were invited to show their equipment on Feb 1st. Election Systems & Software (ES&S) which is headquartered in Canada and the Henry M. Adkins & Son Inc. of Missouri. I was able to attend and evaluate the two proposed systems.
Both systems provide a full paper trail of all ballots. That means, that with the introduction of either vendor’s equipment, requirement #1 will be addressed. One of the two, Henry Adkins, uses open source code. They get my vote based on that very important detail! Selecting this system rather than the ES&S equipment will address requirement #2.
I think these responses constitutes movement towards the light of transparency. We should celebrate it!
I want to extend kudos to Kris Kobach and Tabitha Lehman for listening and responding to these concerns. Hopefully, they will also be willing to listen to my criticisms regarding aspects of their proposals I don’t support.
Deputy Election Commissioner Sandy Gritz has informed me that the selection of vendor for new voting equipment in Sedgwick County has not yet been made. Now is the time to provide input on this issue. If you want to have influence over this decision, let Ms. Lehman and our Sedgwick County Commissioners know that you support the purchase of the Henry Adkins system. Just dropping them a quick email informs them about what is most important to their constituents in selecting a new system. I would also recommend asking them to look into how the costs of these electronic systems compare with the cost of doing a full handcount of all ballots.
The next step is contacting our county commissioners here in Sedgwick County and letting them know your preferences in voting equipment. Open source software and a physical paper ballot for every vote – are minimal requirements to be able to conduct an audit of the system. It doesn’t guarantee the machines can’t or won’t ever by rigged. It’s the minimum system requirement for being able to detect tampering if and when it occurs.
I want to emphasize that either proposed system would be a major improvement over our current equipment. I only wish Ms. Lehman was willing to acquire the equipment and train her staff for the 2016 elections. I can also understand her desire to stick with a system that she knows how to manage and use smaller elections to work out the kinks (there are always kinks!) in implementing a new system. Just remember to ask for a paper ballot when you vote! It’s your right and voting on a paper ballot is the only way to know that your vote can be counted later by a human being.
I’ve written a detailed description of my impressions of the two vendors and the equipment demonstrated, but I’ll put up a quick summary before I delve into the details.
Summary
I think optical scanning equipment is an acceptable approach when combined with rigorous audits regarding accuracy and far superior to our current equipment. Ballot Marking Devices are a reasonable and appropriate accommodation for those unable to mark a ballot without assistance. They may also be more cost effective, but I would want an unbiased assessment of the relative costs before considering them a better choice over voter marked paper ballots.
I have serious concerns about the use of electronic poll books. Like the paper ballots, to shift this important record of the voting process away from a physical paper record created at the time the vote is casts disrupts the auditable paper trail used to assess the accuracy of our elections. They may be useful for the election officials in administering the election and justifiable on that basis but the demonstration seemed to indicate movement towards a entirely electronic method of registering voter who show up to cast their ballots.
Electronic Registration equipment should never be used as a replacement of the physical polling book that records all individuals who showed up to cast a vote that day. If we don’t have the paper record, we don’t have an auditable trail to verify the results. Without that, we cannot have confidence that the reported results are accurate! This principle applies to polling books just as it does to ballots.
My opinion is that given the security issues inherent in all electronic voting systems, including optical scanning machines, we should never trust their results without appropriate and wholly transparent verification of those results. It can be accomplished using a sampling approach to reduce costs. But that sampling plan must be set up and run correctly, or you can end up with results that allow government officials to claim the water in Flint was safe to drink.
For this reason, counting all the paper ballots manually should also be considered as an option. I would be interested in comparing the scanning equipment with appropriate post-election audits to verify results compared to a simple manual count of paper ballots. I sometimes wonder if the use of modern equipment is actually the best choice for the ritual of voting.
The Feb 1st Voting Systems Demonstration.
It was a busy day for the elections office. I was a little disconcerted by their request for my name when I arrived, but I signed in without complaint. It wasn’t a requirement, just a request. Presumably to put me on a mailing list for news about our election equipment.
I visited the Henry Adkins room first. They had several represents, a gentleman by the name of Brent Wagoner was kind enough to go through their equipment with me. Their software is open source, which is one of two major requirements for me. Built on a Linux system – a choice he seemed a bit defensive about and started expounding on its superiority to Windows. I explained that wasn’t necessary. Personally, I can’t imagine why anyone would consider Windows a suitable choice for election software, but I consider it immoral to use proprietary software to report results of a public election.
Henry Adkins representative demonstrated an optical scanning system. This system expects the majority of voters to mark paper ballots and insert them into the scanner/ballot box. For ADA requirements, their touchpad ballot marker machine is designed to enable people with a variety of disabilities to mark their ballot and verify the machine correctly understands their choices. It outputs a paper ballot marked with their choices, shown both in English and in a machine code that the scanner will read. Mr. Wagoner even explained the coding system they use for the scanning equipment, thereby enabling me to verify that the machine coding of my choices was accurate. Another level of transparency!
The machines are claimed to have no wireless connections or modems and check encryption codes before allowing election data transferred via flash drives. While this is all good practice but not always done, those are details that I lack the skills to personally verify. Nor do I feel I trust such claims given findings such as EVEREST: Evaluation and Validation of Election-Related Equipment, Standards and Testing. They found various ports in voting machines that weren’t supposed to have them, including wireless connections. I feel the best approach is to acknowledge the lack of security with respect to computer vote counting and implement a detection/verification system as an integral part of the process.
ES&S also showed an optical scanning system. They had several representatives all dressed in gray suits. The main meaningful difference with their competitor across the hall is that their software is proprietary. Their system is set up to include a wireless modem, but they won’t be on any machines shipped to Kansas as our state law prevents it. As with the Henry Adkins equipment, I feel it’s best to assume the machinery could be comprised and make verification of results part of the process.
The only other notable difference in the two systems was the machine coding of ballots. ES&S used a barcode system. Thus, there’s no way for a voter to verify that the machine readable code is accurately representing their selections. This relatively minor difference (few voters will care enough to check that the machine code is correct) is a point toward the Henry Adkins machines because it improves the transparency of the process for the voter.
There was also a difference between the two vendors in their suggested uses of the ballot marking machines. The ES&S representative said the cost per machine marked ballot was 9 cents compared to approximately 25 center for a paper ballot. Thus, they recommend directing voters to use the ballot marking machine in preference to marking their own ballot as a cost saving measure. I am reluctant to take these figures at face value from a sales representative, but if an unbiased evaluation were to show a lower overall cost for that approach, I would consider that a point in favor of the ES&S equipment.
The county currently uses ES&S equipment, so a transition to new machines will be easier for the office to implement. That is a legitimate point in favor of ES&S, particularly from the point of view of people charged with getting the new equipment up and running. However, it is a point of relatively minor importance to voters.
The Henry Adkins representative focused on the use of the ballot marking machine as a way to accommodate disabled voters and recommended that we purchase only one or two such devices to be available at a polling station. This is the approach I favor as well. The voter marked paper ballot is the best tool we have to ensure that our vote will be properly counted. We should use it!