Post-Election Audits | CTVotersCount.org

Connecticut RLA Working Group, 2nd Meeting

On Wednesday, the CT RLA (Risk Limiting Audit) Working Group held its second meeting and firmed up plans for the 2021 prototype audits. Below is a video of the meeting and an email I sent the Chair Gabe Rosenberg and the chief scientist from UConn, Alex Russell.

Watch here:
https://www.youtube.com/channel/UCzakPMlj0LigvokDrkgiDVw/live

I have watched both meetings of the RLA Working Group and have several Suggestions and comments…

On Wednesday, the CT RLA (Risk Limiting Audit) Working Group held its second meeting and firmed up plans for the 2021 prototype audits. Below is a video of the meeting and an email I sent the Chair Gabe Rosenburg and the chief scientist from UConn, Alex Russell.

Watch here:
https://www.youtube.com/channel/UCzakPMlj0LigvokDrkgiDVw/live

****************************************************

Gabe, Alex,

I have watched both meetings of the RLA Working Group and have several Suggestions and comments.

As you may or may not know, I have been involved in post-election audits and audit laws in CT and nationally since 2007. Since 2009 I have been involved in what become known as RLAs. I was a catalyst in causing a meeting in early 2010 in D.C. and in securing invitations to three officials from CT who attended. After that meeting, they took the lead in initiating what became the UConn Audit Station. I was involved in the planning of the Colorado RLAs and planning, executing, and reporting on the Rhode Island RLA pilot. I was an observer at the Fairfax City RLA demonstration, which I used as a basis to suggest improvements for the RI pilot.

I complement all involved on the successful development of the UConn Audit Station. And the significant improvements in it in the last couple of years. It can save CT hundreds of thousands in an RLA. Note: That all the experts recommended a Ballot Comparison Audit for RI, yet due to the extreme cost of ES&S central count scanners, the RI Board of Elections chose Batch Comparison Audits.
QR codes: The use of QR codes on ballots (mainly on printed ballot marking device ballots) has raised huge integrity and transparency concerns. The concern is that they are not human readable/verifiable for auditing. I would suggest a)the QR code labels also included printed sequence numbers etc. of what is in the QR code and b)the QR code be readable to show that same information by a standard smart phone QR code reader. I would also suggest you assume that QR codes do not need to be scanned in order, to facilitate multiple people/groups putting on stickers and that some stickers will be broken, misapplied, or otherwise discarded in the process.
Are you aware of the recent guidance for the U.S. DOJ that all election materials should be preserved for 22 months for Federal Elections. So things like CVRs, images, paper and computer records (Arlo records) must also be preserved? Since this is not a Federal election the letter does not apply, yet it would be a good practice to do the same for 6 months for other elections as well.
https://www.justice.gov/opa/press-release/file/1417796/download
I would have liked if ABs were included in the prototypes. Also note that there has been no recognition of the need to account for originally hand counted and EDR ballots in an actual RLA.
Two Transparency Requests: I hope you will consider giving at least a week’s notice of the date, time and place of any RLA prototypes, such that there is an opportunity for observers to attend. I would also suggest, like RI, live video stream(s) of the RLAs showing the whole space, so that the pubic can in that way observe the whole process and additional streams showing the details of ballot comparisons, ballot security, and data entry into Arlo (or other software).
Two Small Comments on the Slides: a) National experts have consistently said that there are three RLA methods, yet they include Batch Comparison Audits and view Full Recounts as a possible final round of all the other types of RLAs. b) I believe it is incorrect to state that “All previous pilots have considered single precincts”. As the person planning and leading the RI Batch Comparison Audit, I can state that it included multiple precincts. I also checked with my colleagues more involved in Colorado and they confirmed that Colorado pilots always included multiple precincts.
Going forward, a) I believe that the biggest challenge in conducting an RLA for CT is the town-by-town structure of elections. I would be a strong advocate for the idea of a small number of sites across the state for scanning with the Audit Station, with trained officials responsible for the Audit Station and guiding local officials in the rescanning process. b) Instead of QR stickers, perhaps a device or scanner could be acquired/developed to print QR codes on an appropriate place on ballots – stickering the whole state would be quite an expensive, time-consuming process.

Finally, while I am a supporter of RLAs, they are not a panacea, In my opinion, in many instances they have been oversold as easy and they are all that is necessary to prove that elections are OK. I am not aware of any state laws for RLAs that I or most would consider good, even as far as they go. Often they leave it to an official to choose contests to audit, leaving the suspicion that they made political decisions, where also they often choose uninteresting high margin contests to make the RLA easier. Similarly they often require the auditing of one or two contests, which says nothing about other contests. Doing Ballot Comparison RLAs has the advantage that with marginally more work all large contests can be audited (in CT, statewide and Congressional Races are most suited to RLAs).

RLAs are not well suited to municipal contests in CT, as your statistical table shows. RLAs do not provide much insight into errors. So, there is a place for the batch audits now done in CT. There is also a need for eligibility audits – are the checkoffs accurate? Are those checked actually qualified to vote? Are absentee ballots signed correctly, signatures accurately adjudicated? Etc. Especially in CT there is a need for audits of ballot security, which according to the experts are a prerequisite to trusted, useful RLAs.

Perhaps you would share this letter with the entire membership of the Working Group.

Thanks,

Luther Weeks
Executive Director, CTVotersCount and Connecticut Citizen Election Audit

September 2021 Audit Drawing

This year’s drawing was different. Based on the law passed this year it finally included centrally counted absentee ballots, an item we have been requesting since 2008. We thank Secretary Merrill for initiating the change this year. Better late than never!

The Arizona “Republican Audit”, no so fast

There are many reviews of the Arizona “Republican Audit” <read> and critiques, like this one <read>. I have to admit that I did not attend or watch the audit and have not read the report in detail, yet I have heard from those who have read the report and some who observed parts of it. Democrats and others are celebrating. Don’t rush to any conclusions, consider:

There are many distorted claims in the audit report, yet a few point to weaknesses in our election process, not just in Arizona…

There are many distorted claims in the audit report, yet a few point to weaknesses in our election process, not just in Arizona. For instance we need to be concerned with ballot security, largely from when they leave the polling place until they are destroyed 22 months later – how secure is the storage? Who can access the storage and ballots undetected? How can we be sure?
There are huge questions about the quality of signature comparison, is it even a reliable way to check for fraud? Does it really serve to inappropriately reject legitimate ballots?
Democrats and others celebrate that the audit “proved” that Biden won by a slightly larger margin than the official results. Given the non-transparency of the audit, there is no reason to believe its results are more accurate that the official results; no reason to believe they are accurate at all. Perhaps Biden won by a little more or a little less than the official results, yet this audit provides no reliable evidence.

RLA Working Group holds its first meeting.

This spring there was a bill to initiate a Risk Limit Auditing Working Group. I testified in favor of the concept and suggested several changes to the bill. A similar bill was passed near the end of the session. Last week the group began its work. Here is a video of the meeting:
https://www.youtube.com/watch?v=UhH4JVYcaso

Here is the bill passed by the General Assembly: https://cga.ct.gov/2021/ACT/PA/PDF/2021PA-00002-R00SB-01202SS1-PA.PDF (The RLA section is on page 156.)

The meeting lasted about an hour, primarily consisting of an introduction to RLAs by Alexander Russell from UConn. Assisted by Ben Fuller also from UConn. Also in the group was Brian Macdonald, a statistician from Yale.

I am only familiar with the Chair Gabe Rosenburg, Alex Russel, and one of the registrars of voters in the group. There was no published agenda or list of members. The members were listed by the Chair at the beginning of the meeting. I could be wrong, yet as far as I know none of the members of the group have any experience observing or participating in RLAs. A far cry from RIs plan which included experienced experts from around the country and took much more time and effort that seems to be possible here…

Here is the bill passed by the General Assembly: https://cga.ct.gov/2021/ACT/PA/PDF/2021PA-00002-R00SB-01202SS1-PA.PDF (The RLA section is on page 156.)

Russell’s slides were a reasonable introduction to RLAs. I found three things curious:

He listed three types of RLAs. One was full hand counts usually not listed and omitted from the list was the usually included Batch Comparison Audit.
He indicated that creating a ballot manifest requires a complete hand count of the number of ballots. That is usually not required, although very accurate it would be quite a task to hand could all the ballots in every town for a state-wide RLA.
His plan seems to be to rescan each ballot with the UConn Audit Station would additionally involve stickering each ballot with a QR Code. Once again usually not required, although very accurate quite a task stickering all the ballots in every town for a state-wide RLA.

With their next meeting to be scheduled for early October, they have a lot to accomplish.

Here is my testimony from last year, my concerns and suggestions still stand: (My Testimony)

Summary Recommendations:

Consider doing what CO, VA, RI and perhaps what every other state that has
prototyped or implemented RLAs have done: Convene a team of election officials, and
volunteer national experts to help plan a prototype, bring tested software (typically available
at no cost) to the table to provide tested statistical calculations and public verifiability. And
then report on the results and recommend methods and changes necessary in the law.
Such a group could report to the Task Force such as the Rhode Island RLA Group reported
to the RI Board of Elections.

Do not limit the prototype to 5 to 10 largely redundant municipal RLAs

Move the prototype(s) to the 1st half of 2022. That will give sufficient time to plan the
prototype(s). It will provide more access to statistical experts and those with experience with
RLAs to participate.

Move the reporting deadline to Jan 2023 to provide more time to absorb the lessons of the
prototype, to consider all the issues necessary to implement actual RLAs, and to detail the
changes necessary in the law.

See my detailed testimony for reasoning behind these recommendations.

Risk Limiting Audits: A Guide for Global Use

A recent report, Risk Limiting Audits: A Guide for Global Use is about the most comprehensive and balanced introduction to Risk Limiting Audits that I have seen. Its 38 pages will take an hour or two to read in detail, and well worth it.

I am a fan of Risk Limiting Audits, yet I am concerned that they are misunderstood in several dimensions:

RLAs are not a panacea:…

A recent report, Risk Limiting Audits: A Guide for Global Use <read> is about the most comprehensive and balanced introduction to Risk Limiting Audits that I have seen. Its 38 pages will take an hour or two to read in detail, and well worth it.

I am a fan of Risk Limiting Audits, yet I am concerned that they are misunderstood in several dimensions:

RLAs are not a panacea: They are one type of audit among that others that need to be performed among several. Complementary audits are needed as a prerequisite to trusted RLAs, such as auditing the security of ballots presented for the RLA. RLAs done well, only assure that ballots were counted and tabulated correctly enough. There are other audits needed to determine the legitimacy of the election, such as the accuracy of voting lists and the integrity of the check-in lists and processes.
RLAs are not easy or simple: Some tout RLA benefits claiming that they are easy and simple. They are not. They are complicated and require attention to detail. They require scientific expertise to organize, execute, and understand, and for the most part trust on the part of the public in that science. They may be efficient, yet not simple to implement and understand.
RLAs have not been uniformly done well and backed by sufficient laws and procedures: Most state laws and procedures are insufficient and, at best, add confidence to the very few contests actually subject to such audits.
The larger the contest audited, the more efficient a RLA can be: Statewide contests and Congressional races can be reasonably to highly efficient. Auditing local contests, especially all local races can be expensive and time consuming, approaching the cost of recounting those contests by hand.

While optimistic, the Guide, points to all the details at a high-level, while avoiding all the statistical details. That makes it readable It does not avoid pointing out all other audits and their necessity. It also emphasizes the need for transparency and public verifiability – often neglected in RLAs and other audits.

Testimony on two Elections Bills

Earlier this week we testified on two elections bills.

First a bill for a Task Force to provide a prototype and recommend state laws for Risk Limiting post-election Audits (RLAs). See our testimony and that of the inventor of RLAs, Philip Stark, and John Marion of RI. Phil and I disagree just a bit on our recommendations. I find November is just not the best time to do a prototype and then providing less that two months to make recommendations to the General Assembly is not enough time. Here is all thee testimony <read>

Then on a long bill with several election changes recommended by the Secretary of the State. We had comments on two sections: We asked that two officials empty drop boxes and sign logs listing their content. Also a reform we have been requesting for a long time – including central count absentee ballots and Election Day Registration ballots in post-election audits <testimony>

Earlier this week we testified on two elections bills.

State Audit Working Group comments on H.R.1

This week the State Audit Working Group published a letter sent to Rep Sarbanes regarding the collective concerns with the bill.

Here are the details from the cover letter to Rep Sarbanes:

We write to request critical changes to H.R.1, along with suggested improvements. Without a few key changes, we believe the bill might degrade election integrity and miss opportunities for improvement, rather than meet its well-intended, laudable goals. Our comments are restricted to election administration and integrity issues pp78-407 of the bill.

Attached to this letter is a list of detailed comments. Here we summarize the most critical items:

Requirements for grants should be stronger, to help ensure effective Risk Limiting Audits (RLAs)…

This week the State Audit Working Group published a letter sent to Rep Sarbanes regarding the collective concerns with the bill <read>

Here are the details from the cover letter to Rep Sarbanes:

We write to request critical changes to H.R.1, along with suggested improvements. Without a few key changes, we believe the bill might degrade election integrity and miss opportunities for improvement, rather than meet its well-intended, laudable goals. Our comments are restricted to election administration and integrity issues pp78-407 of the bill.

Attached to this letter is a list of detailed comments. Here we summarize the most critical items:

>Requirements for grants should be stronger, to help ensure effective Risk Limiting Audits (RLAs). We suggest specific improvements to the HR1 grant requirements. Grants should be available to audit compliance and eligibility which are crucial for valid RLAs.

Poll books should be part of the Federal certification program, as proposed. So should other systems used to determine the eligibility of voters or ballot packets. They however, should be tested and certified separately from the voting system. Competition will be stifled if pollbooks are only tested as part of an entire voting system. Election officials will end up with fewer and less innovative purchase choices.

Ballots cast by an in-person voter by hand marked paper ballots may be rejected later under the current text. When a voter appears in person they must be offered an opportunity to be authenticated and, upon authentication, vote on a hand marked paper ballot without further eligibility checks.

Voter Privacy / Ballot Secrecy. Ballots should never be associated with voters, thus compromising ballot secrecy There should be no unique identification numbers on some ballots for voters with disabilities. Voters should not be able to waive their ballot secrecy, a collective right.

Voting over the internet is not secure and does not protect the secrecy of the ballot. For security and integrity, votes should not be transmitted over the internet or by other electronic means such as email or fax.

Last week we posted a three-part series on our concerns with H.R.1. There concerns were largely based on the effects for states like Connecticut.

PS: I am the Moderator of the State Audit Working Group. Members spent hours over a couple of weeks under tight deadlines to publish our concerns.

To trust our elections we need evidence, enough evidence

A recent article in Barons by respected scientists: Elections Should be Grounded in Evidence, Not Blind Trust

Here’s what an evidence-based election would look like:

Voters hand-mark paper ballots to create a trustworthy, durable paper vote record. Voters who cannot hand-mark a ballot independently are provided assistive technologies, such as electronic ballot marking devices. But because these devices are subject to hacking, bugs, and software misconfiguration, the use of such ballot-marking devices should be limited.
Election officials protect the paper ballots to ensure no ballot has been added, removed, or altered…

A recent article in Barons by respected scientists: Elections Should be Grounded in Evidence, Not Blind Trust <read>

Here’s what an evidence-based election would look like:

Voters hand-mark paper ballots to create a trustworthy, durable paper vote record. Voters who cannot hand-mark a ballot independently are provided assistive technologies, such as electronic ballot marking devices. But because these devices are subject to hacking, bugs, and software misconfiguration, the use of such ballot-marking devices should be limited.

Election officials protect the paper ballots to ensure no ballot has been added, removed, or altered. This requires stringent physical security protocols and ballot accounting, among other things.

Election officials count the votes, using technology if they choose. If the technology altered the outcome, that will (with high confidence) be corrected by the steps below.

Election officials reconcile and verify the number of ballots and the number of voters, with a complete canvass to ensure that every validly cast ballot is included in the count.

Election officials check whether the paper trail is trustworthy using a transparent “compliance audit,” reviewing chain-of-custody logs and security video, verifying voter eligibility, reconciling numbers of ballots of each style against poll book signatures and other records, and accounting for every ballot that was issued.

Election officials check the results with an audit that has a known, large probability of catching and correcting wrong reported outcomes—and no chance of altering correct outcomes. The inventory of paper ballots used in the audit must be complete and the audit must inspect the original hand-marked ballots, not images or copies.

None of these steps stands alone. An unexamined set of paper ballots, no matter how trustworthy, provides no evidence. Conversely, no matter how rigorous, audits and recounts of an untrustworthy paper trail provide no evidence that the reported winners won. Auditing or recounting machine-marked ballots or hand-marked ballots that have not been kept secure can check whether the reported outcome reflects that paper trail, but cannot provide evidence that the reported winners won.

We completely and enthusiastically agree.

Elections Should be Grounded in Evidence, Not Blind Trust

Commentary in Barron’s this week Elections Should be Grounded in Evidence, Not Blind Trust <read>

Even though there is no compelling evidence the 2020 vote was rigged, U.S. elections are insufficiently equipped to counter such claims because of a flaw in American voting. The way we conduct elections does not routinely produce public evidence that outcomes are correct.

Commentary in Barron’s this week Elections Should be Grounded in Evidence, Not Blind Trust <read>

Even though there is no compelling evidence the 2020 vote was rigged, U.S. elections are insufficiently equipped to counter such claims because of a flaw in American voting. The way we conduct elections does not routinely produce public evidence that outcomes are correct.

Furthermore, despite large investments since 2016, voting technology remains vulnerable to hacking, bugs, and human error. A report by the National Academies into the 2016 election process concluded that “there is no realistic mechanism to fully secure vote casting and tabulation computer systems from cyber threats.” The existence of vulnerabilities is not evidence that any particular election outcome is wrong, but the big-picture lesson from 2020 is that ensuring an accurate result is not enough. Elections also have to be able to prove to a skeptical public that the result really was accurate.

We need evidence-based elections: processes that create strong public evidence that the reported winners really won and the reported losers really lost, despite any problems that might have occurred. Every step in election administration—from technology choices to voter eligibility checks, physical security, the canvass, and audits—should flow from that requirement…

Here’s what an evidence-based election would look like:

Voters hand-mark paper ballots to create a trustworthy, durable paper vote record. Voters who cannot hand-mark a ballot independently are provided assistive technologies, such as electronic ballot marking devices. But because these devices are subject to hacking, bugs, and software misconfiguration, the use of such ballot-marking devices should be limited.

Election officials protect the paper ballots to ensure no ballot has been added, removed, or altered. This requires stringent physical security protocols and ballot accounting, among other things.

Election officials count the votes, using technology if they choose. If the technology altered the outcome, that will (with high confidence) be corrected by the steps below.

Election officials reconcile and verify the number of ballots and the number of voters, with a complete canvass to ensure that every validly cast ballot is included in the count.

Election officials check whether the paper trail is trustworthy using a transparent “compliance audit,” reviewing chain-of-custody logs and security video, verifying voter eligibility, reconciling numbers of ballots of each style against poll book signatures and other records, and accounting for every ballot that was issued.

Election officials check the results with an audit that has a known, large probability of catching and correcting wrong reported outcomes—and no chance of altering correct outcomes. The inventory of paper ballots used in the audit must be complete and the audit must inspect the original hand-marked ballots, not images or copies.

None of these steps stands alone. An unexamined set of paper ballots, no matter how trustworthy, provides no evidence. Conversely, no matter how rigorous, audits and recounts of an untrustworthy paper trail provide no evidence that the reported winners won. Auditing or recounting machine-marked ballots or hand-marked ballots that have not been kept secure can check whether the reported outcome reflects that paper trail, but cannot provide evidence that the reported winners won…

outsourcing audits, as Georgia did after the November vote, may prevent such process improvements. It is the responsibility of election officials (and not a third party) to ensure and demonstrate that the paper trail includes no more and no less than every validly cast ballot, and that the reported result is what those ballots show.

We note that, to us, ‘Outsourcing’ audits is a distinct concept from ‘Independent’ audits. Outsourcing implies turning all responsibly over to a hired vendor or entity dependent on election officials for funding. Independent auditing means assigning responsibility for the audit, or at least assessment and oversight of the audit to an entity independent of election officials.

November Election 2020 Post-Election Audit Drawing

Yesterday we observed the Post-Election Audit Drawing by the Secretary of the State Denise Merrill. 38 districts in 31 municipalities will be audited by no later than December 4, 2020.
The Secretary’s Office is offering electronic auditing to municipalities, subject to scheduling limitations. Official Press Release <read>