Start Here
Donate

Case 8 — Clarkson v. Lehman, 2013–2018

8.1 — Introduction

The case studies in this collection, with one exception, share a structure. Each begins with a specific election, documents a specific discrepancy between what the voting machines counted and what the official results reported, and shows how the discrepancy was discovered — usually late, usually by accident, usually after the results had already been certified. Each concludes that Actual Vote would have caught the error earlier.

This case is the exception. No error was found, because nobody was allowed to look. A Wichita State University statistician with a Ph.D. in statistics and three decades of professional experience as a certified quality engineer filed a public records request for paper records produced by her county’s voting machines. She intended to compare those records against the official results and write an academic paper on voting machine accuracy. The county election commissioner refused to disclose the records. A trial court ruled for the commissioner. The Kansas Court of Appeals dismissed the appeal as moot. At no point — through five years of litigation, across two pro se filings, one counseled filing, and one appeal — did any court rule on whether the underlying voting machines had counted correctly. That question was never reached, because the evidence needed to answer it was never produced.

This case does not require us to adjudicate the statistical claim at its center. It requires us to examine what happened when a citizen sought the primary evidence to test that claim independently. The answer, in Kansas from 2013 to 2018, was: she could not have it. Not under the state’s open records law, not under its election recount statute, not as a spectator at any county-run audit, not ever. Whatever the voting machines in Sedgwick County had actually recorded on election day would remain between the machines and the officials who ran them.

The case matters because transparency is a pillar of election legitimacy independent of whether any given election contains an error. A system that counts correctly but cannot be independently verified is still a system with a structural vulnerability — not to fraud, but to the collapse of public confidence. The Clarkson litigation is the clearest documented American example of that vulnerability being tested in court and confirmed.

8.2 — Background: The Statistician and the Pattern

Elizabeth “Beth” Clarkson holds a Ph.D. in statistics and has been a certified quality engineer with the American Society for Quality for three decades. At the time of the events in this case study she served as chief statistician for the National Institute for Aviation Research (NIAR) at Wichita State University, where she produces reports for the Federal Aviation Administration and private industry. Her credentials in statistical methodology are conventional and established; she is not, and has never presented herself as, an election specialist.

Her interest in elections began pedagogically. She used data from the 2000 presidential election to enliven her statistics classes. Over the following decade she continued to examine publicly available election returns and began to notice patterns that she initially set aside as anomalies. In 2012 she encountered a paper by Francois Choquette and James Johnson titled “Republican Primary Election 2012 Results: Amazing Statistical Anomalies,” which argued that across jurisdictions using electronic voting machines, the percentage of votes for establishment Republican candidates in primaries tended to increase as precinct size grew — a pattern the authors argued should not occur under honest counting. Clarkson reproduced and extended the authors’ analysis, eventually publishing a version of the work in the Royal Statistical Society’s StatsLife in 2015 that examined the 2012 Ohio presidential race, the 2014 Wisconsin governor’s race, and the 2014 Kansas Senate race.

The statistical interpretation of this pattern — known in the literature as “cumulative vote share” analysis — is contested. Other election researchers have argued that the pattern can be explained by non-fraud factors such as systematic differences between urban and rural voter composition, turnout dynamics specific to particular candidates, and structural correlations that arise when precincts are ordered by size. Clarkson herself has consistently distinguished between the statistical observation and its interpretation. She has written that her findings are “consistent with the hypothesis of voting machine fraud” but do not prove it, and she has repeatedly stated that the only way to resolve the question is through an audit against primary evidence.

Whether Clarkson’s statistical inference was correct is orthogonal to the subject of this case study. What matters for our purposes is what she did next. In 2013 she went to her local election office and asked to inspect the paper records produced by her county’s voting machines. She was refused. She filed a lawsuit to compel disclosure. And the five years of litigation that followed established, as a matter of Kansas law, that she was not entitled to those records under any legal theory available to her.

8.3 — The Physical Evidence: What RTAL Are and Why They Matter

Sedgwick County, which includes Wichita, used voting machines that produced a continuous paper record known by the vendor’s brand name as Real Time Audit Logs (RTAL). The generic industry term is Voter Verifiable Paper Audit Trails (VVPAT). The physical artifact consists of paper strips loaded into each voting machine. As a voter makes selections on the machine’s touchscreen, the machine prints those selections onto the strip in real time, which the voter can see through a plastic window before confirming their vote. The machine retains the strip after the voter leaves.

The parties to Clarkson’s litigation stipulated that RTAL “are not limited to ‘statistical information,’ but contain information such as the identification of the voting equipment used to cast the ballot, the time the vote session started, the precinct, the ballot style and language used, and which candidate or ballot initiative was selected by the voter.” Clarkson separately asserted — and the election commissioner did not dispute — that “another purpose for RTALs is for recounts and audits.”

The physical scale of the 2014 Sedgwick County RTAL corpus gives a sense of what independent auditing of these records would involve. Each paper roll is approximately 385 feet long. The county stored the 2014 records in 42 boxes. Each individual vote occupies roughly 27.5 inches of paper. Votes are not segregated by precinct on the physical rolls — a single roll contains all the votes cast on a given machine during the election, regardless of precinct boundaries. This is relevant because the election commissioner argued in court filings that producing the records would be “unnecessarily burdensome” given their physical format. The court did not ultimately need to reach the burden question to resolve the case.

The only documented prior use of Sedgwick County’s RTAL to verify an election took place in 2006, when challenger Walt Chappell contested a 373-vote loss to Representative Brenda Landwehr in the 91st House District race. Three examiners appointed by the court rolled through the tapes using an improvised reel-to-reel device assembled by an election office worker. The process took a full day. Landwehr, Chappell, their attorneys, and a reporter from the Wichita Eagle observed the count. No discrepancies between the machine count and the reported count were found, and the presiding judge closed the recount before it was completed. That recount, handled by then-Election Commissioner Bill Gale, is the only instance in which Sedgwick County’s VVPAT records have been publicly examined.

8.4 — The First Lawsuit (2013)

On June 18, 2013, Clarkson filed a pro se lawsuit in Sedgwick County District Court against Tabitha Lehman, the county’s election commissioner. She sought access to the RTAL from the November 2010 general election in Precinct 208 — the precinct where she voted. Her stated purpose was to “verify the official outcome” of that election through an independent post-election audit. She invoked the Kansas Open Records Act (KORA), K.S.A. 45-215 et seq., arguing that RTAL were open public records.

Lehman opposed the request. Her position was that RTAL were not public records subject to disclosure under KORA, that Kansas election law treated them as ballots rather than general records, and that disclosure would therefore be prohibited under state statute.

After a bench trial, on August 26, 2013, District Judge Mark Vining entered judgment for Lehman. The ruling, as transcribed in the later appellate record, reasoned that “election results, voting records are in an area of law that is specifically set out in statutes separate and distinct from [KORA] and how elections are run and what records are kept. And while there is no specific exemption that I can find in [KORA] itself, I do find based upon the information that is provided in the election definitions and applications in Article 25 that the request that you have made is outside of the scope of what is considered open records information and should not be disclosed.”

Clarkson did not appeal. She has said publicly that she was “pro se, meaning I was representing myself, and it was pretty easily defeated.” Her attorney in the later litigation, Randall Rathbun — a former United States Attorney for the District of Kansas — would characterize her as “a brilliant statistician” but “a horrible lawyer.” The 2013 ruling, unappealed, became the first in a chain of legal determinations that closed off access to the records.

8.5 — The Second Lawsuit (2015–2016) and the Appeal (2018)

On February 2, 2015, Clarkson filed a second lawsuit, again pro se, this time seeking RTAL from the November 2014 general election in Sedgwick County. As with the first suit, the requested relief was a court order directing Lehman to allow Clarkson and designated assistants access to the RTAL for the purpose of conducting an audit. The amended complaint, filed April 1, 2015, named both Lehman and Kansas Secretary of State Kris Kobach as defendants. Kobach, the state’s chief election officer, responded in court filings that the records were sealed by state statute and that, in any case, he was not their custodian — the records resided with the county, not the state.

During 2015, attorney Randy Rathbun assumed Clarkson’s representation pro bono. A nonprofit, the Show Me The Votes Foundation, and a crowdfunding campaign were established to defray costs. Newspaper editorial boards from Winfield to Kansas City published opinion pieces supporting the release of the records on transparency grounds. Kobach responded to those editorials with his own, arguing that Kansas law required ballots to be kept secret even if personally identifying information was redacted.

On January 6, 2016, Lehman moved for summary judgment. Her three-part argument was that Clarkson’s claim was barred by res judicata and collateral estoppel based on the 2013 ruling; that release of RTAL would violate K.S.A. 25-2422(a)(1), which criminalizes unauthorized disclosure of the contents of a ballot; and that RTAL were not subject to KORA.

After Lehman’s motion was filed, Clarkson’s legal theory shifted. In a pretrial questionnaire submitted January 7, 2016, she recast her request as a demand for an election recount under K.S.A. 25-3107 rather than a records request under KORA. Her theory was that if the court ordered a recount of the 2014 election, she would be able to observe the process — and, through observation, gain visibility into the RTAL content that had been denied her under KORA. Lehman’s reply characterized this as “creative and calculated timing” and argued that “regardless of how [Clarkson] labels this lawsuit, it is clear that her ultimate goal remains the same: to gain personal access to [RTAL] from the November 2014 General Election. This request is clearly one for records, not a recount.”

On February 18, 2016, District Judge Timothy G. Lahey held a hearing on the summary judgment motion. More than one hundred people filled the courtroom, most of them supporters of the lawsuit — by the judge’s account the largest audience he had seen at a motion-to-dismiss proceeding in twenty-six years on the bench. Rathbun’s argument centered on a specific claim: “my client believes that the RTALs are not the same as what is being reported. That’s why [Clarkson] wants to look at the RTALs. That’s why we assume they’re fighting so hard to have us [not] look at the RTALs.” Lehman’s counsel, Assistant County Counselor Michael North, responded that even if a recount were ordered, “individuals are allowed to go ahead and look over [the] shoulder of the election board to see the accuracy of what they are doing. … They are not allowed. They can be there, but they don’t get to participate in the process.”

Judge Lahey expressed the implication directly: “If I allow the recount, [the paper tapes] will not be disclosed to your client.”

On April 22, 2016, Judge Lahey filed an order granting partial summary judgment. The RTAL-access claim was barred by collateral estoppel from the 2013 ruling. The recount-participation claim was barred by K.S.A. 25-3107(b), which provides that “only a special election board appointed by the county election officer may conduct a recount of the ballots.” Clarkson, as a private citizen, was “not legally entitled to participate in the recount process.” A final journal entry of judgment followed on June 23, 2016, denying the recount request in its entirety.

Clarkson appealed. On August 31, 2018 — nearly four years after the November 2014 election whose records she had sought — a three-judge panel of the Kansas Court of Appeals dismissed the appeal as moot. The court’s reasoning was that the 2014 election results had long since been certified, that any recount conducted at that stage would have no effect on the outcome, and that the relief Clarkson sought was therefore “ineffectual for any purpose.” Clarkson argued that two exceptions to the mootness doctrine — the vital-rights exception and the public-interest exception — should apply. The panel rejected both. On the vital-rights claim, the court noted that Clarkson had conceded she would not have been legally permitted to participate in any recount even if one had been ordered. On the public-interest claim, the court declined to issue what it characterized as an advisory opinion on the use of VVPAT in statewide recounts.

The appeal was dismissed. The records remained undisclosed.

8.6 — What the Rulings Establish

Taken together, the 2013 trial court ruling, the 2016 summary judgment, and the 2018 appellate dismissal establish the following propositions of Kansas law as applied to a citizen seeking to independently verify an election’s reporting against its primary evidence.

First, VVPAT records — the paper strips produced by voting machines as voters cast their ballots — are not subject to disclosure under the Kansas Open Records Act. They fall outside KORA’s scope because the framework governing election records in Kansas is statutorily separate and more restrictive.

Second, disclosure of VVPAT to a private citizen is prohibited by K.S.A. 25-2422(a)(1) as unauthorized disclosure of ballot contents, regardless of whether the records contain personally identifiable information. The trial court in 2013 accepted the argument that VVPAT should be treated as ballots for this purpose. The 2016 court did not disturb that conclusion.

Third, the Kansas election recount statute, K.S.A. 25-3107, provides that recounts are conducted by special election boards appointed by the county election officer. A private citizen who petitions for a recount is not entitled to participate in or observe that recount at a level of access that would permit independent statistical analysis of the underlying records. The 2016 court ruled this point explicitly, and the 2018 appellate court noted that Clarkson herself had conceded it.

Fourth, once an election is certified, the passage of time moots any further legal challenge to the records or results. An appellant who pursues access through the ordinary legal process will, in the normal course, find that by the time the appeal is resolved, the controversy is over and no effective relief can be granted. The 2018 appellate dismissal is a direct application of this principle.

The cumulative effect is that a Kansas citizen who wishes to compare the primary paper evidence produced by a voting machine against the officially reported results has, as a matter of law, no route to do so. Not through records requests. Not through recount petitions. Not through participation in government-conducted audits. Not through post-certification appeal. The Kansas courts did not rule that Clarkson’s underlying statistical concerns were wrong. They ruled that she had no legal right to the evidence that would resolve them.

8.7 — The Wider Pattern: Access to Primary Election Evidence

The Clarkson litigation is not a Kansas-specific anomaly. It is the most extensively adjudicated American instance of a broader pattern in which access to the primary evidence produced by voting equipment — paper tapes, ballot images, tabulator records, memory card contents — is legally constrained in ways that prevent independent verification of reported results.

In Georgia, ballot images have been the subject of years of litigation and administrative dispute. The Georgia State Election Board has taken shifting positions on whether ballot images are public records. County election officials have produced them in some cases and refused to produce them in others. Litigation by groups including Coalition for Good Governance has attempted to compel disclosure on transparency grounds.

In Arizona, records requests for ballot images and tabulator logs from the 2020 and 2022 elections produced inconsistent results across counties. Maricopa County’s handling of such requests — and the political controversies that accompanied them — are well documented, though the access disputes are entangled with unrelated election-denial litigation that complicates their use as a clean legal precedent.

In Wisconsin, Pennsylvania, and Michigan, different statutory regimes govern access to voting-equipment outputs, and the legal status of specific record types varies by state and in some cases by county. There is no national standard. The Brennan Center for Justice and Verified Voting have documented the patchwork nature of state-level transparency law as it applies to elections.

In a more direct parallel to Clarkson’s case, voting system vendors have in some jurisdictions asserted that their software’s internal data formats, error logs, and audit trails are proprietary trade secrets not subject to public disclosure even where general election records are. These claims have been contested with mixed success. The practical effect is that even when a state’s open records law would otherwise reach a given record, a layered claim by the vendor can delay or defeat disclosure.

The Kansas ruling in Clarkson v. Lehman is the cleanest, most fully litigated, and most legally definitive instance of the pattern. But the pattern itself — primary election evidence that is produced in the normal course of an election and that exists in physical or digital form, yet is not reliably available to citizens, researchers, or the press — is national in scope. A citizen in most American jurisdictions who wishes to do what Clarkson attempted to do will face some version of the same structural obstacle.

8.8 — What Actual Vote Would Have Contributed

The preceding case studies in this collection describe failures in which an error occurred and AV would have caught it. This case describes a failure of a different kind: no error was documented, because the evidence that would have been needed to document one was never produced. AV’s contribution in cases of this type is structurally different, and in some respects more fundamental.

The evidence exists before the access question arises. Every case in this collection that depends on post-election access to government records is vulnerable to the same structural weakness Clarkson encountered. An AV-collected image of a precinct’s poll tape, captured at the close of polls by a civilian observer in a public polling place, does not become a government record. It is a private photograph of a public document taken in a public space at a lawful moment. The custody, storage, and disclosure of that image are not governed by KORA, K.S.A. 25-2422(a)(1), or the analogous statutes of any other state, because the image does not live in the state’s custody to begin with. The access question is resolved before it can be asked.

The AV corpus enables the analysis that Clarkson attempted. Clarkson’s immediate purpose was to conduct a statistical comparison between precinct-level machine counts and their corresponding paper records to estimate the error rate of voting machines. She could not do this because she did not have access to either input. An AV corpus — systematic civilian photographs of precinct-level results tapes, taken at the close of polls and paired with corresponding official results published by county and state election offices — provides both inputs in public, independently verifiable form. Anyone with statistical training, a laptop, and a research question can perform the kind of analysis Clarkson proposed, using a dataset that no government body can withdraw, reclassify, or seal. The class of legitimate academic inquiry into voting system performance that Kansas foreclosed on Clarkson is opened by the AV dataset to anyone who wants to pursue it.

Transparency without permission. The most important structural feature of the AV approach is that it does not depend on government cooperation. Clarkson’s attorney framed the problem directly during the 2016 summary judgment hearing: the trial court was being asked to rule on whether a Kansas citizen had the right to examine the evidence that would determine whether her county’s voting machines had counted correctly, and the court ruled that she did not. Every element of that ruling turned on the fact that the evidence was in government custody, governed by statutes written primarily to protect the privacy of individual ballots and the administrative control of election officials. AV’s photograph of a precinct tape is not a ballot. It reveals no individual vote. It is a photograph of an aggregate total — the sum of votes cast on a specific machine at a specific precinct — printed on a paper document that was, at the moment of photography, physically displayed or posted in a public polling place. No Kansas statute, no county rule, no court ruling in this line of cases applies to that photograph, because the photograph is a civilian-custody record of a public fact captured at a public moment. The transparency it creates operates outside the legal regime that blocked Clarkson.

Statistical auditing at scale. Clarkson’s analysis covered a handful of elections using publicly available official returns. The structural limit on her work was not her methodology; it was the absence of a parallel dataset of precinct-level machine outputs against which to compare those returns. An AV corpus covering tens of thousands of precincts across multiple states and multiple election cycles would enable a class of statistical audit — cross-jurisdictional, longitudinal, and independent of any government data pipeline — that is currently not possible anywhere in the United States. Whether any given statistical methodology, including Clarkson’s cumulative-vote-share approach, ultimately yields findings that are interpretable as evidence of machine error or as artifacts of voter composition, the presence of the underlying data at sufficient scale is what makes the question answerable at all. AV builds that data.

What AV Would Not Have Done

AV would not have produced, and cannot produce, a court ruling establishing a legal right to access VVPAT or similar government-custody records. The Kansas rulings in Clarkson v. Lehman remain as the law of Kansas. AV sidesteps the ruling rather than overturning it.

AV also does not address the proprietary-software and trade-secret dimensions of election transparency. The internals of voting system firmware, the data formats of memory cards, the contents of vendor-controlled audit logs — none of these are photographed at the precinct, and none of these become available through AV collection. Claims about machine behavior that cannot be inferred from the printed precinct tape remain out of reach.

Finally, AV does not itself conduct statistical analysis or draw conclusions. It captures evidence. The analysis that Clarkson attempted — or that any successor researcher might attempt — is a separate undertaking performed against the corpus. AV’s contribution is to make the corpus exist; what is done with it is determined by the researchers who use it.

8.9 — Transparency as a Value

The case studies in this collection have, in various places, made the case for AV on accuracy grounds: elections contain errors, the errors go undetected, and AV detects them. That argument is valid as far as it reaches, but it does not exhaust the argument for AV. The Clarkson case surfaces a second pillar that the accuracy argument on its own can obscure.

A correctly run election is not the same as a verifiable election. The difference matters because the social function of election verification is not only to catch errors — it is also to provide the losing side with evidence-based reasons to accept the outcome. An election that the losing side cannot independently verify is an election whose legitimacy rests entirely on the credibility of the officials who administered it. When that credibility is strong and uncontested, this works. When it is contested, it does not.

American election administration is increasingly operating under contested credibility conditions. The post-2020 rise of election denial has made the legitimacy of routine administrative processes a subject of active public dispute in ways it was not a decade ago. The usual response to that dispute has been to emphasize the robustness of existing official safeguards: audits, certifications, chain-of-custody procedures. That response has limits. When the credibility of officials is itself part of what is disputed, official assertions about official procedures do not resolve the dispute; they deepen it.

Independent civilian evidence, captured outside the government’s custody, answers a different question. It does not require trusting the officials. It does not require trusting the vendor. It requires only trusting the photograph of a paper document taken in a public space, which is the lowest bar of verification available. Its legitimacy is structural rather than reputational — a property of where the evidence came from, not of who vouches for it.

This is the transparency argument, and it holds regardless of whether any given election contains an error. An election whose official results are perfectly correct, whose machines counted every vote accurately, and whose administration was flawless in every respect still benefits from having been independently verified, because the public record of that verification is what closes off bad-faith attack. A defender of the election result who can point to the AV corpus and say “here is a civilian photograph of the precinct-level tape, and it matches the official total” is in a qualitatively stronger rhetorical and epistemic position than one who can point only to the certified result and ask to be trusted. The existence of the AV evidence, not any particular finding it contains, is what produces this effect.

Clarkson’s lawsuit sought access to evidence that would have served this function. She was denied. In the twelve years since she first filed, neither Kansas nor most other states have changed the legal regime that denied her. AV does not need that regime to change. It operates alongside it.

8.10 — What We Know and What We Don’t

What we know:

Between 2013 and 2018, Elizabeth Clarkson pursued two lawsuits in Kansas seeking access to paper records produced by voting machines in Sedgwick County. She lost at the trial court level in 2013 and again in 2016. The Kansas Court of Appeals dismissed her 2016 ruling on appeal as moot in 2018. At no point did any court order disclosure of the records she sought. Kansas law, as established by this line of cases, does not permit a private citizen to obtain VVPAT records under either the open records act or the election recount statute, and does not entitle a private citizen to participate in a government-conducted recount at a level of access sufficient to perform an independent audit.

Separately, Kansas Secretary of State Kris Kobach proposed legislation in early 2016 that would require post-election audits of voting equipment by bipartisan boards in public settings. Clarkson publicly endorsed the proposal and testified before the Kansas Joint Committee on Ethics and Election in October 2017 in support of audit legislation. As of that testimony, Sedgwick County’s election office confirmed that state statutes had not yet changed to permit post-election audits of voting equipment.

What we don’t know:

Whether the 2014 Sedgwick County voting machines counted correctly, because the evidence to establish that was never produced. Whether the statistical pattern Clarkson observed in her published work reflected voting-machine error, voter demographic correlations, or some combination, because the audit that would distinguish among these explanations was legally foreclosed. Whether Kansas has since enacted audit legislation of the kind Clarkson and Kobach both proposed, and if so whether it has in practice produced the kind of transparency Clarkson sought, because any audits administered by the same bodies whose work they are intended to verify do not eliminate the structural access problem she identified. Whether the class of independent statistical analysis Clarkson attempted would, if conducted at scale against primary evidence, yield findings that bear on public confidence in American elections — because, in the absence of a corpus of precinct-level primary evidence available outside government custody, such analysis has not been possible.

These are not rhetorical unknowns. They are practical ones. Each of them would be answerable, in Kansas and elsewhere, if the corresponding primary evidence were available. AV makes such evidence available.

8.11 — Why It Matters

A credentialed statistician who set out to conduct routine academic research on a question of public importance was, over the course of five years, told by every available level of Kansas government that she could not have the evidence. Not the trial court. Not the elections commissioner. Not the secretary of state. Not the appeals court. The ruling was not that she was wrong on the merits — no court ever reached the merits. The ruling was that the evidence that would permit a ruling on the merits was outside her legal reach.

This case is uncomfortable precisely because it does not produce the kind of crisp conclusion the other case studies in this collection produce. There is no “4,000-vote error” here, no “40% discrepancy,” no “seat was awarded to the wrong candidate.” There is a sequence of legal rulings that, taken together, establish that in Kansas — and by extension, in any American jurisdiction whose law is structured similarly — the machinery of elections runs in a black box as far as any ordinary citizen is concerned, and that the doors to the box are sealed both during elections and after.

The defense of that closure is typically couched in legitimate-sounding terms: ballot secrecy, administrative integrity, the need to prevent the chain of custody from being compromised. These concerns are real. But they explain only why the government’s copy of the records must be handled carefully; they do not explain, and cannot explain, why a civilian photograph of a publicly displayed precinct tape must be foreclosed. The reasons that operate on the government’s copy do not operate on the civilian copy, because the civilian copy is a different kind of thing — a record of a public fact, created lawfully in public, held in private hands, subject to no statutory regime governing government-custody ballots.

The Kansas courts in Clarkson v. Lehman ruled on the government’s copy. They did not, and could not, rule on the civilian copy, because the civilian copy did not exist in 2014. AV is the mechanism by which that civilian copy comes into existence at every precinct in every election where it is deployed. Its legal standing is entirely different from the records Clarkson sought, because its custodial status is entirely different.

This is the structural argument for AV stated in its simplest form: the records we need in order to verify our elections independently are records that the legal system will not provide through existing channels. Clarkson v. Lehman is not a gap in the legal regime; it is the regime. It works exactly as designed. It produces exactly the outcome it is structured to produce. And that outcome is opacity.

The response to structural opacity is structural transparency. AV does not argue against ballot secrecy, does not contest the chain of custody, does not violate any privacy interest that any court has identified. It simply creates a parallel record, held in different hands, subject to different rules. That record is the evidence Clarkson was denied. And in every election where AV is deployed, the question she was unable to answer about the 2014 election in Sedgwick County — did the machines count correctly? — becomes answerable.

8.12 — Further Reading

Litigation

News reporting

Investigative and advocacy coverage

Advocacy site and academic work

8.13 — Related Cases

Clarkson v. Lehman is unusual in this collection in that it documents a transparency failure rather than an accuracy failure. Several other cases in the collection touch adjacent ground:

  • Case 5 — Baldwin County, Alabama, 2002 is the collection’s other case where the underlying question remains permanently unanswerable, though for a different reason: in Baldwin, the ballots were destroyed by law before any audit could occur. In Clarkson, the evidence physically exists but is inaccessible. Both produce the same end state — a question that cannot be resolved — through different mechanisms.
  • Case 6 — Ohio 2004 documents a pattern of record destruction despite a federal preservation order. The defense of closure in Ohio turned on different legal theories than in Kansas, but the structural effect was similar: evidence that would have resolved a post-election dispute became, through administrative action, unavailable.
  • Case 14 — Georgia Presidential Election, 2020 — Fulton County’s unsigned tabulator tapes covering hundreds of thousands of votes and the disappearance of roughly ten tapes covering approximately twenty thousand more. The access question in Fulton was operational rather than legal — the evidence was not withheld but was degraded in the ordinary course of administration — but the downstream effect was the same: a portion of the primary record of the election was not available for independent verification.
  • Case 11 — Fraction Magic Attack — A theoretical demonstration that precisely the kind of reporting-layer attack AV is designed to catch is technically feasible against the systems in use. Clarkson is the companion case from the transparency side: the kind of independent verification that would detect such an attack has been legally foreclosed in at least one state.