Coalition for Good Governance v. Raffensperger

Annotate this Case
Justia Opinion Summary

Petitioners challenged the 2018 election for lieutenant governor, an election in which more than 3.7 million Georgians cast a vote. They alleged defects in electronic voting machines cast doubt on the election in which Geoff Duncan defeated Sarah Riggs Amico by 123,172 votes. To prevail, a party contesting an election must therefore offer evidence, not merely theories or conjecture, that places in doubt the result of an election. "And although the technology our State has used to conduct elections has changed over time, the burden a party carries when challenging the result of an election has not. The Petitioners in this case have not carried that burden."

Download PDF
SUPREME COURT OF GEORGIA October 31, 2019 The Honorable Supreme Court met pursuant to adjournment. The following order was passed: It appearing that the attached opinion decides a second-term appeal, which must be concluded by the end of the August Term, it is ordered that a motion for reconsideration, if any, must be received in the Supreme Court E-Filing/Docket (SCED) System by noon on Friday, November 8, 2019. SUPREME COURT OF THE STATE OF GEORGIA Clerk’s Office, Atlanta I certify that the above is a true extract from the minutes of the Supreme Court of Georgia. Witness my signature and the seal of said court hereto affixed the day and year last above written. , Clerk In the Supreme Court of Georgia Decided: October 31, 2019 S19A0769. MARTIN et al. v. FULTON COUNTY BOARD OF REGISTRATION AND ELECTIONS et al. WARREN, Justice. In this case, Petitioners challenge the 2018 election for lieutenant governor—an election in which more than 3.7 million Georgians cast a vote—alleging that defects in electronic voting machines cast doubt on the election in which Geoff Duncan defeated Sarah Riggs Amico by 123,172 votes.1 Elections are critical to our democratic republic. We give great credence to the choices citizens make when they engage in the democratic process by voting to select their representatives. And This case was originally styled Coalition for Good Governance et al. v Raffensperger et al. But because the trial court dismissed petitioner Coalition for Good Governance and defendant the Georgia Secretary of State, and because no party appeals those rulings, this Court has entered an order correcting the case caption. This opinion reflects those changes. 1 because we place so much value on that exercise of democracy, we afford great weight to election results. Indeed, “[t]he setting aside of an election in which the people have chosen their representative is a drastic remedy that should not be undertaken lightly, but instead should be reserved for cases in which a person challenging an election has clearly established a violation of election procedures and has demonstrated that the violation has placed the result of the election in doubt.” Hunt v. Crawford, 270 Ga. 7, 10 (507 SE2d 723) (1998). Georgia law nonetheless allows elections to be contested through litigation, both as a check on the integrity of the election process and as a means of ensuring the fundamental right of citizens to vote and to have their votes counted accurately. See OCGA § 212-520 et seq. But an election contest is, by statutory design, an expedited proceeding—and one that vests in trial courts broad authority to manage the proceeding, including to “proceed without delay to the hearing and determination of” the election contest. See OCGA § 21-2-525 (b). This system balances citizens’ franchise 2 against the need to finalize election results, which, in turn, facilitates the orderly and peaceful transition of power that is a hallmark of our government. As explained in more detail below, Petitioners claim that they placed in doubt the election for lieutenant governor (and thus established that a new election was required) by offering evidence of a few specific instances of electronic voting machine malfunction, and of statistical differences in voting patterns between the 2018 general election and prior general elections that they say show that Georgia’s “profoundly vulnerable machines caused thousands of voters using electronic machines to either not vote for Lieutenant Governor or for those votes not to be counted.” This Court has long held that “‘the party contesting the election has the burden of showing an irregularity or illegality sufficient to change or place in doubt the result of the election.’” Meade v. Williamson, 293 Ga. 142, 143 (745 SE2d 279) (2013) (citation omitted). To prevail on such a claim, a party contesting an election must therefore offer evidence—not merely theories or conjecture— 3 that places in doubt the result of an election. And although the technology our State has used to conduct elections has changed over time, the burden a party carries when challenging the result of an election has not. The Petitioners in this case have not carried that burden, and the discussion that follows explains why. In Division 1, we chronicle Petitioners’ claims from the time they were filed in the days after the November 6, 2018, statewide general election, until the trial court granted a motion to involuntarily dismiss Petitioners’ then-remaining state law election contest claim after trial in January 2019. In Division 2, we review Petitioners’ four enumerations of error related to pre-trial discovery—including claims that the trial court did not allow reasonable time for discovery, did not permit needed discovery, and wrongly denied Petitioners’ motion to compel and motion for continuance. We conclude that, given the Election Code’s statutory mandates and the broad discretion trial courts are given to manage pretrial discovery, the trial court did not abuse its discretion. In Division 3, we review Petitioners’ two enumerations of error related 4 to the involuntary dismissal of their election contest claim—that the trial court made an erroneous factual finding about the number of potential illegal or irregular votes in the election for lieutenant governor, and that the trial court erred in its legal analysis of whether Petitioners met their burden of presenting sufficient evidence of irregularities related to electronic voting machines used in the 2018 general election. We conclude that, although the trial court made at least one clearly erroneous finding of fact, it reached the correct legal conclusion when it determined that Petitioners failed to meet their burden of presenting evidence that places in doubt the result of the election for lieutenant governor. Finally, in Division 4, we review and reject Petitioners’ unsupported argument that the trial court erred by denying Petitioners’ request for a jury trial. As a result, we affirm the trial court’s dismissal of Petitioners’ petition contesting the election for lieutenant governor. *** On November 6, 2018, a statewide general election was held to elect Georgia’s next Governor and a number of other statewide 5 officials, including (among others) Attorney General and Secretary of State.2 3,780,304 ballots were counted in the election for Georgia’s Lieutenant Governor; candidate Geoff Duncan received 1,951,738 votes and candidate Sarah Riggs Amico received 1,828,566.3 Duncan therefore won the election with a margin of victory of 123,172 votes. On November 23, 2018, Petitioners—the Coalition for Good Governance (a nonprofit organization organized under Colorado law but apparently headquartered in North Carolina), Rhonda J. Martin (an “aggrieved elector”), Jeanne Dufort (an “aggrieved elector”), and Smythe DuVal (a voter and the Libertarian Party candidate for Secretary of State of Georgia in the November 2018 election)—filed A number of candidates for other state and federal offices were also on the ballot. 2 A total of 3,939,328 votes were cast in the election for Georgia’s Governor. As explained more below, Petitioners label the difference between the total votes cast in the election for Governor and the total votes cast in the election for Lieutenant Governor the “undervote.” Here, that number is 159,024 (3,939,328 minus 3,780,304). 3 6 a petition under OCGA § 21-2-520 et seq. contesting the election.4 They sued the Secretary of State of Georgia; the Gwinnett, DeKalb, and Fulton County Boards of Registration and Elections; and thenLieutenant Governor-elect Geoff Duncan (“Defendants”), requesting (among other things) that the lieutenant governor election be declared invalid and a new election ordered that did not use the direct-recording electronic (“DRE”) voting system. After approximately seven weeks of motions practice, multiple hearings, accelerated discovery, and a two-day bench trial5 the trial court granted Defendants’ motion for involuntary dismissal of Petitioners’ petition. See OCGA § 9-11-41 (b) (“After the plaintiff, in an action tried by the court without a jury, has completed the presentation of his evidence, the defendant, without waiving his These were the original petitioners, but—as discussed below in footnote 8—the Coalition for Good Governance was ultimately dismissed as a petitioner with respect to the state law election contest claim, which was Petitioners’ only remaining claim at that time. 4 The Election Code refers to “hearings” in relation to election contests. See OCGA § 21-2-525. But because the parties refer to the final merits hearing in this case as a “trial,” and to distinguish it from prior hearings in the case, we will also refer to the final hearing as a trial. 5 7 right to offer evidence in the event the motion is not granted, may move for dismissal on the ground that upon the facts and the law the plaintiff has shown no right to relief.”). Arguing that the trial court abused its discretion and committed legal error when it dismissed the petition, Petitioners now appeal. Specifically, they contend that 159,024 fewer ballots were cast for lieutenant governor than for governor in the 2018 general election—an undervote of approximately 4 percent—and that when compared to historical voting patterns—which they argue show only a 0.8 percent average undervote between elections for governor and lieutenant governor over the previous four general elections—the difference constitutes an “[e]xtreme [u]ndervote” that places in doubt the result of the election for lieutenant governor and requires a new election under OCGA § 21-2-527 (d). In its most basic form, Petitioners’ argument is that the “proven vulnerability” of Georgia’s electronic voting machines, coupled with this “[e]xtreme [u]ndervote” and with specific instances of voting machine malfunction, place in doubt the election result for the office of lieutenant governor. 8 What follows below is a comprehensive review of the key events that transpired in this case leading up to trial, as well as a summary of the evidence presented at trial. Although these events occurred over the span of only a few months, there was—to say the least—a large volume of communications between the many lawyers on this case, motions made before the trial court, and legal briefing filed in the trial court (and later in this Court). A thorough recounting of those events and arguments is necessary to illuminate the record that was created before and during trial, and therefore constitutes the record we review today on appeal. In light of that record, the relevant statutes, and case law interpreting those statutes, and given our review of the trial court’s factual findings and legal conclusions in this case, we affirm the dismissal of Petitioners’ petition. 1. Procedural and Evidentiary History. (a) The Petition and Other Post-Election Motions. Georgia held a statewide general election on November 6, 2018. On November 23, 2018, Petitioners timely filed a petition contesting 9 the election for lieutenant governor, alleging three counts. Count 1, the state law election contest claim, alleged that the use of “defective, legally non-compliant, and malfunctioning DRE machines to conduct the Contested Election constituted ‘misconduct’ and ‘irregularity’ sufficient to change or place in doubt the result of the Contested Election” under OCGA § 21-2-522 (1), and that the “malfunctioning DRE machines rejected legal votes or received illegal votes sufficient to change or place in doubt the result of the Contested Election” under OCGA § 21-2-522 (3). Count 2, a federal claim under 42 USC § 1983, alleged that the use of defective DRE machines violated Petitioners’ due process rights under the Fourteenth Amendment by infringing on their fundamental right to vote. And Count 3, another federal claim under 42 USC § 1983, alleged that use of the defective DRE machines violated Petitioners’ equal protection right under the Fourteenth Amendment by treating electors who voted on DRE machines differently than other, similarly situated electors who used paper absentee ballots. Five days later, on November 28, 2018, Petitioners filed a 10 “Motion for Clarification and to Request Setting of Trial Date,” in which they noted that they would “be filing a motion to inspect a sample population of the suspect electronic voting equipment to determine whether there is . . . extant electronic evidence of the causes of the election machine malfunctions”; requested that the already-set hearing date of December 6, 2018 be “a preliminary hearing or status conference . . . to address preliminarily issues related to the trial of the matter, including stipulations, evidence, discovery, necessary parties, etc., as well as any special issues raised under the election contest laws”; and asked the trial court to “fix a date for the trial itself, for two days no earlier than the week of December 10, 2018, and taking into consideration realistic dates for necessary forensic examination and submission of pre-trial briefing.” The next day, Petitioners filed an “Emergency Motion for Inspection of Electronic Election Equipment and Production of Documents” (“motion for inspection”). In that motion, Petitioners argued that they needed an “expert forensic examination” of 11 electronic voting equipment and of related election records, including the internal memory of electronic voting machines as well as paper records that could provide evidence of voting irregularities (and their causes) in the November 6 election. Specifically, Petitioners requested the “inspection, sampling and copying of the electronic election equipment and related election records” for [a]ll DRE machines used in the November 6, 2018 election that . . . [were] taken out of service, temporarily or permanently[;] . . . for which voter complaints were received regarding “vote flipping” or “slipping”[;] . . . reported as not permitting the voter to review the summary screen and press the final review/cast vote target area before the machine cast the ballot[;] . . . reported as not displaying the Lieutenant Governor’s race on the initial voting screen[;] . . . reporting one or more blank ballots[; and] . . . for which the public counter number on the machine tape is different than the total ballots cast on the machine tape. Petitioners also requested the same type of discovery for a list of select DRE machines from various precincts in 20 different counties across the state and all the DRE machines from 20 different precincts6; “additional equipment, information or programming 6 Petitioners’ requests were attached to their motion as Exhibits A and B. 12 necessary to make the electronic equipment operative, such as a/c power, internet connections, voter cards, supervisor cards, passwords, etc.”; and that their “representatives . . . be permitted to make copies of any of the electronic files associated with the electronic equipment that is being inspected.” Petitioners argued that a forensic examination of the DRE machines’ internal memory was required because “critical electronic information” regarding “programming errors, machine malfunction, malicious manipulation, or viruses . . . may be available only in the internal memory of the DRE machine,” and that the “electronic records of the presence and introduction of unauthorized or irregular programs or data are most likely to be located through forensic examination of unaltered internal memory data, not merely selected files of ballot images, ballot programming or tallies.” (Emphasis in original). Finally, Petitioners argued that “[s]upporting and related electronic equipment and paper records are also required in order for voting system computer experts and forensic scientists to identify the sources of the problems causing the irregularities.” 13 That same day—November 29, 2018—the case was reassigned to the Seventh Judicial Administrative district; Senior Superior Court Judge Adele Grubbs was appointed to preside over the case on November 30, 2018.7 (b) Motions to Dismiss. On December 4, 2018, the Secretary of State moved to dismiss the petition, arguing (among other things) that the Secretary was an improper party to the state election contest claim; the Coalition lacked standing to bring a state election contest claim; Petitioners failed to state a claim for their state election contest claim; Petitioners failed to name other county election superintendents, who were necessary parties; and Petitioners failed sufficiently to allege federal claims. The trial court held a status conference the next day and set a The case was reassigned in accordance with OCGA § 21-2-523 (d), which provides that “[i]f the administrative judge is a member of the circuit in which the [election contest] proceeding was filed”—which was the case here— “then the administrative judge shall select an administrative judge of an adjoining district” who is “to select a superior court judge from that district, or a superior court judge from the district in which the proceeding was filed, but not the circuit in which the proceeding was filed, or a senior judge who is not a resident of the circuit wherein the proceeding was filed.” 7 14 December 20, 2018 filing deadline for motions; a January 9, 2019 hearing date for motions; and a January 17, 2019 trial date. Petitioners did not object to that timeline, which set the trial date more than a month after the date Petitioners had originally requested. On December 20, 2018, three other defendants—the Gwinnett and Fulton County Boards and Lieutenant Duncan—also moved to dismiss the petition. Governor-elect Various of those defendants joined the Secretary of State’s motion in full or in part, and they separately contended, among other things, that Petitioners failed to serve properly all of the defendants. In a written response, Petitioners conceded that the Coalition for Good Governance—the lead petitioner in the case—lacked standing to bring a state election contest claim, but maintained that the Coalition had standing to bring the two federal claims and that, as “‘electors’ under the statute,” the other individual petitioners were proper parties under OCGA § 21-2-521 to bring the state law election contest claim. On December 27, 2018, the trial court granted Petitioners’ motion to 15 voluntarily dismiss the DeKalb County Board of Registration and Elections as a defendant. At that point, there still remained Petitioners’ single state law election contest claim and two federal claims against defendants the Secretary of State, the Gwinnett and Fulton County Boards, and Lieutenant Governor-elect Duncan. (c) January 9, 2019 Hearing and Related Rulings. On January 9, 2019, the trial court heard argument on Petitioners’ motion for inspection and on Defendants’ motions to dismiss. (i) Motions to Dismiss. At the hearing, the trial court orally granted in part, denied in part, and reserved ruling on in part Defendants’ motions to dismiss, informing the parties that it would deny Defendants’ motions to dismiss for failure to state a claim on the state law election contest count but grant Defendants’ motions to dismiss the Secretary of State as a defendant. The trial court entered an order dismissing Petitioners’ two federal counts for failure to state a claim later that afternoon and issued rulings on the remainder of Defendants’ 16 motions to dismiss as part of the January 11 discovery order discussed more fully below in Division 1 (d).8 The trial court’s various rulings left Martin, Dufort, and DuVal as the remaining petitioners, with only a state law election contest claim against three defendants: Lieutenant Governor-elect Duncan and the election boards of Fulton and Gwinnett Counties. (ii) Motion for Inspection. In support of their motion to inspect, Petitioners argued at the January 9 hearing that they wanted a “full forensic examination” of DRE machines from the various precincts and counties listed in their motion. According to Petitioners, this would require “computer Specifically, the trial court ruled that the Secretary of State was an improper party as to the state law election contest claim because the Secretary was not a permissible defendant under OCGA § 21-2-520 (2), and granted as unopposed the portion of Defendants’ motion to dismiss that contended that the Coalition lacked standing to bring a state law election contest claim. In denying the remainder of Defendants’ motions to dismiss, the trial court concluded that the remaining 157 Georgia counties that were not joined as defendants were not necessary parties to the case; Defendants had been served with process; and Petitioners had sufficiently pleaded their state law election contest claim. On appeal, Petitioners have not challenged the dismissal of their two federal claims. Defendants have not cross-appealed any of the trial court’s rulings in this case. 8 17 experts to look at the internal memory and programming” of the DRE machines. Petitioners also requested “general discovery of documentary evidence relating to the programming of the machines,” including “discovery of the GEMS databases.”9 Defendants, in turn, objected that Petitioners were requesting “unfettered access to highly sensitive information about the state voting machines,” and that although “the data from the GEMS is public record,” “the source code and the means of accessing the GEMS . . . is highly confidential.” Citing “security concerns,” Defendants urged the trial court to tailor any inspection it granted and to ensure that “whatever access is occurring is done in a way in coordination with the Secretary of State’s office . . . to make sure that nothing is altered with the machines themselves” and that other necessary security precautions were taken. After hearing argument on the motion, the trial court expressed concern that According to Petitioners’ expert, Matthew Bernhard, the Global Election Management System (GEMS) database is “an unencrypted set of data for programming and tabulating the election that is built by the election official for each election.” 9 18 Petitioners’ requested discovery sounded like a “fishing expedition” and announced that it would grant Petitioners’ request in part, but limit inspection and production to certain DRE machines and to GEMS servers in certain counties. The trial judge emphasized that going forward, she was “not going to give continuances” and specifically told the parties: “if you run up on a stumbling block” during the discovery process, “you can always [handle] it by emails to me and I will email you back.” After the parties could not agree on language for a proposed order memorializing the trial court’s rulings, both sides emailed proposed orders and additional written argument from January 10 to January 11 for the trial court’s consideration. (d) January 11, 2019 Order. On January 11—six days before the scheduled trial date—the trial court entered a written order granting in part Petitioners’ motion for inspection. But shortly before it did, Petitioners emailed the trial judge and the parties to provide advance notice that they “anticipate[d] filing” that afternoon a request for “continuance of the 19 trial.” Less than one hour later, the trial judge responded by email and informed the parties that “[t]here will b[e] no continuance” because of the “expedited” nature of the “election contest.” The trial judge again stated that if there were “issue[s] as the Plaintiffs conduct[ed] . . . discovery,” she would “take them up by e[-]mail or conference call if necessary.” Shortly thereafter, the trial court entered an Order on Pending Motions, which included rulings on Petitioners’ motion for inspection. Granting in part Petitioners’ motion, the trial court permitted two categories of inspection. First, it permitted the remaining Petitioners “to inspect the ‘GEMS’ reports or complete electronic copies thereof for the November 2018 elections that are maintained by the Gwinnett County Board and the Fulton County Board,” and specifically, the “a. Base Precincts With Races Report[;] b. Ballot Image Report[;] c. Vote Center With Cards Report[;] d. Statement of Votes Cast Report[; and] e. Summary Report.” It also restricted the examination by permitting “[o]nly personnel of the Secretary of State or Fulton or Gwinnett Counties [to] access the 20 GEMS servers directly in connection with this inspection.” Second, the trial court permitted inspection of certain DRE machines in “post-election mode.” Specifically, it authorized Petitioners to inspect DRE units “using post-election memory cards that were used in the voting locations” identified in paragraphs 40, 41, 44, 45, and 46 of the Petitioners’ petition—i.e., DRE machines in specific polling places in Fulton, Henry, and Worth Counties. The trial court ruled that Petitioners could “examine the Internal memory storage of each such DRE unit,” but made clear that Petitioners were “not to in any way damage the DRE machines or the information contained therein.” The trial court also prohibited Petitioners from “copy[ing], imag[ing], sav[ing], or retain[ing] the DRE machines or the information contained therein” and from “upload[ing] or introduc[ing] any information into the DRE machines,” and also ruled that “[d]estructive testing shall not be permitted.” Finally, the trial court required the parties to “enter into an appropriate protective order preserving the confidentiality of confidential information, if any, obtained in this discovery prior 21 to the commencement of any inspection authorized under this Order.” (e) Production and Inspection of DRE Machines and the GEMS Database. Following the January 9 hearing and the trial court’s January 11 order, the parties communicated about—and disagreed about— the proper protocol for conducting DRE machine inspections. Notwithstanding these disagreements, representatives of the Fulton County Board and the Secretary of State agreed to meet with Petitioners’ retained expert on Monday, January 14, to conduct an examination of the County’s voting machines.10 In an email sent later that day, Petitioners’ expert told Petitioners’ counsel that Defendants gave him a compact disc containing “all reports” required by the trial court’s discovery order except for most of the Separately, it appears that non-party Henry County Board of Registration and Elections agreed to make the DRE units it possessed available for inspection the following day, January 15. It appears from the record, however, that Henry County offered to allow inspection in accordance with the same protocol that Fulton County did, and there is no indication that Petitioners actually attempted to conduct any inspection of the Henry County DRE machines. 10 22 Ballot Image Reports; only 2 out of 2,084 Ballot Image Reports were copied onto a CD at the time. But Petitioners later complained that they were “unable to conduct any examination or investigation of the DRE machines or their internal memory” because Defendants did not make “a complete and accurate electronic copy . . . of the internal memory,” did not allow Petitioners’ expert “to instruct County personnel on how to make the copy . . . without damage to the files or machine,” and did not make electronic election records available. To that end, Petitioners’ expert recounted in a January 14 email to Petitioners’ counsel that Defendants were unwilling to implement (without first consulting with their counsel) a proposal to copy the internal memories of selected DRE machines by physically removing the outside cases from the machines, accessing the machines’ motherboards, and then inserting either a chip or an adapter into the machines—a process that Petitioners’ expert acknowledged could create a “security vulnerability.” Petitioners’ expert therefore determined that he was unable to conduct an inspection of the DRE 23 machines and left. According to Petitioners’ motion for continuance filed later that night, “Plaintiffs’ expert left the premises without having been able to examine the internal memory of the DRE machines, which review has to precede any other examination of the DREs.” (f) Motion for Continuance; Motion to Compel and for Additional Discovery; and Demand for Jury Trial. Later that night, Petitioners filed a motion to continue the January 17 trial to a later date.11 They made two main complaints: first, that Defendants were “delay[ing] in making the electronic election records and voting equipment available for examination”; and second, that even if Defendants “were immediately to comply with the Court’s Order on Pending Motions and start copying the selected DRE machines’ memory onto laptops,” it would take so long to copy and analyze the internal memories that Petitioners would The record shows that the parties had been emailing—and arguing about—the scope of, and protocol for, DRE machine inspection since January 9. But the record contains no evidence that Petitioners emailed or requested a conference call with the trial court about this issue between the entry of the trial court’s January 11 order and the filing of Petitioners’ motion for continuance on the night of January 14. 11 24 not be ready to present their case by the scheduled trial date. According to Petitioners’ motion for continuance, “it will physically take days before the copying is complete and it will require several weeks before the memory can be analyzed to produce evidence that can be presented meaningfully in court.” Petitioners thus requested that the trial date be reset “to no earlier than three weeks after the date when the Defendants make available for inspection and examination all the electronic election records and equipment.” The remaining Defendants and the Secretary of State filed responses in opposition to Petitioners’ motion for continuance the next day, arguing that the Secretary’s proposed protocol for DRE machine inspection complied with the trial court’s January 11 order—but that when Defendants “offered to allow Plaintiffs to inspect the DRE machines as the Secretary suggested, Plaintiffs refused.” Noting that Petitioners did not object to the “expedited calendar” when the January 9 hearing date and January 17 trial date were initially set, Defendants argued that Petitioners ignored the trial court’s instructions to bring discovery issues to the court’s 25 attention by email or conference call and that they instead filed a motion for continuance as an “attempt to delay” the trial date, even though Petitioners’ “discovery needs were known to them” when they filed their petition in late November. Early on the morning of January 16, the trial court emailed the parties and once again stated that the case would not be continued, and later that morning, the trial court entered a written order denying Petitioners’ motion. But between the time the trial court had informed the parties that the case would not be continued and entry of a written order denying Petitioners’ requested continuance, Petitioners filed a “Motion to Compel and for Additional Discovery,” arguing that Defendants had refused to allow Petitioners to examine the internal memory of the DRE machines (as required by the January 11 order), and again requesting production of the entire Gwinnett and Fulton County GEMS databases, as opposed to reports generated from those databases. Later that afternoon, Petitioners also filed a demand for a jury trial under OCGA § 21-2526 (a). 26 Defendants opposed both motions, contending that they had complied with the trial court’s discovery order; that Petitioners were not entitled to relief; that Petitioners were employing “attempts to delay” the proceedings; and that Petitioners were actually seeking reconsideration of the trial court’s January 11 discovery order requiring production of certain GEMS reports from Gwinnett and Fulton Counties. On January 17, 2019—the day trial was scheduled to begin— the trial court addressed the parties’ outstanding motions from the bench before starting trial. The court denied Petitioners’ motion to compel and for additional discovery, struck Petitioners’ demand for a jury trial, and denied another request for additional discovery that Petitioners made orally that morning. The trial began as scheduled on January 17. (g) The Evidence Offered at Trial. Over the course of trial, Petitioners presented documentary evidence as well as testimony from seven lay witnesses and one expert witness in support of their claim that the state’s electronic 27 voting system was so defective that it cast doubt on the election for lieutenant governor and required the election to be overturned. One of Petitioners’ lay witnesses was a poll watcher who testified that she had personally observed one DRE machine that would not print the poll tape after the election and that a precinct poll manager had stated that “there were multiple technical issues” on election day.12 She also observed one voter having a problem with a DRE machine, and then learned from the poll manager that the DRE machine initially “self-cast” the voter’s ballot before she made a selection for lieutenant governor and another candidate. The poll watcher acknowledged, however, that the machine was immediately placed out of service and that the voter did not complain that her vote had been improperly recorded. Petitioners also called a voter who testified that when she voted on a DRE machine, the lieutenant governor’s race did not The witness testified that she was tasked with reporting the total number of votes cast at the precinct to the Democratic Party and was never informed one way or the other whether there was a discrepancy regarding that total. 12 28 initially appear on her electronic ballot. But the voter also testified that when she went back to the beginning of the ballot, she was able to find that race and cast her vote for Amico. Petitioners also called Michael Barnes, Director of the Center for Election Systems in the Secretary of State’s Office, a lay witness who testified primarily about DRE machines and ballot design. With respect to the accuracy of DRE machines, Barnes testified that in all of the pre-election-mode and election-mode tests of DRE machines that he had “been involved with” since 2001, he “never encountered a situation where [he] put in one particular vote and . . . it came out differently in the report.” Specifically, for the 2018 election, the pre-election and election-day testing showed the DRE machines were properly counting votes, including for the lieutenant governor’s race. He also testified that Georgia’s electronic voting system was a “closed system” that could not be “hack[ed]” remotely, and that although the Secretary of State’s Office had not conducted a “forensic analysis” of the DRE voting systems either before or after the 2018 election, it did conduct a “recertification” of the voting 29 system in 2017—and examined the equipment in several counties— and identified no problems. Moreover, the Secretary of State’s Office ran reports for the Fulton and Gwinnett County Boards that confirmed that the lieutenant governor’s race “appeared on every electronic ballot” in those counties, and that “all memory cards that were created that would have collected . . . votes [had] been uploaded and accounted for by GEMS.” With respect to the ballot design, Barnes testified that after the election, there was some concern that the ballot design could have “been confusing, especially for new voters,” and caused voters inadvertently to skip the lieutenant governor’s race because they thought they were “voting for a ticket” that included the governor’s race. He contrasted the 2018 general election with general elections conducted in Georgia in 2010 and 2014, testifying that in the 2018 general election, the DRE machines were programmed to display only the candidates for governor and lieutenant governor on the first screen, and that the two sets of candidates appeared side-by-side with the Democratic and Republican candidates for each race 30 positioned horizontally across from each other. By contrast, in 2010 and 2014, the lieutenant governor’s race did not appear side-by-side with the governor’s race on the same DRE machine screen; instead, a “U.S. Senate race and the governor’s race [were] on the first page” and “the race for lieutenant governor and other” races were on the next page. Barnes testified that although election officials had “been very pleased with the two-column ballot,” there was “a possibility that it could have been confusing” in the 2018 general election, “especially for new voters that are voting for the first time in the state of Georgia . . . [who] may feel like because the Republican candidates are line[d] up and the Democratic candidates are lin[ed] up that you make one selection and you’re voting for a ticket,” like in “many states” other than Georgia.13 Additionally, Barnes confirmed that—unlike the layout of the electronic ballots cast on DRE machines—the paper ballots cast in the 2018 general election Barnes also testified that voter turnout was more than 11% higher in 2018 than in 2014, and that over 336,000 first-time voters voted in Georgia’s 2018 general election—an “increase of about three-and-a-half times the number of new voters” who voted in 2014. 13 31 did not include only the governor’s and lieutenant governor’s races on a single page; instead, the candidates for those races, along with candidates for other offices, were listed one below the other, rather than side-by-side. He also testified that the certified returns from Fulton, Gwinnett, and DeKalb Counties showed that “the highest number of write-ins cast for statewide office was in the lieutenant governor’s office.” Petitioners also called as a lay witness the Fulton County Director of Registration and Elections, who testified that Fulton County did not examine its GEMS database or memory cards for malware after 2016, review its GEMS database for coding errors prior to the 2018 election, or examine the DRE machines in Fulton County after the 2018 election. On cross-examination, he clarified that Fulton County conducted “logic and accuracy tests” on all of its DRE machines before the 2018 election, that all of the DRE machines passed those tests, and that Fulton County had never experienced software issues or viruses with its DRE machines or GEMS system. 32 Petitioners also called Marilyn Marks, Executive Director of the by-then-dismissed petitioner Coalition for Good Governance, who also served as a poll watcher for Georgia’s 2018 general election. Petitioners attempted to qualify Marks as an expert in “DRE machines generally and their use in Georgia and the practices and procedures of the State of Georgia for paper ballots and for electronic ballots,” but—after Defendants conducted a voir dire about Marks’s experience and credentials—the trial court denied Petitioners’ motion, allowing Marks to testify only as a lay witness. Among other things, she testified that she observed three DRE machine malfunctions in which the screen displayed “vote cancelled” and a “big red warning sign” before ceasing to work. Also testifying for Petitioners was Christopher Brill, a political consultant whom Petitioners twice tendered as an expert qualified to discuss “under-voting and the particular reasons for undervoting,” including the reasons or possible reasons for the “undervote” in the contested election, but whom the court twice refused to qualify as an expert because opinion regarding the “reasons why” 33 there was an undervote was not “an area of expertise.”14 Though not qualified as an expert, Brill was permitted to offer lay testimony that in the election for “lieutenant governor, the under-vote was about 4 percent lower than the total that was cast for governor”; that he calculated the average undervote for the four prior lieutenant governors’ elections in Georgia as “around 0.8 percent”; and that the four percent undervote in the November 6, 2018 election for lieutenant governor was higher than in the previous four general elections. He also testified that, if the “historical trend of .8 [percent] were applied” to the 2018 contested election, “the under-vote assumption” would result in a total number of under-votes of “around 31,532.”15 Brill also testified about the difference in undervotes between votes cast electronically and those cast on paper absentee ballots. 14 Petitioners do not challenge these rulings on appeal. Based on evidence offered at trial, this calculation was reached by multiplying the total number of votes for governor (3,939,328) by approximately .008 (the number Brill offered as the average historic undervote) to identify the total number of expected undervotes (“around 31,532”). 15 34 He testified that “paper voting” for lieutenant governor “had only a 1 percent under-vote” compared to the “electronic” voting, which “had over 4 percent,” and that a differential of that magnitude did not appear in the races for Secretary of State or Attorney General, where the undervote percentages of electronic and paper voting were “relatively even” and “consistent.” Brill testified that he had “never seen a type of under-vote where voters” skip one race but then vote at higher rates for other races farther down the ballot. On crossexamination, however, Brill agreed that Georgia’s 2018 general election was “a very high turnout election,” admitted that he did not consider the number of new voters in 2018 when he compared that election to prior elections, and acknowledged that Amico received more votes than some other statewide Democratic candidates. He also conceded that he had never conducted any analysis of ballot design in general. Petitioners then sought to have Professor Philip Stark testify. However, Stark was available only by telephone or video conference, and Petitioners had not provided the notice required by Uniform 35 Superior Court Rule 9.2 (c) (“Any party desiring to call a witness by video conference shall file a notice of intention to present testimony by video conference at least thirty (30) days prior to the date scheduled for such testimony.”). The trial court noted that although the case was “an expedited situation,” Petitioners should have at least given notice at the January 9 hearing instead of “suddenly walk[ing] in [on] the afternoon of trial and say[ing] I want to call somebody.” As a result, the court denied Petitioners’ request to examine Stark by video conference, and he was not permitted to testify. Petitioners then tendered, and the court qualified as “an expert on DREs,” Matthew Bernhard, a Ph.D. candidate studying computer science, who testified that the reported undervote pattern in the lieutenant governor’s race had “less than 1 in 10,000 chance of appearing at random . . . in the course of an election.” To support that conclusion, he opined that Georgia’s DRE machines “were defective to begin with but they’ve only become more defective over time” and that “based on the repeated security mistakes by the State 36 of Georgia, my opinion is that the — Georgia’s elections are far more vulnerable than most other states and most other elections in general.” Similarly, he opined that based on the “significant undervote rate in the lieutenant governor’s race” observed in ballots cast on the DRE machines—as well as various other reported errors, such as DRE machines producing error codes and rebooting, and an instance in one precinct “where one voting machine reported . . . significantly different results than every other voting machine in the precinct”—he believed that “there were defects” in the electronic voting system. On cross-examination, Bernhard admitted that he was not aware of “a single instance anywhere in Georgia where there was a piece of malware that was somehow propagated from a server that creates the ballot format, down through [the] memory card, onto the DRE.” He also admitted that he was not aware of “an actual election” in which a vote cast on a DRE machine was not counted, though he explained “that’s because of the way the machine [is] built, it’s impossible to tell. You cannot know.” He also conceded 37 that the configuration of the electronic ballots in the 2018 general election was different than the configuration of the paper ballots, and that among all statewide races in the 2018 general election, Amico’s vote total, as compared to other Democrats, was “somewhere in the middle of the pack.” (h) Involuntary Dismissal. At the close of evidence on January 18, Defendants moved for an involuntary dismissal of Petitioner’s petition under OCGA § 911-41 (b) (“After the plaintiff, in an action tried by the court without a jury, has completed the presentation of his evidence,” defendants “may move for dismissal on the ground that upon the facts and the law the plaintiff has shown no right to relief. The court as trier of the facts may then determine the facts and render judgment against the plaintiff[.]”). The trial court granted the motion orally and issued an order that day. In its order dismissing the petition, the trial court cited Hunt v. Crawford for the proposition that “[i]t is presumed that election results are valid, and the party contesting the election has the burden of showing an irregularity or illegality 38 sufficient to change or place in doubt the result of the election,” and that “[t]he setting aside of an election in which the people have chosen their representative is a drastic remedy that should not be undertaken lightly.” Hunt, 270 Ga. at 8, 10. The trial court also made a number of factual findings. To begin, it found that Petitioners “presented evidence that the DRE system of voting used in Georgia has many problems and irregularities and is regarded as an outdated and inaccurate system of conducting a vote.” It found that Petitioners had shown “five instances of problems with voting at two precincts”—specifically, that “of the 8 voting machines at the Winterville precinct in Clark[e] County[,] Georgia, 7 went decidedly Democratic and 1 went decidedly Republican.” And it found that there was a 4.5 percent undervote between the elections for lieutenant governor and governor.16 Petitioners correctly note that they offered evidence at trial to support an undervote of approximately 4.0 percent. But the trial court’s reference to a 4.5 percent undervote (as opposed to 4.0 percent) does not change its ultimate conclusion. 16 39 But the trial court also found that “[t]here was no evidence of misconduct, fraud, or irregularity by any primary or election official or officials,” and, in acknowledging the vote totals and undervote in the lieutenant governor’s election, concluded that the 123,172 votedifference between Duncan and Amico and the undervote between the lieutenant governor’s election and the governor’s election “do not show any irregularity or illegality in themselves.” The court thus concluded that Petitioners had not “shown any evidence that illegal votes have been received or legal votes rejected at the polls sufficient to change or place in doubt the result in the race for lieutenant governor held on November 6, 2018.” The trial court then went on to cite Fuller v. Thomas, 284 Ga. 397, 397-398 (667 SE2d 587) (2008), for the proposition that “where the focus is on improperly cast ballots or irregularities in the conduct of the election, the number of illegal or irregular ballots necessary to cast doubt on an election is derived by taking the difference between the total votes cast in the election and the race at issue, and adding the margin of victory in the race at issue.” It then made the 40 factual finding, unsupported by citations to the record, that “the most votes that the Plaintiff has shown that could be in any way arguably considered irregular or illegal is approximately 32,000 votes,” assuming that “all such votes would have been cast for Sarah Amico.” Implicit in its order was that Petitioners’ best-case evidentiary showing of irregular or illegal votes could not overcome the 123,172-vote difference between Duncan and Amico. The trial court therefore dismissed Petitioners’ petition. 2. Petitioners’ Enumerations of Error Related to Pre-Trial Discovery. On appeal, Petitioners enumerate a number of alleged errors related to the trial court’s handling of pre-trial discovery. Specifically, Petitioners argue that the trial court erred by: (a) failing to allow reasonable time for discovery; (b) denying Petitioners’ motion for continuance; (c) refusing to allow discovery of the GEMS database; and (d) denying Petitioners’ motion to compel Defendants to allow inspection of the internal memories of designated DRE machines. But after reviewing the record, the 41 appellate briefs, and the relevant law, we cannot say that the trial court abused its broad discretion in managing pre-trial discovery. To address those claims we must examine, as an initial matter, the interplay between Georgia’s Civil Practice Act (“CPA”) and its Election Code. The CPA “shall apply to all special statutory proceedings except to the extent that specific rules of practice and procedure in conflict [with it] are expressly prescribed by law.” OCGA § 9-11-81. Thus, the CPA provides background discovery rules in election contests—which are civil actions—except to the extent the Election Code sets forth “specific rules of practice and procedure” that conflict with the CPA. Id. The Election Code, for its part, vests trial courts presiding over election contests with the “plenary power . . . to make, issue, and enforce all necessary orders, rules, processes, and decrees for a full and proper understanding and final determination and enforcement of the decision of every such case” within the framework of Georgia civil practice or as “necessary and proper to carry out” the chapter of the Georgia Code pertaining to contested elections. OCGA § 2142 2-525 (b). Although the relevant provisions of the Election Code do not mention the term “discovery,” let alone set out election-contestspecific party discovery procedures, see generally OCGA § 21-2-525, the Code does give trial courts the “authority . . . to compel the production of evidence which may be required at such hearing, in like manner and to the same extent as in other civil cases litigated before such court.” OCGA § 21-2-525 (b). And most importantly for this case, the Election Code by its plain terms both authorizes and compels trial courts to conduct election contests expeditiously. As just one conspicuous example, OCGA § 21-2-525 (b) contains the unusual provision that trial courts “shall have authority . . . to proceed without delay to the hearing and determination of such contest, postponing for the purpose, if necessary, all other business.”17 OCGA § 21-2-525 (b) (emphasis supplied). See also Payne v. Chatman, 267 Ga. 873, 875 (485 SE2d 723) (1997) (“Thus, the legislation which provides the right to contest a primary election by Additional analysis of the relevant statutory language is found below in Division 2 (a). 17 43 providing an explicit framework for dealing with the issues of the contest also sets forth the mechanisms for expediting the action in the trial and appellate courts.”). As a result, the CPA’s background rules of discovery—as well as the plenary power the Election Code gives trial courts—must be viewed in light of the specific statutory directives contained in the Election Code indicating that election contests are time-sensitive and should be handled promptly and expeditiously. In sum, “[t]he Election Code gives a trial court ample power and discretion to control the election contest process to insure that the proceedings are resolved in a timely manner.” Payne, 267 Ga. at 875 (citing OCGA § 21-2-525 (c)); cf. Head v. Williams, 269 Ga. 894, 895 (506 SE2d 863) (1998) (“[T]he purpose of the [Election Code] procedural rules [is] to have election contests resolved as quickly as possible.”). And, generally speaking, trial courts have broad discretion over the types of scheduling and discovery-related issues that Petitioners complain about here. See, e.g., Resurgens, P.C. v. Elliott, 301 Ga. 589, 597 (800 SE2d 580) (2017) (“A trial court 44 has broad discretion to control discovery . . . and this Court will not reverse a trial court’s decision on discovery matters absent a clear abuse of discretion.”) (citation and punctuation omitted); Talmadge v. Elson Properties, 279 Ga. 268, 270 (612 SE2d 268) (2005) (“Granting or refusing a continuance is a matter within the sound discretion of the trial court.”) (citation and punctuation omitted); Housing Auth. v. MMT Enterprises, 267 Ga. 129, 129 (475 SE2d 642) (1996) (denial of motion to compel discovery reviewed for abuse of discretion); Woelper v. Piedmont Cotton Mills Inc., 266 Ga. 472, 473 (467 SE2d 517) (1996) (“A trial court has wide discretion to shorten, extend, or reopen the time for discovery.”). (a) First Enumeration: Reasonableness of Time for Discovery. Petitioners contend that the trial court failed to provide them “reasonable time for discovery,” as contemplated by the CPA, when it entered an order permitting discovery only six days (and only three business days) before the scheduled trial in a case “involving technology of this complexity.” We see no abuse of discretion. Petitioners primarily rely on the portion of the CPA that 45 provides that “the court shall in all cases afford to the parties reasonable time for discovery procedures,” OCGA § 9-11-40 (a) (emphasis supplied), to argue that although the Election Code “alters civil practice in some other ways, it does not curtail the right to discovery.” They also cite Mead v. Sheffield, 278 Ga. 268 (601 SE2d 99) (2004), and Taggart v. Phillips, 242 Ga. 454 (249 SE2d 245) (1978), as examples of election contests where the parties were “given reasonable discovery,” which Petitioners contend did not happen here. To begin, Petitioners’ argument disregards the canon that specific statutes prevail over general ones18 and incorrectly treats OCGA § 9-11-40 as supplanting the Election Code, even though the CPA expressly contemplates that it may be forced to yield to “specific rules of practice and procedure in conflict [with it that] are expressly prescribed by law,” OCGA § 9-11-81, and the Election Code contains See Southstar Energy Svcs., LLC v. Ellison, 286 Ga. 709, 712 (691 SE2d 203) (2010) (“For purposes of statutory interpretation, a specific statute will prevail over a general statute, absent any indication of a contrary legislative intent.”) (citation and punctuation omitted); In re C.S., 282 Ga. 7, 8 (644 SE2d 812) (2007) (applying the same canon). 18 46 a litany of specific statutory directives—including procedures designed to expedite time-sensitive election contests. Those directives include that a trial judge presiding over an election contest: “shall promptly begin presiding over such proceedings,” OCGA § 21-2-523 (e) (emphasis supplied); may schedule hearings “as . . . necessary to decide the contest promptly,” § 21-2-525 (a); may, “in its discretion, limit the time to be consumed in taking testimony,” § 21-2-525 (c); and “shall have authority . . . to proceed without delay to the hearing and determination of such contest, postponing for the purpose, if necessary, all other business,” § 21-2525 (b) (emphasis supplied). Other aspects of the Election Code also demonstrate that the timeline for election contests does not mirror the typical timeline for other civil proceedings. For example, a petition contesting the result of an election must be filed “within five days after the official consolidation of the returns of that particular office . . . and certification thereof by the election official having responsibility for taking such action under this chapter,” OCGA § 21-2-524 (a); the 47 court clerk must then issue a notice “requiring the defendant and any other person named in such petition as a candidate . . . to appear and answer such petition . . . not more than ten days nor less than five days after the service of such notice,” id. § 21-2-524 (f); and the “presiding judge shall fix a place and time for the hearing of the contest proceeding” “[w]ithin 20 days after the return day fixed in the notice” required by OCGA § 21-2-524 (a). OCGA § 21-2-525 (a). Indeed, we have interpreted the 20-day requirement in OCGA § 212-525 (a) as “provid[ing] a deadline by which the judge must set a date for a hearing.” Head, 269 Ga. at 895. Compare, e.g., USCR 5.1 (period of compulsory discovery generally expires six months after filing of answer, unless “the court, in its discretion . . . extend[s] . . . shorten[s]” or otherwise alters the time for discovery); OCGA §§ 911-33 (a) (2) (requiring answers or objections within 30 days after service of interrogatories or 45 days after service of summons and complaint); 9-11-34 (b) (2) (requiring response to requests for production within 30 days after service of the request or 45 days after service of the summons and complaint); 9-11-36 (a) (2) 48 (requiring response to requests for admission within 30 days after service of the request or 45 days after service of the summons and complaint). Given all of this, it is apparent that, in this context, there is no definite, one-size-fits-all amount of time that satisfies the “reasonable time for discovery” set forth in the CPA. Instead, what constitutes a “reasonable time” for a contested election must be evaluated in light of the specific facts and circumstances of an expedited proceeding that is markedly different from most other civil proceedings, and in light of the statutory authority given to trial courts presiding over election contests. Now turning to the facts here: from the outset of this case, Petitioners knew they had the burden of presenting evidence to support the allegations in their petition, and they had 55 days between the day they filed their petition (November 23, 2018) and the day of trial (January 17, 2019). Moreover, the trial court scheduled the January 17 trial date during the December 5, 2018 49 status conference, giving Petitioners 43 days’ notice before trial.19 During the December 5 status conference, Petitioners voiced no objection to the trial court scheduling a motion hearing on January 9, 2019—just 8 days prior to the scheduled trial date.20 The motion hearing was held as scheduled on January 9 and the trial court’s oral ruling that it would grant Petitioners’ motion for inspection in part was followed by a written order on January 11. In sum, the record shows that Petitioners had 55 days between filing their petition and the date of trial; 43 days between a trial being set and the trial date itself; 8 days between the trial court’s oral ruling granting part of their motion for inspection and the trial date; and 6 days between the entry of the court’s written order on that motion Soon after filing their petition, Petitioners had asked the trial court, by written motion, to address discovery issues at the December status hearing, and also requested a trial date as early as December 10. Those requests indicate that Petitioners could be ready to try the case as early as 13 days after filing their petition and 4 days after the requested discovery conference. 19 During that status conference, the trial court suggested that the hearing on all motions, including Petitioners’ motion for inspection, be held on January 7, but Petitioners expressed a preference for January 9. The trial court acceded to that two-day delay. 20 50 and the trial. On this record, and in light of the statutory authority given to trial courts to “decide [election] contest[s] promptly,” we cannot say that the trial court abused its discretion in managing the proceedings leading up to trial, including with respect to pre-trial discovery. Petitioners’ citations to Mead and Taggart are unavailing. In Mead, the petitioner contested the result of a statewide Court of Appeals election based on the omission of his name from 481 absentee ballots cast in one county; discovery was not an issue on appeal, and the opinion says nothing about how long the parties engaged in discovery below. See 278 Ga. at 268-279. And in Taggart, the petitioner contested the result of a State House district election based on alleged illegal votes and misalignment of his name on one voting machine; there was no mention of the timing of discovery except for an admonition that “[t]rial courts and litigants should make every effort to dispose of election contests with dispatch.” 242 Ga. at 455. Neither case relied on or cited OCGA § 9-11-40, and neither discusses the reasonableness of time for 51 gathering evidence to support the allegations of an election contest petitioner. We cannot say that the trial court abused its discretion in managing pre-trial discovery here. (b) Second Enumeration: Motion for Continuance. Petitioners parlay their complaints about the length of discovery into a separate enumeration of error pertaining to the trial court’s “[r]efusal” to grant Petitioners’ motion to continue the January 17 trial. We again disagree and conclude that the trial court did not abuse its discretion. Petitioners cite no authority to support this enumeration, and instead argue only that it was error for the trial court to have suggested at the December 5 status conference that the trial date could be moved (“We can move [it]”), and to state in a December 7 email that “[i]f it is deemed necessary to move the trial date” at the January 9 hearing, “the Court will consider it,” before ultimately denying Petitioners’ motion to continue on January 16. But in addition to citing no authority, Petitioners misconstrue the trial court’s initial statements as apparent promises, and they discount 52 the substantial discretion the trial court expressly retained to “consider” whether postponing trial was “necessary.” Indeed, these statements—which were made during some of the first interactions between Petitioners and the trial court—must be viewed in context. So viewed, the statements Petitioners highlight are best characterized as preliminary statements the trial court made at the outset of the election contest that were later superseded by the trial court’s repeated insistence—based on the facts of this case and the statutory mandates of the Election Code—that it would not delay the scheduled trial date. Our precedent also supports the trial court’s authority in this regard. In Head, we held that an election contest hearing (i.e., trial) “may be held on the same day that the defendants are required to file an answer if the parties have had reasonable notice of the hearing date.” 269 Ga. at 894.21 There, the petitioner—who Head was decided under Georgia’s former Municipal Election Code (Chapter 3 of Title 21), which was repealed in 1998. However, the current Election Code was also amended in 1998 to apply to municipal elections, and instructs that “[r]eferences in general and local law to the ‘Georgia Municipal 21 53 contested the results from a city council election and alleged “illegal voting and tampering with absentee ballots”—filed a petition on November 11; the defendants were served with the petition and a summons directing them to answer by November 24; and the petitioner was informed on November 18 that a hearing date was being set for November 24—the very same day defendant’s answer was due. Id. at 895. The petitioner moved for a continuance, claiming that he needed time to “hire an attorney, develop support for his complaint, and subpoena witnesses,” but the trial court denied it. Id. We held that the trial court did not abuse its discretion in denying petitioner’s motion for a continuance, because he had been given reasonable notice of the hearing date: he knew from the Election Code,’ the ‘Municipal Election Code,’ or ‘Chapter 3 of Title 21’ shall be deemed to refer to this chapter.” OCGA § 21-2-1. As explained above in Division 2 (a), we held in Head that the precursor to OCGA § 21-2-525 (a) “provides a deadline by which the judge must set a date for a hearing.” Head, 269 Ga. at 895. Given that OCGA § 21-2-525 (a) also permits trial courts to “fix additional hearings at such other times and places as are necessary to decide the contest promptly,” it is not entirely clear whether the hearing required to be set under the 20-day deadline must also be the actual trial on the election contest, but Head demonstrates that it can be. See Head, 269 Ga. at 894. We note that no party complains that the trial court in this case failed to meet the statutory requirements for setting hearings under OCGA § 21-2-525 (a). 54 outset that “he had the burden of presenting evidence to support his allegations” and had “six days’ notice that a hearing date was being set and three days’ notice of the actual date of the hearing.” Id. See also Broughton v. Douglas County Bd. of Elections, 286 Ga. 528, 528529 (690 SE2d 141) (2010) (“Our legislature put a very short fuse on election contest cases,” and “the swift resolution of election contests is vital for the smooth operation of government.” (citations and punctuation omitted)). Given the facts and circumstances of this particular election contest, and in light of prevailing case law, Petitioners have not carried their burden of establishing that the trial court abused its discretion when it decided to deny Petitioners’ requests for continuance. See Head, 269 Ga. at 895; Payne, 267 Ga. at 876 (election contests should be resolved with “dispatch”). See also Pointer v. Roberts, 288 Ga. 150, 152 (702 SE2d 130) (2010) (“The trial court’s discretion in granting or refusing a continuance will not be interfered with by the appellate courts unless it clearly appears that the judge abused his discretion.”) (citation and punctuation omitted). 55 (c) Third Enumeration: Discovery of GEMS Database. Petitioners argue that the trial court erred by not allowing them discovery of the GEMS database itself, and by instead requiring Defendants only to produce certain reports generated from the GEMS database that Petitioners claim “are already public records” and “would not reveal any programming flaw in the system.” To support their argument, Petitioners point to OCGA § 911-26 (b) (1), which provides that “[p]arties may obtain discovery regarding any matter, not privileged, which is relevant to the subject matter involved in the pending action,” and also to this Court’s case law construing relevance “‘broadly to encompass any matter that bears on, or that reasonably could lead to other matter that could bear on, any issue that is or may be in the case.’” Bowden v. The Medical Center, Inc., 297 Ga. 285, 291 (2015) (quoting Oppenheimer Fund, Inc. v. Sanders, 437 U.S. 340, 351 (98 SCt 2380, 57 LE2d 253) (1978)). According to Petitioners, the trial court’s decision to order production only of certain reports generated by the GEMS database “constituted reversible error” because Petitioners had explained— 56 with support from expert affidavits—that the GEMS database was the “first, most efficient, and easiest place an expert would look for the cause and to quantify the impact of [voting] irregularities.” Petitioners further contend that the GEMS data they requested “could have [been] produced . . . within minutes” and that the GEMS database could have been “examined forensically in a matter of days.” We conclude that the trial court did not abuse its discretion in limiting inspection of the GEMS database. To begin, Petitioners’ relevance argument must be viewed against the backdrop of their counsel’s admission at the January 9 motions hearing that it was “possible,” but “I won’t say likely,” that discovery of the GEMS database would reveal programming mistakes “indicat[ing] that there’s [a] system[-]wide failure.” Moreover, Petitioners’ argument almost completely discounts the factors—apart from relevance—that the trial court was authorized to consider when evaluating Petitioners’ requested discovery. Cf. OCGA § 9-11-26 (c) (authorizing trial courts to issue protective orders and to limit discovery by, among other things, ordering 57 discovery “only on specified terms and conditions”; “only by a method of discovery other than that selected by the party seeking discovery”; “conducted with no one present except persons designated by the court”; or conducted so “that a trade secret or other confidential . . . information not be disclosed or be disclosed only in a designated way.”). Here, for example, the record shows that the court balanced Petitioners’ arguments for discovery of the entire GEMS database against the Defendants’ arguments about “security concerns” raised by the “sensitive” and “highly confidential”22 information contained within the GEMS database and its source code, including a concern that Petitioners’ requested inspection “would expose highly confidential information about Georgia’s voting system that would leave it vulnerable to attack.” See Smith v. DeKalb County, 288 Ga. App. 574, 574-578 (654 SE2d 469) (2007) (where Open Records Act The trial court’s order required the parties to enter into a protective order preserving confidentiality before inspection. The record suggests that the parties entered into a confidentiality agreement, but the agreement itself is not contained in the record. 22 58 request sought a “copy of the [entire] GEMS CD-ROM(S)” to examine the “computer codes” contained in the database and “look for evidence of irregularities resulting from election fraud and malfunctions of the electronic voting equipment and election software,” trial court did not abuse its discretion by permanently enjoining release of the requested information, in part because it contained “passwords, encryption codes, and other security information”). The trial court thus entered an order granting in part Petitioners’ motion for inspection, but limiting the scope and process of that inspection and production in several ways. Specifically, the trial court ruled that Petitioners were entitled to inspect the Base Precincts with Races Report, Ballot Image Report, Vote Center with Cards Report, Statement of Votes Cast Report, and Summary Report that were produced from the GEMS database and maintained by the Gwinnett and Fulton County Boards. But it also limited the inspection by ordering that only “personnel of the Secretary of State or Fulton or Gwinnett Counties may access the GEMS servers directly,” and that those officials could then provide 59 copies of the information to Petitioners. A trial court’s substantial discretion over the discovery process includes balancing competing interests related to a party’s discovery requests. See McGinn v. McGinn, 273 Ga. 292, 293 (540 SE2d 604) (2001) (noting that trial court has an “obligation to assure that the scope of the discovery is restricted to the extent necessary to prevent an unreasonable intrusion into the defendant’s privacy”) (citation and punctuation omitted); Atlanta Journal-Constitution v. Jewell, 251 Ga. App. 808, 812 (555 SE2d 175) (2001) (recognizing “when parties seek discovery of unprivileged but sensitive materials, the trial court must balance the requesting party’s specific need for the material against the harm that would result by its disclosure,” even when the requested material is relevant to the case). Cf. Smith, 288 Ga. App. at 574-578. Here, the record shows that the trial court acted within its authority when it evaluated the relevance of Petitioners’ initial discovery requests and ordered discovery of certain aspects of the GEMS database, thus allowing Petitioners to explore their allegations of programming errors and irregularities, 60 but also limited that discovery in light of other concerns, including election security. To the extent Petitioners were unhappy with the scope of the discovery the trial court granted after weighing the relevant considerations, they could have availed themselves of the discovery that was offered and then gone back to the trial court to make a further showing of why—based on the limited discovery the trial court granted and that the Petitioners conducted—such discovery was inadequate. But instead of taking that tack, the record shows that Petitioners walked away—literally and metaphorically—without engaging in the discovery that the trial court granted, choosing instead to move to compel Defendants to conduct discovery on Petitioners’ desired terms (see more in Division 2 (d) below), even though Petitioners had not availed themselves of much of the discovery offered to them at that point in the expedited proceedings. Under these circumstances, we cannot say that the trial court abused its discretion, and this enumeration of error therefore fails. 61 (d) Fourth Enumeration: Motion to Compel. Petitioners argue that the trial court abused its discretion when it denied their motion to compel. Petitioners’ chief complaint is that the trial court denied their motion after Defendants did not allow Petitioners to “examine the Internal memory storage of each DRE unit,” which the trial court’s January 11 order permitted. According to Petitioners, the trial court’s failure to compel examination of DRE internal memory storage resulted in their being “completely denied discovery.” For the reasons explained below, we disagree. To review, in its January 11 order, the trial court ruled that Petitioners could “examine the Internal memory storage of each such DRE unit.” But it also ordered Petitioners “not to in any way damage the DRE machines or the information contained therein” and prohibited Petitioners from “copy[ing], imag[ing], sav[ing], or retain[ing] the DRE machines or the information contained therein,” and from “upload[ing] or introduc[ing] any information into the DRE machines.” 62 It is clear from the record that Petitioners met with representatives of Fulton County and the Secretary of State’s office on January 14 to inspect DRE machines. As recounted above in Division 1 (e), the parties ardently disagreed about the scope of the examination the trial court had authorized, as well as about the protocol for conducting the examination. Defendants offered to insert post-election memory cards into the DRE machines while they were in post-election mode, thus allowing for the extraction of the “election archive” files from the internal memory, which Petitioners could then examine. But Petitioners wanted to examine the DRE machines’ internal memory by removing the outside cases of the machines and then inserting chips or adapters to copy their entire internal memories. Defendants—who expressed concern that Petitioners’ proposed examination protocol contravened the trial court’s January 11 order—would not allow the DRE machines to be examined using Petitioners’ proposed methods without further consultation with their own legal counsel. At that point, Petitioners simply terminated the scheduled inspection without inspecting the 63 DRE machines. In sum, Petitioners rejected Defendants’ offered examination outright and chose to conduct no examination whatsoever. Nor did Petitioners raise the discovery dispute with the trial court by email, as the trial court had already directed twice during pre-trial proceedings, or by conference call, as the trial court had mentioned at least once. Instead, Petitioners filed a motion for continuance and then a motion to compel two days later, the day before trial. In election contests, as in other civil cases, we review a trial court’s ruling on a party’s motion to compel inspection of evidence for an abuse of discretion. See Housing Authority, 267 Ga. at 129 (denial of motion to compel discovery reviewed for abuse of discretion); Payne, 267 Ga. at 875 (noting the “ample power and discretion” of trial courts to control the election contest process to ensure the timely resolution of election contests); Simon v. Murphy, 350 Ga. App. 291, 296 (829 SE2d 380) (2019) (on review of trial court’s denial of a motion to compel, noting the trial court’s “broad discretion”) (citation and punctuation omitted). 64 On appeal, Petitioners contend that the trial court “directly contra[dicted]” its own January 11 order when it denied the January 16 motion to compel, because the January 11 order authorized examination of DRE machines’ internal memory, yet the court commented (in denying Petitioners’ motion) that it had not ordered “a forensic investigation” of the machines. We see no abuse of discretion, in large part because we discern no such contradiction. Indeed, the trial court’s January 11 order did not expressly authorize a “forensic investigation,” as Petitioners claim; it authorized examination of the DRE machines’ internal memory, but with very specific limitations on how Petitioners could carry out that examination.23 An exchange between Petitioners’ counsel and the court leaves little doubt about this: Judge Grubbs emphatically Petitioners’ argument that the trial court abused its discretion in failing to grant their motion to compel is, in many ways, a veiled attempt to quarrel with the substance of the discovery the trial court did grant, or with the limitations the trial court placed on that discovery. To the extent Petitioners complain about the discovery that was ordered with respect to the GEMS database, we have addressed that above in Division 2 (c). But to the extent Petitioners dispute more generally the scope of the discovery the trial court ordered, they have not separately enumerated that complaint as error— and with good reason; trial courts enjoy broad discretion over the management of discovery. See, e.g., Resurgens, 301 Ga. at 597. 23 65 stated, “I did not order a forensic investigation. I specifically left that word out. I set out certain things to be done.” Given this record, we conclude that the trial court did not abuse its broad discretion to manage discovery, or to evaluate Petitioners’ motion to compel. See Resurgens, 301 Ga. at 597-598; Simon, 350 Ga. App. at 297. We also note that, as a practical matter, it is especially difficult for Petitioners to show that the trial court abused its discretion when Petitioners complained that the discovery the trial court ordered was not sufficient (or not offered in a sufficient form), but did not heed the trial court’s repeated admonition to bring discovery disputes to it by phone or email—and did not even accept the DRE machine inspection that was offered to them. 24 By refusing to move forward with the inspection, or to seek real-time resolution of any dispute about how Defendants offered that inspection, Petitioners also eliminated another potential source of recourse: accepting the Accordingly, Petitioners’ argument on appeal that their refusal to “accept Fulton’s invitation to power up the machines” was motivated by a concern that “operating the DRE machines without first making a back-up copy of the internal memory would alter the electronic memory in violation of the trial court’s order,” is also unavailing. 24 66 discovery the trial court ordered and going back to the trial court to argue that additional discovery was necessary. Under these circumstances, we cannot say that the trial court abused its discretion. (e) Conclusion. After a comprehensive review of Petitioners’ claims of trialcourt error related to pre-trial discovery, we have concluded that— under the facts and circumstances of this election contest—the trial court did not abuse its broad discretion in managing discovery. As a result, Petitioners are not entitled to a remand for more discovery to expand the record, and we consider only the evidence the parties presented at trial in evaluating the trial court’s grant of Defendants’ motion to involuntarily dismiss Petitioners’ petition.25 On appeal, Petitioners rely on certain material that is not evidence. For example, Petitioners assert in their opening appellate brief that they “presented undisputed evidence that, in the run-up to the 2018 general elections, the entire national intelligence and computer science communities were warning the State of Georgia that its DRE voting system was profoundly vulnerable to malicious attack, programming errors, and irregularities,” and support that contention with citations to two congressional reports, two news articles, and two scholarly articles about the issue. But that is misleading. Petitioners did cite a handful of reports and articles—including most of the 25 67 3. Petitioners’ Enumerations of Error Related to Involuntary Dismissal of their Election Contest Claim. Petitioners also argue that the trial court committed reversible error (a) by finding that “the most votes that the [Petitioners have] shown that could be in any way arguably considered irregular or illegal is approximately 32,000 votes,” a factual finding that Petitioners claim is clearly erroneous, and (b) by reciting the mathematical formula set out in Fuller, 284 Ga. at 397-398—which Petitioners claim is the incorrect legal standard—in rejecting their claim of systematic irregularities, instead of determining whether Petitioners showed that Georgia’s electronic voting system was so defective as to cast doubt on the result of the election for lieutenant ones cited in their appellate brief— in their petition, but none of those reports or articles were admitted into evidence at trial. At trial, Petitioners sought to admit a similar report through their expert witness and then again after the close of evidence, but the trial court refused to admit it both times. And although the trial court ruled that Petitioner’s expert could rely on the report in forming his opinions, it expressly ruled that the report itself was inadmissible and that the expert could not quote from it. Those rulings appear to be within the court’s discretion, and Petitioners have not challenged them on appeal. 68 governor.26 (a) Legal Background. We examine these claims of error against a backdrop of statutory law and precedent interpreting those statutes. As relevant here, an election may be contested on the grounds of “[m]isconduct, fraud, or irregularity by any primary or election official or officials sufficient to change or place in doubt the result,” OCGA § 21-2-522 (1), or “[w]hen illegal votes have been received or legal votes rejected at the polls sufficient to change or place in doubt the result,” id. § 21-2-522 (3). But because—as the trial court recognized—“[t]he setting aside of an election in which the people have chosen their representative is a drastic remedy,” it “should not be undertaken lightly, but instead should be reserved for cases in which a person challenging an election has clearly established a Citing OCGA § 21-2-525 (b), Petitioners also contend that the trial court committed reversible error by “[f]ail[ing] to [g]ain [u]nderstanding of the [c]ase.” Although it is, of course, important for trial courts to understand the cases before them, Petitioners’ argument misapprehends OCGA § 21-2-525 (b), which establishes the plenary power trial courts are given to gain a “full and proper understanding” of election contest cases; the statute does not create a legal responsibility, the failure of which results in reversible error. 26 69 violation of election procedures and has demonstrated that the violation has placed the result of the election in doubt.” Hunt, 270 Ga. at 10. For these reasons, we “‘presume[ ] that election returns are valid, and the party contesting the election has the burden of showing an irregularity or illegality sufficient to change or place in doubt the result of the election.’” Meade, 293 Ga. at 143 (quoting Banker v. Cole, 278 Ga. 532, 535 (604 SE2d 165) (2004). But that is not all. We have explained that “[i]t is not sufficient to show irregularities which simply erode confidence in the outcome of the election,” and that “[e]lections cannot be overturned on the basis of mere speculation.” Meade, 293 Ga. at 149 (punctuation and citation omitted); see also Hunt, 270 Ga. at 9 (election outcome cannot “be nullified based merely upon speculation”). Moreover, a “trial court’s findings in an election contest will not be disturbed unless clearly erroneous.” Banker, 278 Ga. at 533 (citation and punctuation omitted). Historically, we have recognized two election-contest paradigms in our case law. See Meade, 293 Ga. at 143 (noting that 70 “[t]his Court has set aside elections under two different circumstances”). The first paradigm—which includes the vast majority of election contest cases—pertains to allegations of illegal votes, irregularly recorded votes, and illegal or irregular ballots. In those cases, which involve the types of allegations that may be proven or disproven by examining or counting a specific number of ballots, “we have required the evidence to ‘show that a sufficient number of electors voted illegally or were irregularly recorded in the contest being challenged to change or cast doubt on the election.’” Id. (quoting McCranie v. Mullis, 267 Ga. 416, 416 (478 SE2d 377) (1996), and citing McIntosh County Bd. of Elections v. Deverger, 282 Ga. 566 (651 SE2d 671) (2007); Whittington v. Mathis, 253 Ga. 653 (324 SE2d 727) (1985); and Bell v. Cronic, 248 Ga. 457 (283 SE2d 476) (1981)). Within that context, the party contesting an election generally must “show a specific number of illegal or irregular ballots,” Middleton v. Smith, 273 Ga. 202, 203 (539 SE2d 163) (2000), or a specific number of “voters [who] voted illegally or were irregularly 71 recorded [or rejected],” Howell v. Fears, 275 Ga. 627, 627-628 (571 SE2d 392) (2002). See also Deverger, 282 Ga. at 566 (in an election contest based on allegations of illegally cast or wrongfully rejected votes in a race for county commissioner, petitioner was required to “establish that sufficient legal votes were rejected to change or place in doubt the result”). That number, in turn, must be “sufficient to place the result” of the contested election in doubt. Taggart, 242 Ga. at 486. The party contesting the election is not, however, required to show how votes would have been cast had the ballots been regular or had the votes not been rejected wrongfully. See Mead, 278 Ga. at 271; Deverger, 282 Ga. at 566. We have applied this legal standard when, for example, a petitioner claims that a specific number of votes were wrongfully rejected, Deverger, 282 Ga. at 566-567; that a certain number of ballots omitted a candidate’s name, Mead, 278 Ga. at 268, 271; or that certain ballots omitted a particular race, Howell, 275 Ga. at 627-628. The second paradigm involves cases where a party alleges systemic irregularities in the election process that may not be 72 measurable in the same discrete manner that is used in cases falling within the first paradigm. Under this second set of circumstances— which we have identified in far fewer cases—we have “recognized that the result of an election may be voided where systemic irregularities in the process of the election are sufficiently egregious to cast doubt on the result.” Meade, 293 Ga. at 143 (citing Stiles v. Earnest, 252 Ga. 260 (312 SE2d 337) (1984)). Implicit in this line of cases is that the alleged systemic irregularities caused the election result to be placed in doubt. See Meade, 293 Ga. at 143.27 We have applied this legal standard when, for example, petitioners have alleged systemic irregularities such as widespread vote-buying or wrongful distribution of absentee ballots, id. at 150 (reversing invalidation of election); a county sheriff improperly campaigned on behalf of particular candidates, Middleton, 273 Ga. at 204 (reversing invalidation of election); election officials engaged in illegal conduct, Stiles, 252 Ga. at 260-262 (holding that evidence of illegality was To conclude otherwise would permit the type of “mere speculation” that we have cautioned against. See, e.g., Meade, 293 Ga. at 149. 27 73 sufficient to warrant new election); or other “irregularities,” such as allegations that a candidate used a name that misled voters, that a candidate campaigned improperly, or that various elections officials were improperly appointed or not appointed at all, Fuller, 284 Ga. at 398 (“None of these alleged irregularities is specific enough to cast doubt on the results of the election.”).28 However, we are aware of only one case in which we have applied this legal standard to order a new election. See Meade, 293 Ga. at 148-149. In Stiles v. Earnest, the petitioner alleged that certain school officials who “checked off voters from the voting lists” within 250 feet of various polling places changed or placed in doubt the result of a referendum to elect members of a county school board, which failed by 17 votes. 252 Ga. at 261. We concluded that the alleged “illegality” was “sufficient to change or place in doubt the result” of the election and ordered another referendum to be held. Id. For the reasons explained below in Division 3 (d), we express doubt about the mathematical formula set forth in Fuller. 28 74 By contrast, in Middleton v. Smith, a case involving an election for county commissioner and court clerk, we reversed a trial court order invalidating the election where the margins of victory were 20 votes and 117 votes, respectively, and a petitioner had alleged that a county sheriff’s misconduct “put in doubt the validity of all 506 votes cast” from a particular precinct. 273 Ga. at 203. There, we concluded that the allegations that “enough electors voted illegally so as to change or cast doubt on the result of the election” were “based on mere speculation,” even though the trial court made multiple findings of “irregularities”—including that the county sheriff “mailed approximately 1,200 letters to voters” on official stationery “urging them to vote” for the candidates who ultimately prevailed and campaigned for those candidates “within 150 feet of the precinct,” and that the sheriff was statutorily “responsible for maintaining order at the polling places.”29 Id. at 202-203. In Middleton, then-Chief Justice Benham dissented, criticizing the majority for viewing “too narrowly the scope of election irregularities which can justify setting aside an election, and giv[ing] too little weight to the trial court’s findings.” 273 Ga. at 204 (Benham, C.J., dissenting). Some of us have 29 75 Similarly, in Meade v. Williamson, a case in which there was a 39-vote margin of victory in a primary runoff election for county sheriff, we concluded that the “evidence of systemic misconduct for vote buying and alleged wrongful distribution of absentee ballots is largely speculative and is insufficient to support the trial court’s conclusion that irregularities in the election process were shown to cast doubt upon the results.” 293 Ga. at 150. There, the petitioner had presented evidence of only one illegally bought vote and the remaining evidence of vote buying was based upon hearsay and gossip. Insufficient evidence was presented to conclude any misconduct relative to the distribution of absentee ballots. No evidence at all was presented to support the conclusion that unqualified persons were wrongly permitted to assist voters. Id. As a result, we concluded that “[t]he evidence [fell] short of demonstrating systemic irregularities in the election process.” Id. concerns—similar to those Justice Benham expressed in his dissent—that the Middleton majority did not convincingly distinguish the conduct in Middleton from the conduct in Stiles that resulted in that election being invalidated. But an analysis of whether Middleton (or for that matter, Stiles) was correctly decided is not necessary to decide the case before the Court today. 76 (b) Petitioners’ Theory of the Case Below and On Appeal. Petitioners alleged a claim under both election contest paradigms before the trial court, offering a few examples of specific instances of irregularities in voting and also claiming that defects in Georgia’s electronic voting system cast doubt on the election as a whole. But the main thrust of Petitioners’ arguments to the trial court was that their case was one of systemic failure and not one where “hard” or “tangible” evidence would prove voting defects. Petitioners in fact argued at trial that this case is “not simple math,” but rather a case that relies on voting patterns—i.e., that “the under-vote shows that the machines simply were not working”—“to cast the entire election in doubt.” (Emphasis supplied.) Before this Court, however, Petitioners have presented a hybrid set of arguments, contending that at least some aspects of the two historical paradigms in this Court’s election contest cases (i.e., specific quantities of irregular votes and systemic irregularities) apply in this case and together require invalidation of the November 6, 2018 election for lieutenant governor. Although 77 Petitioners have fused these arguments in their appellate briefing, we address each enumeration—and the relevant law—in turn. (c) Fifth Enumeration: Trial Court’s Factual Findings. In its order granting Defendants’ motion for involuntary dismissal, the trial court found that “the most votes that the Plaintiff has shown that could be in any way arguably considered irregular or illegal is approximately 32,000 votes.” Petitioners argue that this finding is clearly erroneous, and we agree. After an exhaustive review of the record, we are unable to identify any evidence presented to the trial court that supports its finding, nor have Defendants pointed to any. As best we can tell, the “approximately 32,000” number originated from trial testimony offered by Petitioners’ lay witness, Christopher Brill.30 Brill testified that by his calculations, the historical average undervote for lieutenant governor in the past four general elections was “around In a supplemental brief filed after oral argument, two defendants— the Gwinnett and Fulton County Boards—acknowledged that the 32,000 figure “appears to originate” from Brill’s testimony. 30 78 0.8 percent” lower than the total votes cast for governor in those elections. Brill then testified that if the “historical trend of .8 [percent] were applied” to the 2018 election, the expected number of undervotes between the governor’s election and lieutenant governor’s election would be “around 31,532.”31 But the record shows that Petitioners offered that testimony in an attempt to establish what they believed the undervote should have been in the 2018 election for lieutenant governor, not to establish the number of irregular or illegal votes they claim were cast in the race. The trial court’s factual finding in this regard is not supported by the record and is clearly erroneous. Nevertheless, even a clearly erroneous factual finding does not always require reversing a trial court’s judgment. See, e.g., Deverger, 282 Ga. at 566-568 (although trial court erred in finding certain votes were illegally cast, evidence showed that a sufficient In other words, by multiplying the reported total votes cast in the 2018 governor’s race (3,939,328 votes) by the “historical trend” of undervoting Brill offered at the hearing (“around 0.8 percent”), the undervote would equal “around 31,532” votes. 31 79 number of other votes were wrongfully rejected to place the election results in doubt, resulting in affirmance of trial court’s ruling). That is especially true when the trial court’s ultimate legal conclusions are correct based on the evidence presented. See Banker, 278 Ga. at 535 (trial court’s erroneous conclusion that candidate’s notice of withdrawal from election was sufficient “does not necessarily require reversal of its judgment,” and did not require reversal there because petitioner still failed to carry burden of showing irregularity sufficient to place in doubt result of election); see also Tolbert v. Toole, 296 Ga. 357, 361-362 (767 SE2d 24) (2014) (concluding that habeas court “reached the right result,” and therefore affirming that ruling, even though it was based upon a “mistaken” factual assumption). Our analysis, therefore, continues to determine whether the trial court properly concluded that Petitioners failed to meet their burden of presenting sufficient evidence to place in doubt the election result. (d) Sixth Enumeration: Trial Court’s Legal Analysis. Petitioners also argue that the trial court erred by applying 80 “the incorrect legal standards in rejecting [their] claim that systemic irregularities in the State of Georgia’s electronic voting system rendered the system so defective as to cast doubt on the result of the election.” Specifically, Petitioners complain that the trial court recited the mathematical standard set out in Fuller v. Thomas: “the number of illegal or irregular ballots necessary to cast doubt on an election is derived by taking the difference between the total votes cast in the election and the race at issue, and adding the margin of victory in the race at issue.” 284 Ga. at 397-398. Arguing that they are not required to prove by “rote application of a formula” that the election result would have been different but for the alleged irregularity, see e.g., Stiles, 252 Ga. at 261, Petitioners contend that they need only offer evidence of a material irregularity sufficient to place in doubt the result of the election—a standard Petitioners assert they have met. We disagree that Petitioners have offered sufficient evidence to place in doubt the result of the 2018 general election, and for that reason we affirm the trial court’s dismissal of the petition. 81 We first turn to Petitioners’ argument that the trial court erred by quoting Fuller in its dismissal order. To begin, we conclude that the mathematical formula used in Fuller should not be used to evaluate whether a party has cast sufficient doubt on the result of an election based on allegations of systemic irregularities. Applying the Fuller test (or for that matter, any inflexible mathematical formula) is antithetical to the second paradigm of election contest cases, where allegations of systemic irregularities may not be measurable in the same discrete manner that the majority of election contest cases are.32 Even so, the trial court order shows that We note, however, that the margin of victory is still relevant in evaluating whether a contestant has cast doubt on an election, even when a party alleges systemic irregularities. That is because a party contesting an election under OCGA § 21-2-522 (1) or (3) must always offer evidence “sufficient to change or place in doubt the result” of the election to prevail, whether the allegations at issue fit into the first paradigm (discrete and measurable votes or ballots) or second paradigm (systemic irregularities) of historical election contest cases. Whether evidence of alleged systemic irregularities “place[s] in doubt the result” of the contested election depends, at least as an initial matter, on the margin of victory in the relevant election. In short, under both election case paradigms, the margin of victory serves as a kind of materiality threshold for evaluating whether a party has “place[d] in doubt the result” of an election. If—in light of the particular facts of a given election contest—a party offers evidence of voting irregularities that could be sufficient to overcome the margin of victory, then it may have met its burden by casting doubt on the result of the election. By contrast, if a party has not 32 82 the court merely recited Fuller’s mathematical formula, but did not apply it to the facts of this case. Petitioners acknowledge as much on appeal, conceding that despite the trial court’s recitation of the Fuller formula, it “did not actually make this calculation.” Because the trial court’s legal conclusions did not hinge on application of the Fuller formula, any error the trial court made by reciting the formula is harmless.33 offered evidence of voting irregularities, or if it has offered evidence of voting irregularities that (however real and disturbing) are not prolific enough to overcome the margin of victory in an election, then the party has not cast doubt on the election and has not met its burden under OCGA § 21-2-522. We take this opportunity to express our doubts about the mathematical formula set out in Fuller: that “the number of illegal or irregular ballots necessary to cast doubt on an election is derived by taking the difference between the total votes cast in the election and the race at issue, and adding the margin of victory in the race at issue.” 284 Ga. at 397-398. That formula originated from a one-line footnote in McCranie v. Mullis, where it was recited with no explanation other than that “[t]he trial court found, and the parties agree[d],” on the number of illegal or irregular ballots necessary to cast doubt on that election. 267 Ga. at 416 n.5. It has been cited in one other case, but was not material to the outcome of that election contest because we concluded that the petitioners’ unsupported allegations about improperly cast ballots and irregular conduct were merely “speculation and innuendo.” Scoggins v. Collins, 288 Ga. 26, 28 (701 SE2d 134) (2010). In light of our discussion above in footnote 32, some of us are unsure why trial courts should add to the margin of victory in a given race the number of undervotes for that particular race—i.e., the Fuller formula—instead of looking to the margin of victory as the touchstone inquiry for evaluating 33 83 As a result, we must review the trial court’s legal conclusions— including those pertaining to Petitioner’s claims of quantifiable irregular votes, as well as their more generalized claims of systemic irregularities—to determine whether the court erred in dismissing Petitioners’ petition. (i) Conclusions Regarding Specific Numbers of Illegal or Irregular Votes. To the extent Petitioners offered evidence of specific instances of irregular votes, the trial court did not err in concluding that Petitioners failed to meet their burden of placing in doubt the result of the 2018 general election on the basis of that evidence. Notably, the trial court made specific factual findings about alleged voting irregularities and credited certain evidence Petitioners presented at trial: it found that Petitioners had showed “five instances of problems with voting at two precincts” and that “of the 8 voting materiality in an election contest. Indeed, Fuller is an outlier among more recent election contest cases where we have looked to the margin of victory to establish the threshold of materiality that a petitioner must cross to cast sufficient doubt on an election based on allegations of illegal or irregular votes or ballots. See, e.g., Meade, 293 Ga. at 148; Deverger, 282 Ga. at 568; Mead, 278 Ga. at 268-271. 84 machines” at a particular precinct, “7 went decidedly Democratic and 1 went decidedly Republican.” But it also found that “[t]here was no evidence of misconduct, fraud, or irregularity by any primary or election official or officials.” Given these factual findings and the small number of irregularities involved, we cannot say that the trial court erred as a matter of law when it concluded that the Petitioners had “not shown any evidence that illegal votes have been received or legal votes rejected at the polls sufficient to change or place in doubt the result in the race for Lieutenant Governor.” Whether the trial court should have cited a case other than Fuller does not affect the court’s ultimate legal conclusion, especially where there is no serious contention that the evidence offered at trial established enough instances of illegal or irregular votes to overcome a margin of victory of 123,172 votes. (ii) Conclusions Irregularities. Regarding Claims of Systemic The heart of Petitioners’ case, and their theory of the case with respect to systemic voting irregularities, is best summarized by their 85 contention on appeal that they “established to a near metaphysical certainty that” Georgia’s “profoundly vulnerable machines caused thousands of voters using electronic machines to either not vote for Lieutenant Governor or for those votes to not be counted.” (Emphasis supplied.) But because that assertion is wholly unsupported by the record Petitioners created below, the trial court did not err when it concluded that Petitioners failed to meet their burden of showing an irregularity in Georgia’s electronic voting system sufficient to cast doubt on the 2018 general election. As an initial matter, we recognize that the trial court specifically found that Petitioners “did present[ ] evidence that the DRE system of voting used in Georgia has many problems and irregularities and is regarded as an outdated and inaccurate system of conducting a vote.” The court also found that there was “a difference of 123,172 votes” received between Duncan and Amico; that there was a 4.5 percent undervote34 between the lieutenant governor’s race and the governor’s race; and that Amico “received 34 See footnote 16 above. 86 more votes than those cast for the Democrat in the State-wide races for Commissioner of Agriculture, Commissioner of Insurance, State School Superintendent, and Commissioner of Labor.” The trial court then concluded that “[t]hese number[s] do not show any irregularity or illegality in themselves.” It is this legal conclusion we review when assessing whether the trial court erred in evaluating Petitioners’ claims of systemic voting irregularities. Importantly, that conclusion acknowledges and rejects the Petitioners’ core argument to the trial court, and again on appeal: that the number of votes cast in the 2018 election for lieutenant governor itself proves a voting irregularity sufficient to cast doubt on the election. Put simply, Petitioners contend that “if the historical average percentage of voters”—which they calculate to be 99.2 percent over the past four general elections—“actually voted [in] the Lieutenant Governor’s race in 2018, then over 127,000 votes for Lieutenant Governor were not counted.”35 Petitioners In their briefing, Petitioners calculate this number by subtracting 31,532 (the expected undervote according to Brill’s calculations at trial) from 35 87 apparently use this math both to illustrate the magnitude of the alleged systemic problem, and also to offer a theory that could meet a materiality threshold—i.e., a theory that identifies a number of votes higher than the margin of victory in the contested race. Yet the trial court’s dismissal order does not mention the 0.8 percent number that is the linchpin of Petitioners’ efforts to mathematically “prove” systemic irregularities; instead, the court made findings about the margin of victory (123,172) and the magnitude of the undervote (4.5 percent)36 and concluded that those “numbers do not show any irregularity or illegality in themselves.” Even considering the trial court’s general finding “that the DRE system of voting used in Georgia has many problems and irregularities and is regarded as an outdated and inaccurate system of conducting a vote,” we cannot say that the cursory evidence Petitioners presented about the 159,124 (the reported undervote between the 2018 elections for governor and lieutenant governor). 36 See footnote 16 above. 88 numbers of votes cast in the 2018 lieutenant governor’s election proved an irregularity or illegality as a matter of law.37 Moreover, inherent in Petitioners’ argument is a set of assumptions that must be correct for Petitioners to prevail. Those assumptions include, for example, that the 0.8 percent average is, in fact, the correct number to use for the “historical undervote,” as opposed to using the highest percentage undervote in the prior four general elections (which, according to the evidence Petitioners offered at trial, would be 1.2 percent38), and that even accepting 0.8 percent as the correct metric, the observed undervote in the 2018 election was caused by an impermissible influence such as DRE machine malfunction, rather than by a factor—or combination of factors—like ballot design, increase in the number of new voters, or To the extent Petitioners argue that any lack of evidence was caused by Defendants’ failure to cooperate or by trial-court error in pretrial discovery, we have already rejected those arguments in Division 2. 37 If applied here, a 1.2 percent undervote would result in 111,752 “missing” votes—a number of votes that would not be sufficient to overcome the 123,172-vote margin of victory in the 2018 election for lieutenant governor. 38 89 another of the rationales Defendants raised at trial. But the trial court did not make findings about—let alone credit—those assumptions, and Petitioners did not establish causation at trial.39 Just as in Meade, the “evidence of systemic misconduct . . . is largely speculative and is insufficient to support [a] conclusion that irregularities in the election process were shown to cast doubt upon the results.” 293 Ga. at 150; see also Fuller, 284 Ga. at 398 (“None of these alleged irregularities is specific enough to cast doubt on the results of the election.”). And, as we have said before, “[e]lections cannot be overturned on the basis of mere speculation, or an appearance of impropriety in the election procedures.” Middleton, 273 Ga. at 203 (citations omitted). Because the evidence Petitioners presented at trial “failed to To that end, the record contradicts Petitioners’ argument on appeal that “based solely on the undisputed election results, to a 99.9% certainty, there was something about the electronic machines that caused people’s vote in the Lieutenant Governor’s race to not be counted, or to not be counted correctly.” (Emphasis supplied.) Even accepting as true testimony from Petitioners’ witness that the reported undervote in the lieutenant governor’s race had “less than 1 in 10,000 chance of appearing at random . . . in the course of an election,” that testimony did not provide an answer for what caused the undervote. 39 90 carry the burden of demonstrating the election results should be invalidated either by establishing a sufficient number of specific irregular or invalid votes to change or place in doubt the results, or by establishing sufficient irregularities in the election process to cast doubt upon the results,” Meade, 293 Ga. at 150, we affirm the trial court’s ruling involuntarily dismissing Petitioners’ petition at the conclusion of trial.40 4. Seventh Enumeration: Petitioners’ Jury Trial Request. Finally, Petitioners contend on appeal that the trial court erred in rejecting their demand for a jury trial under OCGA § 21-2-526 (a). We disagree. The Georgia Constitution states that “[t]he right to trial by We are aware that a federal district court recently prohibited “any use of the GEMS/DRE system after 2019” as the result of a federal 42 USC § 1983 lawsuit in which the Coalition for Good Governance (originally the lead petitioner in this case) was a plaintiff. Curling v. Raffensperger, No. 1:17-CV2989-AT, 2019 WL 3822123, at *63 (N.D. Ga. Aug. 15, 2019). We express no view about the conclusions that court reached in its order partially granting preliminary injunctive relief. But we note that Curling was not an election contest litigated under Georgia’s Election Code, and that the plaintiffs in that case sought only prospective relief. As a result, the plaintiffs in Curling, unlike the Petitioners here, did not seek to invalidate an election. 40 91 jury shall remain inviolate.” Ga. Const. of 1983, Art. I, Sec. I, Par. XI (a). In Bell v. Cronic, 248 Ga. 457, 458, 459 (283 SE2d 476) (1981), we held that, under the substantively identical provision of our prior Constitution, “there is no constitutional right to a jury trial in an election contest proceeding,” and that “the right can only exist by statute.” In this regard, our Election Code provides: All issues of a contest shall be fully tried and determined by the court without the aid and intervention of a jury, unless a litigant to the contest shall demand a trial by jury at any time prior to the call of the case; and the court shall determine that it is an issue which under other laws of this state the litigant is entitled to have tried by a jury. OCGA § 21-2-526 (a) (emphasis supplied). Therefore, “unlike other situations where demand alone is sufficient, here there are two requirements: 1) demand; and 2) determination that there are issues which under other laws of this State the litigant is entitled to have trial by a jury.” Henderson v. County Bd. of Registration & Elections, 126 Ga. App. 280, 285 (190 SE2d 633) (1972). Here, although Petitioners did make a demand for jury trial on the eve of trial, they did not point to any “laws of this state,” OCGA 92 § 21-2-526 (a), that required a jury trial with respect to the types of issues presented in this election contest. Likewise, other than relying on the general proposition that “factual issues [are] for the jury,” Petitioners point to no such “other laws” on appeal, and we have found none. Indeed, the types of issues Petitioners raise regularly have been adjudicated by trial courts without juries, as all of the election contest cases Petitioners cite in their appellate briefs—and those cited in this opinion—demonstrate. Accordingly, we conclude that the trial court did not err in denying Petitioners’ demand for a jury trial. Judgment affirmed. All the Justices concur, except Melton, C. J., not participating. 93