RECOMMENDED FOR PUBLICATION Pursuant to Sixth Circuit I.O.P. 32.1(b) File Name: 21a0204p.06
UNITED STATES COURT OF APPEALS FOR THE SIXTH CIRCUIT
┐ SEAN DECRANE, │ Plaintiff-Appellee, │ > No. 20-3620 │ v. │ │ EDWARD ECKART, in his official and individual │ capacities; CITY OF CLEVELAND, OHIO, │ Defendants-Appellants. │ ┘
Appeal from the United States District Court for the Northern District of Ohio at Cleveland. No. 1:16-cv-02647—Christopher A. Boyko, District Judge.
Argued: April 29, 2021
Decided and Filed: September 1, 2021
Before: GUY, DONALD, and MURPHY, Circuit Judges. _________________
COUNSEL
ARGUED: David R. Vance, ZASHIN & RICH CO., L.P.A., Cleveland, Ohio, for Appellants. Subodh Chandra, THE CHANDRA LAW FIRM LLC, Cleveland, Ohio, for Appellee. ON BRIEF: David R. Vance, Jon M. Dileno, ZASHIN & RICH CO., L.P.A., Cleveland, Ohio, William Menzalora, CITY OF CLEVELAND, Cleveland, Ohio, for Appellants. Subodh Chandra, Donald Screen, Patrick Haney, THE CHANDRA LAW FIRM LLC, Cleveland, Ohio, for Appellee. _________________
OPINION _________________
MURPHY, Circuit Judge. Someone in the City of Cleveland’s fire department leaked embarrassing information to the media about its chief’s insufficient training hours. No. 20-3620 DeCrane v. Eckart, et al. Page 2
Sean DeCrane, the director of the Fire Training Academy, was not that person. But, according to DeCrane, Edward Eckart thought he was. DeCrane alleges that Eckart facilitated several harmful personnel actions against him because of this mistaken belief—all in violation of the First Amendment. Eckart’s response? He argues that the First Amendment does not protect speech made as part of an employee’s government job, see Garcetti v. Ceballos, 547 U.S. 410, 421–22 (2006), and that DeCrane would have leaked the information pursuant to his job duties as the training director. The district court denied Eckart qualified immunity on this argument, holding that our precedent clearly established that DeCrane would have tipped off the media as a private citizen rather than a public employee. We agree and affirm this part of the court’s decision. We dismiss Eckart’s other two arguments for lack of jurisdiction in this appeal’s interlocutory posture.
I
Given this case’s procedural posture, we recite the facts (many of which are disputed) in the light most favorable to DeCrane. See Beck v. Hamblen County, 969 F.3d 592, 598 (6th Cir. 2020). DeCrane started as a firefighter with the City of Cleveland’s Division of Fire in the 1990s, eventually working his way up to battalion chief. In August 2012, DeCrane became the director of training at the city’s Fire Training Academy.
The same year, DeCrane applied to be the chief of the Division of Fire. He went through a two-step interview process—an interview with a panel followed by one with the mayor. The panel included Eckart, an assistant director in the Department of Public Safety that oversees the Division of Fire. After the panel rated DeCrane second behind Daryl McGinnis, the mayor chose McGinnis to be the next chief. In early 2013, Eckart informed DeCrane of the decision. During their call, DeCrane said he was surprised by this choice. He told Eckart that McGinnis had fallen behind in his required continuing-education hours. When confronted with this claim, McGinnis assured Eckart that his training was up to snuff.
McGinnis was lying. Someone tipped off the media about McGinnis’s deficient training. Six months after his promotion, a reporter asked for his training records. McGinnis then came No. 20-3620 DeCrane v. Eckart, et al. Page 3
clean that his training was, in fact, inadequate. The city put him on leave the next month; he resigned a day or two later. The ensuing media coverage reflected poorly on the city.
DeCrane did not leak the tip about McGinnis’s deficient training, which was an open secret in the Division of Fire. According to DeCrane, however, Eckart mistakenly believed that he was the leak’s source. DeCrane contends that Eckart (among others) subjected him to a three- year campaign of retaliation for this misperceived leak. DeCrane was not disciplined or demoted. But he says that he endured four types of retaliatory actions.
1. Lack of Promotions. DeCrane alleges that he received no further promotions because of the mistaken belief that he was the leaker. Chief McGinnis’s abrupt departure left a vacancy at the top. Although DeCrane applied for the position, Patrick Kelly began serving as the interim chief in August 2013. Kelly received the promotion even though DeCrane’s interview score from a few months earlier had been higher. In late 2013, DeCrane interviewed to be the permanent chief, but Kelly received that appointment too.
DeCrane was next passed over for assistant chief in early 2015. A panel ranked him last of eleven candidates. In May 2015, DeCrane confronted Eckart about his lack of promotions. When DeCrane suggested that Eckart thought that he had disclosed McGinnis’s training deficiencies to the media nearly two years earlier, Eckart allegedly slammed his fist on a table and shouted “[y]ou have to admit it’s pretty coincidental” that the leak occurred a few months after DeCrane alerted him of the issue. DeCrane Dep., R.145, Page 260. (Eckart disputes this account.)
Chief Kelly resigned later that year. DeCrane interviewed to be the interim chief. Angelo Calvillo was appointed instead. DeCrane was passed over a final time when the city named Calvillo the permanent chief in April 2016.
2. Misconduct Charges. DeCrane next alleges that he faced unfounded misconduct charges because of the mistaken belief that he leaked McGinnis’s training deficiencies. In January 2015, a firefighter named Larry Moore filed a complaint alleging that, on DeCrane’s orders, a captain asked him to record Training Academy information inaccurately. Eckart directed the Office of Integrity Control, Compliance, and Employee Accountability (which goes No. 20-3620 DeCrane v. Eckart, et al. Page 4
by “OIC”) to investigate the complaint. He tasked the OIC with reviewing not just Moore’s specific allegations but also the Training Academy’s general recordkeeping practices.
The OIC recommended that the city bring charges against DeCrane (among others) for the wrongdoing alleged by Moore and the failure to keep accurate records. Yet, after the OIC interviewed DeCrane in October 2015, Jim Votypka, the OIC head, informed Eckart that Moore’s allegations of intentional wrongdoing “could not be substantiated.” Mem., R.139-17, PageID 9309. DeCrane was nevertheless brought in for a second interview (an unusual occurrence) about two months later in December 2015. Eckart questioned DeCrane about McGinnis: “How/why did you know Daryl McGinnis was deficient at continuing education for EMT?” Eckart Dep., R.140, PageID 9641.
The same month as this interview, state auditors reviewed the Training Academy’s records and concluded that they were “exceptionally well kept and complete.” Review, R.147-19, PageID 11693. Given the Academy’s history of sloppy recordkeeping, this finding came as good news. A short time later, Votypka told Eckart that the Academy’s recordkeeping was “now efficient, organized and in keeping with a professional organization.” Mem., R.139- 18, PageID 9320. Rather than dismiss the charges against DeCrane, however, Eckart sat on them. He did not notify DeCrane of their dismissal until December 2016, a year after the favorable audit.
3. Other Academy Issues. DeCrane next suggests that the city undermined his work at the Training Academy in retaliation for the alleged leak of McGinnis’s deficient training. After news broke of the issue in August 2013, the mayor ordered an audit of all training records for all firefighters. On the day of the order, a group that included Eckart seized the Academy’s records. The city had never before confiscated records in such dramatic fashion. Despite requests that the records be returned, the Academy also did not receive them back for months. DeCrane claims that the seizure slowed his efforts to improve the Academy’s poor recordkeeping.
In 2014, the city also tried (but failed) to outsource firefighter training to a local college. Eckart initiated and oversaw that process. According to a declaration from Chief Kelly, Eckart offered to drop the outsourcing efforts if Kelly replaced DeCrane as the Training Academy’s No. 20-3620 DeCrane v. Eckart, et al. Page 5
director. (Kelly disputed his own account in a second declaration, but we must resolve this conflict in DeCrane’s favor.)
4. Last-Day Event. DeCrane alleges a final retirement-related slight tied to the belief that he was the leaker. When he retired from the Division of Fire in September 2016, his coworkers threw him the standard last-day party. Calvillo, who by then was the chief, had a policy requiring certain events to receive prior approval. DeCrane’s party was shut down because it purportedly did not comply with this policy.
* * *
A month later, DeCrane sued Eckart, Votypka (the OIC head), Christopher Chumita (an OIC investigator), and the City of Cleveland. As relevant for this appeal, DeCrane brought constitutional claims under 42 U.S.C. § 1983. He alleged that the individual defendants retaliated against him in violation of the First Amendment, and he sought to hold the city liable for this retaliation under Monell v. Department of Social Services, 436 U.S. 658 (1978). The district court granted summary judgment to Votypka and Chumita on the First Amendment retaliation claim and to the city on the Monell claim. DeCrane v. Eckart, 2020 WL 2475357, at *6–11 (N.D. Ohio May 13, 2020). But it denied summary judgment to Eckart on the First Amendment claim. Id. at *7–11. Eckart now appeals this part of the court’s decision.
II
DeCrane’s complaint alleged that Eckart violated the First Amendment by retaliating against him because of the media leak about Chief McGinnis’s deficient training. To hold Eckart liable under § 1983, DeCrane needed to establish two things. See District of Columbia v. Wesby, 138 S. Ct. 577, 589 (2018). He first needed to make out a constitutional violation by establishing the three elements for his claim: that the First Amendment protected the media leak, that Eckart took an adverse action against him, and that a causal connection existed between the leak and the action. See Rudd v. City of Norton Shores, 977 F.3d 503, 513 (6th Cir. 2020). He next needed to overcome Eckart’s qualified-immunity defense by showing that Eckart’s conduct violated clearly established law. See Wesby, 138 S. Ct. at 589. The district court held that DeCrane created a No. 20-3620 DeCrane v. Eckart, et al. Page 6
jury question on his First Amendment claim against Eckart and that Eckart’s conduct, if proved, would satisfy qualified immunity’s demanding test. See DeCrane, 2020 WL 2475357, at *6–11.
Eckart raises three challenges to the court’s reasoning, one for each element of DeCrane’s First Amendment claim. He asserts that the media leak was not protected speech and that he is at least entitled to qualified immunity on the issue. He also asserts that DeCrane has sued over adverse actions that fall outside the statute of limitations. He lastly asserts that any suspicion that DeCrane tipped off the media did not cause any alleged adverse action. We find that Eckart’s first argument fails on its merits and that his other two arguments fall outside our jurisdiction.
A. Protected Speech
A First Amendment claim requires DeCrane to identify protected speech. See Rudd, 977 F.3d at 513. DeCrane faces what might look like an obvious problem in this regard: he did not engage in the media leak that he claims to be protected. Instead, he alleges that Eckart wrongly believed that he was the leaker. Do public employees have a right not to be disciplined for perceived speech that they do not engage in? The Supreme Court recently answered “yes.” If an official punishes an employee for speech protected by the First Amendment, the employee may sue under § 1983 even if the official made a mistake about the speaker’s identity. See Heffernan v. City of Paterson, 136 S. Ct. 1412, 1418 (2016); see also Dye v. Off. of Racing Comm’n, 702 F.3d 286, 299–300 (6th Cir. 2012). Eckart thus does not contend that DeCrane’s claim fails because DeCrane did not tip off the media. Nor does Eckart seek qualified immunity on this basis.
But that threshold answer merely begins the analysis. DeCrane must still show that the tip to the media would have been protected if he had, in fact, provided it. See Heffernan, 136 S. Ct. at 1419. Public-employee speech must survive three inquiries to fall within the First Amendment. See Buddenberg v. Weisdack, 939 F.3d 732, 739 (6th Cir. 2019). The First Amendment protects this speech only if it addresses a matter of public (rather than private) concern. Connick v. Myers, 461 U.S. 138, 146–47 (1983). The amendment also protects the speech only if an employee speaks as a private citizen rather than pursuant to the employee’s job No. 20-3620 DeCrane v. Eckart, et al. Page 7
duties. Garcetti v. Ceballos, 547 U.S. 410, 421–22 (2006). And even if the speech survives these two steps, the First Amendment protects the speech only if the employee’s interest in speaking outweighs the employer’s interest in operating—a test that has come to be called “Pickering balancing.” Pickering v. Bd. of Educ. of Twp. High Sch. Dist. 205, 391 U.S. 563, 568 (1968).
Eckart makes no arguments under the first or third of these steps. We thus may assume that the media leak addressed a matter of public concern under Connick and that the speech interest in airing McGinnis’s training problems outweighed the city’s interest in operating under Pickering. Eckart relies only on the second inquiry under Garcetti. He asserts that DeCrane’s claim fails because DeCrane would have been speaking as part of his job if he had aired McGinnis’s deficient training with the media. At the least, Eckart says, qualified immunity should shield him from liability on this issue. We take his arguments in turn.
1. Would the media leak have been part of DeCrane’s job duties under Garcetti?
In Garcetti, the Supreme Court held that the government did not “abridg[e]” the “freedom of speech” when it fired an employee for a memo he wrote as part of the job it paid him to perform. U.S. Const. amend. I; Garcetti, 547 U.S. at 421–22. This rule follows from basic First Amendment principles. The government has a broad right to control its own “government speech.” Pleasant Grove City v. Summum, 555 U.S. 460, 467 (2009). When speaking itself, it may even engage in what would otherwise be the most pernicious of First Amendment sins—the “viewpoint discrimination” that allows one perspective on a subject but prohibits others. See Walker v. Tex. Div., Sons of Confederate Veterans, Inc., 576 U.S. 200, 207–08 (2015). The Supreme Court has suggested that this discrimination is automatically invalid when the government regulates private speech. See Rosenberger v. Rector & Visitors of Univ. of Va., 515 U.S. 819, 829 (1995). But how could the government accomplish anything if it could not favor its own views in its public programs? Summum, 555 U.S. at 468. That would mean, for example, that the government could not run ads encouraging COVID-19 vaccines without running ads discouraging them. See Walker, 576 U.S. at 207–08. Unlike government coercion, this government speech does not trigger the First Amendment because it does not “abridge” the speech of others. See id. No. 20-3620 DeCrane v. Eckart, et al. Page 8
Yet the “government” is an inanimate entity that cannot “speak” itself. It must rely on human agents to convey its messages—whether those agents are private contractors chosen to undertake discrete initiatives, see Rust v. Sullivan, 500 U.S. 173, 193 (1991), or public employees chosen to perform recurring tasks, see Garcetti, 547 U.S. at 422. When these agents undertake the programs that they are assigned to perform, the government may require them to stick to its message rather than express their own without offending the First Amendment. See Rust, 500 U.S. at 196. Conversely, the Supreme Court’s “unconstitutional-conditions doctrine” leaves the government with less room to regulate a government agent’s “off the job” private speech as a condition of the agent receiving a public contract or public job. See Agency for Int’l Dev. v. All. for Open Soc’y Int’l, Inc., 570 U.S. 205, 214 (2013); Pickering, 391 U.S. at 568.
Garcetti implements the distinction between the government engaging in its own speech (which does not trigger the First Amendment) and the government regulating private speech (which does). When public employees speak as agents of public employers, the employers may discipline the employees for saying something unacceptable because this speech is effectively the government’s. See Garcetti, 547 U.S. at 421–22. When, by contrast, employees speak “as private citizens” on issues of public concern, employers may discipline employees for the speech only if they satisfy Pickering because that discipline is effectively regulating private speech. Id. at 419.
This test requires courts to distinguish an individual’s speech as a public employee from the individual’s speech as a private citizen. Yet it can be unclear whether an employee spoke while wearing a “public” or “private” hat. See Mayhew v. Town of Smyrna, 856 F.3d 456, 464 (6th Cir. 2017). After all, employees regularly speak on their own time about events that happen during their workday or matters that relate to their job. See Lane v. Franks, 573 U.S. 228, 238– 40 (2014). So how should courts decide whether expression falls within the public or private bucket? Similar to the common law’s “scope of employment” test, courts have looked to all of the circumstances surrounding the speech. See Fledderjohann v. Celina City Sch. Bd. of Educ., 825 F. App’x 289, 293–98 (6th Cir. 2020); cf. Restatement (Second) of Agency §§ 228–37 (Am. L. Inst. 1958). No. 20-3620 DeCrane v. Eckart, et al. Page 9
When doing so, courts have asked several recurring questions. Most notably, what was the “impetus” for or “motivation[]” behind the speech? See Fledderjohann, 825 F. App’x at 294 (citation omitted); Handy-Clay v. City of Memphis, 695 F.3d 531, 540 (6th Cir. 2012). If an employee speaks to fulfill a job duty (say, the employee is conducting a government audit), the speech is more likely as a government agent. See DeWyse v. Federspiel, 831 F. App’x 759, 763 (6th Cir. 2020). If, by contrast, the employee speaks to fulfill an unrelated goal (say, the employee is acting as a union rep to obtain better terms), the speech is more likely as a private citizen. See Boulton v. Swanson, 795 F.3d 526, 534 (6th Cir. 2015). This question, of necessity, requires us to identify the employee’s job duties. Because the test is a “practical one,” Garcetti, 547 U.S. at 424, courts should consider the duties that the employee actually performs on a day- to-day basis, Buddenberg, 939 F.3d at 740. That said, an employee’s official duties listed in a formal job description are relevant too. See Mayhew, 856 F.3d at 465.
Another common question: What was the speech’s setting? On-the-clock speech at the employer’s place of business is more likely to be speech as a government agent as compared to off-the-clock speech away from the office. See Handy-Clay, 695 F.3d at 540–42. So a fire chief acted as a government agent when emailing firefighters about a job-related issue from his official account using his official title. See Holbrook v. Dumas, 658 F. App’x 280, 288–89 (6th Cir. 2016). But a fire captain acted as a private citizen when he called city council members from his home as a concerned taxpayer. See Stinebaugh v. City of Wapakoneta, 630 F. App’x 522, 527–28 (6th Cir. 2015); see also Westmoreland v. Sutherland, 662 F.3d 714, 719 (6th Cir. 2011).
A third common question: Who was the speech’s audience? Speech to supervisors is more likely to be speech as a government agent as compared to speech to outside individuals. So a police officer acted as a public employee when he complained to his chief about training cutbacks, as did a public-records employee when she complained to supervisors about obstacles to obtaining records. See Haynes v. City of Circleville, 474 F.3d 357, 364 (6th Cir. 2007); Handy-Clay, 695 F.3d at 541–42; see also Bushong v. Delaware City Sch. Dist., 851 F. App’x 541, 544 (6th Cir. 2021); Fox v. Traverse City Area Pub. Schs. Bd. of Educ., 605 F.3d 345, 349 (6th Cir. 2010). Yet a police officer acted as a concerned citizen when he talked to the FBI about corruption in the ranks, as did an officer who issued his own “press release” about discrimination No. 20-3620 DeCrane v. Eckart, et al. Page 10
in the department. See Barrow v. City of Hillview, 775 F. App’x 801, 813 (6th Cir. 2019); Miller v. City of Canton, 319 F. App’x 411, 413, 417 (6th Cir. 2009); see also See v. City of Elyria, 502 F.3d 484, 493 (6th Cir. 2007).
When considering these questions in the course of resolving an employer’s summary- judgment motion, we take any disputed historical facts in the light most favorable to the employee. See Pucci v. Nineteenth Dist. Ct., 628 F.3d 752, 768 (6th Cir. 2010). After doing so, however, we treat the ultimate question whether the employee spoke as a government agent or private citizen as a legal one to be decided without deference to the factfinder. See Mayhew, 856 F.3d at 462–64.
*
Invoking Garcetti here, Eckart argues that DeCrane would have been acting as the director of the Training Academy if he had informed the media about McGinnis’s training issues. DeCrane responds that he would have been acting as a private citizen. Both parties thus assume that the usual Garcetti test fully applies even when, as in this case, an employee is allegedly punished for speech that the employee did not undertake and so must rely on Heffernan’s protections for perceived speech. See 136 S. Ct. at 1418. Given that the parties do not disagree on this point, we can simply assume that the ordinary Garcetti test applies in this context.
Yet we flag one issue with this assumption. As noted, Garcetti requires us to consider the “impetus” for or “motivations” behind the employee’s speech. Fledderjohann, 825 F. App’x at 294 (citation omitted). Is this a subjective test that depends on the employee’s state of mind or an objective test that asks what a reasonable person would identify as the speaker’s intent? The common law of agency would suggest a subjective approach, as the Restatement notes that the “state of the servant’s mind” is “material” and that the “external manifestations” of that mindset are only “evidence” of an intent to serve the employer. Restatement (Second) of Agency § 235 cmt. a. Yet this understanding would make no sense here. How can we identify DeCrane’s intent behind the leak when he was not the leaker? So the parties assume that Garcetti establishes an objective test asking what a reasonable person would take to be the intent behind the speech. Cf. United States. v. Leon, 468 U.S. 897, 919 n.20 (1984). This view might better No. 20-3620 DeCrane v. Eckart, et al. Page 11
comport with our cases given that they treat Garcetti as raising a legal issue. See Mayhew, 856 F.3d at 462–64. But we have not expressly addressed this objective-versus-subjective debate and can save it for another day here too. Because the parties apply an objective approach, we can do the same for this appeal.
Objectively speaking, then, what would have been the reason for the leak if DeCrane had been the one who tipped off the media? Would he have been fulfilling one of his job duties? Or would he have been furthering a personal goal? The factors on which we have relied in other cases all lead to the same conclusion: DeCrane would have been acting as a private citizen. His official duties (which are relevant but not dispositive) are a good place to start. See Mayhew, 856 F.3d at 463. As the director of the Training Academy, he was “responsible for all training activities within the Division of Fire including maintaining appropriate training records” as well as “oversee[ing] and coordinat[ing] all required internal and external reporting related to training.” Gen. Order, R.145-3, PageID 11115. Nothing in this description lists speaking to the media, let alone secretly doing so. What is more, the order that contains the training director’s duties identifies a different employee—the public information officer—as bearing the responsibility to “act[] as the liaison between the Division of Fire and the Media.” Id., PageID 11107.
DeCrane’s “de facto” duties point in the same direction. Weisbarth v. Geauga Park Dist., 499 F.3d 538, 544 (6th Cir. 2007). No evidence shows that DeCrane’s superiors ever gave him any secret orders to speak with the media off the record about the Division of Fire. Nor is there any evidence that he ever engaged in such “authorized” leaks. Eckart even admitted that “[p]roviding information outside the chain of command is forbidden” and that employees must “receive permission from their appointing authority or direct supervisor” to speak to the media. Eckart Dep., R.140, PageID 9512. DeCrane shared the same understanding that the city barred him from making disclosures to the media without approval. If DeCrane had tipped off the media, then, he would have been acting against the directions of his public employer—not pursuant to them. See Garcetti, 547 U.S. at 421.
The leak’s audience and setting confirm that DeCrane would have provided it as a private citizen, not a public employee. Unlike when he informed Eckart about McGinnis’s training No. 20-3620 DeCrane v. Eckart, et al. Page 12
issues, he would have been going well outside the chain of command by giving this information to a journalist. Cf. Mayhew, 856 F.3d at 465. That type of disclosure looks a lot more like the officer’s personal “press release” to the media about discrimination in the police department, see Miller, 319 F. App’x at 417, than the officer’s complaint to his chief about training cutbacks, see Haynes, 474 F.3d at 364. The leak’s secretive nature reinforces that it was not an official communique. Cf. Barrow, 775 F. App’x at 813. Unlike the fire chief who sent an email using his official title, see Holbrook, 658 F. App’x at 288–89, this type of leak would presumably occur without official imprimatur. All told, then, the relevant factors show that DeCrane would have been acting as a private citizen—not a government agent—if he had secretly spoken to the media.
Eckart’s responses do not change things. He argues that the leak would have fallen within DeCrane’s job duties because it concerned training and DeCrane was the director of training. This argument flouts precedent. The Supreme Court has made clear that employees do not speak “on the job” simply because they speak about the job. See Lane, 573 U.S. at 239–40. Indeed, “speech by public employees on subject matter related to their employment holds special value precisely because those employees gain knowledge of matters of public concern through their employment.” Id. at 240. That this media leak concerned training thus does nothing to bring it within Garcetti’s rule so long as it would not have been conveyed in DeCrane’s capacity as the director of training.
Citing our cases holding that even “ad hoc” (that is, atypical) duties can fall within Garcetti’s rule, Eckart next argues that DeCrane occasionally spoke to journalists about the Division of Fire. Weisbarth, 499 F.3d at 544. DeCrane responds by suggesting that Lane overruled our cases extending Garcetti to ad hoc duties. Lane described the “critical question” as “whether the speech at issue is itself ordinarily within the scope of an employee’s duties, not whether it merely concerns those duties.” 573 U.S. at 240 (emphasis added); cf. Boulton, 795 F.3d at 534. An “ad hoc” duty, DeCrane’s argument goes, is the opposite of an “ordinary” duty. Even after Lane, however, we have held that speech pursuant to a “special” duty falls outside the First Amendment when the duty (and speech) arose within the scope of the public No. 20-3620 DeCrane v. Eckart, et al. Page 13
job. See DeWyse, 831 F. App’x at 763; see also Davidson v. Arlington Cmty. Schs. Bd. of Educ., 847 F. App’x 304, 309 (6th Cir. 2021); Henderson v. City of Flint, 751 F. App’x 618, 623–24 (6th Cir. 2018) (lead opinion). Garcetti’s premise—that speech as a public employee effectively qualifies as government speech—also does not hinge on whether the speech arose pursuant to a regular or atypical duty. It hinges on whether the speech “owes its existence to a public employee’s professional responsibilities[.]” Garcetti, 547 U.S. at 421.
Regardless, we need not decide Lane’s effect on our precedent because leaking information to the media cannot even be described as one of DeCrane’s “ad hoc” duties. True enough, DeCrane occasionally spoke to the media. In 2012, for example, the Division of Fire hosted an event for politicians and journalists at which DeCrane’s supervisors asked him to give a presentation on modern firefighting. He later got permission to follow up with journalists about news articles on the subject. Yet these occurrences—in which he conversed with the media in his official capacity with approval from his superiors—do not suggest that he would have been acting officially by leaking embarrassing information about the chief. If anything, they prove the contrary: he spoke to the media pursuant to his duties only when he got approval to do so.
Eckart lastly highlights DeCrane’s testimony that, in later years, he got permission to speak with a journalist about training issues. But this authorization came from Chief Patrick Kelly—who became the chief only after the leak. This testimony says nothing about DeCrane’s authorization to speak to the media at the correct time—the time of the leak. So none of Eckart’s arguments alter our conclusion that DeCrane would have spoken as a private citizen.
2. Did the media leak fall clearly outside DeCrane’s duties under Garcetti?
Even so, Eckart responds, qualified immunity should still shield him from liability. This defense protects a public official so long as the official does not violate “clearly established” law. Wesby, 138 S. Ct. at 589. An official’s conduct flunks this “clearly established” test only if the conduct’s unconstitutionality was “beyond debate” when the official acted, such that any reasonable person would have known that it exceeded constitutional bounds. See id. (citation omitted). While qualified immunity does not force plaintiffs to have a “case directly on point,” No. 20-3620 DeCrane v. Eckart, et al. Page 14
Ashcroft v. al-Kidd, 563 U.S. 731, 741 (2011), its “demanding” test nevertheless requires them to define the “clearly established” constitutional rule on which they rely with a “high ‘degree of specificity.’” Wesby, 138 S. Ct. at 589–90 (citation omitted); see also Reichle v. Howards, 566 U.S. 658, 665 (2012). A plaintiff has defined the legal rule too generically whenever “the unlawfulness of the [defendant’s] conduct ‘does not follow immediately from the conclusion that [the rule] was firmly established.’” Wesby, 138 S. Ct. at 590 (citation omitted).
In this speech context, for example, plaintiffs generally cannot defeat qualified immunity simply by arguing that they have a clearly established right not to suffer an “abridgment” of the “freedom of speech.” U.S. Const. amend. I. Such a rule (defined at the highest possible level of generality) would not have immediately alerted public officials of the First Amendment’s limits in the specific situations that they confronted. See Lane, 573 U.S. at 243–45; Reichle, 566 U.S. at 664–70. Even Garcetti’s rule (that speech can receive protection if taken as a citizen rather than employee) sometimes cannot provide sufficient guidance because we have recognized that it can be “challenging” to distinguish public from private speech. Mayhew, 856 F.3d at 464. Eckart thinks that this fact resolves the qualified-immunity issue in his favor. He argues that the then-existing caselaw did not clearly establish that the media leak would have fallen outside DeCrane’s job duties within the meaning of Garcetti. See Wesby, 138 S. Ct. at 589.
Eckart is mistaken because our cases at the relevant time had already set more specific ground rules to distinguish public from private speech. We had held that employees speak as private citizens (not public employees) at least when they speak on their own initiative to those outside their chains of command and when their speech was not part of their official or de facto duties. See Handy-Clay, 695 F.3d 542–43; see also Westmoreland, 662 F.3d at 719; Pucci, 628 F.3d at 768–69; See, 502 F.3d at 493–94; cf. Devlin v. Kalm, 630 F. App’x 534, 539 (6th Cir. 2015) (per curiam). Would this “firmly established” rule have “immediately” alerted a reasonable person that the media leak would have been in DeCrane’s private capacity? Wesby, 138 S. Ct. at 590 (citation omitted). We think so. Such a leak would have fallen outside his duties. The speech also would have been to journalists, not supervisors. For these reasons, we had already held that a police officer’s disclosure to the media fell outside Garcetti precisely No. 20-3620 DeCrane v. Eckart, et al. Page 15
because the officer engaged in the speech on his own time and had no media-related duties. See Miller, 319 F. App’x at 417.
Nor do the two circuit cases on which Eckart relies—Haddad v. Gregg, 910 F.3d 237 (6th Cir. 2018) (per curiam), and Hurst v. Lee County, 764 F.3d 480 (5th Cir. 2014)—muddy the waters. Haddad, a two-paragraph opinion that merely attached the district court’s decision as an appendix, looks nothing like this case. 910 F.3d at 240. A state department of insurance fired an employee allegedly for speech that was designed to end “what he believed to be an unfair insurer practice.” Id. at 241. Yet the employee conceded “that he was acting pursuant to his official duties” when he engaged in the speech. Id. Although he consulted with individuals outside his office about the challenged practice, he did so as part of an “investigation” to fulfill a “central function” of his job. Id. at 249. Here, by contrast, the leak furthered no function of DeCrane’s job.
In Hurst, by comparison, a sheriff fired an officer for statements made to a journalist about an arrested college football player. 764 F.3d at 482–83. The Fifth Circuit held that the officer was speaking as a public employee when he spoke to the journalist. Id. at 485. The journalist had traveled to the jail to get information about the player, and the officer had conversed with him in an official capacity while working there. Id. at 482–83. The sheriff’s media policy authorized the officer to reveal certain information to this journalist, but the officer disclosed more than the policy permitted. Id. at 482–83, 485. The speech thus was part of the officer’s job, and he performed that job poorly by exceeding the parameters that the sheriff had given him. Id. at 485. Here, by contrast, the leak to the media cannot be characterized as part of DeCrane’s duties.
Eckart lastly cites two district-court decisions that found speech unprotected when the plaintiff spoke either to the media or to a county board about matters learned on the job. See Meggison v. Charlevoix County, 2008 WL 5411896, at *7–9 (W.D. Mich. Dec. 23, 2008); Omokehinde v. Detroit Bd. of Educ., 563 F. Supp. 2d 717, 730 (E.D. Mich. 2008). It is not obvious that we should even consider these district-court decisions when identifying the rules that our cases have clearly established. See Camreta v. Greene, 563 U.S. 692, 709 n.7 (2011). Nor is it obvious that the cases survived Lane’s clarification that speech does not fall within No. 20-3620 DeCrane v. Eckart, et al. Page 16
Garcetti’s rule merely because it relates to an employee’s job. See 573 U.S. at 240. Be that as it may, the decisions have sufficiently distinguishable facts that they would not have allowed Eckart to reasonably conclude that DeCrane’s job duties included leaking information to the media.
We close by reiterating our ruling’s narrow scope. Eckart made just one argument as to why the media leak was not protected speech: because it would have fallen within DeCrane’s job duties. Eckart did not argue that he could reasonably believe that he could discipline employees for violating a seemingly neutral policy banning unauthorized speech to the media. Cf. Heffernan, 136 S. Ct. at 1419. Nor did he ask us to consider the circumstances in which a public employer may have such a media policy or the manner in which to analyze this constitutional question. We thus do not consider these issues; we resolve only the Garcetti argument presented to us.
B. Adverse Action
After DeCrane identifies protected speech, he next must show that Eckart took an “adverse action” against him. See Rudd, 977 F.3d at 514–15. The district court identified a medley of harmful actions, ranging from the denial of promotions and the misconduct charges to the attempt to outsource training and the lack of a last-day party. Eckart does not dispute that these actions were sufficiently “adverse” to trigger the First Amendment’s protections. He instead raises an affirmative defense different from qualified immunity, arguing that § 1983’s two-year statute of limitations bars relief for some of these actions. See Browning v. Pendleton, 869 F.2d 989, 992 (6th Cir. 1989) (en banc).
His argument has a fatal jurisdictional problem. We would normally lack jurisdiction over the district court’s decision to proceed to trial on DeCrane’s retaliation claim because we typically have jurisdiction only over “final decisions” that end a case. 28 U.S.C. § 1291; Stojcevski v. Macomb County, 827 F. App’x 515, 520–21 (6th Cir. 2020). Yet, under the collateral-order doctrine, the Supreme Court allows parties to appeal decisions that are “final” on certain collateral issues even if the entire case is not over. Stojcevski, 827 F. App’x at 520–21. No. 20-3620 DeCrane v. Eckart, et al. Page 17
This doctrine permits an official like Eckart to appeal a decision denying qualified immunity. See id. at 521 (citing Mitchell v. Forsyth, 472 U.S. 511, 530 (1985)). Qualified immunity is a defense not just against having to pay a judgment but also against having to litigate the dispute at all. Mitchell, 472 U.S. at 526–29. If officials could not immediately appeal the denial of immunity, the defense would be lost because they would have to litigate the case in order to appeal the right not to litigate it. Id. at 526–27. That is why we have jurisdiction over Eckart’s claim that he is entitled to qualified immunity on the legal question whether DeCrane identified protected speech under Garcetti.
Yet this logic does not extend to Eckart’s separate claim about the statute of limitations. A statute of limitations is not an immunity from suit; it is a defense to liability. A defendant thus usually may not use the collateral-order doctrine to appeal an otherwise nonfinal decision denying a statute-of-limitations defense. See United States v. Davis, 873 F.2d 900, 908–09 (6th Cir. 1989); see also Watkins v. Healy, 986 F.3d 648, 659 (6th Cir. 2021).
Does it change things, though, that this statute-of-limitations issue arose from the same order that denied Eckart qualified immunity? We do not see why. The collateral-order doctrine’s interpretation of the word “final” in 28 U.S.C. § 1291 does not bring up the entire decision for appeal. Cf. BP P.L.C. v. Mayor & City Council of Baltimore, 141 S. Ct. 1532, 1537–40 (2021). Rather, it allows us to review the legal issues that the decision resolves in the course of denying qualified immunity—not other unrelated issues. See Swint v. Chambers Cnty. Comm’n, 514 U.S. 35, 43–51 (1995); cf. Ashcroft v. Iqbal, 556 U.S. 662, 672–75 (2009). Put another way, the resolution of the qualified-immunity issue is “final” within the meaning of 28 U.S.C. § 1291 in a way that the resolution of unrelated issues is not. And a statute-of- limitations defense (which asks whether a plaintiff timely sued) has nothing to do with a qualified-immunity defense (which asks whether a defendant’s conduct violated clearly established law). See Watkins, 986 F.3d at 658–59; see also Johnson v. Johnson, 694 F. App’x 945, 947 (5th Cir. 2017) (per curiam); Sanchez v. Hartley, 810 F.3d 750, 761 (10th Cir. 2016); Ochoa Lizarbe v. Rivera Rondon, 402 F. App’x 834, 837 (4th Cir. 2010) (per curiam); Garnier v. Rodríguez, 506 F.3d 22, 25 (1st Cir. 2007). No. 20-3620 DeCrane v. Eckart, et al. Page 18
That said, we sometimes have exercised what has been dubbed “pendent appellate jurisdiction” over other issues if they are “inextricably intertwined” with qualified immunity. See Watkins, 986 F.3d at 659–60; Stojcevski, 827 F. App’x at 523–24. But we have also held that a statute-of-limitations defense does not satisfy this “intertwined” test because our resolution of the immunity issue will not affect the defense’s viability. See Watkins, 986 F.3d at 659–60. In any event, Eckart did not even invoke our discretionary pendent appellate jurisdiction or claim that the statute of limitations was intertwined with his immunity defense. We thus need not address this jurisdictional theory. Because we lack jurisdiction to consider Eckart’s statute-of- limitations argument in this appeal’s interlocutory setting, we dismiss this portion of his appeal.
C. Causation
This conclusion leaves DeCrane with a final task—to show a link between the media leak and the adverse actions. See Rudd, 977 F.3d at 513, 515. A First Amendment retaliation claim under § 1983 ultimately requires a “but-for” causal connection—meaning that the public employer would not have taken the harmful action “but for” the protected speech. See Nieves v. Bartlett, 139 S. Ct. 1715, 1722 (2019). Our cases include seemingly conflicting statements on who bears the burden to prove but-for causation. Some say that the employee must prove that the employer would not have taken the adverse action absent the protected speech; others say the employer must prove that it would have still taken the action even absent the speech once the employee shows that the speech was a motivating factor for it. See Spithaler v. Smith, 803 F. App’x 826, 829–30 (6th Cir. 2020) (collecting cases); cf. Nieves, 139 S. Ct. at 1722–23. Either way, our cases leave no doubt that this causation element (unlike the protected-speech element) presents a pure fact question about the reason(s) for the adverse action. See Farkhoury v. O’Reilly, 837 F. App’x 333, 341 (6th Cir. 2020); Hartsel v. Keys, 87 F.3d 795, 803 (6th Cir. 1996); Matulin v. Village of Lodi, 862 F.2d 609, 613 (6th Cir. 1988). Eckart argues on appeal that DeCrane introduced insufficient evidence to allow a reasonable jury to decide this fact question in his favor.
His argument has a different (but similarly fatal) jurisdictional problem. We lack jurisdiction not just over legal arguments unrelated to qualified immunity (like Eckart’s statute- of-limitations argument) but also over pure sufficiency-of-the-evidence arguments (like this No. 20-3620 DeCrane v. Eckart, et al. Page 19
causation argument). The Supreme Court has held that qualified immunity falls within the collateral-order doctrine, in part, because the question whether existing precedent would have “clearly established” that a defendant’s conduct violated the Constitution is collateral to the plaintiff’s claim on the merits. Mitchell, 472 U.S. at 527–28. And it has held that this “clearly established” inquiry necessarily encompasses an initial inquiry into related legal questions, such as whether the defendant’s conduct even violated the Constitution at all. See Iqbal, 556 U.S. at 673–74; Hartman v. Moore, 547 U.S. 250, 257 n.5 (2006). So, for example, the collateral-order doctrine allows us to consider whether a given set of facts amounts to “excessive force” under the Fourth Amendment or whether the existing caselaw would clearly establish that conclusion. See Liogghio v. Township of Salem, 766 F. App’x 323, 326 (6th Cir. 2019); DiLuzio v. Village of Yorkville, 796 F.3d 604, 609 (6th Cir. 2015). In the process of deciding these legal questions, moreover, courts must also identify the facts that the record would allow a reasonable jury to find. See DiLuzio, 796 F.3d at 611; cf. Ouza v. City of Dearborn Heights, 969 F.3d 265, 276–77 (6th Cir. 2020).
That said, the Supreme Court has held that the theory for treating qualified immunity as collateral to the merits does not cover a defendant who seeks to appeal merely the district court’s decision that a plaintiff has presented enough evidence to prove a factual element of the claim. DiLuzio, 796 F.3d at 609 (quoting Johnson v. Jones, 515 U.S. 304, 313 (1995)); see Bey v. Falk, 946 F.3d 304, 319 (6th Cir. 2019). Admittedly, an argument that the summary-judgment record does not create a “genuine dispute” as to a “material fact” raises a legal question—hence, why we review that ubiquitous issue de novo. Fed. R. Civ. P. 56(a); see Iqbal, 556 U.S. at 674; Hartsel, 87 F.3d at 799. For better or worse, however, this question about “which facts a party may, or may not, be able to prove at trial” is not the type of legal issue over which our interlocutory jurisdiction independently extends. DiLuzio, 796 F.3d at 609 (quoting Johnson, 515 U.S. at 313).
Eckart’s challenge falls on the wrong side of this line. He argues only that the summary- judgment record would not permit a reasonable jury to find a causal connection between the media leak and the adverse actions. He thus raises a pure “evidence sufficiency” challenge of the sort that falls outside our jurisdiction. Bey, 946 F.3d at 319–20. Indeed, in this interlocutory No. 20-3620 DeCrane v. Eckart, et al. Page 20
setting, we have repeatedly refused to consider whether a plaintiff presented enough evidence to create a genuine dispute of fact on the causation element of a First Amendment retaliation claim. See, e.g., Farkhoury, 837 F. App’x at 340–41; Liogghio, 766 F. App’x at 325–26; Sigler v. City of Englewood, 424 F. App’x 449, 454–55 (6th Cir. 2011). Many other courts have issued similar decisions. See, e.g., Hunter v. Town of Mocksville, 789 F.3d 389, 400 (4th Cir. 2015); Cooley v. Grimm, 272 F. App’x 386, 392 (5th Cir. 2008) (per curiam); Brown v. Town of LaBarge, 97 F. App’x 216, 227 (10th Cir. 2004); Herts v. Smith, 345 F.3d 581, 586 (8th Cir. 2003); Locurto v. Safir, 264 F.3d 154, 168 (2d Cir. 2001); Stella v. Kelley, 63 F.3d 71, 77 (1st Cir. 1995).
To be sure, legal arguments about the proper meaning of the First Amendment’s causation element would fall within our jurisdiction if raised in a qualified-immunity defense. The Supreme Court, for example, held that it had jurisdiction to consider the legal question whether this causation element contains a “lack-of-probable-cause” requirement when a plaintiff alleged that officials engineered a criminal prosecution in retaliation for protected speech. See Hartman, 547 U.S. at 257 n.5. And Eckart makes a passing head fake at a potential legal issue. He argues that his supposed retaliatory animus could not have caused many of the adverse actions because he was not the official with the final decisionmaking authority for them. Our caselaw makes clear that § 1983 defendants can be held liable only for their own conduct, Rudd, 977 F.3d at 512, so a plaintiff must show that a defendant was sufficiently involved in the adverse action, cf. Spithaler, 803 F. App’x at 830; Sigler, 424 F. App’x at 455. What involvement is required? Does § 1983 permit a so-called “cat’s paw” theory of liability for non- decisionmakers? Cf. Staub v. Proctor Hosp., 562 U.S. 411, 419 (2011); Bose v. Bea, 947 F.3d 983, 991 n.4 (6th Cir. 2020). These inquiries ask legal questions. The problem for Eckart is that he does not raise them. He simply asserts that “DeCrane can cite to no credible evidence that Eckart influenced [the relevant] decisions[.]” Appellants’ Br. 35. He nowhere suggests that the district court failed to apply the proper law on causation. For that matter, he does not cite any law on this issue. Eckart thus raises a pure sufficiency-of-the-evidence challenge of the kind that we lack jurisdiction to consider. See Bey, 946 F.3d at 319. No. 20-3620 DeCrane v. Eckart, et al. Page 21
We affirm the district court’s decision in part and dismiss Eckart’s appeal in part for lack of appellate jurisdiction.