OPINION
Kevin Lindke claims that James Freed, the City Manager of Port Huron, Michigan, violated Lindkes free-speech rights by blocking him on Facebook and deleting his Facebook comments. We originally ruled for Freed on state-action grounds, but the Supreme Court adopted a different test. Because the factual record isnt developed enough for us to apply the Supreme Courts revised test, we remand this case to the district court for further proceedings.
I.
We spelled out the facts of this dispute in our last opinion, so well provide an abbreviated version here. See Lindke v. Freed, 37 F.4th 1199 (6th Cir. 2022), vacated, 601 U.S. 187, 144 S.Ct. 756, 218 L.Ed.2d 121 (2024). James Freed was the City Manager of Port Huron, Michigan. He maintained a public Facebook page where he posted updates about his personal life and job. Kevin Lindke left comments on these posts that were critical of the citys handling of the COVID-19 pandemic. In response, Freed deleted Lindkes comments and “blocked” Lindke from accessing Freeds Facebook page. Lindke sued under 42 U.S.C § 1983, alleging that Freeds actions violated Lindkes First Amendment rights.
We affirmed the district courts judgment in favor of Freed on the ground that Freed wasnt engaged in state action. Id. at 1207. We held that an officials social-media activity could count as state action only when the activity (1) is part of an officeholders “actual or apparent duties” or (2) couldnt happen in the same way without “the authority of the office.” Id. at 1203 (cleaned up). Applying that test, we concluded that Freed hadnt engaged in state action when he blocked Lindke and deleted his comments. Id. at 1204. The Supreme Court then granted certiorari, articulated a state-action test that was different from ours, and vacated our judgment to the extent that it was inconsistent with the new test. See Lindke v. Freed, 601 U.S. 187, 144 S.Ct. 756, 218 L.Ed.2d 121 (2024).
Back in this court, Lindke filed a motion to remand the case to the district court. He argues that he should have an opportunity to conduct additional discovery that is tailored to the Supreme Courts new test. In response, we ordered supplemental briefing on various questions related to the Supreme Courts test and the need to remand.
II.
Under the Supreme Courts new test, an officials social-media activity counts as state action only if the plaintiff can show that the official (1) “possessed actual authority to speak on the States behalf” and (2) “purported to exercise that authority when he spoke on social media.” Id. at 198, 144 S.Ct. 756. This new test differs from the one we previously applied in two ways: its narrower in one respect and broader in another. At prong one, we previously held that actual or apparent authority could support a finding of state action. Lindke, 37 F.4th at 1204. But the Supreme Court made clear that only actual authority could suffice. Lindke, 601 U.S. at 198, 144 S.Ct. 756. And on prong two, we focused on the appearance and administration of Freeds page as a whole. Lindke, 37 F.4th at 1203. But the Supreme Court explained that this prong requires a post-by-post inquiry. Lindke, 601 U.S. at 204, 144 S.Ct. 756. For the reasons we discuss below, a limited remand is necessary to apply this revised test.
A.
The first step of the state-action inquiry is to determine whether Freed had “actual authority” to speak on the States behalf “on a particular matter.” Id. The Supreme Court identified three key features Lindke must show to satisfy this element.
First, Freeds authority to speak for the state must have been actual, not simply apparent. It doesnt matter whether Freed acted like he had authority to speak for the government. Nor does it matter whether others thought that Freed had such authority. Thats because Freeds social-media activity cannot be “attributable to the State” unless he was actually “possessed of state authority” to speak on the states behalf. Id. at 198, 144 S.Ct. 756 (citations omitted).
Second, to be state action, Freeds posts must relate to a specific matter within his portfolio of responsibilities. Id. at 199, 144 S.Ct. 756. Its not enough for Lindke to show that Freed had “some authority to communicate with residents on behalf of Port Huron.” Id. Rather, Lindke must show that Freeds posts pertained to topics “within Freeds bailiwick.” Id. The Supreme Courts hypothetical example is instructive. Imagine that Freed had posted a list of local restaurants with health-code violations and then deleted negative user comments. Id. Freed would not be engaging in state action unless responsibility for public health was “within the portfolio of the city manager.” Id.
Third, the grant of actual authority must come from one of the sources mentioned in 42 U.S.C. § 1983: “statute, ordinance, regulation, custom, or usage.” The first three options are straightforward: they refer to “written law” that “empower[s] [Freed] to make official announcements.” Id. at 200, 144 S.Ct. 756. An example might be a city mayor who uses social media to satisfy a statutory public-notice provision for a budgetary hearing. Or a city press manager whose statutory responsibilities include issuing official documents.
“Custom” and “usage” are a bit harder to pin down. Those terms refer to “ ‘persistent practices of state officials’ that are ‘so permanent and well settled’ that they carry ‘the force of law.’ ” Id. (quoting Adickes v. S. H. Kress & Co., 398 U.S. 144, 168, 90 S.Ct. 1598, 26 L.Ed.2d 142 (1970)). The idea is that a state may assign job duties to an official through unwritten practices that have become so entrenched as to carry the force of law.
Freed might have actual authority in the absence of written law if “prior city managers have purported to speak on [the states] behalf and have been recognized to have that authority for so long that the managers power to do so has become ‘permanent and well settled.’ ” Id. (quoting Adickes, 398 U.S. at 167–68, 90 S.Ct. 1598). For example, imagine that three successive fire marshals have issued statements designating certain days on which residents are prohibited from outdoor burning. Even if no written law authorizes the fire marshals to issue these statements, the fact that they have consistently done so might be evidence that the fire marshals office has that authority via custom and usage. If the next fire marshal issued a burn ban via social media, that would count as state action.
Municipal-liability caselaw under § 1983 can also help flesh out the meaning of “custom” and “usage.” An actionable custom must be sufficiently “widespread” to have the force of law. See Bd. of Cnty. Commrs of Bryan Cnty. v. Brown, 520 U.S. 397, 404, 117 S.Ct. 1382, 137 L.Ed.2d 626 (1997). Employees’ patterns of conduct can establish a custom. See id. at 407–08, 117 S.Ct. 1382; City of Canton v. Harris, 489 U.S. 378, 389, 109 S.Ct. 1197, 103 L.Ed.2d 412 (1989). Alternatively, “formal rules or understandings” that “establish fixed plans of action to be followed under similar circumstances consistently and over time” can create a custom. Pembaur v. City of Cincinnati, 475 U.S. 469, 480–81, 106 S.Ct. 1292, 89 L.Ed.2d 452 (1986). These municipal-liability cases shed light on how settled but unwritten practice can give rise to “custom” and “usage” under § 1983.
Whats more, for a post to constitute state action, an official must have the power to make statements on the matter over which he has responsibility. It isnt enough that “making official announcements could fit within the job description” of an official. Lindke, 601 U.S. at 201, 144 S.Ct. 756. Rather, issuing official announcements must be “actually part of the job that the State entrusted the official to do.” Id. In this case, for instance, Freed must have been “possessed of state authority to post city updates.” Id. at 199, 144 S.Ct. 756 (cleaned up). This makes sense. After all, a social media post is merely a statement. And a grant of authority to perform job responsibilities doesnt necessarily come with authority to make official statements about those responsibilities. 1
There is, however, one situation when a grant of authority is accompanied by a power to issue statements about those duties. If a state grants “broad responsibility” for a major project to a “high-ranking official,” then this authorization might come with the authority to “make official announcements on that subject.” Id. at 201, 144 S.Ct. 756. A director of a state transportation department, for example, would possess authority to issue statements on the state highway system that he oversees. Id. But the Supreme Court emphasized that this scenario would be the exception, not the rule. See id. Courts must not rely on “excessively broad job descriptions to conclude that a government employee is authorized to speak for the State.” Id. (cleaned up). Unlike the transportation director above, a construction worker employed by the city would not be engaging in state action if he took to Facebook to post updates about his work. Why? The State didnt authorize the worker to make official announcements about his construction duties.
B.
If Freed lacked actual authority to speak on the states behalf, thats the end of the inquiry—theres no state action. But if Freed did have such authority, then we move to the second prong: whether Freed purported to exercise that authority when he spoke on social media. This is an objective test, although the “appearance and function” of the social-media activity are relevant. Id. at 198, 144 S.Ct. 756. For example, if a government website links to an officials social-media posts, its likely that such posts exercised state authority.
But not all posts concerning a public officials job are exercises of state power. After all, “[s]tate employees do not work for the State every hour of the day.” Mackey v. Rising, 106 F.4th 552, 554 (6th Cir. 2024). Public officials “have the right to speak about public affairs in their personal capacities.” Lindke, 601 U.S. at 203, 144 S.Ct. 756. So when a state official doesnt purport to exercise his actual authority, he speaks in his private capacity rather than as a state actor.
Resolving this second prong requires a post-by-post analysis. Id. at 203–04, 144 S.Ct. 756. Accounts like Freeds, which arent clearly labeled as either personal or official, may contain some posts made in a personal capacity and others in an official capacity. Id. at 202, 144 S.Ct. 756. Consequently, Lindke must identify “specific posts” in which Freed purported to exercise authority to speak on the states behalf. Id. at 203, 144 S.Ct. 756. Moreover, its “crucial” that Lindke show that Freed was purporting to exercise state authority in the specific posts that Lindke identifies. Id. Its not the district courts job to comb through the offending account.
Of course, the posts Lindke identifies must be ones that were the subject of alleged censorship. Otherwise, Lindke wouldnt be able to allege a rights violation. Here, Lindke alleges two categories of posts that were the subject of censorship: (1) posts from which Lindkes comments were deleted, and (2) posts that Lindke couldnt access because he was “blocked” from Freeds page.
When it comes to the deletion of comments, the courts task is straightforward: look at the allegedly offending posts that Lindke identifies and determine whether any of those posts represented state action. Id. at 204, 144 S.Ct. 756. For deletion, “the only relevant posts are those from which Lindkes comments were removed.” Id.
But when it comes to Freeds wholesale “blocking” of Lindke from accessing his page, Lindke need only identify one single post on Freeds account that represented state action. Id. Indeed, “[b]locking” someone on Facebook is a “blunt[ ] ․ tool.” Id. This means that when an official blocks someone from accessing a page on which both personal and official business is conducted, the official risks liability if the blocked user is prevented from commenting on even a single state-action post—even if the majority of the accounts content is personal. Id.
Context is everything in analyzing this second prong. Whether each posts “content and function” invoke personal or official authority is a “fact-specific undertaking.” Id. at 203, 144 S.Ct. 756. The same substantive message may alternatively qualify as state or private action, depending on the circumstances. For example, imagine that a school board president announces that the school board has lifted pandemic-era restrictions on public schools. See id. at 201–02, 144 S.Ct. 756. If the president delivers that message at a school board meeting, its state action. Id. But if he subsequently shares that message at a backyard barbecue with friends, its private action. Id. All told, its unlikely that an official is purporting to exercise the authority of his office when he makes a statement that merely “relates to” or “concerns information learned during” his public employment. Id. at 203, 144 S.Ct. 756 (cleaned up). Indeed, an official “might post job-related information for any number of personal reasons,” such as a “desire to raise public awareness.” Id.
To be sure, public officials can reduce ambiguity about whether their social-media posts are public or private. Officials may choose to display a disclaimer on their account, explaining that that their page is personal or that the accounts posts contain only personal views. Id. at 202, 144 S.Ct. 756. Such a label would entitle the official to a “heavy” presumption that “all of the posts on his page were personal.” Id. That presumption could only be rebutted through the most obvious exercises of state action: that is, conduct that only a state actor would be capable of doing, such as performing notice-and-comment rulemaking on a social-media platform. Id. at 203 & n.2, 144 S.Ct. 756. And, to the extent this presumption can be overcome, it would be on a post-by-post basis: one official-action post wouldnt invalidate the disclaimers presumption for all other posts on a page.
By contrast, posts from an account that clearly belongs to the government—like a page labeled “City of Port Huron”—presumptively speak for the government. Id. at 202, 144 S.Ct. 756. Officials may only access those pages by virtue of their state office, so its obvious that posts from such accounts will almost always be state action.
But when an account neither has a disclaimer nor belongs to the government, the lines are not as bright. Discerning whether a post exercises state authority will be a context-specific, totality-of-the-circumstances inquiry. But a few factors can guide courts undertaking this task. For instance, posts that expressly invoke an officials legal authority will likely be exercises of that authority. As will posts that have some legal consequence in the real world, like a “burn ban” issued on social media by our hypothetical fire marshal. Likewise, if a government official uses their staff or office funding to make a post, its more likely the post is made pursuant to state authority.
On the flip side, the mere subject matter of a post isnt enough to show that a post is state action—especially when the information contained in the post is available in other places (whether online or in-person). After all, any citizen can post about a school policy, road closure, or press conference. None of that requires authorization by state law. Indeed, when a state official posts about government happenings in a way that any other citizen could, plaintiffs will need to show that those posts were made pursuant to the officials job duties. Mackey, 106 F.4th at 561 (explaining that even if a State authorizes an official to engage in certain conduct, the conduct “will not qualify as state action” if the State “has permitted anyone to engage in the activity”).
III.
Because the Supreme Courts test changes the relevant inquiries, Lindke argues that he needs to further develop the record before a court can resolve his claims. We agree. The discovery he conducted couldnt have been tailored toward satisfying a legal standard that had eluded all the parties as well as this court. Thus, we remand this case to the district court for further proceedings. We now turn to consider what the remand might look like.
As an initial matter, the district court should determine whether Lindkes claim for declaratory and injunctive relief is moot. Freeds summary-judgment motion asserted that it was. As of 2021, Freed represented that his Facebook page had been deactivated for almost a year. He also testified that he wouldnt reinstate the page without assurances that he could operate it in his personal capacity. Of course, “a defendants voluntary cessation of a challenged practice does not deprive a federal court of its power to determine the legality of the practice.” Friends of the Earth, Inc. v. Laidlaw Envt Servs. (TOC) Inc., 528 U.S. 167, 189, 120 S.Ct. 693, 145 L.Ed.2d 610 (2000) (citation omitted). Because Freeds claim for mootness rests on his own voluntary decision to deactivate his Facebook page, he bears a “heavy burden of persuading” the court that the challenged conduct “cannot reasonably be expected to start up again.” Id. (cleaned up). So, if the relevant facts are unchanged from three years ago, Lindkes claim for forward-looking relief likely isnt moot. But its possible that the facts have changed in some way that would allow Freed to meet his burden of establishing mootness. We leave it to the district court to determine whether any such facts have emerged. 2
Additionally, Freed sought summary judgment on several grounds unrelated to the state-action question. The district court had no occasion to pass on these theories the first time around because it granted Freeds motion on state-action grounds. Lindke v. Freed, 563 F. Supp. 3d 704, 708 (E.D. Mich. 2021). But on remand, the district court should rule on these alternative theories before opening the door to additional discovery.
If the district court ultimately tackles the state-action question, it must give Lindke an opportunity to establish facts that became legally relevant only after the Supreme Court articulated its test. But the district court doesnt need to re-open discovery for issues on which Lindke already had a chance to develop the record the first time around.
Start with the first prong of the state-action test, which asks whether Freed had authority to speak for the state on the matters in question. Lindke, 601 U.S. at 198–200, 144 S.Ct. 756. Our prior test asked a different question: whether Freed had authority to run a Facebook page. Lindke, 37 F.4th at 1203. So, the relevant object of authority has changed from authority to run a Facebook page to authority to speak for the state. As explained above, this authority can come from written law, custom, or usage. And permissible sources of custom could range from use of social media by Freeds predecessors to any other patterns established by Port Huron. If the district court determines that Lindke hasnt had an opportunity to show that these sources granted Freed authority to speak on the states behalf, then discovery on this issue would be warranted.
If Freed lacked actual authority to speak for the state, then theres no state action, and Freed is entitled to summary judgment. The district court would proceed to prong two only if Freed had actual authority. This prong asks whether Freed purported to exercise that authority in specific posts. Lindke, 601 U.S. at 203, 144 S.Ct. 756. To be sure, Lindke has already discovered a trove of Freeds Facebook posts. But on this prong, context is everything, so a fact-intensive analysis of individual posts might be necessary. Id. Lindke might need additional discovery to develop facts that could be important to understanding the context of Freeds posts.
Lastly, even if discovery uncovers that some of Freeds posts were state action, a host of other First Amendment issues remain: what kind of forum Freeds social-media accounts are, what level of scrutiny his deletion or block decisions receive, and whether hes entitled to qualified immunity. We leave these interesting questions for another day.
* * *
We remand this case to the district court for further proceedings.
FOOTNOTES
1
. Of course, a grant of actual authority to speak on the states behalf need not mention social media as the method of speaking. See id. at 200, 144 S.Ct. 756. Thus, actual authority may exist even if the relevant laws or customs dont mention social media.
2
. The fact that Lindke has a live claim for damages doesnt tell us anything about whether his claim for injunctive relief is moot. A plaintiff must demonstrate a live controversy “separately for each form of relief sought.” Friends of the Earth, 528 U.S. at 185, 120 S.Ct. 693; see also Uzuegbunam v. Preczewski, 592 U.S. 279, 284, 141 S.Ct. 792, 209 L.Ed.2d 94 (2021) (observing that claim for injunctive relief was moot even though claim for nominal damages remained live).
THAPAR, Circuit Judge.