McKEAGUE, Circuit Judge.

NOT RECOMMENDED FOR PUBLICATION
File Name: 20a0029n.06 ON APPEAL FROM THE UNITED STATES DISTRICT COURT FOR THE SOUTHERN DISTRICT OF OHIO

OPINION

BEFORE: McKEAGUE, GRIFFIN, and NALBANDIAN, Circuit Judges.

McKEAGUE, Circuit Judge. Keith Crabbs had his DNA collected after he was acquitted for voluntary manslaughter. The Sheriff's Office in Franklin County, Ohio, where Keith was tried, arguably has a policy that allows for post-acquittal DNA swabs from felony arrestees. All the evidence before us seems to indicate that the Sheriff's Office employees swabbed Crabbs' DNA because they thought he was a felony arrestee. But he was not; he was actually arrested for a bond revocation. So now the question is this: can Franklin County Sheriff Zach Scott, in his official capacity, still be held liable for the DNA swab? The district court said no. We agree and AFFIRM.

I. Background

Back in December 2010, Keith Crabbs turned himself in for voluntary manslaughter, a first-degree felony under Ohio law. Ohio Rev. Code § 2903.03(c). He spent a night in jail before being released on bond. Then in March 2012, his case went to trial. One day, during his trial, Crabbs arrived late and was caught arguing with a witness just outside the courthouse, so the judge revoked his bond and had him placed in protective custody. "So far as the record shows, violating bond is not a standalone crime." Crabbs I, 786 F.3d at 430. But under the bond violation, Crabbs had to stay in the Franklin County jail throughout the rest of the trial.

The Franklin County Sheriff's Office has certain procedures for arrestee intake. At issue here is the procedure for DNA collection. According to the county's intake procedures, individuals arrested for certain listed offenses—including all felonies—are processed through the Identification Bureau. Most of the data points collected are routine: name, date of birth, address, eye and hair color, and others. In addition to these more routine matters, Franklin County also collects DNA samples from certain arrestees.

Sometimes these DNA collections are required under state law. In 2011 (after Crabbs was released on bond but before he was rearrested for violating the bond), Ohio started requiring "a person who is eighteen years of age or older and who is arrested on or after July 1, 2011, for a felony offense [to] submit to a DNA specimen collection procedure administered by the head of the arresting law enforcement agency." Ohio Rev. Code § 2901.07(B)(1)(a). Shortly before this law was to go into effect, Franklin County Sheriff's Office employees received a memo from Chief Deputy Mark J. Barrett. The memo discussed the Ohio statute and issued a directive: "[O]n or after July 1, 2011, for purposes of [Ohio Rev. Code] section 2901.07(B)(1), we (Sheriff's Office) will only collect DNA specimens from persons eighteen years of age or older arrested on a felony charge . . . ." In addition, the Sheriff's Office intake procedures identify which arrestees are processed through the Identification Bureau, including "All felons" and persons arrested for certain misdemeanors. Bond-revocation arrestees are not mentioned.

Also relevant is the question of when qualifying arrestees will have their DNA swabbed. The Ohio statute directs law enforcement to collect the DNA specimens "during the intake process." Id. Normally this means that arrestees will have all their DNA collected when they are first brought into jail. But that can't always happen. Sometimes arrestees are unruly or uncooperative, or some other factor makes it impossible for the ID technician to collect the DNA. To deal with these situations, the Sheriff's Office allows for "ID holds." An arrestee with an ID hold will instead have his identifying information collected as part of the release process. Although an ID hold is not enough on its own to keep an inmate in jail, if an arrestee with an ID hold is about to be released, the supervisor is directed to order the completion of that arrestee's ID processing as soon as possible. None of the evidence of Franklin County's ID hold policy makes any distinction between acquitted and convicted inmates.

Here, under the judge's order, Crabbs was to be kept in protective custody until his trial was over. Jail records indicated that he had been charged with voluntary manslaughter and had his bond revoked. But under "Charge Data," the entry listed only the voluntary-manslaughter charge. For whatever reason, Crabbs was placed on an ID hold, and he was taken to protective custody without having had his DNA swabbed. The Identification Bureau's paperwork accompanying the ID hold said that he needed to be "processed on case #10CR7478," which was the manslaughter charge. The jail's electronic file also pointed to the manslaughter charge as the case under which Crabbs was to be processed. The trial continued, and the jury ended up acquitting him of the voluntary-manslaughter charge; so he could be released from jail. Because he still had the ID hold, an ID technician took a DNA specimen from Crabbs (over his objection) as part of his release process. The notes accompanying his discharge again suggested that the DNA was collected in connection with the underlying felony charge: "DNA collected 3-16-12. Processed on charge 10CR7478," which is the manslaughter charge.

Crabbs then sued. He filed a § 1983 action against Sheriff Zach Scott, among other defendants, claiming that the DNA swab violated his constitutional rights. Originally, Scott claimed sovereign immunity as a defense: he argued that his DNA collection policy, as applied to Crabbs, merely carried out what the Ohio statute required. This was the issue in the first appeal before our court. We held that Sheriff Scott was not shielded by sovereign immunity. Crabbs I, 786 F.3d at 430-31. The statute required DNA collection only from felony arrestees, and Crabbs was not a felony arrestee. Id. True, Crabbs had previously been arrested for a felony—and, indeed, he would not have been out on bond had he not been arrested for a felony. However, as we noted:

So far as the record shows, violating bond is not a standalone crime. Nor does violating bond make the resulting arrest one "for" the underlying charge simply because the bond could not have been initially imposed without that charge. That would defy how we normally talk about arrest. We would not say that a prisoner recaptured after escape was "arrested for" the offense that first landed him in jail. Because the new facts leading to the prisoner's second arrest have nothing to do with his original crime, he was not arrested "because of" or "on account of" that crime.

The issue now is whether Scott can be liable for the swab of Crabbs' DNA under a municipal policy. On a motion for summary judgment, the district court said no. The court determined that even if Sheriff Scott's office had an "official policy authorizing the post-acquittal collection of DNA from felony arrestees who were on an ID hold," and even if that alleged policy were unconstitutional, it still could not give rise to liability. The alleged policy, like the Ohio statute, would apply only to felony arrestees, which Crabbs was not. Moreover, the district court held that even if the Sheriff's Office employees thought they were implementing the policy when they swabbed Crabbs, that mistake was not reasonably foreseeable, so the policy was not the proximate cause of the swab. Accordingly, the district court granted summary judgment in favor of Scott. Crabbs then appealed.

II. Standard of Review

This case comes to us from a grant of summary judgment. Accordingly, we review de novo. Range v. Douglas, 763 F.3d 573, 587 (6th Cir. 2014). Summary judgment is proper when "there is no genuine dispute as to any material fact and the movant is entitled to judgment as a matter of law." Fed. R. Civ. P. 56(a). We review the facts "in the light most favorable to the party opposing the motion." Matsushita Elec. Indus. Co. v. Zenith Radio Corp., 475 U.S. 574, 587 (1986) (quoting United States v. Diebold, Inc., 369 U.S. 654, 655 (1962) (per curiam)). However, the nonmovant must still provide enough evidence "for a jury to return a verdict for that party." Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 249 (1986).

III. Discussion

There are essentially two ways in which Sheriff Scott could be liable. First, he could be liable if his office's policy authorized the collection of Crabbs' DNA. Second, he could be liable if it was reasonably foreseeable that the policy would cause the collection of Crabbs' DNA, even if technically the policy did not allow the swab. Before moving on, it's worth mentioning that the district court did not conclusively find that the Franklin County Sheriff's Office had any official policy regarding post-acquittal DNA collection. In the same way, we do not purport to hold that such a policy exists; we merely assess whether any alleged policy could create liability for Scott. With that said, we proceed to analyze the two theories of Scott's potential liability in turn.

A. Did the Policy Allow the DNA Swab?

Sheriff Scott could be liable if (1) his office had a formal policy authorizing the swab of Crabbs' DNA and (2) that policy was unconstitutional. Crabbs sued Scott under 42 U.S.C. § 1983. That statute makes it illegal for any "person" to, "under color of any statute, ordinance, regulation, custom, or usage, of any State . . . subject[], or cause[] to be subjected, any citizen of the United States or other person within the jurisdiction thereof to the deprivation of any rights, privileges, or immunities secured by the Constitution." Id. A municipality can be liable under § 1983. Monell v. Dep't of Soc. Servs., 436 U.S. 658, 694 (1978). But to hold a municipality liable, it is not enough to show that a municipal employee violated someone's rights; there is no respondeat superior municipal liability under § 1983. Id. at 691. Instead, plaintiffs must point to a municipal "policy or custom" and show that it was the "moving force" behind the constitutional violation. Id. at 694. Crabbs is suing Scott in his official capacity as Franklin County Sheriff, so the claim "is, in all respects other than name, to be treated as a suit against the entity." Kentucky v. Graham, 473 U.S. 159, 166 (1985).

Here, Scott has no formal policy permitting the taking of Crabbs' DNA, because Crabbs was not a felony arrestee at the time the swab was taken. According to the 2011 Memo from Chief Deputy Barrett, Sheriff's Office employees could "only collect DNA specimens from persons . . . arrested on a felony charge." And a bond-violation arrest is not the same thing as a felony arrest in the statutory context, as we noted in Crabbs I. 786 F.3d at 430. Crabbs' "felony voluntary-manslaughter charge was irrelevant to his second arrest." Id. at 431. If Crabbs was not "arrested on a felony charge" under the statute's definition, then it follows that he also was not "arrested on a felony charge" under a memo that uses the exact same phrasing. Also, the Sheriff's Office intake procedures listed the offenses for which arrestees would be processed through the Identification Bureau—bond revocation was not listed. Therefore, the alleged policy did not give Sheriff's Office employees any authority to swab Crabbs' DNA.

Going beyond the written documents, Crabbs points to Sheriff Scott's earlier position in this litigation and his deposition testimony. Both reveal, according to Crabbs, that Scott thought swabbing Crabbs' DNA was required under the policy. According to this argument, we should interpret the policy to mean what the Sheriff, who made the policy, said it means. Now, Monell does contemplate municipal liability for the actions of officials "whose edicts or acts may fairly be said to represent official policy." 436 U.S. at 694. So it's at least possible that, if Scott understood the policy to cover Crabbs and then applied it to him, then that application could amount to office policy. In Pembaur v. City of Cincinnati, a plurality of the Supreme Court recognized that municipal liability should attach when the decision comes from an official who "possesses final authority to establish municipal policy with respect to the action ordered." 475 U.S. 469, 481 (1986) (plurality); see also Paige v. Coyner, 614 F.3d 273, 284 (6th Cir. 2010) (citing the Pembaur plurality). But there, liability attached only because the policymaker's "decision directly caused the violation" of constitutional rights. Pembaur, 475 U.S. at 484; see Bd. of Cty. Comm'rs v. Brown, 520 U.S. 397, 406 (1997) (noting that, in Pembaur, "[n]o questions of . . . causation arose").

Therein lies the problem. Scott did not take the swab, nor did he direct anyone to do so. So his involvement was limited to the policy. Neither Scott's deposition testimony nor his litigation position as to the meaning of the policy is what "directly caused" the swab of Crabbs' DNA. By the time Scott took a litigation position and was deposed, the deed had already been done—Crabbs' DNA had already been taken. Even assuming Scott had the type of final policymaking authority contemplated by Pembaur, there's no way his after-the-fact interpretation of the policy, correct or incorrect, could have "directly caused" something that had already happened. See Pembaur, 475 U.S. at 481, 484. So really, Crabbs is saying that, when it comes to interpreting the policy, and thus why the swab was taken, we should trust Scott's reading. But we are not bound by Scott's after-the-fact interpretation of how the alleged policy should or should not have applied to Crabbs. Because we find that Crabbs was not a felony arrestee, the alleged policy simply did not apply to him, and what Scott says after the DNA swab doesn't change that conclusion.

Crabbs also points to the deposition testimony of Gary Goodrich, the sergeant on duty the night Crabbs' DNA was taken. In the deposition, Crabbs' attorney described the facts as he understood them: after Crabbs was acquitted, he was not allowed to leave the jail until his DNA had been collected. The attorney then asked Sgt. Goodrich whether, if the facts were as he described them, that would be consistent with "standard procedure." Sgt. Goodrich responded, "[i]n a nutshell, yes." But this statement refers to the Sheriff's Office standard procedure only as Goodrich understood it. It turns out he was wrong; "standard procedure" would not have permitted the swab of Crabbs' DNA. What Sgt. Goodrich doesn't point to is any statement from a superior officer directing him to apply the policy to Crabbs. And Crabbs makes no argument that Sgt. Goodrich had the type of final policymaking authority for his individual actions to constitute municipal policy under Pembaur. So Sgt. Goodrich is simply another municipal employee who misread the policy. All in all, Crabbs cannot establish that there was a policy requiring—or even permitting—the swab of his DNA.

B. Can Scott Still Be Liable for the Mistaken Application of the Policy?

The district court's analysis did not end when it found that the policy did not allow the Sheriff's Office to collect Crabbs' DNA. The court instead proceeded to consider whether Scott could still be liable for his employees' incorrect application of the policy. In doing so, the district court made several assumptions: "(1) that the Sheriff's Office has a policy permitting the post-acquittal collection of DNA from felony arrestees on ID holds, (2) that such a policy was unconstitutional, and (3) that the [employees who collected Crabbs' DNA] misapplied such a policy to Plaintiff as the authority for the collection of his DNA." We make the same assumptions here. And we determine that no reasonable juror could find that the alleged policy proximately caused Crabbs' injury.

Even if the policy was not followed here, Crabbs must still show that the policy was the "moving force" behind the injury he suffered. See Monell, 436 U.S. at 694. In other words, he must show a "direct causal link" between the policy and the DNA swab. See Powers v. Hamilton Cty. Pub. Def. Comm'n, 501 F.3d 592, 607 (6th Cir. 2007). For § 1983 claims, this causation analysis is informed by "traditional tort concepts of causation." Id. at 608. Accordingly, to give rise to Monell liability, the policy must have been the factual and proximate cause of Crabbs' injury. Id.

The first element of causation under § 1983 is factual causation. Factual causation "is typically assessed using the 'but for' test, which requires us to imagine whether the harm would have occurred if the defendant had behaved other than it did." Id. Here, the district court assumed that Scott had a policy allowing the Sheriff's Office to collect DNA samples from felony arrestees after they had been acquitted. Given this assumption, the factual-causation issue is this: if Scott did not have a post-acquittal DNA collection policy, would Crabbs' DNA sample have been taken? The answer is clearly no; the parties do not even dispute this point. The Sheriff's Office employees all thought that Crabbs was a felony arrestee, so they thought the policy applied to him. The court records all suggest that they thought Crabbs was being processed for the felony voluntary-manslaughter charge. Under "Charge Data," the jail records listed the manslaughter charge, not the bond revocation. Both the jail records and the Identification Bureau records stated that Crabbs needed to be "processed on" only the manslaughter charge. We can therefore conclude that the Sheriff's Office employees took Crabbs' DNA simply because they thought he was a felony arrestee eligible for DNA collection under the policy. In other words, but for the policy, Crabbs' DNA would not have been swabbed. So factual causation is satisfied.

But factual causation is not enough. Crabbs must also prove the second element, proximate causation. The cornerstone of the proximate-cause analysis is foreseeability; we ask "whether it was reasonably foreseeable that the complained of harm would befall the § 1983 plaintiff as a result of the defendant's conduct." Id. at 609. Foreseeability is measured as of the time of the defendant's wrongdoing—thus, we avoid hindsight bias. See 1 Restatement (Third) of Torts: Liability for Physical and Emotional Harm § 29 cmt. j (Am. Law Inst. 2010) (noting how proximate cause excludes liability for harms that were not foreseeable "at the time of the actor's tortious conduct"). So here, the proximate-cause issue is whether, when Sheriff Scott's office passed the alleged policy, it was reasonably foreseeable that the policy could be mistakenly applied to someone like Keith Crabbs—that is, someone who was arrested for a felony, then released on bond, then rearrested for violating that bond, then placed on an ID hold, then tried for and acquitted of the underlying felony, then swabbed for a DNA sample upon release.

Proximate cause can be resolved by the court if the plaintiff has not provided enough evidence for a reasonable juror to conclude that the relevant injury was reasonably foreseeable. See Hunley v. DuPont Auto., 341 F.3d 491, 499 n.5 (6th Cir. 2003); see also Anderson, 477 U.S. at 249. That is the case here. Crabbs cannot point us to any other instances where the Sheriff's Office collected DNA from non-felony arrestees. Nor can Crabbs show us that anyone else—even a felony arrestee—has had their DNA collected by the Sheriff's Office after being acquitted. Indeed, to show reasonable foreseeability, Crabbs essentially just reiterates that the Sheriff's Office employees thought they were applying the policy. But the issue is no longer whether Scott's employees thought they were applying the policy; that was the question under factual causation. The issue now is whether their doing so was reasonably foreseeable. And on that issue, Crabbs does not point us to any additional evidence. Without more, a reasonable juror could not conclude that the policy was the proximate cause of Crabb's injury. Therefore, summary judgment in favor of Scott was proper.

Our assumption that the underlying policy was unconstitutional does not change this outcome, because even an unconstitutional policy must proximately cause the plaintiff's injury. To be sure, in cases involving a municipal actor departing from a municipal policy, the underlying policy is usually constitutional on its face. For example, in Berry v. City of Detroit, the city's formal deadly-force policy was facially sound, but the plaintiff was alleging a systemic misapplication of that policy such that police officers were regularly using deadly force in an unconstitutional manner. 25 F.3d 1342, 1345-46 (6th Cir. 1994). Crabbs has not pointed us to any cases where the policy being incorrectly applied was facially unconstitutional.

But we do not go so far as to say that a municipality is always liable if its unconstitutional policy leads to a deprivation of rights, regardless of how unreasonably the policy is applied. Doing so would conflict with traditional tort principles. In tort law, even if a defendant has breached her legal duty, she is not liable unless her breach proximately caused the alleged injury. See 1 Restatement (Third) of Torts: Liability for Physical and Emotional Harm § 29 (2010) ("An actor's liability is limited to those harms that result from the risks that made the actor's conduct tortious."). So too for breaches of constitutional duty. See Powers, 501 F.3d at 608. Limiting the scope of liability here is consistent not only with tort doctrine, but also with basic fairness; municipalities should not have to anticipate and preemptively account for every possible incorrect reading of their policies in order to avoid Monell liability. Otherwise we would be drifting too close to respondeat superior liability for § 1983 claims against municipalities, an idea that has been repeatedly and resoundingly rejected. See Brown, 520 U.S. at 403. Accordingly, we would hold Scott liable for Crabbs' injury only if Scott's alleged policy was both the factual and proximate cause of the injury—even though we are assuming that the underlying policy was unconstitutional. See Powers, 501 F.3d at 608. And Crabbs has not presented sufficient evidence for a reasonable juror to find in his favor on proximate cause.

IV. Conclusion

For those reasons, we AFFIRM the judgment of the district court.

NALBANDIAN, Circuit Judge, dissenting. The majority looks at this case the wrong way. The question is not whether, when the Sheriff's office passed the policy, "it was reasonably foreseeable that the policy could be mistakenly applied to someone like Keith Crabbs." Maj. Op. at 10 (emphasis added). Rather, the correct question is simply whether the policy applied to someone like Crabbs. It did. Sheriff Scott admitted that the policy applied to Crabbs. And he never claimed that his employees made a mistake under the policy. So the majority bases its question on a false premise—that Sherriff Scott's employees in fact made a mistake. But when we frame the question correctly, focusing on the district court's assumptions and Sheriff Scott's admissions, we should conclude that Sheriff Scott's DNA collection policy caused the collection of Crabbs's DNA.

To be sure, we find ourselves in a strange situation: on the record in front of us, with the district court's assumptions and Sheriff Scott's admissions, the Monell requirements go undisputed. Sheriff Scott admitted that he had a policy of collecting DNA from inmates arrested on a felony charge. The district court assumed that Sheriff Scott's DNA policy is unconstitutional because it arguably allowed for the collection of DNA after a jury acquitted an inmate. The Sheriff's Office indisputably took Crabbs's DNA post-acquittal. And Sheriff Scott said that his office collected Crabbs's DNA under his policy. So although I am skeptical of Crabbs's Monell claim and the underlying constitutional issues, the district court's assumptions and Sheriff Scott's admissions force our hand, in my opinion. In this strange case—at this unique stage of the litigation—Crabbs's Monell claim should move forward. I would reverse.

I.

The only question before us is whether Sheriff Scott's policy of collecting DNA proximately caused the collection of Crabbs's DNA. Sheriff Scott asks us to adopt the district court's reasoning, arguing that his staff misapplied his policy, so we cannot hold him responsible under Monell for his staff's mistake. If we did, Sheriff Scott argues that we would create respondeat superior liability, which the Supreme Court has forbidden under Monell.

Sheriff Scott is both right and wrong. He is correct that, under Monell, an employee's mistake cannot transform an employer's good policy into a bad one. But Sheriff Scott is wrong on that rule's application here for two reasons. First, Sheriff Scott's policy is not "otherwise sound." The opposite is true: the district court assumed that Sheriff Scott's DNA collection policy is unconstitutional. And second, there was no mistake. Sheriff Scott admitted that his policy of collecting DNA caused the collection of Crabbs's DNA.

A.

"Under the doctrine of respondeat superior, an employer is liable, despite having no fault whatsoever, for the acts of its employees taken within the scope of their employment." Hamilton v. Carell, 243 F.3d 992, 1001 (6th Cir. 2001). This type of strict liability does not apply to constitutional injuries, such as § 1983 claims or Monell claims. Powers v. Hamilton Cty. Pub. Defender Comm'n, 501 F.3d 592, 607 (6th Cir. 2007) (citing Monell v. Dep't of Soc. Servs., 436 U.S. 658, 691 (1978)) ("A municipality cannot be liable for the constitutional torts of its employees; that is, it cannot be liable on a respondeat superior theory."). In other words, "a local government may not be sued under § 1983 for an injury inflicted solely by its employees or agents." Monell, 436 U.S. at 694.

So to make sure that a municipality is liable only for constitutional injuries that it caused, Monell limits liability to situations "where the plaintiff establishes that the municipality engaged in a 'policy or custom' that was the 'moving force' behind the deprivation of the plaintiff's rights." Powers, 501 F.3d at 607 (quoting Monell, 436 U.S. at 694). Applying this principle, we have developed a line of "mistake" cases, which explain that "[t]here can be no municipal liability where 'an otherwise sound program has occasionally been negligently administered.'" Graham ex rel. Estate of Graham v. County of Washtenaw, 358 F.3d 377, 385 (6th Cir. 2004) (quoting City of Canton v. Harris, 489 U.S. 378, 391 (1989)).

This rule makes sense. "The fact that individual actors [or employees] may 'occasionally make mistakes . . . says little about the . . . legal basis for holding the [County] liable.'" Id. (quoting Harris, 489 U.S. at 391) (ellipses and second alteration in original). Think of it this way. In a "mistake" case, the employer has a good, constitutionally-sound policy—i.e., it was doing what it was supposed to do. But the plaintiff still wants to hold the municipality responsible for some independent mistake made by an employee. Said differently, the plaintiff tries to turn the municipality's good policy into constitutional liability, even when it did nothing wrong. That is precisely the type of respondeat superior claim that Monell forbids.

For example, "mistake" cases often attempt to hold a municipality responsible for the inaction of the municipality. Take Doe v. Claiborne County, 103 F.3d 495 (6th Cir. 1996), which we relied on in Powers. See 501 F.3d at 607. The plaintiff in Doe wanted to hold a school board responsible for the sexual abuse she suffered at the hands of a teacher. 103 F.3d at 500. But we refused to hold the school board responsible for the isolated and tortious conduct of an employee. Yet in doing so, we recognized an important point: "Doe does not claim that the School Board had a custom [or policy] of affirmatively condoning sexual abuse." Id. at 508. Of course it didn't—what school board would? Still, this absence of a policy shielded the school board from Monell liability. But imagine if the opposite was true; that would be a troublesome policy. And we would have little hesitation holding the school board responsible for sexual abuse if it had an official policy telling its teachers to engage in sexual misconduct.

Other "mistake" cases share this same feature as Doe: a lack of any troublesome policy. In Harris, for example, the plaintiff wanted to hold the prison responsible for an inmate's death. The plaintiff claimed that a nurse's medical negligence was to blame. Harris, 489 U.S. at 386. The jail, of course, had no policy telling its nurses to engage in medical malpractice. Id. ("There can be little doubt that on its face the city's policy regarding medical treatment for detainees is constitutional."). And any medical mistake was isolated to just the nurse. But imagine if the jail had an official policy telling its nurses to provide substandard care. If it did, Harris would be a different case that would unquestionably create municipal liability.

The same is true in Berry v. City of Detroit, another "mistake" case. 25 F.3d 1342, 1348 (6th Cir. 1994). In Berry, the plaintiff wanted to hold the City of Detroit responsible for a police officer's excessive force when he shot and killed the plaintiff's son. Id. at 1343. But we recognized the problem with the plaintiff's Monell claim: "No one disputes that [the city's deadly force] policy, as written, is constitutional." Id. at 1345. The plaintiff's claim failed, then, because we are "prohibited from making the master liable for the tort of the servant, unless it goes through the 'custom or policy' analysis and ties in a city in that manner." Id. at 1348.

But in rejecting the claim, we gave a helpful example of a hypothetical policy that would create Monell liability: "To illustrate, if a city . . . announced [a] policy that deadly force could be used against fleeing felons, regardless of the circumstances, one would have little problem holding that city responsible if a person is wrongfully injured or killed by a police officer using deadly force." Id. In this situation, liability would flow to the city for its unconstitutional policy. "On the other hand, if a city has a proper policy [on the use of force] and it is not followed . . . [by] one of that city's officers[,]" Monell protects the city from liability. Id. In this latter situation, the officer deviated from the city's otherwise constitutionally-sound policy.

The Supreme Court evaluated a similar excessive force case in City of Oklahoma City v. Tuttle, 471 U.S. 808, 823-24 (1985), when the plaintiff wanted to hold Oklahoma City responsible for an officer's decision to shoot the plaintiff's husband. The plaintiff claimed the shooting amounted to excessive force. The Supreme Court explained, however, that the city had no unconstitutional policy on the use of force. See 471 U.S. at 823-24; see also, e.g., Brown v. Cuyahoga County, 517 F. App'x 431, 436 (6th Cir. 2013) (explaining that, even if correctional officers had "blanket parties"—i.e., "a euphemistic term for the beating of prisoners"—there was no policy of beating prisoners). But to be sure, if Oklahoma City had an official policy telling its officers to use excessive force (or if Cuyahoga County had an official policy telling its officers to beat prisoners), each plaintiff would have had a very different (and much easier) case against the municipality.

In sum, these cases show that a municipality's liability depends on the existence of a good policy versus a troublesome policy. And our "mistake" cases protect a municipality only when it has an "otherwise sound policy." It is this existence of a constitutional policy (or, at least, the lack of an unconstitutional policy), that saves a municipality from liability when its employee makes a mistake. But on the other hand, if a municipality has an unconstitutional policy on its face, such as encouraging excessive force, sexual abuse, beating prisoners, or providing substandard care, there is no problem holding the municipality liable under Monell. See 436 U.S. at 694-95 (explaining that a bad policy, which compelled pregnant women to take early and unpaid maternity leave, "unquestionably involve[d] [an] official policy as the moving force of the constitutional violation").

Here, Crabbs's appeal falls under the latter scenario: the district court assumed Sheriff Scott implemented an unconstitutional policy of collecting DNA—i.e., it was a bad policy on its face. Indeed, throughout this litigation, the district court has maintained this assumption to avoid "a thorny constitutional question of first impression[.]" (R. 129 at 5 (citation omitted).) So at this stage of the litigation, the district court's assumption means that Sheriff Scott had a written policy telling his staff to engage in constitutional violations. This point alone distinguishes this appeal from our "mistake" cases under Monell.

Now, of course, this does not mean that Crabbs would have ultimately won his case if we did reverse. As I explained, this is a unique case with a strange record. The district court's assumptions might have ultimately been wrong: a jury could have decided that no official policy existed, or the district court could have determined that the policy was perfectly constitutional. But based on the record in this appeal, Crabbs's Monell claim should continue.

B.

Sheriff Scott's argument should have failed for another reason: there was no mistake when the Sheriff's employees implemented the policy, contrary to the majority's interpretation. Sheriff Scott testified that his policy of collecting DNA caused the collection of Crabbs's DNA. His staff confirmed this. And his office paperwork shows that his staff followed policy. Indeed, in a previous appeal to our court, Sheriff Scott openly admitted this: "Sheriff Scott . . . collect[ed] Crabbs'[s] DNA on March 16, 2012, pursuant to a 2011 DNA collection policy his office adopted . . . ." (No. 14-4068, Reply Br. at 1 (emphasis added).)

In contrast, the record contains no evidence that Sheriff Scott or his staff considered the collection of Crabbs's DNA a mistake. Even though the majority implies we cannot assess foreseeability based on hindsight, it does by focusing on the legal mistake of Crabbs's arrestee label. But unlike the legal definition, the policy itself posed no mistake. This lack of evidence of a mistake is enough to survive summary judgment: "[if] [n]othing in the record, and no evidence supplied by the defendants, contradicts [a plaintiff's] record-supported claim [of causation]," the plaintiff "satisfies the direct-cause component of a Monell claim." Hardrick v. City of Detroit, 876 F.3d 238, 246 (6th Cir. 2017).

Our opinion in Crabbs I does not change the result. See Crabbs v. Scott, 786 F.3d 426 (6th Cir. 2015) ("Crabbs I"). Nor can it change the record evidence, including Sheriff Scott's own testimony, that his office collected Crabbs's DNA under his policy. Sheriff Scott does not cite (nor could I find) any authority when a court dismissed a Monell claim for lack of causation when the policymaker admitted it had an official policy—and conceded that the policy caused the alleged constitutional injury. This makes sense, too. So it is unsurprising that the DNA collection policy caused the collection of Crabbs's DNA.

Sheriff Scott would prefer that we just ignore the evidence. He argues that his testimony is "not relevant" because, when he testified, he was "under the assumption" that Crabbs was a felony arrestee when he returned to jail and his DNA was collected. (See R. 133 at 7.) But this is precisely the point. When his staff took Crabbs's DNA, Sheriff Scott believed that his staff followed policy. At no point did he testify that his staff made a mistake—or that his policy did not apply to Crabbs. Instead, Sheriff Scott explained that his staff was "within policy" when it took Crabbs's DNA before releasing him from jail. (See R. 67-1 at 68-70.) And he consistently testified that his office could take Crabbs's DNA. (See id. at 57 ("He still was in the category of a person who was arrested for a felony."); id. at 59 ("He was arrested. He's in the category.").) Indeed, Sheriff Scott provides no reason why his staff would collect Crabbs's DNA other than that they were following his DNA collection policy.

Sheriff Scott's missing evidence is glaring when we compare our record to classic "mistake" cases under Monell. Those cases include evidence of actual mistakes—as identified by the policymaker or municipality. And once the municipality identifies a mistake, it generally fixes the situation. See, e.g., Colvin v. Caruso, 605 F.3d 282, 287-88, 292 (6th Cir. 2010) (explaining that once the supervisor learned of "multiple mistakes in administering the kosher-meal program," which caused the denial of kosher food to a Jewish inmate, the supervisor told employees to fix the problem "as soon as possible"); Cardinal v. Metrish, 564 F.3d 794, 802 (6th Cir. 2009) ("[T]he evidence established that as soon as the [warden] was made aware of [a mistake that caused] the plaintiff's situation, she instituted a transfer to a facility that could accommodate plaintiff's kosher needs and security level."); Graham, 358 F.3d at 380, 384-85 (detailing a nurse's alleged mistakes when providing medical care to an inmate).

But here, no one thought it was important that Crabbs lost his bond. (See, e.g., R. 67-3 at 53-54 (explaining how the Sheriff's Office repeats the intake process, even if an inmate just lost his bond, each time the inmate "come[s] back into the building").) The district court recognized this: "it is clear from Sheriff Scott's deposition testimony that neither he, nor, frankly anyone whose testimony was presented to this Court, appreciated the fact that [Crabbs] was not arrested for a felony . . . at the time his DNA was collected." (R. 155 at 3 n.3.) Instead, the driving fact that caused the collection of Crabbs's DNA was his arrest for felony manslaughter. And so, unsurprisingly, Sheriff Scott admitted that his staff followed policy when it took Crabbs's DNA. Our post-hoc observation about Crabbs's technical status does not retroactively change this fact. Nor can it transform a policy-driven decision into a mistake.

II.

In sum, I find it foreseeable that Crabbs would be arrested, released on bond, re-arrested for violating his bond, and placed on an ID hold. It was foreseeable that when officers re-arrested Crabbs, Sheriff Scott's employees would label Crabbs under the same case number as his felony arrest, because he violated the bond in place from his felony arrest. Sheriff Scott admitted that Crabbs fell under the policy and that the policy classified him as a felony arrestee. And there is no evidence that the officers made any factual mistake under the policy.

We should read the policy and interpret it as the employees foreseeably would on the day they processed Crabbs back to jail. And when I do, I see that a jury could find it reasonably foreseeable that employees would label Crabbs as a felony arrestee and take his DNA. Indeed, the policy required classifying Crabbs as a felony arrestee, given the options. So while it may ultimately be true that Sheriff Scott did nothing wrong—and his policy was constitutionally sound—at a minimum, there remains a disputed issue of fact on causation. And a reasonable jury could find that Sheriff Scott's policy on DNA collection caused his staff to collect Crabbs's DNA. As a result, summary judgment was inappropriate. I respectfully dissent.