OPINION

ROGERS, Circuit Judge.

Plaintiff Candido Romo was sitting in the driver's seat of a parked car, intoxicated, when he was approached by defendant Officer Jeff Largen. Largen arrested Romo for operating a vehicle while intoxicated, assertedly based on Largen's observation of Romo's having driven the car minutes earlier. Romo contends that he had not been driving. Romo filed a § 1983 suit claiming that Largen violated his constitutional rights and committed several intentional torts. The district court denied Largen's qualified-immunity motion for summary judgment, finding among other things that a jury could disbelieve Largen's claim that he had previously seen a very similar car being driven in violation of traffic laws. Without such prior observation, Largen lacked probable cause to arrest Romo. Because this is an interlocutory appeal, we are bound by the district court's finding that a genuine dispute of material fact existed as to whether Largen had observed such activity, in the absence of objectively incontrovertible evidence that he had. We therefore affirm the district court's denial of qualified immunity on Romo's § 1983 claim. The district court also properly permitted some of Largen's state-law claims to proceed.

On November 6, 2010, Candido Romo was supposed to meet his brother and brother-in-law at a bar. However, because the bar's parking lot was full, he left his vehicle, a silver Dodge Ram pickup, at a restaurant across the street. Romo and his brother-in-law began drinking, while his brother served as the designated driver. They drove in Romo's brother's truck to several bars, and Romo consumed approximately nine beers over the course of the night. At the end of the night, Romo's brother left Romo back at his vehicle. Although it was 35 degrees outside, Romo insisted on spending the night in his truck, instead of staying with his relatives or being driven home, because he was afraid the truck would be vandalized. To ensure that Romo would not drive drunk, his brother took Romo's keys, planning to return them in the morning when the three met up again for breakfast. Romo sat in his truck, leaned on his center console, and fell asleep.

Approximately forty-five minutes later, Romo awoke to a tapping on his window. Because Romo's car had power windows and he did not have the keys, he opened the door to see Officer Jeff Largen standing outside. Largen asked Romo what he was doing, and Romo responded that he was "waiting this out." Over the next ten minutes, Largen asked Romo about his night, how he was doing, and whether and how much he had been drinking. Finally, Largen asked "do you know why I pulled you over?" Romo responded by stating that the officer did not pull him over because he was not driving. Largen asked Romo if he had passed a semi tanker by the railroad tracks, and Romo responded that he had "been here for a while." Romo also told Largen that his brother had taken his keys some time before. At this point, Largen activated his microphone and offered to "start this conversation over" because "the more you lie to me, the more you're likely to go to jail." Largen then continued to accuse Romo of passing a semi tanker at a railroad crossing and Romo repeatedly answered "I don't know what you're talking about."

After this line of questioning proved unfruitful, Largen asked Romo to step out of the truck and administered sobriety tests. Romo responded by stating again that he was not driving, saying "my keys aren't even in the trunk."1 Largen did not respond to this statement. Largen took Romo through several sobriety tests and, after Romo failed, Largen arrested Romo for operating a vehicle while intoxicated. They then had the following exchange:

After putting Romo in his car, Largen asked if Romo wanted his wallet or his phone from the truck, and Romo declined. Largen then stated: "Okay. Do you have — you don't have the keys on you. Are they in the vehicle?" Romo responded: "Umm, yeah. I've been sitting there for a while."

As Largen explained why he was arresting Romo, Romo continued to protest his innocence. Largen responded by telling Romo to "[s]ave it for the Judge." Romo promised to do so and asked Largen, who had been rifling through the car, if he had found the keys. Largen responded that Romo had said that he did not want the car locked, and Romo stated that he had given the keys to his brother over an hour previously. Largen then waved his hand over the car's hood without touching it and said "I'm sure. That's why the engine's still warm to the touch." Largen then drove Romo to the jail. All charges against Romo were later dismissed.

Largen filed a police report that described the incident. In the police report, Largen claimed to have seen in his rearview mirror "a tan Dodge Ram pass a semi tanker in the railroad crossing no passing zone." He stated that he "turned on the truck and then pulled behind it" and that the "truck had pulled into the parking lot behind Big T and stopped." The police report noted that Romo had denied driving and had stated "I wasn't driving my kids aren't even in the car." In the police report, Largen claimed that he "placed [his] hand on the hood of the truck and it was warm" although since it was "less than 35 degrees ... the hood would have been cold [had Romo not recently driven the truck.]" Largen would later clarify that he did not follow the Dodge Ram into the Big T parking lot. Instead, he "pulled a U-turn and lost sight of it for a second, couple seconds, whatever." He then did another U-turn, to face the direction the Dodge was going, but it was nowhere in sight. At that point, he decided that it likely headed west down a different street, and turned down that street. It was only at that point that he saw Romo's vehicle.

Romo filed suit under 42 U.S.C. § 1983 and Michigan law. His complaint included federal claims of false arrest and malicious prosecution. The complaint also stated, without elaboration, that Largen "violated [Romo]'s right to equal protection." Romo also claimed that Largen committed the intentional torts of false arrest, false imprisonment, and malicious prosecution. Largen claimed qualified and governmental immunity and moved for summary judgment.

The district court determined that there were "at least three basic narratives possible from what's going on here." The first is that Largen fabricated the background story.

In that case, the district court noted that "a reasonable fact-finder could conclude no reasonable officer, mistaken or otherwise, could conclude there's probable cause on that. And that's, I think, on a record like this, enough to send the case to the jury." The district court later noted that "a jury conceivably could discount" Largen's story.

The second possible narrative is that Largen's story, including the warm hood, is entirely correct. In that case, the district court noted that probable cause may have existed. However "that's not a story [the court] can accept for purposes of evaluating a defense motion for summary judgment."

The court, like the parties, spent the most time on the third possible narrative. In that version of the story, Largen "ma[d]e up the part about the warm hood and the warm engine, but then he really did see a Dodge truck pull alongside the semi on the railroad tracks, saw it in his rear-view mirror, ... [but] didn't have continuous contact, [and] the colors [of the trucks] were different." Id. "[E]ven if that's the scenario," the court determined, "summary judgment would be inappropriate."

The district court stated that while a jury may "ultimately agree with the officer on the defense theory, ... it's one of those cases that should be the jury's call as opposed to my call" and denied summary judgment on the federal false-arrest claim. The court also denied summary judgment on the state-law claims because they are similar, albeit with a malice requirement that the court determined was fact oriented. Finally, the district court denied summary judgment on the federal malicious-prosecution claim, finding that bond conditions may be enough of an additional deprivation of liberty to fulfill the requirements for that claim.

Largen now appeals. He argues that he is entitled to qualified immunity for his "reasonable but mistaken belief that Romo was driving the pickup truck [he] observed commit a traffic violation." He also argues that the state-law claims should have been dismissed because he acted in good faith, that the federal malicious-prosecution claim should have been dismissed because the charges against Romo were dismissed, and that the equal-protection claim, which was not discussed below, should be dismissed.

While the parties argue extensively about the third scenario, the district court's treatment of the first scenario is sufficient for affirmance. In adjudicating this appeal, we are required by the limitations on interlocutory appeals of qualified immunity denials to accept the district court's finding that a genuine dispute of material fact existed as to whether Largen fabricated the whole or a part of his story about seeing a Dodge Ram pickup pass a semi tanker, and we refuse to consider Largen's factual disputations to the contrary.2 In the Supreme Court's words, "a defendant, entitled to invoke a qualified immunity defense, may not appeal a district court's summary judgment order insofar as that order determines whether or not the pretrial record sets forth a `genuine' issue of fact for trial." Johnson v. Jones, 515 U.S. 304, 319-20, 115 S.Ct. 2151, 132 L.Ed.2d 238 (1995); see also Kirby v. Duva, 530 F.3d 475, 481 (6th Cir.2008) ("[T]his court may simply ignore defendants' attempts to dispute plaintiffs' version of the facts, obviating the need to dismiss the entire appeal for lack of jurisdiction." (internal quotation marks omitted)).3 It follows that Largen's qualified-immunity defense clearly fails. No reasonable officer would believe that he could constitutionally arrest a person found sleeping in a truck and take that person to jail for drunk driving, particularly after the person offers a cogent explanation for sleeping in the truck and shows that he does not even have the truck's keys. We therefore uphold the judgment of the district court denying summary judgment to Largen on the basis of qualified immunity.

Relying on Scott v. Harris, 550 U.S. 372, 127 S.Ct. 1769, 167 L.Ed.2d 686 (2007), and policy considerations, the concurrence suggests a "reading" of Johnson under which defendants may generally challenge on interlocutory appeal a district court's determination that the summary judgment standard has been met with respect to factual inferences (although, in concept at least, not to facts that underlie such inferences). Such an approach is facially contradicted by the Supreme Court's instructions in Johnson to "take, as given, the facts that the district court assumed when it denied summary judgment" and, when that is unclear, to "determine what facts the district court ... likely assumed." Johnson, 515 U.S. at 319, 115 S.Ct. 2151. The Court considered and dismissed the criticisms of this approach that the concurrence raises. See id.

Thus, to accept the reading of Johnson advocated by the concurrence, we would have to read Scott to have foreshadowed the overruling of Johnson's explicit holding.4 It may be proper in very clear situations for a lower court to anticipate the Supreme Court's overruling of one of its precedents.5 But such instances should be limited to the most compelling of inferences from intervening cases. See United States v. Koch, 383 F.3d 436, 441-43 (6th Cir.2004). Indeed, the Supreme Court — albeit in dictum — has admonished the lower courts not readily to anticipate the overruling of its prior holdings. See, e.g., Rodriguez de Quijas v. Shearson/Am. Express, Inc., 490 U.S. 477, 484, 109 S.Ct. 1917, 104 L.Ed.2d 526 (1989). In the words of Learned Hand, it is not "desirable for a lower court to embrace the exhilarating opportunity of anticipating a doctrine which may be in the womb of time, but whose birth is distant." Spector Motor Serv. v. Walsh, 139 F.2d 809, 823 (L.Hand, J., dissenting) (2d Cir.1943), vacated sub nom. Spector Motor Serv. v. McLaughlin, 323 U.S. 101, 65 S.Ct. 152, 89 L.Ed. 101 (1944). The instant appeal hardly presents a situation in which to disregard such admonitions. The Scott opinion did not even mention the Johnson holding, and Scott is easily limited as an exception for blatantly contradicted facts. It is up to the Supreme Court to overrule Johnson, and Scott does not provide us a basis for anticipating — much less predicting with compelling confidence — that the Supreme Court will do so.

We next consider Largen's appeal of the denial of governmental immunity on Romo's state-law claims. Viewing the facts in the light most favorable to Romo, Largen did not have probable cause to arrest Romo for operating while intoxicated and a jury could find that Largen acted with malice. The district court properly denied governmental immunity to Largen. Largen did not have any evidence that Romo was operating the truck. Largen found Romo intoxicated and sleeping in his truck. Since people rarely sit in their vehicles without driving them, Largen could reasonably suspect that Romo either had driven or was planning to drive while intoxicated. However, Romo's responses to Largen's questioning should have dispelled those suspicions. Romo explained that he was sleeping off a night of drinking, that he had not been driving, and that he could not drive because he did not have the keys. Although Largen claims that he heard Romo say that his kids, rather than his keys, were in the truck, he later states "you don't have the keys on you," suggesting that he had heard Romo say he did not have the keys at some point during the conversation. Romo also stated in his deposition that he told Largen that his brother had taken his keys before Largen even turned on the microphone. Romo had a reasonable explanation and Largen had no inculpatory evidence.

Drunk driving is a serious danger to the public, and police are properly suspicious of anyone sitting in his car drunk. However, they need more than presence in the driver's seat of a vehicle for probable cause. For example, in Nettles-Nickerson v. Free, the plaintiff had started her car, her brake lights had turned on, and she sat in it while it was running. 687 F.3d 288, 291 (6th Cir.2012). This court determined that in such a situation, a police officer can reasonably believe that the plaintiff was operating the car. The court noted that "nothing impeded Nettles-Nickerson's ability to move the car" and that she was "able, at any moment, to drive away." Id. at 291-92. In contrast, Romo could not possibly drive anywhere because he did not have the keys. The Michigan Court of Appeals has found that a person operates a car once it is put into gear. See City of Plymouth v. Longeway, 296 Mich.App. 1, 818 N.W.2d 419, 424 (2012). However, if the car is in "park" or not even running, and the person is asleep, then he is not operating the car. See Michigan v. Burton, 252 Mich.App. 130, 651 N.W.2d 143, 151 (2002); Michigan v. Andres, No. 258280, 2006 WL 448811, at *5 (Mich.Ct.App. Feb. 23, 2006). The only exception is if there is evidence that the person had driven the car while intoxicated to the location where the officers found it. See Michigan v. Stephen, 262 Mich.App. 213, 685 N.W.2d 309, 313 (2004). However, Largen had no evidence that Romo had driven the car to the parking lot after drinking, and Romo explained that his brother had driven him there and took his keys so that he could not drive.

To be entitled to governmental immunity under Michigan law, a lower-level government official accused of an intentional tort must show that:

Odom v. Wayne County, 482 Mich. 459, 760 N.W.2d 217, 228 (2008). The only question here is whether Largen was acting in good faith or with malice. A police officer is entitled to governmental immunity "if he acted in good faith and honestly believed that he had probable cause to arrest, even if he later learned that he was mistaken." Id. at 229. This test is subjective, not objective. Id. Largen argues that since the district court stated that "the officer sounds like he believes what he's talking about," he is entitled to governmental immunity. However, the court also found that "a reasonable fact-finder could still draw an inference of actual malice." Moreover, if Largen lied on his police report and to the prosecutor in order to obtain a prosecution of Romo, that is strong evidence of malice. Accordingly, Largen is not entitled to governmental immunity on Romo's state-law claims.

On the other hand, Largen is entitled to summary judgment on Romo's federal malicious-prosecution and equal protection claims because Romo has stated that he is no longer pursuing these claims. See Romo Br. at 38-39.

The district court's denial of qualified immunity with respect to Romo's federal malicious-prosecution claim is reversed. The judgment of the district court is otherwise affirmed.

SUTTON, J., concurring in part and concurring in the judgment.

One aspect of this case is easy: deciding whether Romo's false-arrest claim should go to a jury. Once we give Romo's testimony all of the inferences to which it is entitled at summary judgment, it is clear that a reasonable jury could grant him relief on this claim. Another aspect of this case is not: deciding how to apply Johnson v. Jones, 515 U.S. 304, 115 S.Ct. 2151, 132 L.Ed.2d 238 (1995), and deciding whether we have jurisdiction over this interlocutory appeal. The majority offers one way of applying Johnson to this appeal. I write separately to offer another way of looking at it.

In Johnson, Houston Jones filed an excessive force claim against five policemen who allegedly beat him. Three of the officers moved for summary judgment. Their argument? "We didn't do it," an argument supported by the three officers' affidavits but in tension with Jones's evidence-supported arguments. In a quintessential application of Civil Rule 56, the district court held that the jury should resolve this material fact dispute. The officers filed an appeal, insisting once more that they didn't do it. "When asked at oral argument if they could lose the factual dispute and still prevail, defendants' lawyer answered no." Jones v. Johnson, 26 F.3d 727, 728 (7th Cir.1994). In a 6-paragraph per curiam opinion, the Seventh Circuit dismissed the appeal for lack of jurisdiction on the ground that "defendants may urge only legal issues on [interlocutory] appeals." Id. at 727.

The Supreme Court followed the Seventh Circuit's approach and elaborated on it. In the normal course, the Court explained, a denial of summary judgment is not final, and appellate courts must wait for a district court to issue a final order before reviewing the case. 515 U.S. at 309, 115 S.Ct. 2151. That patience saves appellate resources, prevents delay and keeps pretrial matters where they belong — with the district court. Id. at 317, 115 S.Ct. 2151. But qualified immunity is not a typical summary judgment defense; it is an immunity from standing trial. Forcing an official who loses in the district court to wait for a jury verdict before appellate review would defeat an essential component of the defense. See id. at 315-17, 115 S.Ct. 2151. The prospect of interlocutory qualified immunity appeals in this area thus pits the desire to protect government employees from needless trials against the fear of clogging appellate dockets with delay-creating and essentially frivolous appeals.

In a unanimous decision, Johnson tried to resolve that tension. A defendant may immediately appeal a denial of qualified immunity "to the extent that it turns on an issue of law," such as whether the defendant violated a clearly established constitutional right. Id. at 313, 115 S.Ct. 2151 (quoting Mitchell v. Forsyth, 472 U.S. 511, 530, 105 S.Ct. 2806, 86 L.Ed.2d 411 (1985)). But defendants who merely want to argue evidentiary sufficiency have to wait until after trial. Id. District courts have more experience sifting through convoluted factual records, the Court explained, and asking the courts of appeals to take a second look at discovery documents risks unduly postponing a plaintiff's day in court. Id. at 316, 115 S.Ct. 2151. And all of that time would be wasted if, after affirming, the jury finds in favor of the plaintiff, and the defendant appeals yet again. Id. at 316-17, 115 S.Ct. 2151. In view of these concerns and others, Johnson told appellate courts to limit their interlocutory resources to questions of law.

Keeping in mind that courts do not read judicial opinions like statutes, I submit that there are two ways to read Johnson. One applies it only to prototypical "he said, she said" fact disputes, in which the defendants (usually government employees) refuse to accept the truth of what the plaintiffs (usually individual claimants) say happened. When the appeal boils down to dueling accounts of what happened and when the defendants insist on acknowledging on appeal only their accounts, the underlying basis for an interlocutory appeal disappears.

The other applies the decision not just to whether the defendant officers accept the plaintiff's evidence-supported version of what happened but also to whether the defendants accept the district court's reading of the inferences from those facts: here, whether Officer Largen lied about seeing a Dodge Ram on the road. Under that view (and the majority's view), when a district court determines that there is a "genuine issue of fact" for trial by drawing an inference in favor of the plaintiff, the appellate court may not second-guess that inference, indeed lacks jurisdiction to do so. I favor the former reading.

First, my proposed reading respects Scott v. Harris, 550 U.S. 372, 127 S.Ct. 1769, 167 L.Ed.2d 686 (2007), which post-dates Johnson. In Scott, Deputy Timothy Scott followed Victor Harris in a high-speed chase. Worried that Harris's reckless driving posed a threat to the public, Scott ran Harris off the road, which caused a crash, which left Harris a quadriplegic. Harris sued, alleging excessive force, and the parties introduced a videotape of the chase at summary judgment. According to the district court, reasonable minds could disagree about whether Harris posed a real risk, creating a fact issue for the jury. It wrote: "A rational fact finder could find that a reasonable officer would not have believed that because of his traffic offense, Harris posed an immediate threat to the safety of others.... Harris did not use his vehicle in an aggressive manner.... A jury could also find that Scott's use of force — ramming the car while traveling at high speeds — was not in proportion to the risk that Harris posed, and therefore was objectively unreasonable." Harris v. Coweta Cnty., Ga., No. CIVA 3:01CV 148 WBH, 2003 WL 25419527, at *5 (N.D.Ga. Sept. 25, 2003). The Eleventh Circuit affirmed on the merits and rejected the "argument that we are without jurisdiction over this interlocutory appeal." Harris v. Coweta Cnty., Ga., 433 F.3d 807, 811 n. 3 (11th Cir.2005).

The Supreme Court reversed on the merits. "At the summary judgment stage," it started, "facts must be viewed in the light most favorable to the nonmoving party only if there is a `genuine' dispute as to those facts." 550 U.S. at 380, 127 S.Ct. 1769 (quoting Civil Rule 56(c)). Harris's account — that he remained relatively in control — was "blatantly contradicted by the record," namely the videotape, which showed him driving "shockingly fast," "swerv[ing]" and "forc[ing] cars ... to their respective shoulders to avoid being hit." Id. at 379-80, 127 S.Ct. 1769. "[N]o reasonable jury could have believed" Harris's "version of events," and, the Court held, neither the district court nor the Eleventh Circuit should have accepted his version either. Id. at 380, 127 S.Ct. 1769.

The details of Harris's driving were not in dispute; the video laid those to rest. What mattered were the inferences those details could support: Was Harris driving safely? Those inferences, Scott illustrates, remain subject to interlocutory review. A contrary reading of Johnson — that in no event does an appellate court have jurisdiction to say a district court, drew the wrong inferences — cannot co-exist with Scott. This reading of Scott also respects Behrens v. Pelletier, 516 U.S. 299, 116 S.Ct. 834, 133 L.Ed.2d 773 (1996), which cautions that "[d]enial of summary judgment often includes a determination that there are controverted issues of material fact, and Johnson surely does not mean that every such denial of summary judgment is nonappealable." Id. at 312-13, 116 S.Ct. 834.

Second, the only way to avoid this reading of the holdings of Johnson and Scott is to create distinctions that are not driven by meaningful differences or that would be difficult to implement in practice. One might for example limit Scott to videos (and audio recordings), reasoning that appellate courts need not defer to the eyes (or ears) of district courts when the relevant evidence comes in the form of a video or audio of the encounter as opposed to dueling eyewitness accounts of the encounter disclosed in depositions. But one need not be an expert in the senses to appreciate that the capacity of appellate judges to identify untenable inferences in a district court's assessment of a video and an audio depiction of an event may apply just as often to their capacity to see untenable inferences in a district court's assessment of a deposition transcript. In both cases, the district and appellate court judges are looking at the same thing.

Or one could say, as today's majority does, that Scott trumps Johnson only when the district court's assessment of the inferences to draw is "blatantly contradicted by the record." Scott, 550 U.S. at 380, 127 S.Ct. 1769. Only if the district court was really wrong would we thus have authority to review the district court's reading of the inferences. But the most straightforward way to say that a district court did not make such a mistake is to say that a reasonable jury could rule for the plaintiff on that record. That is the traditional question we consider in reviewing qualified-immunity interlocutory appeals. If Johnson is designed to protect appellate courts from record review, how does a "blatantly contradicted by the record" standard help? It does not allow us to ignore depositions and documents we would otherwise read, as those same items will be the source of any such clear-error argument. That explains why the majority in this case must review that same sort of record evidence here — to assure itself that the district court did not blatantly contradict the record. See Maj. Op. at 674-75 n. 3. A defendant making a Scott claim and a defendant raising a de novo challenge to the lower court's inferences are asking us to do much the same thing. Even if it turns out to be only the occasional case that meets the Scott standard, I doubt it will be only the occasional lawyer who argues that the district court badly misread the record. This reading of Johnson thus trades a long-established direct approach to qualified-immunity appeals for a complicated indirect approach that ultimately returns to the same question (or something very close to it). And this reading implies that we have jurisdiction to reverse badly drawn inferences but no jurisdiction to affirm less than badly drawn inferences — a new and strange version of appellate power. What, in short, has been gained?

Third, much may be lost with a broad interpretation of Johnson, as it implicates a jurisdictional inquiry that raises the stakes and complexities of the debate. Johnson says that a defendant who loses a qualified immunity motion "may not appeal a district court's summary judgment order," and that the court of appeals lacks "jurisdiction" over such an appeal, "insofar as that order determines whether or not the pretrial record sets forth a `genuine' issue of fact for trial." 515 U.S. at 309, 319-20, 115 S.Ct. 2151. Since Johnson, it is true, the Court "has endeavored ... to bring some discipline to the use of the term `jurisdictional.'" Gonzalez v. Thaler, ___ U.S. ___, 132 S.Ct. 641, 648, 181 L.Ed.2d 619 (2012) (internal quotation marks omitted); see also Arbaugh v. Y & H Corp., 546 U.S. 500, 510, 126 S.Ct. 1235, 163 L.Ed.2d 1097 (2006). But it is difficult to see how Johnson does not implicate core subject matter jurisdiction. Congress has granted courts of appeals "jurisdiction of appeals from all final decisions of the district courts." 28 U.S.C. § 1291. That speaks to subject matter jurisdiction. See Bowles v. Russell, 551 U.S. 205, 215, 127 S.Ct. 2360, 168 L.Ed.2d 96 (2007). In Cohen v. Beneficial Industries Loan Corp., 337 U.S. 541, 69 S.Ct. 1221, 93 L.Ed. 1528 (1949), the Court construed § 1291 to give appellate courts jurisdiction over some interlocutory — non-final — decisions. That too speaks to subject matter jurisdiction. See Mohawk Indus. v. Carpenter, 558 U.S. 100, 106, 114, 130 S.Ct. 599, 175 L.Ed.2d 458 (2009). And Mitchell v. Forsyth, 472 U.S. 511, 530, 105 S.Ct. 2806, 86 L.Ed.2d 411 (1985), construed Cohen to give appellate courts jurisdiction over interlocutory appeals from a district court's denial of a motion for qualified immunity. That too speaks to subject matter jurisdiction. See Mohawk Indus., 558 U.S. at 106, 130 S.Ct. 599. It follows that Johnson raises an unyielding jurisdictional inquiry, one that the courts of appeals must raise whether they like it or not, whether the parties have raised it or not.

That is a tall order if courts embrace a broad interpretation of Johnson. It is difficult to think of qualified immunity appeals that do not involve inference drawing by the district courts, whether implied or explicit. And there are few days of oral argument in the life of an appellate judge that do not involve at least one qualified immunity interlocutory appeal. This case is a good example of the task that awaits us. Neither party mentioned Johnson in its briefs. All that appears in the plaintiff's brief on this score is this stand-alone sentence: "In fact, given that the court ruled that these credibility disputes created a factual issue regarding what actually happened that must be determined by a jury, which directly affects the evidence allegedly supporting probable cause, this should not even be an issue that is subject to an appeal. Gregory v. City of Louisville, 744 [sic] F.3d 725 (6th Cir.2006)." App'ee Br. at 30. If we followed the normal rules for identifying error, we would say that the plaintiff forfeited any Johnson argument. The argument is not identified (or embedded) in the issues presented on appeal, it is undeveloped, and it leaves the reader guessing what the appellee means to argue. See Johns v. Holder, 678 F.3d 404, 408-09 (6th Cir.2012). To its credit, the majority nonetheless takes on the Johnson issue, as indeed it must if it implicates our jurisdiction. In view of the underlying interpretation of § 1291 implicated by all interlocutory appeals and in view of the prohibition on "hypothetical" jurisdiction, Steel Co. v. Citizens for a Better Env't, 523 U.S. 83, 101, 118 S.Ct. 1003, 140 L.Ed.2d 210 (1998), that is the only path available to us. Yet the majority's approach pulls jurisdiction into many, if not most, qualified immunity appeals, whereas my interpretation of Johnson — confining it to "I didn't do it" appeals — will create few jurisdictional disputes.

Fourth, a broad interpretation of Johnson not only requires us to police subject matter jurisdiction in a great many cases, but it also requires us to do so by reviewing the record in new and unusual ways. At first glance, the reader might think that separating authority to review law-based appeals from authority to review fact-based appeals is easy. It is not. Yes, a qualified-immunity appeal based solely on what the Constitution means or whether an interpretation of it was clearly established at a given point in time raises purely legal issues. But the classic summary judgment question — could a reasonable jury rule for the plaintiff on this record as construed in his favor? — also raises a "legal" question, even though it is intertwined with the facts, which is why appellate courts traditionally give fresh review to district court decisions in this area. Milligan v. United States, 670 F.3d 686, 696 (6th Cir.2012). The rub is that Johnson appears to divide this last category into appealable and non-appealable decisions. I would construe the decision as applying only when the defendant refuses to accept the truth of the plaintiff's evidence at summary judgment by maintaining for example that he didn't do it when the plaintiff produces evidence that he did. The majority would construe the decision as applying whenever the defendant challenges the inferences drawn by the district court in denying the summary judgment motion.

The majority's approach creates two unpalatable options when it comes to reviewing summary judgment decisions. The first is to comb the record to figure out the inferences that the district court likely drew, then to defer to those assumed inferences. Johnson, 515 U.S. at 319, 115 S.Ct. 2151. If those inferences create a genuine issue of material fact, no jurisdiction exists. The other option is to defer only to the inferences the district court actually drew on the record, see, e.g., Lewis v. Tripp, 604 F.3d 1221, 1226 (10th Cir.2010), and no jurisdiction exists if those inferences create a genuine issue of material fact. While the latter approach has the virtue of being real rather than imagined, it creates problems of its own. In my experience, district court judges generally do not write summary judgment opinions in this way. Take this decision, which the judge delivered orally, presumably because he placed a (legitimate) premium on resolving this summary judgment motion promptly. The district court explained why the plaintiff could win based on the court's reading of the record, intermingling verbatim statements from the various depositions, paraphrases from those depositions and inferences from the depositions. That is not unusual. I indeed have not come across trial judges who distinguish between facts and inferences or who explain the difference material fact by material fact. That perhaps is because Johnson speaks to what appellate judges should do, not to what district court judges should do, and I doubt the point of Johnson was to pile new tasks onto trial judges' work loads or to tell them how to do their jobs. Much simpler, I think, to let appellate courts do what they long have done by taking a fresh look at the traditional (non-jurisdictional) summary judgment question: After giving the non-movant (usually the plaintiff) the benefit of all reasonable inferences from the record, does the case create a material triable issue of fact?

Fifth, efforts to work around these problems will create problems of their own. Courts for example might construe Johnson as giving appellate judges the option of dismissing appeals for lack of jurisdiction. Consistent with this approach, one court has called the Johnson inquiry a "prudential" one, M.M.R.-Z. v. Puerto Rico, 528 F.3d 9, 13 (1st Cir.2008). From what I can tell, that seems to be the approach the majority has taken. After engaging in the Johnson inquiry and after noting that the appellant (Officer Largen) has failed to accept the district court's inferences from the record, the court nonetheless "affirm[s]" and "uphold[s]" the district court's decision rather than dismissing this part of it for lack of jurisdiction. Maj. Op. at 675, 677. The majority insists that Johnson remains a jurisdictional rule in our circuit, but offers an explanation for its judgment with which I am unfamiliar: "The fact that at the end of this opinion we `affirm' rather than `dismiss' reflects that we are ruling on what is properly before us, and says nothing about what is jurisdictionally not before us." Id. at 674 n. 2. As shown, Cohen and Mitchell (and thus Johnson as well) interpret § 1291, which makes any application of those decisions (whether narrowing or broadening) jurisdictional. The baseline norm at any rate is that courts should resolve all disputes — including appeals — that Congress authorizes them to resolve. Cohens v. Virginia, 6 Wheat. 264, 404, 5 L.Ed. 257 (1821). We should not lightly create exceptions to that "unflagging" obligation. Deakins v. Monaghan, 484 U.S. 193, 203, 108 S.Ct. 523, 98 L.Ed.2d 529 (1988) (internal quotation marks omitted). This cure, in short, seems worse than the disease.

Johnson itself acknowledged another way around these knotty issues, though it is not one the Court seemed to endorse. It "assum[ed], for the sake of argument, that it may sometimes be appropriate to exercise `pendent appellate jurisdiction' over" a district court's genuine-issue-of-material-fact decision. 515 U.S. at 318, 115 S.Ct. 2151 (citing Swint v. Chambers Cnty. Comm'n, 514 U.S. 35, 50-51, 115 S.Ct. 1203, 131 L.Ed.2d 60 (1995)). An appeal that raised purely legal questions under prong one of the qualified immunity inquiry (what does the Constitution mean in this setting?) or prong two (was the right clearly established when the officers acted?) thus would create appellate jurisdiction over these issues and pendent appellate jurisdiction over any other issues raised by the district court's decision. After acknowledging this possibility, Johnson said it was "unlikely that Courts of Appeals" would exercise pendent appellate jurisdiction "where the appealable issue appears simply a means to lead the court to review the underlying factual matter." Id. Fair enough.

But in our circuit and, best I can tell, in most circuits, the test for pendent appellate jurisdiction — whether the two issues are "inextricably intertwined," Chambers v. Ohio Dep't of Human Servs., 145 F.3d 793, 797 (6th Cir.1998) (citing Swint, 514 U.S. at 49-50, 115 S.Ct. 1203) — will not be demanding in the context of a qualified immunity appeal and thus will not require gamesmanship to put all issues in front of the appellate court. Will it not usually be the case, at prong one of the qualified immunity inquiry, that the meaning of the Constitution and the underlying facts are inextricably intertwined? How could they not be? They often implicate "mixed questions of law and fact," Williams v. Mehra, 186 F.3d 685, 689 (6th Cir.1999) (en banc), the epitome of inextricably intertwined law and fact issues. They are called mixed, as opposed to segregated, questions of law and fact for a reason. And even when that is not the case, how does one decide what the Constitution means without regard to the facts? So too with the clearly established inquiry, one that looks at whether the officers should have known that their fact pattern did not permit them to act the way they did under then-existing law. Keep in mind, moreover, that the question is whether a genuine issue of material fact exists. Even if the genuine-issue question somehow is purely factual in nature, the issue of materiality is not. Which facts are material to a constitutional claim will always be a legal question, intertwining the genuine-issue fact question and the constitutional standard implicated by the plaintiff's claim. Cf. Livermore v. Lubelan, 476 F.3d 397, 407 (6th Cir.2007). Rather than engage in debates about this or that path to pendent appellate jurisdiction in today's case, I would affirm on the traditional ground that, once the officer accepts the truth of the plaintiff's evidence-supported factual allegations, his appeal fails to rebut the argument that a reasonable jury could rule against him on this record.

Why, moreover, should the officer willing to concede the truth of the plaintiff's evidence-supported version of the facts not get interlocutory review over the district court's genuine-issue-of-fact determination (itself a legal question), but the officer who challenges that and the district court's understanding of the meaning of the Constitution gets interlocutory review? If the difference turns on efficiency, that is an unusual explanation for a court-made jurisdictional rule and at all events not necessarily true. The capacity of district courts and appellate courts to identify triable issues of material fact based on the review of paper records strikes me as about equal — and as being done by both just as frequently. And while interlocutory review delays a potential trial, the appeal will potentially avoid unnecessary constitutional rulings, see Pearson v. Callahan, 555 U.S. 223, 237, 129 S.Ct. 808, 172 L.Ed.2d 565 (2009), end the case (if the officer wins), often prompt a settlement (if the officer loses) or often eliminate a later appeal (if the officer loses the appeal and the jury trial). As to this last point, while the summary judgment record and the jury record often will differ, they normally will differ in ways that make it more difficult for an officer to win on appeal, as the appellate court will be asked to second guess not just inferences that could be drawn from a paper record but inferences that may be drawn from eyeballing the witnesses as well. In the end, if the district court is wrong about permissible fact inferences at summary judgment, even if not "blatantly" wrong, I do not understand why we should force these officers to undertake a trial.

Sixth, a broad reading of Johnson is in tension with related principles of qualified immunity, jurisdiction and judicial administration. The courts are reluctant to make appealability decisions in this area turn on the novelties of individual cases as opposed to principles that cut across groups of traditional appeals. "Making appealability depend upon ... factor[s] particular to the case at hand[] would violate the principle ... that appealability determinations are made for classes of decisions not individual orders in specific cases." Behrens, 516 U.S. at 312, 116 S.Ct. 834. By making jurisdiction turn on the facts in every case, every case will become a class of an appealable or a non-appealable decision unto itself. For related reasons, courts strive to separate jurisdictional inquiries from merits-driven ones, Arbaugh, 546 U.S. at 510-11, 126 S.Ct. 1235; see also Chafin v. Chafin, ___ U.S. ___, 133 S.Ct. 1017, 1024, 185 L.Ed.2d 1 (2013); Primax Recoveries, Inc. v. Gunter, 433 F.3d 515, 519 (6th Cir.2006), an objective thoroughly undermined by a broad reading of Johnson.

Pearson v. Callahan suggests another way of looking at this problem. That case overruled Saucier v. Katz, 533 U.S. 194, 121 S.Ct. 2151, 150 L.Ed.2d 272 (2001), which required courts to resolve prong one (the current meaning of the Constitution) of a qualified immunity defense before moving to prong two (the clearly established meaning of the Constitution at the time of the incident). One of the problems with that "rigid order of battle," Pearson, 555 U.S. at 234, 129 S.Ct. 808, was the frequency with which courts ignored it. Saucier was mandatory — but not when the case was overly factbound, id. at 237, 129 S.Ct. 808, or when the Supreme Court granted certiorari on a related issue, id. at 238, 129 S.Ct. 808, or when state law came into play, id. These exceptions undermined, not proved, the rule, and any rule ought not to be honored in the breach. A broad reading of Johnson, I fear, raises similar problems. It forbids record review — but not when the district court made a blatant error or when the district court was not explicit or when intent or mixed questions of law and fact come into play or when pendent appellate jurisdiction is exercised.

The majority's reading of Johnson also creates tension with Ashcroft v. Iqbal, 556 U.S. 662, 129 S.Ct. 1937, 173 L.Ed.2d 868 (2009), which requires courts to determine whether the inferences a plaintiff wants a jury to draw are "plausible." Id. at 678, 129 S.Ct. 1937. No doubt, it may be easier for appellate courts to draw reasonable inferences from complaints than from summary judgment records. But how strange to think that the distinction is jurisdictional — that a court of appeals has jurisdiction to overturn a district court's reading of the inferences at the motion to dismiss stage but not at the summary judgment stage — a theory that may have the incidental (and unintended) effect of encouraging defendants to file appeals at each stage.

Seventh, a broad reading of Johnson creates other oddities with respect to mixed questions of law and fact and questions of intent, which frequently come up in qualified immunity appeals. Some issues, such as the presence (or absence) of exigent circumstances or the exculpatory (or not) nature of evidence, go to a jury. Ewolski v. City of Brunswick, 287 F.3d 492, 501 (6th Cir.2002); Gregory v. City of Louisville, 444 F.3d 725, 744 (6th Cir. 2006). Other issues, such as the existence of probable cause or deliberate indifference, may or may not go to a jury under our case law. See McKenna v. Edgell, 617 F.3d 432, 441 (6th Cir.2010); id. at 447-50 (Rogers, J., dissenting). Compare Williams, 186 F.3d at 689; with Phillips v. Roane Cnty., 534 F.3d 531, 539 (6th Cir. 2008). And issues of intent will often require courts to draw inferences. If a district court says reasonable minds could differ on one of these questions, say exigent circumstances, does that tie our hands and remove any jurisdiction over the appeal, even when the underlying facts are undisputed? See Dixon v. Donald, 291 Fed.Appx. 759, 762 (6th Cir.2008) ("To the extent that Donald is arguing the existence of exigent circumstances ... he is raising genuine issues of material fact that Johnson prohibits us from reviewing."); Elliott v. Leavitt, 105 F.3d 174, 175 (4th Cir.1997) (Wilkinson, J., concurring in the denial of rehearing en banc) (noting that, under a broad reading of Johnson, "[a]ll plaintiffs need do to receive an automatic trial is simply assert, without any evidence, that `the police are lying'"). If so, how does that interpretation advance the efficiency concerns of Johnson? One would have thought that an earlier, rather than a later, appellate ruling on the meaning of law in an accepted setting of facts would help everyone, the parties and the district court included.

A broad reading of Johnson requires appellate courts to separate (unreviewable) factual inferences from (reviewable) legal conclusions. This assignment calls to mind the ill-starred distinction nineteenth century pleading rules drew between ultimate facts (which plaintiffs could, indeed had to, plead) and conclusions of law (which they could not). See Charles E. Clark, The Complaint in Code Pleading, 35 Yale L.J. 259, 259-60 (1926). An example illustrates the perplexities that await. Suppose that a district court faced with an exigent circumstances case concludes that the plaintiff's behavior did not manifest a risk of "imminent destruction of evidence." Brigham City v. Stuart, 547 U.S. 398, 403, 126 S.Ct. 1943, 164 L.Ed.2d 650 (2006). Has the court drawn a factual inference from the plaintiff's behavior? Or has it reached a legal conclusion about the applicability of a particular justification for a warrantless search? Whatever the answer, does it make sense to hinge jurisdiction on such a will-o'-the-wisp distinction?

Eighth, all of these distinctions and complexities increase the odds that appellate jurisdiction will turn on the quality of the lawyer a party happens to hire. It is the rare qualified-immunity appeal that does not implicate the meaning of the Constitution, and it is the rare capable attorney who thus cannot frame an appeal that invokes that core appellate jurisdiction. What happens as a result is that lawyers schooled in Johnson know how to avoid its shoals (by raising a colorable legal question, which is not a high bar, see Steel Co., 523 U.S. at 89, 118 S.Ct. 1003) and lawyers unschooled in Johnson unwittingly approach its shoals, requiring appellate courts to resolve a difficult, sometimes excruciating, jurisdictional question, made more difficult by the flexible opt-in of pendent appellate jurisdiction, instead of simply affirming on the merits because the plaintiff's claim implicates a triable issue of material fact. This case could have been resolved by a three-page summary order to that exact effect. Once the officers acknowledged that they had to accept the reasonable inferences from plaintiffs' record-supported evidence, as their attorney did at oral argument, appellate jurisdiction existed, and we could have affirmed on that basis then and there.

On a related score, it is not unusual for lawyers to pay tribute to Johnson, suggesting they are willing to accept the plaintiff's version of the facts or the district court's inferences from those facts but nonetheless shading the recitation of those facts in their client's favor, whether at the briefing stage or at oral argument. See Estate of Carter v. City of Detroit, 408 F.3d 305, 310 (6th Cir.2005) ("[T]his court can ignore the defendant's attempts to dispute the facts and nonetheless resolve the legal issue, obviating the need to dismiss the entire appeal for lack of jurisdiction."). Lawyers will be lawyers when it comes to this kind of thing, not because there is anything wrong with them but because it is human nature to shade facts and inferences in favor of the advocate's position. We ought to be wary of a rule for processing a high percentage of appeals that is a trap for some and easy to evade for others — and that requires a degree of restraint that is difficult for most lawyers, indeed most human beings, to muster. Little is gained, and much is lost, and the difference from the party's perspective often will turn on nothing more than the acumen of the attorney he can afford to hire.

The option of pendent appellate jurisdiction and the ease with which most lawyers can avoid Johnson jurisdictional problems might suggest that I should not be troubled by the majority's reading of it. That may be true in one sense. If ingenuity is a key that invariably unlocks the gateway to appellate jurisdiction, that is no less true for judges than for lawyers. But it does not flatter a legal rule that it is so often implicated and so easily avoided, particularly when it goes to jurisdiction.

Nor does it seem helpful that, nearly twenty years after Johnson, every circuit in the country has some decisions that adopt my reading of it and some that adopt the majority's. Start with our circuit. Some support the majority's view. See Morrison v. Bd. of Trs. of Green Twp., 583 F.3d 394, 404 (6th Cir.2009); Phillips v. Roane Cnty., 534 F.3d 531, 538 (6th Cir.2008); Johnson v. Karnes, 398 F.3d 868, 876 (6th Cir.2005). Others support my view or variations on it. See Chappell v. City of Cleveland, 585 F.3d 901, 906 (6th Cir.2009) (stating that a court of appeals can consider "legal errors as to whether the factual disputes (a) are genuine and (b) concern material facts"); see also Bishop v. Hackel, 636 F.3d 757, 769 (6th Cir.2011); Leary v. Livingston Cnty., 528 F.3d 438, 441-42 (6th Cir.2008).

The Seventh Circuit, affirmed in Johnson, seems to adopt a narrow interpretation of Johnson, See Anderson v. Cornejo, 355 F.3d 1021, 1023 (7th Cir.2004). But that is not always the case. See Sain v. Wood, 512 F.3d 886, 891 (7th Cir.2008). In every other circuit, save possibly for the D.C. and Federal Circuits, there are opinions supporting my view or otherwise involving appellate court review of inferences on the merits. See Camilo-Robles v. Hoyos, 151 F.3d 1, 15 (1st Cir.1998); Salim v. Proulx, 93 F.3d 86, 89-90 (2d Cir.1996); Schieber v. City of Philadelphia, 320 F.3d 409, 420 (3d Cir.2003); Winfield v. Bass, 106 F.3d 525, 533 (4th Cir.1997) (en banc); Brown v. Callahan, 623 F.3d 249, 254-55 (5th Cir.2010); Nelson v. Shuffman, 603 F.3d 439, 451 (8th Cir.2010); Jeffers v. Gomez, 267 F.3d 895, 907-10 (9th Cir.2001); Lewis v. Tripp, 604 F.3d 1221, 1226-28 (10th Cir.2010); Morton v. Kirkwood, 707 F.3d 1276, 1284 (11th Cir.2013).

That would be a powerful point in my favor if not for the reality that there are also decisions in every other circuit, save for the D.C. and Federal Circuits, that suggest the opposite. See Diaz v. Martinez, 112 F.3d 1, 4-5 (1st Cir.1997); Locurto v. Safir, 264 F.3d 154, 167 (2d Cir. 2001); Ziccardi v. City of Philadelphia, 288 F.3d 57, 61-62 (3d Cir.2002); Culosi v. Bullock, 596 F.3d 195, 201-02 (4th Cir. 2010); Smith v. Brenoettsy, 158 F.3d 908, 913 (5th Cir.1998); Parks v. Pomeroy, 387 F.3d 949, 956 (8th Cir.2004); Chateaubriand v. Gaspard, 97 F.3d 1218, 1223 (9th Cir.1996); Fogarty v. Gallegos, 523 F.3d 1147, 1154 (10th Cir.2008); Ratliff v. DeKalb Cnty., 62 F.3d 338, 341 (11th Cir. 1995). In view of these intra-circuit conflicts, it should not go unmentioned that additional litigation over appealability, an inevitable outcome of any uncertainty over the scope of it, adds to appellate work loads: one hundred and eighty degrees away from Johnson's goal. I would resolve that problem and the others mentioned above by adopting a narrow interpretation of Johnson: Once the appellant accepts the record-supported facts alleged by his opponent, the court of appeals has jurisdiction to give fresh review to whether one party should win as a matter of law or whether the case should go to a jury.

Ninth, today's case confirms many of these complications and potential sources of disagreement. Broad or narrow reading of Johnson? My view: narrow. The majority's view: broad. Is the issue forfeitable? All three of us seem to agree it is not. Does Johnson apply to state-law claims? My view: if there is no interlocutory jurisdiction over the federal claims, there is none over the state law claims. The majority's view: even when there is a Johnson defect with the federal claims, the court of appeals has authority to review the state law claims and may view those claims without applying the Johnson rule to them. Do we even need to get into this? All three of us feel the need to do so — so much so that the majority claims my interpretation of Johnson amounts to an anticipatory overruling of it. No doubt my reading of Johnson will make it frequently inapplicable, but that is in order to salvage, not bury, the decision, the latter of which I suspect happens more often than not when busy appellate courts confront burgeoning appellate dockets. Does Johnson impose a jurisdictional or claim-processing inquiry? My view: jurisdictional. The majority's view: jurisdictional — except that it affirms rather than dismisses for lack of jurisdiction.

A narrow reading of Johnson eliminates all of these sources of disagreement, and I would prefer to adopt it here.