Toby Cordova was shot and killed while fleeing from police in an early-morning car chase in Commerce City, Colorado. His survivors brought this 42 U.S.C. § 1983 action against Officer Derek Aragon and Commerce City, alleging that Officer Aragon had violated Mr. Cordova’s Fourth Amendment rights by using excessive force to end the police chase. Officer Aragon asserted qualified immunity and the defendants filed a motion for summary judgment. The district court granted their motion, holding that even when viewed in the light most favorable to the Cordovas the facts did not constitute ex*1186cessive force. We disagree that the facts could not constitute a constitutional violation. We affirm the grant of summary judgment to Officer Aragon, however, on the grounds that the law was not clearly established, and also affirm as to Commerce City on the grounds that the Cordovas have not created a genuine issue of material fact as to whether a policy or custom of the city was the moving force behind any violation that might have occurred.
I. Background
The parties dispute a number of facts. Because the case was decided on summary judgement, we construe these facts in the light most favorable to the non-moving party — here, the Cordovas.
On May 3, 2006, at 12:58 A.M., Officer James Zamora of the Commerce City Police Department (“CCPD”) spotted a truck driving in a subdivision where several new homes were being built. The truck was pulling a skid-steer loader (a piece of heavy excavation equipment) on a trailer behind it. Officer Zamora found this suspicious, as a number of thefts from construction sites had recently been reported in the area. He ran the trailer’s license plate and determined that it was not registered to an address in the vicinity. The skid-steer loader was later determined to be stolen. Officer Zamora attempted to pull over the truck, but the driver- — Mr. Cordova — refused to stop and ran through a solid red light. At this point, the pursuit began.
When the police radio announced that Officer Zamora was attempting to stop a suspicious vehicle, Officers Derek Aragon, Dax Nance, and Janae Rubino headed toward his location. The officers attempted to corner Mr. Cordova after he pulled into a business parking lot, but Mr. Cordova evaded capture by driving straight at them. Officer Rubino announced over the radio that the truck had attempted to ram her and that she had made an evasive turn to avoid being hit.
The pursuit continued, now moving beyond the parking lot and involving multiple police cars with lights and sirens in action. During the pursuit, Mr. Cordova twice drove off the road to avoid spike strips, once more ran a red light, repeatedly refused to stop for patrol cars with lights and sirens activated, and generally proceeded at speeds between 30 and 50 miles per hour. He eventually turned onto eastbound 1-76. Officer Aragon became concerned that Mr. Cordova might cross over to the wrong side of the highway in an attempt to escalate the danger of the pursuit and avoid apprehension. He decided to cross onto the wrong side of the highway himself so that he could “warn traffic of the potential danger.” Dist Op., 560 F.Supp.2d 1041, 1044.1 Mr. Cordova then crossed onto the wrong side of the highway, too. Officer Aragon slowed his patrol car, at which point Mr. Cordova tried to ram it, according to statements by both Officer Aragon and Officer Nance, who was riding in Officer Aragon’s car. Officer Aragon then accelerated past the truck, stopped his car, and attempted to deploy stop sticks. Although the car’s exact position is not known because Officer Nance later moved it, Officer Aragon testified that he positioned -it sideways so as to force Mr. Cordova either to move back across the highway into the proper lanes or to exit the highway via an on-ramp.
Both officers exited the police car, with Officer Nance positioned at the median in the center of the highway and Officer Ara*1187gon closer to the side. At this point there are key differences in each side’s version of the story. Officer Nance says that the truck appeared to be heading in his direction, so he drew his gun and pointed it at the truck. He claims that the truck then swerved left, away from him and toward Officer Aragon. Officer Aragon says that he then attempted to deploy the stop sticks, but quickly realized the truck was too close. He drew his gun. He claims that he was in immediate danger, about to be run over, and therefore rapidly fired at the vehicle while simultaneously trying to move out of the way. He fired either four or five shots, one of which hit Mr. Cordova in the back of the head, fatally wounding him and causing the truck to crash into a tree.
Several of the facts cast doubt on Officer Aragon’s claims of immediate danger, however. Only one bullet hit the front of the truck; the rest hit the side. The fatal shot, in fact, entered the truck from the side and went through the back of Mr. Cordova’s head. This strongly suggests that Mr. Cordova had turned the truck and was no longer bearing down upon Officer Aragon at the moment the officer fired the fatal shot. The Cordovas also contend that Officer Aragon was never in any immediate danger in the first place and that he could have easily avoided any and all risk by simply remaining behind his patrol car as the truck approached. Furthermore, they suggest that it was Officer Aragon’s own attempt to deploy the stop sticks in the truck’s only path of escape — a maneuver that might have violated CCPD policies — that created whatever danger might have existed. In short, it is not at all clear whether Officer Aragon was ever in any immediate danger at any point during the chase, and it in fact seems quite likely that whatever danger he might have perceived had passed by the time he fired the fatal shot.
Given these facts, the district court agreed with the Cordovas that a reasonable juror could find that Officer Aragon fired the fatal shots at a point when Mr. Cordova posed no immediate danger to the officers’ safety, and that Officer Aragon’s own actions had recklessly contributed to a situation where he perceived a need to shoot Mr. Cordova. Even so, the district court held that the sheer danger demonstrated by Mr. Cordova, who had persisted in a car chase in which he had repeatedly refused to stop, had run through at least two red lights, had driven off the road at least twice to avoid the deployment of stop sticks, had allegedly attempted to ram multiple officers’ patrol cars, and who was now towing a large piece of machinery on the wrong way down a highway at one in the morning, justified a reasonable officer in using deadly force to terminate the chase. Finding no Fourth Amendment violation, the district court granted the defendants’ motion for summary judgment.
On appeal, Officer Aragon and Commerce City do not challenge the factual assumptions the district court made when granting summary judgment — that is, they accept the district court’s premise that Officer Aragon was not in immediate danger and that no innocent bystanders were in the vicinity. See Aple. Br. 2 (“Officer Aragon and the City assume, for purposes of appeal, the factual determinations made by the lower Court.”). Like the district court, they focus instead on what is undisputed: that “Cordova had fled a lawful stop, attempted to ram multiple police vehicles, run two red lights, and had begun driving the wrong way down an interstate highway,” thereby “endangerfing] innocent motorists and police officers,” and argue that because of this the shooting was reasonable. Aple. Br. 26. For this reason, we accept the district court’s findings that a reasonable juror could find that Officer Aragon was not in immediate danger at *1188the time of the shooting, and we ask only whether the potential risk to third parties created by Mr. Cordova’s driving was alone sufficient to justify Officer Aragon’s shooting him.
II. Discussion
A. Did the Use of Force Constitute a Constitutional Violation?
A Fourth Amendment claim of excessive force is analyzed under the “objective reasonableness” standard that governs other Fourth Amendment inquiries. See Graham v. Connor, 490 U.S. 386, 395, 109 S.Ct. 1865, 104 L.Ed.2d 443 (1989) (“[AJll claims that law enforcement officers have used excessive force ... in the course of an arrest, investigatory stop, or other ‘seizure’ of a free citizen should be analyzed under the Fourth Amendment and its ‘reasonableness’ standard.”). We thus ask “whether the officers’ actions are ‘objectively reasonable’ in light of the facts and circumstances confronting them, without regard to their underlying intent or motivation.” Id. at 397, 109 S.Ct. 1865. Reasonableness “must be judged from the perspective of a reasonable officer on the scene,” who is “often forced to make split-second judgments — in circumstances that are tense, uncertain, and rapidly evolving — about the amount of force that is necessary in a particular situation.” Id at 396-397, 109 S.Ct. 1865.
There is no easy-to-apply legal test for whether an officer’s use of deadly force is excessive; instead, we must “slosh our way through the fact-bound morass of ‘reasonableness.’ ” Scott v. Harris, 550 U.S. 372, 383, 127 S.Ct. 1769, 167 L.Ed.2d 686 (2007). This sloshing requires us to weigh “the nature and quality of the intrusion on the individual’s Fourth Amendment interests against the importance of the governmental interests alleged to justify the intrusion.” Id. Some of the factors that we have found useful when conducting this balancing act include “the severity of the crime at issue, whether the suspect poses an immediate threat to the safety of the officers or others, and whether he is actively resisting arrest or attempting to evade arrest by flight.” Weigel v. Broad, 544 F.3d 1143, 1151-52 (10th Cir.2008) (citation omitted). Additionally, we have considered “whether the officers’ own reckless or deliberate conduct during the seizure unreasonably created the need to use such force.” Medina v. Cram, 252 F.3d 1124, 1132 (10th Cir.2001) (citation omitted). Once the relevant facts are determined with all reasonable inferences drawn in favor of the non-moving party at the summary judgment stage, “whether [the suspect’s] actions have risen to a level warranting deadly force ... is a pure question of law.” Scott, 550 U.S. at 381 n. 8,127 S.Ct. 1769.
The Supreme Court recently applied this balancing approach in Scott v. Harris, where the Court had “little difficulty in concluding” that an officer had acted reasonably when he terminated a car chase by ramming the bumper of the fleeing car. Id. at 384, 127 S.Ct. 1769. On one side of the scale, the suspect “posed an actual and imminent threat to the lives of any pedestrians who might have been present, to other civilian motorists, and to the officers involved in the chase,” and had also demonstrated his own culpability by “unlawfully engaging in the reckless, high-speed flight that ultimately produced the choice between two evils that [the officer] confronted.” Id. On the other side of the scale was the life of the suspect and the fact that the ramming “posed a high likelihood of serious injury or death to” the suspect. Id. Faced with the apparent equipoise of “the perhaps lesser probability of injuring or killing numerous bystanders against the perhaps larger probability of injuring or killing a single person,” id., the Court found that the culpability of the *1189suspect and innocence of the bystanders and officers tipped the balance in favor of reasonableness.
Importantly, though, while the threat to the suspect in Scott posed a “high likelihood” of serious injury or death, it did “not [pose] the near certainty of death posed by, say, shooting a fleeing felon in the back of the head, or pulling alongside a fleeing motorist’s car and shooting the motorist.” Id. (internal citations omitted). “A police car’s bumping a fleeing car is, in fact, not much like a policeman’s shooting a gun so as to hit a person.” Id. at 383, 127 S.Ct. 1769 (quoting Adams v. St. Lucie County Sheriff’s Dept., 962 F.2d 1563, 1577 (11th Cir.1992) (Edmondson, J., dissenting), adopted by 998 F.2d 923 (11th Cir.1993) (en banc) (per curiam)). The Court highlighted the difference with a comparison to Vaughan v. Cox, 343 F.3d 1323 (11th Cir.2003), a case that involved the “near certainty” of serious injury or death. In Vaughan, the officers allegedly drove alongside a fleeing truck and ended the flight by shooting three rounds into the truck without warning. Id. at 1327. The Eleventh Circuit found that if the jury believed this version of the facts, the shooting could constitute excessive force. Id. at 1332.
When discussing excessive force, we sometimes use the term “deadly force” as if it is a unitary concept. See, e.g., Sevier v. City of Lawrence, 60 F.3d 695, 699 (10th Cir.1995) (“Defendants’ use of deadly force was justified under the Fourth Amendment if a reasonable officer in Defendants’ position would have had probable cause to believe that there was a threat of serious physical harm to themselves or to others.”). As Scott makes clear, however, the term encompasses a range of applications of force, some more certain to cause death than others. It includes force that is “likely” to cause serious injury or death, such as ramming, and also includes force that is nearly certain to cause death, such as a shot to the head. Scott strongly suggests that the reasonableness balancing must take into account that there is a spectrum of “deadly force,” and that just because a situation justifies ramming does not mean it will justify shooting a suspect in the head.
This is not a case of ramming. Officer Aragon shot Mr. Cordova in the back of the head while he was driving — an application of force closer to that in Vaughan than to that in Scott. But there are other differences that cut the other way. The motorist in Vaughan was exceeding the speed limit by ten miles per hour and attempting to evade the police, but he remained in his lane of traffic and did not engage in highly reckless maneuvers. Id. at 1330. Mr. Cordova’s behavior — towing a large piece of machinery down the wrong way of a highway in the dark, and employing evasive maneuvers such as running red lights, driving off the road, and attempting to ram police cars — posed a much greater risk to police and innocent bystanders. Because there were no other motorists in the immediate vicinity, however, this risk was presumably less “imminent” than that posed by the driver in Scott. Cf. Scott, 550 U.S. at 379, 127 S.Ct. 1769 (describing how the suspect “swerve[d] around more than a dozen other cars, crossfed] the double-yellow line, and force[d] cars traveling in both directions to their respective shoulders to avoid being hit”). In this respect, the case falls somewhere in between Scott and Vaughan. We are therefore forced to consider whether the substantial but not imminent risk imposed on innocent bystanders and police by a motorist’s reckless driving justifies a reasonable officer to use a level of force that is nearly certain to cause the motorist’s death. This is not an easy question, or one that any court could feel confident in answering.
*1190This court has stated that “[w]here an officer has probable cause to believe that a suspect poses a threat of serious physical harm, either to the officer or others, it is not constitutionally unreasonable to prevent escape by using deadly force,” Weigel, 544 F.3d at 1152 (quoting Carr v. Castle, 337 F.3d 1221, 1227 (10th Cir.2003)), but that statement must not be read too broadly. It does not mean that any risk of physical harm to others, no matter how slight, would justify any application of force, no matter how certain to cause death. The Supreme Court’s careful distinction of Vaughan in Scott belies that sweeping a proposition. We do not believe it would be reasonable for an officer to shoot any motorist who ran a red light or swerved through lanes, simply because reckless driving poses some threat of physical harm to a bystander who might be down the road. Car chases inherently risk injury to persons who might happen along their course, and if that risk alone could justify shooting the suspect, every chase would end much more quickly with a swiftly-fired bullet. We do not mean to minimize that risk, or suggest that the risk to others must always be imminent in order to justify the use of deadly force, cf. Scott, 127 S.Ct. at 1778 (use of deadly force, though not of a level nearly certain to cause death, justified by “threat to lives of any pedestrians who might have been present”) (emphasis added), but “the Court’s decision in Scott did not declare open season on suspects fleeing in motor vehicles.” Lytle v. Bexar County, 560 F.3d 404, 414 (5th Cir.2009). When an officer employs such a level of force that death is nearly certain, he must do so based on more than the general dangers posed by reckless driving.
To the extent that the district court held that the hypothetical risk Mr. Cordova posed to fellow motorists who might happen along was itself enough to render the shooting reasonable, it was in error. The threat must have been more than a mere possibility. The facts show that Mr. Cordova was driving recklessly down the wrong side of the highway. The facts do not, however, show that any other motorists were in the vicinity, or that other motorists would not be able to spot Mr. Cordova and avoid an accident themselves. Mr. Cordova’s behavior did, of course, create risks for other motorists who might come along, but that risk of future harm was not enough to justify the near certainty of Mr. Cordova’s death.
The threat to the officers themselves — if actual and imminent — could of course shift the calculus in the direction of reasonableness. If a reasonable officer in Officer Aragon’s position would have feared for his life or the life of his fellow officers, then on one side of the scales would sit not only the potential (even if distant) risk to any motorist who might wander along, but also the very immediate risk of death to the pursuing officers. The urgency of terminating the chase would increase and the balance would tip in the officers’ favor. The use of deadly force — even of the level nearly certain to cause death — would likely be justified. Here, however, the threat to the officers is a disputed fact, and for purposes of the summary judgment motion the district court explicitly assumed the officers to be in no immediate danger.
It is with reluctance that we second-guess the split-second decisions of trained officers reacting to difficult situations in the line of duty. We are not well-suited to act as a police supervisory board, making finely calibrated determinations of just what type of misbehavior justifies just what level of response. We cannot say, however, that the general risks created by a motorist’s fleeing from the police are, without more, enough to justify a shooting that is nearly certain to cause that suspect’s death.
*1191Our esteemed colleague in dissent criticizes us for not delving more deeply into the record and instead accepting the district court’s construction of the undisputed facts of this case. But the dissent rests entirely on parts of the record the parties have not cited in their briefs, and in some cases on arguments they have expressly disavowed. It is not our role to sift through the record to find evidence not cited by the parties to support arguments they have not made. Gross v. Burggraf Constr. Co., 53 F.3d 1531, 1546 (10th Cir.1995) (“Without a specific reference, we will not search the record in an effort to determine whether there exists dormant evidence which might require submission of the case to a jury.” (internal quotation omitted)); State Farm Fire & Cas. Co. v. Mhoon, 31 F.3d 979, 984 n. 7 (10th Cir.1994) (if a party does not raise an argument in its opening brief, the argument is waived).
In support of its claim that Officer Aragon was indisputably in imminent danger, the dissent argues that the district court “went overboard” in construing the facts favorably to the nonmoving party, Dissent at 1200 n. 15, and criticizes us for “[wjith no analysis ... accept[ing][the] assumption” that Officer Aragon was not in immediate danger. Id. at 1200. Officer Aragon and Commerce City have not contested the district court’s assumptions, though, and indeed have expressly stated that “Officer Aragon and the City assume, for purposes of appeal, the factual determinations made by the lower Court.” Aple. Br. 2.2 Perhaps the dissent would have litigated the appeal differently, but the defendants’ decision not to challenge the district court’s analysis of the facts bearing on Officer Aragon’s danger constitutes a waiver. Mhoon, 31 F.3d at 984 n. 7.
Even aside from waiver, the undisputed fact that at least some of the shots hit the truck from the side, showing that the truck was no longer bearing down on Officer Aragon, and the testimony of a Commerce City police officer that “one trained in the use of that pistol in tactical shoot/no shoot scenarios would have no difficulty ceasing firing when the danger ceased,” Aple. Br. 18, suggest that a reasonable jury could disagree with the dissent’s confident determination that Officer Aragon was indisputably in immediate danger when he fired the fatal shot.
In support of its view that Mr. Cordova’s conduct posed an imminent danger to innocent motorists, the dissent challenges the district court’s conclusion that the record did not clearly show that other vehicles were in the vicinity. The dissent relies on a statement in Officer Aragon’s affidavit — referred to in neither party’s brief — that he had seen “three or four cars.” Dissent at 1205. But again, this was an issue the defendants explicitly chose not to contest. They state: “Plaintiffs also claim that the District Court’s conclusion that ‘no evidence suggests that westbound traffic was immediately approaching when Officer Aragon fired’ effectively ‘rules out danger to innocent motorists.’ ... Whether or not there were vehicles visible in the immediate area, there is no question that Cordova’s flight posed enormous danger to motorists and police officers.” Aple. Br. 23-24. In other words, rather than claim that the record indisputably shows that other vehicles were visible in the immediate area, the defendants accept that this is a disputed factual issue and argue that the danger was “enormous” under either version of the facts. When even the defendants treat *1192as an open question whether there were vehicles visible in the immediate area, we cannot join the dissent in treating the contrary as an undisputed fact.
Perhaps defense counsel would have been able to persuade a jury that the danger to Officer Aragon or to the public was indeed imminent, as the dissent believes, but that is not the issue. We do not hold that “Aragon’s conduct in shooting Cordova was unreasonable,” as the dissent asserts, Dissent at 1200, but hold only that the contrary has not been established as a matter of law, based on the undisputed facts in the record and in light of the parties’ arguments on appeal.
B. Was the Law Clearly Established?
Even if Mr. Cordova’s Fourth Amendment rights were violated, Officer Aragon is still entitled to qualified immunity if the right was not clearly established. See Christensen v. Park City Mun. Corp., 554 F.3d 1271, 1277 (10th Cir.2009). “Ordinarily, in order for the law to be clearly established, there must be a Supreme Court or Tenth Circuit decision on point, or the clearly established weight of authority from other courts must have found the law to be as the plaintiff maintains.” Weigel, 544 F.3d at 1153 (quoting Cruz v. City of Laramie, 239 F.3d 1183, 1187 (10th Cir.2001)). When a clearly established right depends on the application of a general principle of law to a particular set of facts, “its contours must be sufficiently clear that a reasonable official would understand that what he is doing violates that right.” Id. at 1154 (quoting Hope v. Pelzer, 536 U.S. 730, 739, 122 S.Ct. 2508, 153 L.Ed.2d 666 (2002)).
There is no question that the general principle governing the use of force is clearly established: deadly force is justified only if a reasonable officer in the officer’s position would have had probable cause to believe that there was a threat of serious physical harm to himself or others. See Graham, 490 U.S. at 396, 109 S.Ct. 1865; Sevier, 60 F.3d at 699; Wilson v. Meeks, 52 F.3d 1547, 1552-53 (10th Cir.1995). What Officer Aragon contests is whether a reasonable officer should have known that this general principle prevented him, in this case, from ending the car chase by shooting Mr. Cordova. Not every constitutional principle that involves a balancing test requires that the precise factors involved in the present case have previously been quantified and weighed, of course. The Supreme Court has “expressly rejected a requirement that previous cases be ‘fundamentally similar’ ” in order for the law to be clearly established, Hope v. Pelzer, 536 U.S. 730, 741, 122 S.Ct. 2508, 153 L.Ed.2d 666 (2002), and has held that “a general constitutional rule already identified in the decisional law may apply with obvious clarity to the specific conduct in question, even though the very action in question has not previously been held unlawful.” Id. (quoting United States v. Lanier, 520 U.S. 259, 271, 117 S.Ct. 1219, 137 L.Ed.2d 432 (1997)). In this case, however, substantial differences from prior case law convince us that the result of the balancing test had not been clearly established.
Two of the inputs into the balancing equation have particular importance in this case: the level of force used by the officer and the degree of risk posed to innocent bystanders. In Scott v. Harris, the risk to potential third parties was substantial, but the risk to the suspect, while high, was not certain to cause death. The Court found no constitutional violation. See also Cole v. Bone, 993 F.2d 1328, 1333 (8th Cir.1993) (officers aiming at engine, rather than driver, of semi truck in high-speed chase on crowded interstate did not act unreasonably, though the driver was hit). In Vaughan, in contrast, the risk to the suspect was high and death nearly certain, *1193but the risk to third parties was insubstantial. There, the court found a constitutional violation. See also Tennessee v. Garner, 471 U.S. 1, 11, 105 S.Ct. 1694, 85 L.Ed.2d 1 (“shooting nondangerous fleeing suspects is [not] so vital as to outweigh the suspect’s interest in his own life”) (emphasis added); Smith v. Cupp, 430 F.3d 766, 773 (6th Cir.2005) (driving a stolen police car did not by itself pose a grave enough danger to the public to justify shooting). Our case is a hybrid — the risk to potential third parties was as substantial, but less imminent, as in Scott and the level of force was nearly certain to cause the death of the suspect. Whether that situation is more like the police using deadly force (but not of a level certain to cause death) to prevent a substantial risk of harm to others, or more like the police shooting a suspect who poses minimal risk of harm to others, is not immediately clear. Officer Aragon was confronted with a situation that fell between these two lines of cases, and the result was uncertain.
In Lytle v. Bexar County, the Fifth Circuit recently addressed a case similar to ours. That court concluded that the general principle governing excessive force was sufficiently well-established that a reasonable officer would have known that “it is unreasonable for a police officer to use deadly force against a fleeing felon who does not pose a sufficient threat of harm to the officer or others.” 560 F.3d at 417. While this general principle is correct, it still begs the question of what constitutes a sufficient threat. In Lytle, the police shot at a fleeing Taurus which posed “some threat of harm.” Id. at 415. That threat, however, consisted of driving at high speeds through a residential area. While that level of threat might be obviously insufficient, it does not compare to the threat posed by Mr. Cordova, who was making evasive maneuvers while driving with extreme recklessness down the wrong way of a highway at night. Lytle is closer to Vaughan, where the certainty of the suspect’s death was weighed against the smaller risk of harm to the general public. It does not resolve how the reasonableness calculus should come out in the situation faced by Officer Aragon, where the risk of death is certain and the risk to third parties is substantial.
“The principles of qualified immunity shield an officer from personal liability when an officer reasonably believes that his or her conduct complies with the law.” Pearson v. Callahan, — U.S. -, 129 S.Ct. 808, 823, 172 L.Ed.2d 565 (2009). The law in our circuit and elsewhere has been vague on whether the potential risk to unknown third parties is sufficient to justify the use' of force nearly certain to cause death. Given that our precedent does authorize the use of deadly forcé when a fleeing suspect poses a threat of serious harm to others, Officer Aragon was not unreasonable in believing that a potential threat to third parties would justify such a level of force.
C. Municipal Liability
Though we hold that Officer Aragon is shielded from liability on the grounds that; even if a constitutional violation occurred, the law was not clearly established, the same qualified immunity analysis does not apply to municipalities. See Christensen, 554 F.3d at 1278. To succeed in a § 1983 claim against a municipality, a plaintiff must show two elements: “(1) a municipal employee committed a constitutional violation, and (2) a municipal policy or custom was the moving force behind the constitutional deprivation.” Walker v. City of Orem, 451 F.3d 1139, 1152 (10th Cir.2006) (internal citation omitted). As discussed above, we have found a genuine issue of material fact as to whether a constitutional violation occurred such that we cannot dispose of that issue on *1194summary judgment. We must therefore remand for the district court to consider the City of Commerce’s liability unless we find that the Cordovas have failed to create a genuine issue of material fact as to whether a municipal policy or custom was the moving force behind the constitutional deprivation.
A municipality is not liable for the constitutional violations of its employees simply because such a violation has occurred; a policy or custom must have actually caused that violation. See Christensen, 554 F.3d at 1279 (“The doctrine of respondeat superior cannot be employed to hold governmental entities liable under § 1983 for the constitutional torts of their employees.”); cf. Duffield v. Jackson, 545 F.3d 1234, 1239 (10th Cir.2008) (supervisor status alone insufficient to establish affirmative link between defendant and constitutional violation). The Cordovas admit that Commerce City’s official policy, which authorizes officers to use deadly force only “in protection of themselves or other persons from the immediate threat of death or serious bodily injury,” Dist. Op., 560 F.Supp.2d at 1050 (citing CCPD Policy and Procedures Manual § 2.1.11), mirrors the constitutional standard. Aple. Br. 32. Their claim is that the city’s actions deviated from this policy both in its training of officers and in failing to discipline Officer Aragon for his conduct.
As for any failure to discipline Officer Aragon, basic principals of linear time prevent us from seeing how conduct that occurs after the alleged violation could have somehow caused that violation. A subsequent cover-up might provide circumstantial evidence that the city viewed the policy as a policy in name only and routinely encouraged contrary behavior, see, e.g. Grandstaff v. City of Borger, 767 F.2d 161, 171 (5th Cir.1985) (“The disposition of the policymaker may be inferred from his conduct after the [violation occurred].”), though that hardly seems to be the case here, since the County Critical Incident Team conducted a full investigation and the CCPD police chief simply called off a second investigation as being unnecessarily duplicative. A failure to investigate or reprimand might also cause a future violation by sending a message to officers that such behavior is tolerated. It does not, however, in and of itself constitute a causal connection in the immediate case.
If the city promulgates a constitutional policy but trains its officers to violate that policy, however, a facially constitutional policy will not shield the city from liability and a causal connection could be established. The Cordovas point to only two statements, both made by Sergeant Robert McCoy, testifying as spokesman for the city, which they contend show a definite disregard for the city policy. First, when asked to describe when officers are taught they can apply deadly force, Sergeant McCoy said the suspect would have to have shown “the ability to commit an act against somebody” who was “actually in jeopardy or can be foreseen [to be] in jeopardy,” and that the suspect had “the opportunity to cause the harm to that person.” App. 434. Then, when asked if the word “immediate” as it appears in the policy actually meant “imminent,” he said he would agree. Id. 434-35. This is not the smoking gun the Cordovas make it out to be. First of all, as Scott v. Harris shows, the constitution permits deadly force even when the dangers to third parties are remote. It is that subcategory of force nearly certain to cause death that requires a more immediate threat. Even as applied to the specific case of shooting at a suspect, though, Sergeant McCoy’s agreeing that “immediate” and “imminent” mean the same thing does not strike us as a matter of constitutional significance. *1195The point is that the threat to another has to be real and concrete in order to justify taking the suspect’s life, and Sergeant McCoy interprets both the words “immediate” and “imminent” as meaning “ ‘Right here, right now,’ ” App. 435, a meaning perfectly consistent with the constitutional standard.
The Cordovas also point to a second portion of Sergeant McCoy’s testimony where he discusses the specific question of when an officer can shoot at fleeing suspects:
A. [Sgt. McCoy]. If you believe that the person is a danger to other officers and the public based upon the actions that they’ve done.
Q. Does it have to be immediate danger, or can it just be some danger in the far distant future?
A. I believe if the actions are such that he’s-he or she’s shown the propensity of doing that violence to the point of trying to hurt someone, kill someone, that you can stop that threat to the public, and it doesn’t have to be immediate.
Q. And that’s the policy of the Commerce City Police Department, correct? A. Yes, sir.
App. 379. Again, though, the Cordovas make more of this statement than it supports. Sergeant McCoy does not say that officers can shoot at any fleeing suspect who poses some general danger to the public. Instead, after being pressed repeatedly on the question of whether an officer might ever be justified in shooting, he concedes that there are some instances that might warrant it, as when a suspect has “shown the propensity of doing that violence to the point of trying to hurt someone, kill someone.” There is a difference between a suspect who is fleeing in an attempt to kill someone and a suspect whose reckless flight risks killing someone. The former suspect would pose a real and concrete threat to an innocent party, even though that threat might not materialize within the course of the chase; the latter’s danger is much more speculative. Had Mr. Cordova been a serial killer whose escape would be nearly certain to result in future victims, we have no doubt the constitutional calculus would change and a shooting might be justified, whether other motorists were in his path or not.
It bears emphasizing that the constitutional standard that we elaborate upon today is hardly a mathematical formula with easily defined lines. We do not hold that an officer would never be justified in shooting a fleeing suspect, and Sergeant McCoy’s reluctance to frame Commerce City’s policy in such absolute terms shows why: policemen encounter a variety of unexpected situations that pose a range of dangers to the public they are sworn to protect. We hold simply that a distant risk to hypothetical motorists posed by a suspect’s reckless driving does not alone justify shooting that suspect. Commerce City’s policy, neither as written nor as described by Sergeant McCoy, does not teach otherwise.
III. Conclusion
When construed in the light most favorable to the non-moving party, the facts show that Officer Aragon used a level of force nearly certain to cause Mr. Cordova’s death, though without any immediate threat to himself or others. Those facts, if true, would constitute a violation of Mr. Cordova’s Fourth Amendment right to be free from unreasonable seizure. Because the law was not clear on how high the risk of harm to third parties must be before an officer can use a level of force nearly certain to cause death, however, we AFFIRM on qualified immunity grounds the district court’s grant of summary judgment as it applies to Officer Aragon. We also AFFIRM the grant of summary motion as it applies to Commerce City on the ground *1196that the Cordovas have not created a genuine issue of material fact as to whether a policy or custom of the city was the moving force behind any violation that might have occurred.
. The district court found that a reasonable juror could consider this decision to cross the median to have been reckless.
. Because this case was decided on summary judgment, the district court did not make any true "factual determinations,” but it carefully set forth its understanding of which facts were undisputed. It is to these determinations the appellees are referring.