concurring:
I concur fully in Judge Thomas’s opinion and write separately to make two points, one brief and one less so.
I.
Judge Thomas understates the problem with the eponymous and notorious Edward Fink, if that is possible. In the case that Judge Thomas cites, Thompson v. Calderon, 120 F.3d 1045 (9th Cir.1997) (en banc), rev’d, 523 U.S. 538, 118 S.Ct. 1489, 140 L.Ed.2d 728 (1998), Thomas Thompson was executed on the basis of Fink’s perjured testimony. It is unlikely that Thompson was death-eligible for his part in the crime, if he was guilty at all of any offense. See generally Stephen Reinhardt, The Anatomy of an Execution: Fairness vs. ‘Process”, 74 N.Y.U. L. Rev. 313, 322-26 (1999). At Thompson’s trial, the prosecutor committed prosecutorial misconduct that constituted a constitutional violation and required reversal. See Thompson, 120 F.3d at 1055. Additionally, Thompson’s lawyer provided woefully inadequate representation, which constituted another constitutional violation that required reversal. See id. at 1053-54.
Despite a request to reverse Thompson’s conviction by seven California prosecutors with. extensive death penalty experience, including the author of California’s death penalty statute, the Supreme Court refused to consider Fink’s perjured testimo*763ny or any of the constitutional violations, on the dubious ground that our court abused its discretion in recalling the mandate, on a basis that the Court had never before recognized. Calderon v. Thompson, 523 U.S. 538, 118 S.Ct. 1489, 140 L.Ed.2d 728 (1998).
Although Thompson was executed as a result of Fink’s perjury (as well as the other unfortunate judicial matters described above), the innocent Mr. Goldstein was fortunate enough to avoid that fate. See Goldstein v. Harris, 82 Fed.Appx. 592, 593 (9th Cir.2003) (affirming the district court’s grant of habeas relief). He now seeks civil damages for spending twenty-four years of his life in prison, as a result of the Los Angeles County District Attorney’s Office’s failure to adopt necessary and reasonable internal administrative policies and procedures. I agree that he is entitled to pursue his claim.
II.
One of the principal arguments on which the County relies is the California Supreme Court’s decision in Pitts v. County of Kern, 17 Cal.4th 340, 70 Cal.Rptr.2d 823, 949 P.2d 920, 923 (1998), in which that court held that “the district attorney represents the state, not the county, when preparing to prosecute and when prosecuting crimes, and when establishing policy and training employees in these areas.” I agree with Judge Thomas that we are not bound by Pitts and that the policy at issue in Pitts differs in kind from the administrative procedure that the District Attorney failed to implement in this case. I write separately, however, to explain why I find the reasoning in Pitts to be unpersuasive. In short, the California Supreme Court’s reasoning in Pitts is imprecise on a question that demands precision.
To begin, the California Supreme Court never clearly states the scope of its holding. On multiple occasions, the state court writes that the setting of policy and training of employees “in these areas” is a state function. E.g., Pitts, 70 Cal.Rptr.2d 823, 949 P.2d at 923; id., 70 Cal.Rptr.2d 823, 949 P.2d at 934. I can only presume that “in these areas” refers to the prosecution of crimes. The problem is that virtually every “policy” in a district attorney’s office has some relationship to the prosecution of crimes. Some policies are directly related to the prosecutorial function — e.g., a policy that prosecutors must instruct their witnesses to tell the truth at all times when testifying. Some procedures are indirectly related, if at all — e.g., a procedure governing the disposal of confidential material in the office. I would call the latter category of procedures “administrative.” The real question then is whether the matter at issue is prosecutorial or administrative. If prosecutorial, then the district attorney— under the circumstances present in Pitts— serves as a state actor. If administrative, then the district attorney — as Judge Thomas’s opinion for the court explains— serves as a county actor.
The California Supreme Court in large part avoided this question in Pitts. It stated that drawing lines would “require impossibly precise distinctions.” Pitts, 70 Cal.Rptr.2d 823, 949 P.2d at 935. It further declared that “no meaningful analytical distinction can be made between” individual prosecutions and the setting of policy and training of employees. Id. Thus, it suggested that the establishing of all policies (including dress codes and the like) and the training of all employees (presumably including secretaries and janitors) is a state function, as well as the prosecution of all cases.
I find the California Supreme Court’s imprecise answer unpersuasive, for three reasons. First, the California Supreme Court failed to follow the process set forth *764in McMillian, which is for courts to make an independent determination “whether governmental officials are final policymakers for the local government in a particular area or on a particular issue.” McMillian v. Monroe County, 520 U.S. 781, 785, 117 S.Ct. 1734, 138 L.Ed.2d 1 (1997) (emphasis added). In Pitts, however, the California Supreme Court made no function-specific determination. Rather, it relied primarily on its conclusion that a district attorney is a state actor when conducting individual prosecutions and asserted that it “logically follow[ed]” that a district attorney was also a state actor when setting policy or training employees. Pitts, 70 Cal.Rptr.2d 823, 949 P.2d at 934-35.1 In contrast, in our opinion for the court, we conduct a close examination of the relevant constitutional and statutory provisions and conclude that the “Los An-geles County District Attorney represents the county when establishing administrative policies and training related to the general operation of the district attorney’s office, including the establishment of an index containing information regarding the use of jailhouse informants.” Maj. Op. at 759-60.
Second, the California Supreme Court’s discussion is not internally consistent. Despite asserting that distinguishing policy and training decisions from individual prosecutorial decisions would require “impossibly precise distinctions.” Pitts, 70 Cal.Rptr.2d 823, 949 P.2d at 935, the state court stated that its conclusion “might differ were plaintiffs challenging a district attorney’s alleged action or inaction related to hiring or firing an employee ... or some other administrative function.” Pitts, 70 Cal.Rptr.2d 823, 949 P.2d at 935. This distinction eludes me. The training of employees is no more related to individual prosecutions than the hiring, firing, or disciplining of those employees. If “no meaningful analytical difference” exists between individual prosecutions and employee training, then it should not exist between individual prosecutions and employee hiring or firing; yet the California Supreme Court suggested that the latter might yield a different conclusion. A better explanation for the California Supreme Court’s inconsistency is that it must have recognized that its conclusion would otherwise call into question a vast swath of Section 1983 jurisprudence, in which counties are generally held liable for failure to hire or fire decisions. See Bd. of Cnty. Commis. v. Brown, 520 U.S. 397, 117 S.Ct. 1382, 137 L.Ed.2d 626 (1997). Unlike the California Supreme Court, we reach a consistent conclusion here: counties should be held liable for their administrative policies and procedures, including the training of their employees and the hiring and firing of those employees. See, e.g., Connick v. Thompson, — U.S.-, 131 S.Ct. 1350, 1359, 179 L.Ed.2d 417 (2011).
Third, the California Supreme Court was simply incorrect when it stated that its conclusion follows as a matter of “logic.” The entirety of our opinion contradicts this assertion of logical inference. As we have demonstrated through a close examination of the various constitutional and statutory provisions, a district attorney may act on behalf of the state when making prosecutorial decisions but act on behalf of the county when setting administrative policy or training employees. See Maj. Op. at 754-60. I therefore agree with the highly respected California Supreme Court Justice Stanley Mosk who wrote:
*765[T]here is no insurmountable analytical difficulty to concluding that a county cannot be held liable under section 1983 when the district attorney or one of his or her deputies, as an agent of the state, commits prosecutorial misconduct, but can be held liable when the district attorney’s hiring, training and supervision program, which the district attorney undertakes as a local policymaker, results in injury to a person’s civil rights.
Pitts, 70 Cal.Rptr.2d 823, 949 P.2d at 940 (Mosk, J., dissenting). Indeed, several circuits have come to the same conclusion that we reach here, that district attorneys act as county officers when deciding administrative policy and procedures related to training or supervision, even though they act as state officers when conducting prosecutions. E.g., Walker v. City of New York, 974 F.2d 293, 296 (2d Cir.1992); Carter v. City of Philadelphia, 181 F.3d 339, 352 (3d Cir.1999); Esteves v. Brock, 106 F.3d 674, 678 (5th Cir.1997); Owens v. Fulton County, 877 F.2d 947, 952 (11th Cir.1989). Thus, I believe that there can be and is a “meaningful analytical distinction” between prosecutorial decisions and the creation of administrative policy or the training of employees.
I do not suggest that it is always easy to distinguish between policies that are pros-ecutorial in nature and procedures that are administrative in nature. That point notwithstanding, it is the proper inquiry. Line drawing is frequently a difficult task for jurists. It is, however, one we perform regularly. As Justice Oliver Wendell Holmes wrote almost a century ago:
Neither are we troubled by the question where to draw the line. That is the question in pretty much everything worth arguing in the law. Day and night, youth and age are only types.
Irwin v. Gavit, 268 U.S. 161, 168, 45 S.Ct. 475, 69 L.Ed. 897 (1925) (internal citation omitted); see also Dominion Hotel v. State of Arizona, 249 U.S. 265, 269, 39 S.Ct. 273, 63 L.Ed. 597 (1919) (Holmes, J., for the court) (“[T]he constant business of the law is to draw such lines.”). Nor is he the only jurist of the Supreme Court to recognize this point.2 The Pitts court abdicated its judicial function by adopting a rule that avoids a case-by-case inquiry into whether a particular function at issue is prosecuto-rial or administrative.
. It made only a passing reference to one provision of the California Constitution, Article V, Section 13, which is not specific to the function at issue in Pitts or in this case. See Maj. Op. at 755-57.
. 10 E. 40th St. Bldg. v. Callus, 325 U.S. 578, 584-85, 65 S.Ct. 1227, 89 L.Ed. 1806 (1945) (Frankfurter, J., for the court) (“On the terms in which Congress drew the legislation we cannot escape the duty of drawing lines. And when lines have to be drawn they are bound to appear arbitrary when judged solely by bordering cases. To speak of drawing lines in adjudication is to express figuratively the task of keeping in mind the considerations relevant to a problem and the duty of coming down on the side of the considerations having controlling weight. Lines are not the worse for being narrow if they are drawn on rational considerations.”); Rosenberger v. Rector & Visitors of Univ. of Virginia, 515 U.S. 819, 847, 115 S.Ct. 2510, 132 L.Ed.2d 700 (1995) (O’Connor, J., concurring) ("Reliance on categorical platitudes is unavailing. Resolution instead depends on the hard task of judging— sifting through the details and determining whether the challenged program offends.... Such judgment requires courts to draw lines, sometimes quite fine, based on the particular facts of each case.”). These are but a handful of the examples in which Justices of the Supreme Court have recognized that line-drawing is inherent in the task of judging.