[Nos. 56341-1, 56398-5. En Banc. January 9, 1992.]
TAGGART v. STATE
VICTORIA TAGGART, Appellant, v. THE STATE OF WASHINGTON, Respondent.
SHANE SANDAU, Appellant, v. THE STATE OF WASHINGTON, Respondent.
 Judgment – Summary Judgment – Review – In General. In reviewing a summary judgment, an appellate court engages in the same inquiry as the trial court and must consider all facts submitted and all reasonable inferences that can be drawn from those facts in the light most favorable to the nonmoving party.
 Government – Torts – Action Against Governmental Entity – Quasi-Judicial Conduct – Judicial Immunity – When Applied. Governmental agencies and executive branch officials are protected by judicial immunity from civil damages suits when in the course of their official duties they perform quasi-judicial functions, i.e., activities functionally resembling those performed by a judge.
 Government – Torts – Action Against Governmental Entity – Quasi-Judicial Conduct – How Determined. Whether an administrative action qualifies as being functionally comparable to judicial action is determined by looking at whether (1) a hearing to resolve an issue or controversy was held, (2) objective standards were applied, (3) a binding determination of individual rights was made, (4) it was an action that courts have historically performed, and (5) there were safeguards to protect against errors.
 Government – Torts – Action Against Governmental Entity – Parole Board – Judicial Immunity – When Applied. The Board of Prison Terms and Paroles (now the Indeterminate Sentence Review Board) is absolutely immune under the doctrine of judicial immunity for its decisions to deny, grant, or revoke parole.
 Government – Torts – Action Against Public Officer – Quasi-Judicial Immunity – When Applied – Supervisory or Administrative Activities. A public official who performs quasi-judicial functions is not shielded from all claims by the doctrine of quasi-judicial immunity. There is no immunity protecting against claims arising from purely supervisory or administrative activities.
 Government – Torts – Action Against Public Officer – Parole Officers – Parole Supervision – Quasi-Judicial Immunity. A parole officer's failure to perform a supervisory activity is not protected by judicial or quasi-judicial immunity.
 Government – Sovereign Immunity – Discretionary Acts – Parole Supervision. A claim against a parole officer for failing to supervise a parolee is not barred by the discretionary immunity exception to the State's waiver of sovereign immunity.
 Government – Torts – Action Against Public Officer – Parole Officers – Parole Supervision – Qualified Immunity. Parole officers are qualifiedly immune from liability for claims of negligent parole supervision if their actions are in furtherance of a statutory duty and are in substantial compliance with the directives of superiors and relevant regulatory guidelines.
 Government – Torts – Action Against Public Officer – Parole Officers – Parole Supervision – Public Duty Doctrine. A parole officer's authority under RCW 72.04A.080 to supervise and report on parolees creates a definite, established, and continuing relationship between the parole officer and parolee. Under the statute, the parole officer has "taken charge" of the parolee such that the parole officer has a duty to exercise reasonable care to control the parolee and to prevent the parolee from causing physical harm to other persons when the parolee's criminal and parole history demonstrate the foreseeability that bodily harm will be inflicted on others if not controlled. Liability for failing to carry out this duty arises only after it is shown that neither absolute nor qualified immunity bars an action against the defendant parole officer.
 Government – Torts – Action Against Public Officer – Parole Officers – Parole Supervision – "Taking Charge". A parole officer "takes charge" of a parolee under the officer's supervision regardless of there not being a custodial relationship or continuing hourly or daily dominance and dominion.
 Negligence – Proximate Cause – Elements. Proximate cause is composed of two elements: (1) cause in fact, i.e., "but for" a certain act, events flowing in a direct, unbroken sequence from the act would not have occurred; and (2) legal causation, i.e., how far the defendant's responsibility for the consequences of the defendant's actions should extend based on considerations of policy and common sense.
 Government – Torts – Action Against Governmental Entity – Parole Officers – Parole Supervision – Proximate Cause. A parole officer's breach of the duty to control a parolee when it is reasonably foreseeable that the parolee will inflict bodily harm on others can be a proximate cause of injuries inflicted by that parolee against another person.
BRACHTENBACH, J., concurs in the result only; UTTER, J., concurs by separate opinion; GUY, J., dissents in part by separate opinion; ANDERSEN and JOHNSON, JJ., did not participate in the disposition of this case.
Nature of Action: Plaintiff Taggart brought claims of negligent parole release and supervision and plaintiff Sandau brought claims of negligent parole supervision against the State and its agents after each plaintiff had been injured by parolees in separate assaults.
Superior Court: The Superior Court for King County, No. 86-2-17127-4, Faith Enyeart, J., on June 30, 1989, entered a summary judgment dismissing plaintiff Taggart's claims; the Superior Court for King County, No. 86-2-16252-6, John M. Darrah, J., on July 13, 1989, entered a summary judgment dismissing plaintiff Sandau's claims.
Supreme Court: Holding that the Board of Prison Terms and Paroles (now the Indeterminate Sentence Review Board) is entitled to absolute immunity for its release decisions but that the negligent parole supervision claims against the individual parole officers should not have been dismissed, the court affirms the judgment dismissing the negligent parole release claim against the Board in the Taggart case, reverses the judgments dismissing the negligent parole supervision claims in both cases, and remands both cases for further proceedings.
Mark Leemon and Fury Bailey, for appellants.
Kenneth O. Eikenberry, Attorney General, Michael E. Tardif, Senior Assistant, and Jon P. Ferguson, Francois L. Fischer, and Steven L. Abel, Assistants, for respondent.
Fred Diamondstone on behalf of Families and Friends of Missing Persons and Violent Crime Victims, amicus curiae for appellants.
CALLOW, J.* «1»
«1» *Justice Keith M. Callow is serving as a justice pro tempore of the Supreme Court pursuant to Const. art. 4, § 2(a) (amend. 38).
– Shane Sandau and Victoria Taggart were injured by parolees in separate assaults. Taggart raised claims against the State of Washington and its agents for allegedly negligent parole release and supervision. Sandau raised only negligent parole supervision claims. The trial courts in each case granted the defendant's motion for summary judgment and dismissed. The cases have been consolidated and are before this court on direct review. The questions presented are: (1) whether immunity shields the Indeterminate Sentence Review Board «2»
«2» Prior to July 1, 1986, the Indeterminate Sentence Review Board was known as "the Board of Prison Terms and Paroles". RCW 9.95.001.
(Board) or individual parole officers from claims alleging negligent parole release or supervision; (2) whether the public duty doctrine bars Taggart's and Sandau's claims alleging negligent parole supervision; and (3) whether as a matter of law neither the State nor its agents proximately caused Sandau's or Taggart's injuries.
We hold that the Board is entitled to absolute immunity for its release decisions, and affirm the trial court's dismissal of Taggart's negligent release claim. We hold that parole officers' supervisory actions are entitled to only qualified immunity, the absence of which Taggart and Sandau should have been given the opportunity to demonstrate at trial. As regards the public duty doctrine and proximate causation, we hold that neither provided a proper basis for dismissing the negligent supervision claims. We reverse and remand those claims for trial.
 The trial courts in these cases dismissed the plaintiffs' claims on the State's motions for summary judgment. This court must therefore engage in the same inquiry as the trial court, considering all facts submitted and all reasonable inferences therefrom in the light most favorable to the nonmoving parties. Wilson v. Steinbach, 98 Wn.2d 434, 437, 656 P.2d 1030 (1982).
Taggart v. State
Parolee Louie Brock assaulted Taggart on April 16, 1985. Brock had an extensive juvenile and adult criminal history. At the age of 15 he was referred to the juvenile court on charges of first degree burglary, having been initially charged with attempted rape. While still a juvenile, Brock was also charged with assault and another attempted rape. Brock's first conviction as an adult was in 1970, at the age of 18, for assault with intent to commit rape. The victim was a 70-year-old woman. At that time, Brock was diagnosed as exhibiting sexual deviation, episodic excessive drinking, antisocial personality, and passive aggressive personality. Brock was released from prison on parole in 1974, only to be arrested 2 months later for the assault and attempted rape of a 75-year-old woman. Both the 1970 and the 1974 attempted rapes involved alcohol abuse. In 1981, Brock was again paroled, and again was arrested within a year, this time for stealing a car, driving while intoxicated, and resisting arrest. At that time Brock was evaluated as being highly susceptible to alcohol abuse, and was reported to have missed counseling appointments and to have no interest or motivation for counseling.
Just prior to his third parole release in September 1984, Brock completed a 6-month counseling program that addressed sexual offender tendencies and substance abuse. Brock did well in this program. The program's deputy executive director described him as "amicable, cooperative, energetic, hard-working, [and] self-sufficient." Prior to his release, Brock prepared a preparole referral plan, which was investigated by parole officer Leda Richardson. Richardson recommended approval of the plan, subject to special conditions, including that Brock complete a substance abuse program and submit to urinalysis testing to ensure that he was not using drugs or alcohol during parole.
A Board meeting was held on September 10, 1984, to evaluate Brock's eligibility for parole. Two members of the Board were present. At that meeting, the Board considered extensive materials, including Richardson's report, Brock's preparole referral plan, and Brock's criminal history, psychological evaluations, and history of institutional conduct. On the basis of this material, the Board decided to release Brock on parole, effective September 24.
Upon his release, Brock entered a halfway house for recently released prisoners. He left the halfway house after 4 months. Richardson, who became Brock's parole supervisor, did not require that he obtain further counseling after that. At no time did Richardson require that Brock submit to urinalysis. Her monitoring of Brock consisted of seeing him weekly in her office. Richardson never contacted Brock's employers or girl friend about Brock. If she had, she probably would have learned that Brock was drinking regularly.
Brock's assault on Taggart occurred in April 1985, after he had been on parole approximately 7 months. Not previously acquainted, they met in a bar and socialized briefly. Brock left the bar shortly before Taggart. When Taggart left and walked out to her car, she found Brock in the parking lot waiting for her. He demanded that she give him a ride to a bus stop, and she agreed. Once they were in the car, he ordered her to turn down a side street. Taggart instead pulled into a 7-eleven store parking lot and asked Brock to get out of the car. He then attacked her, causing serious injuries.
Sandau v. State
Parolee Keith Geyman repeatedly raped Shane Sandau on September 21, 1984. Sandau was 9 years old at the time. Like Brock, Geyman had an extensive juvenile and adult criminal history. At the time of his parole on February 15, 1984, he was serving a sentence for second degree assault in which he had stabbed a man in the chest. An alcoholic, Geyman was usually intoxicated when he committed his crimes. The Board directed him to enter the Kitsap County Alcoholism Recovery Program upon release. Geyman completed the program on March 14, 1984, and reported to his parole officer, Richard Van Stralen, the next day. On March 20, Van Stralen received a report that Geyman was drinking, contrary to the conditions of his parole, and that he was threatening his ex-wife's husband. Van Stralen did not follow up on the report, and on March 21 Geyman's supervision was transferred to James Kairoff.
Geyman met with Kairoff on April 5, 1984. Kairoff asked Geyman about the report of drinking and threats. Geyman said the report was false and Kairoff did not investigate further. After this April 5 meeting, Geyman did not report on a monthly basis as required. Although Geyman was a "maximum supervision" parolee, Kairoff took no further action until July 27, 1984, when he called Geyman's brother Gilbert. Geyman's parole conditions required that he live and work with Gilbert. Gilbert told Kairoff that Geyman was no longer living or working with him, and that he had left no forwarding address. Geyman was thus violating parole. Kairoff completed a violation report and suspended Geyman's parole. The parole warrant, however, was never entered into the state computer.
On September 18, Kairoff received a phone call from a woman identifying herself as "Diana". Diana told Kairoff that Geyman was in Missoula, Montana, that he had been beating his girl friend and her children, and that he was drinking and bragging of his status as a parolee and of his crimes. The girl friend was Wanda Hazel, Sandau's mother. Kairoff then spoke with Detective Rick Newlon from the Missoula sheriff's department. Newlon confirmed Diana's report, and said Hazel feared Geyman but that no local charges had been filed. Newlon also said that Geyman could be arrested on a local misdemeanor charge, but was concerned that Geyman could make bail on that charge. Newlon, therefore, said he would prefer to make the arrest on the basis of a Washington parole warrant. Kairoff received a phone call from Hazel the same day. She indicated that she was the victim described by Diana, and that she was undecided whether to file charges against Geyman. She also wanted to know about Geyman's convictions.
In response to these calls, Kairoff immediately attempted to obtain a "fast entry" warrant. He contacted a supervisor, Art Wheeler, who told him to prepare another violation report and written request for warrant. Kairoff expressed concern to Wheeler that the matter should be handled more quickly, but proceeded as Wheeler recommended. Also on September 18, Montana law enforcement authorities sent a teletype to the Board indicating that they had spoken with Kairoff that morning and believed a teletype parole warrant would be forthcoming. They stated that they had officers "standing by" to arrest Geyman. This teletype was received and routed to hearing officer James Prentice. Prentice's usual practice was to act on such a teletype the day he received it and forward it to the members of the Board immediately. For some reason, however, Prentice took no immediate action. Geyman remained free, and 2 days later, on September 20, he raped Sandau. Unaware of the rape but concerned about the risk Geyman posed, the Montana authorities arrested Geyman on the outstanding local misdemeanor on September 23. Also unaware of the rape, Prentice responded to the Montana authorities' extradition request on September 27, recommending against extradition. The Board overruled him the next day, and an extradition warrant was immediately issued. Geyman meanwhile had been sentenced to a 6-month jail term on the Montana misdemeanor charge, but was released on October 4. In mid-October, Sandau told his mother about the rape. Geyman was arrested for that crime on October 22. He was charged with and convicted of deviate sexual conduct, a felony, and was sentenced to 40 years' imprisonment with 10 years suspended.
A. The Board's Immunity.
 Judges are immune from civil damages suits for acts performed within their judicial capacity. See, e.g., Pierson v. Ray, 386 U.S. 547, 18 L. Ed. 2d 288, 87 S. Ct. 1213 (1967); Adkins v. Clark Cy., 105 Wn.2d 675, 717 P.2d 275 (1986); Burgess v. Towne, 13 Wn. App. 954, 538 P.2d 559 (1975); cf. Pulliam v. Allen, 466 U.S. 522, 80 L. Ed. 2d 565, 104 S. Ct. 1970 (1984) (judicial immunity is not a bar to civil rights suits under 42 U.S.C. § 1983 seeking prospective injunctive relief or to an award of attorney fees in such a suit). The purpose of this immunity is not to protect judges as individuals, but to ensure that judges can administer justice without fear of personal consequences. Adkins, 105 Wn.2d at 677. If disgruntled litigants could raise civil claims against judges, then "judges would lose 'that independence without which no judiciary can either be respectable or useful.'" Butz v. Economou, 438 U.S. 478, 509, 57 L. Ed. 2d 895, 919, 98 S. Ct. 2894 (1978) (quoting Bradley v. Fisher, 80 U.S. (13 Wall.) 335, 347, 20 L. Ed. 646 (1872)).
The purpose of the judicial immunity doctrine is illuminated by an examination of its history. Prior to the development of an appellate system in early English law, suits against courts, or "actions of false judgment", were a mechanism for correcting judicial errors. Block, Stump v. Sparkman and the History of Judicial Immunity, 1980 Duke L.J. 879, 881-85 (hereinafter Block). As the hierarchical appellate system developed, the need for such suits diminished. Block, at 885. Judicial immunity then developed as a way "to protect the appellate system from collateral attacks on judgments, thus channeling actions upward through the appellate hierarchy for the correction of error." Block, at 924. In short, the historical purpose of judicial immunity indicates the importance of mechanisms to protect against judicial errors. Ideally, the doctrine should be applied only when the system is otherwise structured to provide safeguards against judicial errors. Cf. Engelmohr v. Bache, 66 Wn.2d 103, 105-06, 401 P.2d 346 (1965) (absolutely privileged communications not recognized in administrative proceedings lacking safeguards similar to those in judicial proceedings).
Judicial immunity also extends to governmental agencies and executive branch officials performing quasi-judicial functions. For example, Butz v. Economou, supra, held that judicial immunity shields federal administrative agency officials who participate in agency adjudications. 438 U.S. at 512-13. The courts of this State have reached similar conclusions. See, e.g., Pleas v. Seattle, 112 Wn.2d 794, 809-10, 774 P.2d 1158 (1989) (recognizing judicial immunity for a city council's rezoning decision); Layne v. Hyde, 54 Wn. App. 125, 773 P.2d 83 (holding that judicial immunity shields administrative law judges), review denied, 113 Wn.2d 1016 (1989); Rayburn v. Seattle, 42 Wn. App. 163, 709 P.2d 399 (1985) (granting quasi-judicial immunity to police pension board's decision denying a disability claim), review denied, 105 Wn.2d 1007 (1986). The essential question in cases applying the quasi-judicial immunity doctrine is whether the challenged actions were functionally similar enough to those performed by a judge to warrant the immunity. See, e.g., Butz, 438 U.S. at 513; see generally W. Keeton, D. Dobbs, R. Keeton & D. Owen, Prosser and Keeton on Torts 1057 (5th ed. 1984). When an administrative action resembles judicial action, the rationale behind granting judges immunity – the need for independent and impartial decisionmaking – applies with equal force.
 Washington courts have formulated various tests for determining whether administrative action is functionally comparable to judicial action and therefore quasi judicial. Compare Rayburn, 42 Wn. App. at 165 (stressing only basic principle of similarity to judicial process) with Layne, 54 Wn. App. at 132 (stressing functional similarity and expectations of the parties) and Williams v. Seattle Sch. Dist. 1, 97 Wn.2d 215, 218, 643 P.2d 426 (1982) (stating a 4-part test, each element of which articulates concern with similarity to judicial action). In order to determine whether an administrative action is functionally comparable to judicial action, however, one must first define judicial action, a precise definition of which is probably neither possible nor desirable. Although the proceedings properly called "judicial" share similarities, no one attribute is essential to qualify an action as judicial, provided the action has enough other relevant attributes. Raynes v. Leavenworth, 118 Wn.2d 237, 243, 821 P.2d 1204 (1992); see generally L. Wittgenstein, Philosophical Investigations ¶¶ 65-67 (1958) (explaining concepts defined only in terms of "family resemblances"). Therefore whether a challenged administrative action is functionally comparable to judicial action depends on various factors, such as whether a hearing was held to resolve an issue or controversy, whether objective standards were applied, whether a binding determination of individual rights was made, whether the action is one that historically the courts have performed, and whether safeguards exist to protect against errors.
Other jurisdictions that have addressed the issue presented here have recognized the quasi-judicial nature of parole boards, and so have held that a parole board and its members are entitled to quasi-judicial immunity from suits raised by private citizens injured by parolees. For example, in Tarter v. State, 68 N.Y.2d 511, 503 N.E.2d 84, 510 N.Y.S.2d 528 (1986), persons shot by a parolee less than 2 months after his release sued the state parole board. In holding that the parole board was absolutely immune from suit, the New York Court of Appeals stated:
The functions of the impartial Board in deciding whether to grant parole are classically judicial tasks. The Board must measure the facts of a particular inmate's case against the backdrop of the guidelines. Just as a Judge performs the original sentencing function, weighing the defendant's particular situation against case law, sentencing statutes and the Judge's prior experience, the Board must fit the inmate's factual circumstances within the guidelines and use its discretion in its disposition of the matter. Both are decisions which involve the officials' expertise, an application of law and an exercise of their judgment. We do not suggest that every official act involving discretion will be considered a judicial function conferring absolute immunity. It is the peculiar nature of the duties of the Board of Parole with respect to the weighing of evidence, deciding the relative importance of the determining factors and the ultimate discretionary disposition which render it so. Therefore, the Board's actions are entitled to absolute immunity, notwithstanding that discovery could prove claimants' allegations.
(Citations omitted.) Tarter, at 518-19; see also State v. Mason, 724 P.2d 1289 (Colo. 1986) (quasi-judicial immunity from negligent release claim granted to both parole board and its members); Harlow v. Clatterbuck, 230 Va. 490, 339 S.E.2d 181 (1986) (same); Nelson v. Balazic, 802 F.2d 1077 (8th Cir. 1986) (same); see generally Annot., Governmental Tort Liability for Injuries Caused by Negligently Released Individual, 6 A.L.R.4th 1155 (1981); Annot., Immunity of Public Officer From Liability for Injuries Caused by Negligently Released Individual, 5 A.L.R.4th 773 (1981). Similarly, parole boards are commonly granted quasi-judicial immunity from civil rights suits brought by prisoners whose parole applications were denied. See, e.g., Johnson v. Rhode Island Parole Board Members, 815 F.2d 5 (1st Cir. 1987); Douglas v. Muncy, 570 F.2d 499 (4th Cir. 1978); Johnson v. Kegans, 870 F.2d 992 (5th Cir.), cert. denied, 492 U.S. 921 (1989); Anderson v. Boyd, 714 F.2d 906 (9th Cir. 1983).
 We agree that when parole board members decide whether to deny, grant, or revoke parole, their action is essentially quasi judicial. Here, two members of the Board conducted a hearing to determine whether to parole Brock. The Board considered Brock's preparole plan, a parole officer's report, medical, psychological, and psychiatric evaluations, a history of Brock's institutional and criminal conduct, and reports from Department of Corrections staff. The Board then applied objective standards to the facts before it in order to arrive at the decision to parole Brock. See Washington State Board of Prison Terms and Paroles, Guidelines for Reconsideration of Length of Confinement, in WAC Title 381 (Supp. 1984-1985). Moreover, determining a prisoner's actual term of confinement is a task historically performed by courts. Furthermore, the rationale underlying judicial immunity applies forcefully to the Board's parole decisions. Like judges, board members make decisions that without immunity would leave them vulnerable to suits either from inmates or from members of the public injured by parolees. Immunity is necessary to preserve Board members' independent and impartial judgment just as it is necessary to preserve judges' independence.
In short, parole decisions are essentially judicial in nature and, like judges' decisions, require freedom from personal fears of litigation. We hold that the Board is entitled to quasi-judicial immunity for its decision to release Brock.
Taggart makes two arguments against granting the Board quasi-judicial immunity. First, Taggart argues that a parole board hearing is functionally dissimilar to a judicial proceeding in several respects. For example, the Board need hold no meeting or hearing to determine parole, but can make the decision after administrative review only. Washington State Board of Prison Terms and Paroles, Policies and Procedures Manual Rule 3.080, in WAC Title 381 (Supp. 1981-1982) (hereafter Parole Board Manual). In addition, attorneys are not permitted at parole hearings, Parole Board Manual Rule 3.100, and no provision is made for direct appeal of the Board's decision by members of the public. This argument is unpersuasive because none of the factors Taggart identifies, either alone or in conjunction with the others she mentions, is necessary for finding the Board's action to be quasi judicial. The presence of other attributes of a judicial process makes the Board's action quasi judicial despite the absence of the features Taggart identifies. We recognize that the absence of any mechanism by which the Board's decisions can be challenged and reviewed represents a significant departure from the ideal form of quasi-judicial action. This departure, however, is not enough to warrant denying quasi-judicial immunity to the Board.
Taggart's other argument against granting the Board quasi-judicial immunity is based upon Grimm v. Arizona Bd. of Pardons & Paroles, 115 Ariz. 260, 564 P.2d 1227, 5 A.L.R.4th 757 (1977). «3»
«3» We note that the Arizona Legislature severely limited Grimm in 1984 by passing a statute granting immunity to employees such as members of the parole board and parole officers. Ariz. Rev. Stat. Ann. § 12-820.02 (Supp. 1990).
There, a parolee killed one man and seriously wounded another during a robbery. The Arizona Supreme Court held that the parole board was entitled only to qualified, not absolute, quasi-judicial immunity. The court reached this conclusion partly on the basis of an Arizona statute which provided that "[p]arole may be authorized '[i]f it appears to the board * * * that there is reasonable probability that the applicant will live and remain at liberty without violating the law * * * .'" Grimm, at 265 (quoting Ariz. Rev. Stat. Ann. § 31-412). The court held that this statute implies "that parole may not be authorized if there is no such reasonable probability, and therefore immunity should adhere only for acts within the guidelines." Grimm, at 265. Taggart argues that a parallel argument applies in the instant case because the Arizona statute relied upon in Grimm is comparable to RCW 9.95.100, which provides that "[t]he board shall not . . . until his maximum term expires, release a prisoner, unless in its opinion his rehabilitation has been complete and he is a fit subject for release."
Grimm is distinguishable from the present case because the Arizona statute requires a "reasonable probability" that the parole applicant is rehabilitated, whereas RCW 9.95.100 only requires that the Board form the opinion that rehabilitation is complete. This is not to say that the Board may form its opinions in an irresponsible fashion. The Board must act in conformity with its own regulatory guidelines, and if the Board wishes to claim the benefit of quasi-judicial immunity, its release decisions must actually be the result of a quasi-judicial process. RCW 9.95.100, unlike the Arizona statute, provides no basis for the court to second-guess the Board as to whether there was a "reasonable probability" that the prisoner would not violate the law if released.
Taggart does not contend that the Board failed to act within its own guidelines. Since we have determined that the Board's decision was quasi judicial, we hold that the Board is absolutely immune for its release decision. Because we so hold, we need not reach the State's argument that the discretionary immunity exception to the waiver of sovereign immunity in RCW 4.92.090 also shields the Board's decision. «4»
«4» In Noonan v. State, 53 Wn. App. 558, 769 P.2d 313, review denied, 112 Wn.2d 1027 (1989), victims of a parolee's crimes brought a tort action against the parole board for an allegedly negligent release decision. The court held that the Board was immune, applying the discretionary immunity exception as stated in Evangelical United Brethren Church v. State, 67 Wn.2d 246, 407 P.2d 440 (1965). Here, we reach the same conclusion as the Court of Appeals, but do so relying only on the doctrine of quasi-judicial immunity. We do not decide whether discretionary immunity also shields the Board's release decisions. For a critical discussion of Noonan, see Comment, Washington's Discretionary Immunity Doctrine and Negligent Early Release Decisions: Parole and Work Release, 65 Wash. L. Rev. 619 (1990). For development of the thesis that quasi-judicial immunity, not discretionary immunity, provides the proper basis for immunizing parole boards from negligence actions, see Comment, Governmental Immunity and the Release of Dangerous Inmates from State Institutions: Can the State Get Away with Murder?, 33 Buff. L. Rev. 491 (1984).
B. The Parole Officers' Immunity.
1. Quasi-Judicial Immunity.
In both Taggart and Sandau, the State contends that parole officers are entitled to immunity for their supervision of parolees. The State's argument for this is based on the premise that if an agency is entitled to quasi-judicial immunity, then so are the agency's essential administrative employees. The State then argues that since the Board is an agency entitled to quasi-judicial immunity, and since parole officers are essential administrative employees of the Board, parole officers are also immune.
[5, 6] We reject this argument. Quasi-judicial and judicial immunity shield against claims arising from the performance of a quasi-judicial function; they do not shield an official from any claim whatsoever just because that official sometimes performs judicial or quasi-judicial functions. Thus Forrester v. White, 484 U.S. 219, 98 L. Ed. 2d 555, 108 S. Ct. 538 (1988) held that a state court judge was not absolutely immune when he demoted and discharged a state employee for discriminatory reasons. The Court stated that "immunity is justified and defined by the functions it protects and serves, not by the person to whom it attaches." 484 U.S. at 227. Accordingly, the Court held that the judge's actions were not immune because they were administrative, not judicial, in nature. Similarly, in Adkins v. Clark Cy., 105 Wn.2d 675, 717 P.2d 275 (1986), a plaintiff who lost a personal injury action sued the County and State in a second suit because of alleged misconduct on the part of the bailiff. We held that the bailiff's conduct was immune because it was "intimately associated with the judicial process." 105 Wn.2d at 678. Likewise, in Robichaud v. Ronan, 351 F.2d 533 (9th Cir. 1965), prosecuting attorneys claimed they were immune from a suit alleging that they had filed a complaint with malicious motive and without probable cause. The court declared that prosecuting attorneys are entitled to quasi-judicial immunity when they act in their official capacity, but that "when a prosecuting attorney acts in some capacity other than his quasi-judicial capacity, then the reason for his immunity . . . ceases to exist." Robichaud, at 536. Therefore the function performed and not the person who performs it must be the focus of the inquiry as to whether parole officers are immune for acts performed in the course of parole supervision.
Some courts have granted parole or probation officers quasi-judicial immunity when they prepare parole or presentencing reports. For example, in Harlow v. Clatterbuck, 230 Va. 490, 339 S.E.2d 181 (1986) the Supreme Court of Virginia held that quasi-judicial immunity shielded probation officers who were responsible for investigating and submitting recommendations regarding the appropriateness of the release of a probationer who subsequently assaulted the plaintiff. Similarly, in Hulsman v. Hemmeter Dev. Corp., 65 Hawaii 58, 647 P.2d 713 (1982), a probation officer allowed a probationer to leave an interview knowing he was carrying a rifle. The next day, the probationer shot the plaintiff. The Hawaii Supreme Court declared that because the interview was part of the officer's investigations in preparing a presentence report, the officer "performed a function integral to the judicial process and was acting as an arm of the court." Hulsman, at 65. Quasi-judicial immunity has also been held to protect probation officers from civil rights suits for their actions in submitting or preparing presentencing or probation reports. Burkes v. Callion, 433 F.2d 318 (9th Cir. 1970), cert. denied, 403 U.S. 908, 29 L. Ed. 2d 685, 91 S. Ct. 2217 (1971); Spaulding v. Nielsen, 599 F.2d 728 (5th Cir. 1979).
On the other hand, parole officers have been denied quasi-judicial immunity from suits arising from actions taken outside any judicial or quasi-judicial process. For example, in A.L. v. Commonwealth, 402 Mass. 234, 521 N.E.2d 1017 (1988), a thrice-convicted child molester obtained employment as a teacher at a middle school, in violation of the terms of his probation. The child molester's probation officer never made any inquiry as to where he was employed. Parents whose children were subsequently molested sued the Commonwealth of Massachusetts, and the Commonwealth argued that quasi-judicial immunity barred the suit. The Supreme Court of Massachusetts disagreed, stating that quasi-judicial immunity was available only if the probation officer "acted pursuant to a judge's directive or otherwise in aid of the court. . . . Any claim to immunity which the Commonwealth might have asserted ceased when [the probation officer] failed to aid in the enforcement of the conditions of . . . probation." A.L., at 247.
In Ray v. Pickett, 734 F.2d 370 (8th Cir. 1984), a probationer sued his probation supervisors alleging they had violated his constitutional rights by intentionally falsifying a report to secure a parole violator's warrant. The probation officers claimed that quasi-judicial immunity barred the suit, but the court disagreed:
[A] probation officer does not perform an adjudicatory function by filing a report with the parole commission. The effect of filing such a report is merely to trigger an inquiry by another officer that may or may not lead to an administrative proceeding. The probation officer is not acting as closely with the court as in the presentence report process. We cannot conclude that this function is so intimately associated with the judicial process that it entitles probation officers to an absolute immunity.
Ray, at 373. The court denied the probation officers absolute immunity, granting them only qualified immunity. Ray, at 374-75.
In Acevedo v. Pima Cy. Adult Probation Dep't, 142 Ariz. 319, 690 P.2d 38, 44 A.L.R.4th 631 (1984), a child molester was granted probation subject to the condition that he was not to have any contact whatsoever with children. His probation officers subsequently allowed him to live on premises where children resided, and to obtain employment in a program bringing him into contact with children. Parents whose children were then molested sued the probation officer. The Arizona Supreme Court held that quasi-judicial immunity provided the officer no protection. The court stated:
Probation officers, in preparing and submitting pre-sentence reports to the court, should be entitled to absolute immunity because the pre-sentence report is an integral part of the sentencing process. We also believe that a probation officer is entitled to absolute protection from suit for actions which are necessary to carry out and enforce the conditions of probation imposed by the court. We do not agree, however, that all the activities of a probation officer in supervising a probationer are entitled to immunity. Much of the work of a probation officer is administrative and supervisory. Such activities are not part of the judicial function; they are administrative in character.
(Citations omitted.) Acevedo, at 322. See generally Annot., Probation Officer's Liability for Negligent Supervision of Probationer, 44 A.L.R.4th 639 (1986).
We hold that parole officers are entitled to quasi-judicial immunity only for those functions they perform that are an integral part of a judicial or quasi-judicial proceeding. Thus when a parole officer performs functions such as enforcing the conditions of parole or providing the Board with a report to assist the Board in determining whether to grant parole, the officer's actions are protected by quasi-judicial immunity. But when the officer takes purely supervisory or administrative actions, no such protection arises.
In the present case, Taggart and Sandau allege that Brock's and Geyman's parole officers failed substantially to perform their supervisory functions. Taggart claims that Richardson never required Brock to submit to drug testing and never contacted Brock's friends or employers to inquire as to his progress. These are supervisory failings and hence unprotected by judicial or quasi-judicial immunity. Likewise, Sandau claims that although Geyman was a maximum supervision parolee, Geyman's parole officer, Kairoff, took no steps to supervise Geyman's parole from April 5, 1984, until July 27, 1984. This neglect occurred immediately after Kairoff had received a report that Geyman was drinking and threatening his ex-wife's new husband. In addition, weeks before Geyman raped Sandau, Kairoff knew that Geyman had violated the conditions of his parole by quitting his job, changing his address without notifying his parole officer, failing to report to his parole officer, and absconding from parole. On September 18, 2 days before the rape, Kairoff received reports that Geyman was reportedly drinking and beating his girl friend and her children; yet, according to Sandau, Kairoff failed to provide his superior with relevant information necessary to enable the superior to make an informed decision whether to issue a parole warrant on Geyman. Finally, there was an unexplained delay from September 18, when the Montana authorities sent a teletype to the Board indicating that they had spoken with Kairoff and understood that a teletype parole warrant was forthcoming, until September 27, when a hearing officer finally presented the teletype to members of the Board. These failures to act cannot be characterized as themselves quasi judicial, and they were not a part of any quasi-judicial process performed by the Board. They were administrative and supervisory failings that fall outside the absolute protection afforded agency employees by quasi-judicial immunity.
2. Discretionary Immunity.
We next turn to the question whether the discretionary immunity exception to the State's waiver of sovereign immunity affords parole officers a shield to claims arising from parole supervision. The Washington Legislature abolished state sovereign immunity in 1961 by enacting RCW 4.92.090, which provides that "[t]he state of Washington, whether acting in its governmental or proprietary capacity, shall be liable for damages arising out of its tortious conduct to the same extent as if it were a private person or corporation." Responding to the abolition of sovereign immunity, in Evangelical United Brethren Church v. State, 67 Wn.2d 246, 407 P.2d 440 (1965), this court created a discretionary governmental immunity exception. Under this exception, "discretionary" governmental acts are immune from tort liability whereas "ministerial" or "operational" acts are not. 67 Wn.2d at 254-55. The exception has been narrowed in later decisions. For example, in King v. Seattle, 84 Wn.2d 239, 525 P.2d 228 (1974), we explained that the exception's purpose is "to assure that courts refuse to pass judgment on policy decisions in the province of coordinate branches of government", and we held that the State is immune only if it can show that the decision was the outcome of a conscious balancing of risks and advantages. 84 Wn.2d at 246. In addition, we have said that discretionary immunity is narrow and applies only to basic policy decisions made by a high-level executive. For example, in Chambers-Castanes v. King Cy., 100 Wn.2d 275, 282, 669 P.2d 451, 39 A.L.R.4th 671 (1983), we held that the decision whether to dispatch a police officer to the scene of a crime was not protected under discretionary immunity because it was not a basic policy decision by a high-level executive.
 In the present case, we hold that the discretionary immunity exception does not shield parole officers from claims alleging negligent supervision. We recognize that parole officers' supervisory decisions require the exercise of discretion. The crucial point, however, is that the discretionary immunity exception applies only to basic policy decisions. Parole officers' supervisory decisions, however much discretion they may require, are not basic policy decisions. Such decisions are ministerial in nature. Therefore, even if the discretionary immunity exception applies to the Board's release decisions, an issue we have found it unnecessary to resolve, the exception does not shield parole supervision decisions.
3. Qualified Immunity for Parole Officers' Supervisory Decisions.
 In holding that parole supervision is not entitled to absolute immunity, we nonetheless recognize that parole officers perform a difficult job under exacting conditions. They are routinely asked to make decisions that can significantly affect both a parolee's liberty and the safety of members of the public. Completely stripping them of immunity would make them more concerned with avoiding lawsuits than doing their jobs. At the same time, because of the dangers violent parolees present, adequate supervision must be provided. We therefore hold that when parole officers act outside any judicial or quasi-judicial proceeding and so are not entitled to absolute immunity, they nonetheless may be shielded by qualified immunity. Our recent decision in Babcock v. State, 116 Wn.2d 596, 809 P.2d 143 (1991) suggests the parameters of this qualified immunity. In Babcock, caseworkers with the Department of Social and Health Services placed several young girls in the home of a relative who subsequently raped them. We held that the caseworkers were entitled to qualified, but not absolute, immunity. We stated that "[i]n order to qualify for the [qualified] immunity, the caseworker must (1) carry out a statutory duty, (2) according to procedures dictated by statute and superiors, and (3) act reasonably." Babcock, 116 Wn.2d at 618 (citing Guffey v. State, 103 Wn.2d 144, 690 P.2d 1163 (1984) (holding that police officers charged with false arrest are entitled to qualified immunity)). We are persuaded that a similar immunity should shield parole officers from claims arising from violent acts committed by parolees. We, therefore, hold that parole officers are immune from liability for allegedly negligent parole supervision if their action is in furtherance of a statutory duty and in substantial compliance with the directives of superiors and relevant regulatory guidelines. Unlike the immunity Babcock created for caseworkers, the immunity we announce here for parole officers does not require an additional showing that the parole officers' actions were reasonable once it has been shown that the officers performed a statutory duty in compliance with the directives of superiors and relevant guidelines. In addition, the immunity requires only that the parole officers' conduct must be in substantial compliance, not strict compliance, with the directives of superiors and regulatory procedures. The standard we announced in Babcock was appropriate because of the unique circumstances inherent in the foster care placement of children. We believe the standard we announce here for parole officers provides the best balance of the competing needs to protect the public, to provide parole officers the immunity they need to perform their jobs effectively, and to allow recovery in appropriate cases to the victims of parolees' crimes.
The trial courts in these cases dismissed the plaintiffs' negligent parole supervision claims without giving the plaintiffs an opportunity to show that the parole officers failed to carry out statutory duties according to prescribed procedures. In this, the trial courts erred. On remand, the plaintiffs must be permitted to attempt to pierce the parole officers' qualified immunity. We next consider whether the plaintiffs' negligent parole supervision claims are barred by the public duty doctrine.
 The State argues that it owed no duty to Taggart or Sandau as opposed to the general public, and that therefore the public duty doctrine bars their claims based upon negligent parole supervision. We hold that the State has a duty to take reasonable precautions to protect against reasonably foreseeable dangers posed by the dangerous propensities of parolees, and that if injury to Taggart and Sandau was a reasonably foreseeable consequence of paroling Brock and Geyman, then this duty extended to Taggart and Sandau. We therefore reverse and remand the trial courts' dismissals of Taggart's and Sandau's negligent supervision claims. In each case, unless qualified immunity is otherwise found to bar those claims, a jury should be allowed to determine whether injury to Taggart or Sandau was reasonably foreseeable.
The public duty doctrine provides that "no liability may be imposed for a public official's negligent conduct unless it is shown that 'the duty breached was owed to the injured person as an individual and was not merely the breach of an obligation owed to the public in general (i.e., a duty to all is a duty to no one).'" Taylor v. Stevens Cy., 111 Wn.2d 159, 163, 759 P.2d 447 (1988) (quoting J&B Dev. Co. v. King Cy., 100 Wn.2d 299, 304, 669 P.2d 468, 41 A.L.R.4th 86 (1983), overruled in Taylor). Numerous exceptions to the doctrine are recognized. Bailey v. Forks, 108 Wn.2d 262, 268, 737 P.2d 1257 (1987) (identifying four exceptions). These exceptions generally embody traditional negligence principles, and may be used as "focusing tools" to determine whether the public entity had a duty to the injured plaintiff. Taylor, 111 Wn.2d at 166. The question whether an exception to the public duty doctrine applies is thus another way of asking whether the State had a duty to the plaintiff.
The controlling case is Petersen v. State, 100 Wn.2d 421, 671 P.2d 230 (1983). There, the plaintiff had been injured when her car was struck by another vehicle driven by a recently released Western State Hospital psychiatric patient. We held that "the State had a duty to take reasonable precautions to protect against the dangerous propensities of a state hospital patient". 100 Wn.2d at 432. In reaching this holding, we recognized an exception to the common law rule that a person has no duty to prevent a third person from causing physical injury to another. 100 Wn.2d at 426. The exception is stated in the Restatement (Second) of Torts § 315:
There is no duty so to control the conduct of a third person as to prevent him from causing physical harm to another unless
(a) a special relation exists between the actor and the third person which imposes a duty upon the actor to control the third person's conduct, or
(b) a special relation exists between the actor and the other which gives to the other a right to protection.
Restatement (Second) of Torts § 315 (1965) (hereinafter § 315). «5»
«5» Although we did not discuss the public duty doctrine in Petersen, imposing liability on the State for the psychiatrist's negligence presupposes that the doctrine is inapplicable. Petersen's adoption of § 315, in other words, effectively created another exception to the doctrine, and it has been so regarded in later cases. Bailey, 108 Wn.2d at 268; Champagne v. Spokane Humane Soc'y, 47 Wn. App. 887, 893, 737 P.2d 1279, review denied, 108 Wn.2d 1035 (1987).
Petersen thus stands for the proposition that a "special relation" exists between a state psychiatrist and his or her patients, such that when the psychiatrist determines, or pursuant to professional standards should determine, that a patient presents a reasonably foreseeable risk of serious harm to others, the psychiatrist has "a duty to take reasonable precautions to protect anyone who might foreseeably be endangered". 100 Wn.2d at 428. We stated that the scope of this duty is not limited to readily identifiable victims, but includes anyone foreseeably endangered by the patient's condition. 100 Wn.2d at 429; see also Kaiser v. Suburban Transp. Sys., 65 Wn.2d 461, 398 P.2d 14, 401 P.2d 350 (1965) (plaintiff, injured when a bus driver lost consciousness because of medication, had a cause of action against the driver's physician).
We hold that the relationship between a parole officer and the parolees he or she supervises creates a similar duty for the officers. As a preliminary matter, we note that a duty will be imposed under § 315 only upon a showing of a "definite, established and continuing relationship between the defendant and the third party." Honcoop v. State, 111 Wn.2d 182, 193, 759 P.2d 1188 (1988). Under RCW 72.04A.080, parolees "shall be subject to the supervision of the department of corrections, and the probation and parole officers of the department shall be charged with the preparation of progress reports of parolees and to give guidance and supervision to such parolees within the conditions of a parolee's release from custody." RCW 72.04A.080. This statute is sufficient to establish that parole officers have a "definite, established and continuing relationship" with their parolees.
In the Restatement (Second) of Torts, sections 316 through 320 define various "special relations" that, in accordance with the general principle stated in § 315, give rise to a duty to control a third person. Section 319 is most relevant in the present case: "One who takes charge of a third person whom he knows or should know to be likely to cause bodily harm to others if not controlled is under a duty to exercise reasonable care to control the third person to prevent him from doing such harm." Restatement (Second) of Torts § 319 (1965) (hereinafter § 319). Various features of the relationship between parole officer and parolee are relevant to the question whether parole officers have "taken charge" of the parolees they supervise. Parole officers have the statutory authority under RCW 72.04A.080 to supervise parolees. The State can regulate a parolee's movements within the state, require the parolee to report to a parole officer, impose special conditions such as refraining from using alcohol or undergoing drug rehabilitation or psychiatric treatment, and order the parolee not to possess firearms. The parole officer is the person through whom the State ensures that the parolee obeys the terms of his or her parole. Additionally, parole officers are, or should be, aware of their parolees' criminal histories, and monitor, or should monitor, their parolees' progress during parole. Because of these factors, we hold that parole officers have "taken charge" of the parolees they supervise for purposes of § 319. When a parolee's criminal history and progress during parole show that the parolee is likely to cause bodily harm to others if not controlled, the parole officer is under a duty to exercise reasonable care to control the parolee and to prevent him or her from doing such harm.
Other courts have also recognized this or a similar duty. For example, in Sterling v. Bloom, 111 Idaho 211, 723 P.2d 755 (1986), a drunken probationer, originally sentenced for driving while intoxicated, crashed his truck into a motorcyclist. Applying § 319, the court held that the probation officer involved owed a duty to all motorists foreseeably endangered by the probationer's dangerous propensities. Sterling, at 225-26. In Division of Corrections v. Neakok, 721 P.2d 1121 (Alaska 1986), the survivors and personal representatives of three persons murdered by a parolee sued the State of Alaska, claiming that the parolee had been negligently supervised. The Alaska Supreme Court stated that the State "stands in a special relationship with a parolee, both because of its increased ability to foresee the dangers the parolee poses and because of its substantial ability to control the parolee", and held that "[g]iven this special relationship, it is not unreasonable to impose a duty of care on the state to protect the victims of parolees." 721 P.2d at 1126-27. The Alaska court cited Petersen as direct authority for this holding. 721 P.2d at 1127. Similarly, in Semler v. Psychiatric Inst., 538 F.2d 121 (4th Cir.), cert. denied, 429 U.S. 827, 50 L. Ed. 2d 90, 97 S. Ct. 83 (1976), a probation officer and psychiatric institution failed to enforce a court order requiring the probationer be confined and supervised as a day-care patient. The probationer, known to be dangerous to young girls, killed the plaintiff's daughter. The court applied § 319 and declared that "[t]he special relationship created by the probation order . . . imposed a duty on the [defendants] to protect the public from the reasonably foreseeable risk of harm at [the probationer's] hands that the state judge had already recognized." 538 F.2d at 125. In Gibson v. United States, 457 F.2d 1391 (3d Cir. 1972), an instructor in a Federal Job Corps program sought recovery from the United States for injuries he suffered in an attack by a trainee in the program. The court applied § 319 and held that the federal agency managing the program had a duty to prevent the trainees, allegedly known to be violent juvenile offenders and drug addicts, from harming others.
Other courts have found a similar duty without relying upon § 319. For example, in Doe v. Arguelles, 716 P.2d 279 (Utah 1985), a juvenile offender with a history of sexual assaults attacked and injured a child after being conditionally released into the community from a youth detention center. The court stated that after releasing the juvenile, the superintendent of the detention center had an ongoing duty to control him, and that this duty required an "active, ongoing concern" for the juvenile and the community. In Johnson v. State, 69 Cal. 2d 782, 447 P.2d 352, 73 Cal. Rptr. 240 (1968), a youth placed in a foster home by the state youth authority assaulted his foster parent. The California Supreme Court held that the State owed a duty to inform the foster parent "of any matter that its agents knew or should have known that might endanger" her and her family. Johnson, at 786. In State v. Silva, 478 P.2d 591 (Nev. 1970), an inmate at a state "honor camp" escaped and committed a rape. The Nevada Supreme Court held that the State was not immune from a suit alleging negligence in the control and supervision of the facility, and implicitly recognized the State's duty to supervise and control the inmates in a nonnegligent manner. 478 P.2d at 593.
The State seeks to distinguish Petersen v. State, 100 Wn.2d 421, 671 P.2d 230 (1983), on the basis that there the patient who caused the plaintiff's injuries was released from the hospital, where the psychiatrist had a high degree of control over him, whereas in the present case the parolees were living on their own, under circumstances in which the parole officers had very little control over them. The State argues that nothing less than a full custodial relationship justifies holding parole officers to a duty to protect against parolees.
Some courts have adopted such a view, holding that parole officers do not have sufficient control over parolees as to justify imposing on the officers a duty to control the parolees. Fox v. Custis, 236 Va. 69, 372 S.E.2d 373 (1988); Small v. McKennan Hosp., 403 N.W.2d 410 (S.D. 1987); Lamb v. Hopkins, 303 Md. 236, 492 A.2d 1297 (1985). In Fox, for example, the victims of a parolee's crimes sued the state parole officers responsible for the parolee's supervision. The case was analyzed under § 319, and the court held that the parole officers did not "take charge" of the parolee because the statute empowering the officers to supervise parolees "does not contemplate continuing hourly or daily dominance and dominion by a parole officer over the activities of a parolee." Fox, at 75. Similarly, in Lamb the Maryland court expressly adopted § 319, but held that probation officers do not "take charge" of probationers such as to give rise to a duty to exercise due care in controlling the probationers because of the lack of a custodial relationship and the relative freedom the probationers have in conducting their day-to-day affairs. Lamb, at 248-49. The Maryland court distinguished Petersen on the basis that Petersen concerned the negligent release of patients from psychiatric institutions. Lamb, at 250. Likewise, the South Dakota Supreme Court in Small declared that "because there was no custodial relationship involved in this case, we conclude that the officers did not take charge of the probationer", and so held that § 319 did not give rise to any duty on the part of probation officers to control the dangerous propensities of probationers. 403 N.W.2d at 414 (quoting Lamb, at 248).
 We reject this approach and hold that a parole officer takes charge of the parolees he or she supervises despite the lack of a custodial or continuous relationship. The duty we announced in Petersen is not limited to taking precautions to protect against mental patients' dangerous propensities only when those patients are being released from the hospital, as suggested by the Maryland court in Lamb. The duty requires that whenever a psychiatrist determines, or according to the standards of the profession should have determined, that a patient presents foreseeable dangers to others, the psychiatrist must take reasonable precautions to protect against harm. Whether the patient is a hospital patient or an outpatient is not important. Thus in Tarasoff v. Regents of Univ. of Cal., 17 Cal. 3d 425, 551 P.2d 334, 131 Cal. Rptr. 14 (1976), which we followed in Petersen, negligent release from the hospital was not an issue; the patient who murdered the plaintiffs' daughter was not a hospital patient. Similarly, in Lipari v. Sears, Roebuck & Co., 497 F. Supp. 185 (D. Neb. 1980), which we also followed in Petersen, a patient at a psychiatric clinic fired a shotgun into a crowded nightclub. The patient was a day-care patient – so, again, release was not an issue – yet the court found that the defendant therapist had a duty to anyone foreseeably endangered by the patient's negligent care. 497 F. Supp. at 194.
In addition, we recognize that the Washington statute empowering parole officers to supervise parolees contemplates neither a custodial relationship, such as the Maryland court required in Lamb, nor continuous supervision, such as the Virginia court demanded in Fox. In Tarasoff and Lipari, however, which we followed in Petersen, the defendant therapists had neither custodial nor continuous relationships with their patients. We hold that a parole officer may "take charge" of a parolee, thereby assuming the duty to protect against reasonably foreseeable dangers, despite the absence of a custodial relationship and without exercising the "continuing hourly or daily dominance and dominion" over that parolee the Fox court required. Fox, at 75.
We conclude that parole officers have a duty to protect others from reasonably foreseeable dangers engendered by parolees' dangerous propensities. We emphasize, however, that we are not unmindful of the extremely difficult supervisory tasks that parole officers must perform. Nor in holding that parole officers "take charge" of parolees do we exaggerate the degree of control they exercise over parolees. The duty we announce here arises only once it has been shown that the defendant parole officer lacks both absolute and qualified immunity. To pierce these immunities, the plaintiff must show, first, that the officer's actions were not an integral part of any judicial or quasi-judicial process, and, second, that the officer failed to perform a statutory duty according to procedures dictated by statute and superiors. Only then does the question of duty arise. We also emphasize that the Legislature is free to limit or eliminate the duty we announce here by passing a statute broadening parole officers' immunity so as to include any actions taken in their official capacities.
In the present case, the parole officers had a duty to take reasonable precautions to protect anyone foreseeably endangered by Brock's and Geyman's dangerous propensities. To establish that this duty extends to them, Taggart and Sandau must show that they were foreseeably endangered. "Foreseeability is normally an issue for the jury, but it will be decided as a matter of law where reasonable minds cannot differ." Christen v. Lee, 113 Wn.2d 479, 492, 780 P.2d 1307 (1989). Here, the questions of foreseeability were not so one-sided that they should have been decided by the trial courts. Reasonable juries could have found that Taggart's and Sandau's injuries were foreseeable. Brock had a history of alcoholism and violent attacks against women, and a poor prognosis for recovery from his mental illness. In light of these facts, a jury might conclude that it was reasonably foreseeable that if Brock were not supervised closely he would commit another similar crime. The fact that Taggart herself was not the foreseeable victim of Brock's criminal tendencies does not establish as a matter of law that her injury was not foreseeable. Similarly, Geyman had an extensive criminal history that included violent assaults. He violated the conditions of his parole by quitting his job, changing his address without notifying his parole officer, failing to report to his parole officer, and absconding from parole. Geyman's parole officer knew of these violations, and knew as well that Geyman was reportedly drinking alcohol, which increased the risk of violent behavior, and beating Hazel and her children. A jury might conclude that it was reasonably foreseeable that unless Geyman were arrested for violating parole, he would commit further violence against Hazel or her children, including Sandau. The fact that the violence took the form of raping Sandau, when Geyman's criminal history did not include rape, does not show that injury to Sandau was not foreseeable. Violence against Sandau may have been foreseeable, even though the form of that violence may not have been.
To the extent that the trial courts in these cases dismissed the actions because of the absence of any duty, the courts decided the questions whether Taggart's and Sandau's injuries were foreseeable. In this the trial courts erred. Juries could have reasonably concluded that the duty of Brock's and Geyman's parole officers to protect others foreseeably endangered by Brock and Geyman extended to Taggart and Sandau.
The State argues that no actions of the State or its agents proximately caused Taggart's or Sandau's injuries. We are unpersuaded that this provides a sound basis on which to uphold the trial courts' dismissals of Taggart's and Sandau's actions.
 Proximate causation consists of two elements: cause in fact and legal causation. Hartley v. State, 103 Wn.2d 768, 777, 698 P.2d 77 (1985). Cause in fact concerns the "but for" consequences of an act: those events the act produced in a direct, unbroken sequence, and which would not have resulted had the act not occurred. Hartley, at 778. Legal causation rests on considerations of policy and common sense as to how far the defendant's responsibility for the consequences of its actions should extend. Hartley, at 779. The question of legal causation is so intertwined with the question of duty that the former can be answered by addressing the latter. Hartley, at 779-80 (quoting W. Prosser, Torts 244-45 (4th ed. 1971)).
 In Taggart, the State argues that parole officer Richardson's monitoring of Brock did not legally cause Taggart's injuries. The State argues that there was nothing to distinguish Brock from other parolees, that Brock's last incarceration was for car theft, and that the assault occurred almost 7 months after his release. The State argues that Brock's assault on Taggart occurred "without any warning to responsible correctional officials". Brief of Respondent (Taggart v. State), at 49. The State contends that since the state correctional system does not have the resources to monitor parolees constantly in order to prevent them from committing unpredictable crimes, neither the State nor its agents should be deemed the legal cause of Taggart's injury.
We disagree. Brock had a long history of violent attacks against women and a consistently unfavorable prognosis for recovery from his psychiatric problems. A Washington State Reformatory staff psychologist who treated Brock during his incarceration in 1980 stated that effective treatment for Brock was unlikely, and reiterated another psychologist's evaluation of 10 years earlier that "unless Brock does participate in a successful program of psychotherapy and psychiatric treatment . . . he will again be involved in criminal activity similar in nature to that of his last two (now three) offenses and commitments." Clerk's Papers, at 63. History had also shown that Brock's attacks on women usually involved alcohol. In light of such facts, the assertion that Brock's assault upon yet another woman occurred without warning is not credible. At most, the facts the State cites support the contentions that Brock's parole officer was entitled to relax her vigilance, that her actions to protect against Brock's dangerous propensities were reasonable under the circumstances, and so that she did not breach any duty. We are unwilling to declare as a matter of law that no actions of the State or its agents were the legal cause of Taggart's injuries.
In Sandau, the State argues that neither the cause in fact requirement nor the legal causation requirement is met. The cause in fact requirement is not met, says the State, because it would be wholly speculative to find that any act or omission of the State or its agents led to Sandau's injury. We disagree. We recognize that when the connection between a defendant's conduct and the plaintiff's injury is too speculative and indirect, the cause in fact requirement is not met. Walters v. Hampton, 14 Wn. App. 548, 543 P.2d 648 (1975) (holding that failure of police to prosecute a person who later shot the plaintiff was not the factual cause of the plaintiff's injury). We do not believe that this is such a case. Here, a reasonable jury might conclude that if the Washington officials had issued the parole warrant the day they received the teletype from the Montana authorities informing them that Montana police were standing by to arrest Geyman, Sandau never would have been raped. Similarly, the jury might conclude that if the Washington officials had responded to that teletype by telling the Montana police that no parole warrant would be issued, then the Montana police would have arrested Geyman immediately on the outstanding misdemeanor violation, in which case, again, Sandau's injury would have been avoided.
The State argues that the legal causation requirement is not met in Sandau because the State cannot monitor parolees to the extent that would have been required in order to prevent Geyman from harming Sandau. We do not accept the assertion that Geyman's parole officer, Kairoff, would have had to monitor Geyman more intensively than state resources allow in order to have prevented the attack on Sandau. Before the assault, Kairoff knew that Geyman was in violation of the conditions of his parole. In addition, Kairoff was contacted by "Diana", a friend of Sandau's mother, Wanda Hazel, and told that Geyman was beating Hazel and her children. Later the same day, Kairoff received a call from Hazel, who confirmed what Diana had told him. Also the same day, Montana law enforcement authorities sent the teletype to the Board stating that they were standing by to arrest Geyman for violation of parole. The information the State needed in order to determine whether Geyman should be arrested, as well as a convenient means to arrest him, were thus readily available. No minute-by-minute monitoring was required.
The State also argues that finding that some act of the State or its agents caused Sandau's injuries would have the effect of making state officials overly concerned with avoiding lawsuits. This point is sufficiently refuted by the fact that whenever an action or proceeding for damages is instituted against any state official, including parole officers, the State bears all the costs of the litigation, provided that the State Attorney General finds that the officer's actions were, or were purported to be, in good faith and within his or her official capacity. RCW 4.92.070. Similarly, the costs of any judgment against the defendant officer in such a proceeding will be satisfied by the State. RCW 4.92.075. Because indemnity is available in these ways, we do not believe recognizing that a parole officer's negligent supervision may be the legal cause of the injuries suffered by the victims of parolees' violent crimes will have an undue chilling effect upon parole officers' performance of their duties.
We hold that the Board is entitled to quasi-judicial immunity for its release decisions. Therefore we affirm the Taggart trial court's dismissal of the claim alleging negligent parole release against the Board. We also hold that neither immunity, nor the absence of any duty, nor lack of proximate causation warranted dismissal of the plaintiffs' negligent parole supervision claims. We reverse the decisions of the trial courts as regards the negligent supervision claims, and remand both cases for further proceedings consistent with this opinion.
DORE, C.J., and UTTER, DOLLIVER, DURHAM, and SMITH, JJ., concur.
BRACHTENBACH, J., concurs in the result.
UTTER, J. (concurring) – The majority opinion sets forth a coherent and persuasive analysis on the issues of immunity and the public duty doctrine. Justice Guy's opinion, concurring in part and dissenting in part, fails to articulate an acceptable alternative to the majority's analysis and runs contrary to established precedent. Justice Guy also says that he would overrule Babcock v. State, 116 Wn.2d 596, 809 P.2d 143 (1991), without offering persuasive legal reasons besides "public policy concerns." This concurrence responds to Justice Guy's concurrence/dissent.
First, the concurrence/dissent would have us extend quasi-judicial immunity to the parole board's administrative employees because those employees are acting "pursuant to the authority of the board in carrying out parole decisions of the board." This stretches the coverage of quasi-judicial immunity, which is an absolute form of immunity, too far.
As we noted in Babcock, even the United States Supreme Court has been reluctant to extend absolute prosecutorial immunity to anyone but prosecutors and judges because of the extraordinary sweep of absolute immunity. Babcock, 116 Wn.2d at 608 (citing Achterhof v. Selvaggio, 886 F.2d 826, 829 (6th Cir. 1989) (citing Malley v. Briggs, 475 U.S. 335, 89 L. Ed. 2d 271, 106 S. Ct. 1092 (1986)); Imbler v. Pachtman, 424 U.S. 409, 429, 47 L. Ed. 2d 128, 96 S. Ct. 984 (1976)). More recently, the Supreme Court in Hafer v. Melo, ___ U.S. ___, 116 L. Ed. 2d 301, 112 S. Ct. 358 (1991) emphasized the limited instances where absolute immunity is appropriate:
This Court has refused to extend absolute immunity beyond a very limited class of officials, including the President of the United States, legislators carrying out their legislative functions, and judges carrying out their judicial functions, "whose special functions or constitutional status requires complete protection from suit."
112 S. Ct. at 364 (quoting Harlow v. Fitzgerald, 457 U.S. 800, 807, 73 L. Ed. 2d 396, 102 S. Ct. 2727 (1982)).
In Babcock, both the majority and Justice Andersen's concurrence/dissent closely examined the Washington Department of Social and Health Services (DSHS) caseworkers' actions to determine whether an extension of absolute immunity was justified. Similarly, the majority in this case has employed a functional analysis in determining what sorts of conduct are entitled to quasi-judicial immunity. Majority opinion, at 210. As the United States Supreme Court stated in Forrester v. White, 484 U.S. 219, 227, 98 L. Ed. 2d 555, 108 S. Ct. 538 (1988), "immunity is justified and defined by the functions it protects and serves, not by the person to whom it attaches." This is a principled, sensible way of extending judicial immunity to protect those officials who perform quasi-judicial functions. But that immunity applies to officials' actions only when they are playing a judicial role.
The majority in this case applies its functional test to the duties of parole officers when it holds that they are entitled to quasi-judicial immunity only for those functions that are an integral part of a judicial or quasi-judicial proceeding. Majority opinion, at 213. Thus, parole officers have quasi-judicial immunity when they enforce the conditions of parole or they provide the Indeterminate Sentence Review Board (the Board) with reports in parole proceedings. These types of roles are an integral part of the Board's proceedings. Administrative and supervisory actions of parole officers, however, are a different matter. When the parole officers in this case allegedly failed to supervise their parolees, they were not acting in a quasi-judicial capacity. They were not assisting the Board in deciding whether or not prisoners should be paroled. The decision to parole Louie Brock and Keith Geyman already had been made. The parole officers allegedly acted, or failed to act, in their supervisory capacity. For example, Leda Richardson, Brock's parole officer, never required him to submit to drug testing after his release. She never contacted Brock's friends or employers to inquire as to his progress. These are supervisory failures, unrelated to judicial or quasi-judicial proceedings. Similarly, James Kairoff, Geyman's parole officer, failed to supervise him between April 5, 1984, and July 27, 1984, even though Geyman was a maximum supervision parolee, and Kairoff had received negative reports about him. Later, Kairoff failed to act even after he learned that Geyman had violated his parole and was beating his girl friend and her children. All of these alleged actions, or failures to act, were supervisory, not judicial, in nature.
The concurrence/dissent, however, fails to distinguish between the different roles that parole officers play. Instead, it lumps all of those roles together by finding that absolute immunity attaches "inasmuch as the administrative employees are acting pursuant to the authority of the board in carrying out parole decisions of the board." In essence it asserts that parole officers deserve absolute immunity for virtually everything they do. This runs contrary to this court's functional analysis. No precedent is cited for such an extension of absolute immunity. Given the broad measure of protection that absolute immunity gives officials, this court has properly declined to extend absolute immunity in such a wholesale fashion.
The concurrence/dissent derides the qualified immunity that the majority gives parole officers as "illusory and nothing more than a negligence standard wrapped like a gift box which contains sticks and ashes." This criticism is harsh, but unfounded. The qualified immunity that the majority grants parole officers for their supervisory actions in this case is actually quite generous – even more generous than the qualified immunity we gave to DSHS caseworkers in Babcock. Babcock, 106 Wn.2d at 618. All that parole officers need show to be immune from a suit is that they acted in furtherance of a statutory duty and in substantial compliance with the directives of superiors and relevant regulatory guidelines. Majority opinion, at 216. Unlike the immunity for caseworkers announced in Babcock, parole officers need not show they acted reasonably. They also need to show only that they acted in substantial compliance, rather than strict compliance, with the directives of superiors and regulatory procedures. In fact, the majority gives parole officers a qualified immunity for their supervisory actions that more closely resembles the qualified immunity proposed by Justice Andersen in his concurrence/dissent in Babcock, 116 Wn.2d at 634-35. This gives parole officers ample protection from liability for their supervisory actions.
The public policy analysis in the concurrence/dissent focuses solely on the unfairness of holding parole officers individually liable, while ignoring the competing policy considerations in this case. The majority's formulation of qualified immunity fairly balances the needs of protecting the public, allowing recovery for victims of parolees' crimes, and providing the necessary protection for parole officers to perform their jobs. Granting absolute immunity to parole officers would disrupt that delicate balance. Ultimately, if the Legislature finds that we have not given parole officers adequate protection from liability, it can choose to grant them greater protection.
Justice Guy also claims he would overrule Babcock. His concurrence/dissent, however, does not contain a functional analysis of the caseworkers' actions in that case. It suggests that DSHS workers should be absolutely immune for all of their actions, regardless of the functions they are performing. Once again, his opinion is contrary to our functional approach, and unduly expands the coverage of absolute immunity.
In addition, the concurrence/dissent fails to note that the Legislature has granted DSHS caseworkers only qualified immunity when they remove children from child abuse in an emergency. RCW 26.44.056(3). It would have been anomalous for us to grant DSHS caseworkers an absolute immunity for foster care placement decisions in Babcock, while those workers only have a qualified immunity for removing children from abusive situations. Babcock, 116 Wn.2d at 606-07. We properly declined to create such a glaring inconsistency in Babcock.
Although Justice Guy does not explicitly mention the public duty doctrine, his opinion suggests that he disagrees with the majority's conclusion that the parole officers may have had a duty to the victims in this case. The majority's conclusion that parole officers have a duty to prevent foreseeable injuries to victims is based on its conclusion that parole officers have a "special relation" to parolees within the meaning of Restatement (Second) of Torts § 315 (1965). The majority also concluded that parole officers have "taken charge" of parolees within the meaning of Restatement (Second) of Torts § 319 (1965). The concurrence/dissent counters that parole officers do not have a duty to victims because parole officers do not have control "over convicted felons who are free on parole." This is nothing more than word play. Parolees are not "free" from the control of parole officers. As the majority demonstrates, parole officers exercise extensive supervisory control over prisoners. See majority opinion, at 219-20. Therefore, the majority properly concluded that parole officers take charge of the parolees they supervise. The majority cites numerous cases in which other courts have concluded that parole officers may owe a duty to the victims of parolees. See, e.g., Semler v. Psychiatric Inst., 538 F.2d 121 (4th Cir.), cert. denied, 429 U.S. 827, 50 L. Ed. 2d 90, 97 S. Ct. 83 (1976); Sterling v. Bloom, 111 Idaho 211, 723 P.2d 755 (1986); Division of Corrections v. Neakok, 721 P.2d 1121 (Alaska 1986).
Another recent case, Mahomes-Vinson v. United States, 751 F. Supp. 913 (D. Kan. 1990), also illustrates that absolute control over an individual is unnecessary to establish a legal duty. In Mahomes-Vinson, the court found that a plaintiff could sue the federal government because a Veterans Administration Medical Center failed to control a psychiatric patient receiving outpatient services. 751 F. Supp. at 923. That patient only reported to the hospital every 2 weeks for outpatient treatment. Nevertheless, the court found this sufficient to establish a special relationship under Restatement (Second) of Torts § 315, so the federal government could be liable. Therefore, continuous and absolute control over an individual is not, as the concurrence/dissent suggests, a prerequisite of a legal duty.
Our conclusion that a duty may exist does not mean that parole officers will invariably be held liable for the actions of parolees. Victims must still prove that the harm was reasonably foreseeable to establish a duty. In addition, they must overcome any claim of immunity that the parole officer may have before they even reach the issue of whether a duty exists.
The majority's approach does not make parole officers the guarantors of parolees' conduct. Instead, it fairly balances the competing concerns of protecting the public, compensating victims, and giving public officials enough protection so that they can do their jobs.
GUY, J. (concurring in part, dissenting in part) – I respectfully dissent in part. In doing so, I state I would also overrule Babcock v. State, 116 Wn.2d 596, 809 P.2d 143 (1991). I agree that members of the parole board are entitled to quasi-judicial immunity from suits raised by private citizens injured by parolees. I would, however, extend that immunity to the parole board's administrative employees, inasmuch as the administrative employees are acting pursuant to the authority of the board in carrying out parole decisions of the board.
The majority holds that parole board decisions, being essentially judicial in nature, like judges' decisions, require freedom from personal fears of litigation. Independence of action and impartial judgment are important public policy concerns. Such rationale also applies to parole officers.
My dissent in this case is based upon public policy concerns. The same rationale applies to the issues raised in Babcock. To hold such public employees liable for negligence in the performance of their duties will not enhance performance but will enhance inaction. Legislative mandate to Department of Social and Health Services caseworkers is to apply services, protect children until the services have been applied, and to reunite the family. Philosophically this policy assumes a concerned "family" having temporary problems. The reality is that often no family exists. The caseworker is directed to find placement for a child in a foster home with caring and conscientious foster parents. Foster parents' commitment is remarkable given the limited authority they have over the children and the knowledge that their care is temporary, the goal being to reunite the children with their natural parents. This goal seems to assume children are property. The reality is that there are too few foster parents and too many children to be placed. Sometimes, like in Babcock, caseworkers make mistakes and children who come from a sad environment go into a sadder one, with further victimization. If there is liability, then it is one we all share if adequate resources are not made available. We should not make the caseworker a scapegoat because there are limited resources and the caseworker makes an honest, time-pressured but faulted decision.
The same public policy argument applies to parole officers. The parole board employs the parole officers to implement its administrative decisions. The board will feel a responsibility to those officers and may not wish to place them in jeopardy by paroling a person eligible for parole if the officer employed by the board is going to be legally liable for the parolee's conduct. Thus, opportunities will be lost where parole might be warranted. Successful parole requires assisting a parolee in making his or her own decisions and thus allowing some freedom of action. If the parole officer has liability for the actions of a reoffending parolee, the supervision will not be sufficiently flexible to accomplish its rehabilitative goal. See Ferree v. State, 784 P.2d 149, 151 (Utah 1989); Rollins v. Petersen, 813 P.2d 1156, 1161-62 (Utah 1991).
Individuals on parole are generally not single offenders. They have been convicted of multiple felonies, and recidivism is high. To expect complete rehabilitation with no further offenses is unrealistic. To create liability on the parole officer is to place blame somewhere and thereby simplify a complicated social problem. The grant of "qualified" immunity by the majority is illusory and nothing more than a negligence standard wrapped like a gift box which contains sticks and ashes.
The majority opinion admits that generally one does not have liability for the criminal acts of a third party unless one "takes charge" over that person. The issue is the parole officer's control over the parolee's actions. It is a naive fiction to say parole officers have control over convicted felons who are free on parole. A person who does not have the ability to control another's conduct should not have liability imposed upon him or her for the tortious acts of that other person. See, e.g., Lamb v. Hopkins, 303 Md. 236, 492 A.2d 1297 (1985); Fox v. Custis, 236 Va. 69, 372 S.E.2d 373 (1988); Small v. McKennan Hosp., 403 N.W.2d 410 (S.D. 1987); Hartford Fire Ins. Co. v. State, 413 So. 2d 1360 (La. Ct. App. 1982).
Assuming services for the protection of children are valuable to society and parole services are valuable to society, we need public employees motivated, not fearful. Mistakes, lack of action, failure to deliver on promises made should, where warranted, result in discharge from public service. I would affirm the trial court, granting absolute immunity to the parole officers for supervisory actions.
My final point concerns duties of mental health professionals and involves the discussion of Petersen v. State, 100 Wn.2d 421, 671 P.2d 230 (1983). I express no opinion as to the validity of the discussion. My objection to the discussion is that it was not required to be addressed, was not fully presented, and no opportunity was given to amicus to assist the court in this evaluation. It should be regarded as dicta, although it may reflect the position of the court when the issue is ultimately presented.
We have now created differing standards of immunity for three state employees: caseworkers (Babcock), parole officers (Taggart), and police officers (Guffey v. State, 103 Wn.2d 144, 690 P.2d 1163 (1984)). The extent to which there is immunity for acts of public employees is a matter best examined by the Legislature for a uniform determination.