On November 7, 1980, Joseph Dugan, the appellant, applied to become a federal administrative law judge. His previous experience in the practice of administrative law was limited. But, he believed he could qualify by showing sufficient trial experience. The Office of Personnel Management, in its statement of requirements for the ALJ position, says applicants may count towards the seven necessary years of qualifying experience their "experience in preparation of cases for and their presentation at trial in courts of record of unlimited and original jurisdiction." It adds that the applicant must show that he spent at least two years "in the actual preparation and trial of cases"; and, to do so, he should list "in chronological order ... court cases which he has prepared and tried ... to demonstrate 2 full years (400 workdays) within the 7-year period immediately preceding the date of the application." Office of Personnel Management, Administrative Law Judge, Announcement No. 318, October 1979, pp. 4, 5, 11 (emphasis in original). Dugan had practiced as a trial lawyer with Neighborhood Legal Services in Washington, D.C. from 1965 to 1969; he had served as Deputy Director and Director of Rhode Island Legal Services from 1970 to 1977; he practiced privately as a trial lawyer from 1977 to 1980. He estimates that he devoted to the preparation of cases for trial and their trial about 95 percent of his time in the first period, about 60 percent in the second period, and 100 percent in the third period. He had other relevant experience as well. Thus he sent to OPM a lengthy application detailing this experience, as required, with case titles, citations, descriptions, dates, names and addresses of courts, judges, counsel, and the number of workdays spent on each case.

In early December Dugan received a printed form from OPM stating that OPM would not process this application further because, among other things, he had shown only 188 days, not 400 days, of acceptable case work. After considerable effort to find out which cases OPM did not count and why, Dugan concluded (and parties here agree) that OPM based its rejection, in relevant part, on its practice of counting towards the necessary two years of trial experience only those trial preparation days belonging to cases actually tried. It will not count a trial preparation day belonging to a case that settles, even if it settles on the eve of trial.

Dugan attacked this practice as arbitrary, (and also unconstitutional). The ALJ Rating Appeals Panel, within OPM, rejected his argument. The federal district court also seems to have rejected it, although it dismissed Dugan's action for failure to exhaust his administrative remedies--a point that the court raised sua sponte. Dugan now appeals, claiming that the district court, 560 F. Supp. 1230, erred in believing it lacked jurisdiction and that OPM's practice must be set aside as "arbitrary, capricious, an abuse of discretion," under the Administrative Procedure Act, 5 U.S.C. Sec. 706(2)(A). We agree with Dugan on both counts.

The district court's dismissal of this case for failure to "exhaust remedies" rests upon its view that the Civil Service Reform Act of 1978 provides Dugan with an administrative remedy that he failed to invoke, namely, the right to petition OPM's Special Counsel for redress. The statute says the Special Counsel can look into allegations of a "prohibited personnel practice," report his findings or recommendations, and, if necessary, petition the Merit System Protection Board to consider the matter. 5 U.S.C. Sec. 1206. Dugan did not bring his problem to the attention of the Special Counsel. The district court, however, was wrong in believing that "exhaustion law" required him to do so.

For one thing, exhaustion of remedies requirements are normally waivable. Mathews v. Diaz, 426 U.S. 67, 76-77, 96 S. Ct. 1883, 1889-1890, 48 L. Ed. 2d 478 (1976); Mathews v. Eldridge, 424 U.S. 319, 328, 96 S. Ct. 893, 899, 47 L. Ed. 2d 18 (1976). And the government here "waives" (to the extent permitted by statute), for it specifically says that Dugan need not petition the Special Counsel. For another thing, the Court of Appeals for the District of Columbia Circuit has held that the Special Counsel provisions do not provide employees (or applicants) with an exhaustable avenue for further agency review. Frazier v. Merit Systems Protection Board, 672 F.2d 150, 162-63 (D.C.Cir.1982). Rather, the Special Counsel is a type of agency ombudsman. "Corrective action petitions" sent to the Special Counsel are not like individual appeals. Rather, they are "comparable to criminal prosecutions designed to vindicate the public interest." Id. The government, responding to our request at oral argument that it tell us whether it thinks the district court was right or wrong on the "exhaustion" point, has filed a supplementary brief agreeing with Frazier (and with Dugan) and disagreeing with the district court's opinion that "appeal to the [Special Counsel] ... is mandatory in these circumstances." In its view, the "corrective action authority of the Special Counsel does not have 'anything to do' with [the] appeal right of individual employees for grievances ...." Given the government's concession, we see no reason not to follow Frazier here.

The district court's discussion suggests, however, that, in referring to "exhaustion," it may have had a deeper problem in mind. The Administrative Procedure Act makes clear that "final agency action for which there is no other adequate remedy in a court [is] ... subject to judicial review." 5 U.S.C. Sec. 704. But, there are two exceptions. A court cannot review an agency action under the APA "to the extent that (1) statutes preclude judicial review; or (2) agency action is committed to agency discretion by law." 5 U.S.C. Sec. 701. The district court seems to have felt that one or the other of these exceptions may apply here. And, the government's supplementary brief appears to make a similar argument.

We are not aware, however, of any statute that precludes review of the "final" agency action here at issue. The government's suggestion to the contrary seems to depend upon implying a congressional intent to deny review from the Civil Service Reform Act itself. That Act provides for review of most personnel actions (but not this one) by the Merit Systems Protection Board. It goes on to create two channels for court review of MSPB decisions--one (for most cases) leading to the Court of Appeals for the Federal Circuit, 5 U.S.C. Sec. 7703(b)(1), and the other (for special cases involving, for example, claims of discrimination) leading to federal district court, 5 U.S.C. Sec. 7703(b)(2). The government would imply, from the fact that Congress wished to unify much court review of MSPB decisions in the Federal Circuit, a congressional prohibition against court review of final personnel decisions that do not find their way to the MSPB. This conclusion, however, does not follow logically from the premise. It runs counter to the strong presumption in the law that favors reviewability and almost never implies statutory preclusion of review from congressional silence. See Johnson v. Robinson, 415 U.S. 361, 373-74, 94 S. Ct. 1160, 1168-69, 39 L. Ed. 2d 389 (1974) ("clear and convincing" evidence of congressional intent is required to preclude review); Barlow v. Collins, 397 U.S. 159, 167, 90 S. Ct. 832, 838, 25 L. Ed. 2d 192 (1970) (same); San Juan Legal Services, Inc. v. Legal Services Corp., 655 F.2d 434, 438 (1st Cir.1981) (similar); Virginia ex rel. Commissioner v. Marshall, 599 F.2d 588, 592 (4th Cir.1979) (same). It runs counter to a history of court review (as presented here by the government) both before and after the enactment of the CSRA. Friedman v. Devine, No. 81-756 (D.D.C. Feb. 25, 1982) (review by district court of decision of ALJ Rating Appeals Panel); Bromberg v. Civil Service Commission, No. 73 C 537 (N.D.Ill. March 25, 1977) (review by district court of decision of Civil Service Commission), aff'd, No. 75-1485 (7th Cir. Dec. 19, 1975). And, it strikes us as implausible to believe that Congress wished to withdraw court review of even egregious agency behavior in the area--even if, for example, an agency were to discard applications as a result of bribery or the roll of the dice. We therefore reject it.

The district court also suggested that the second exception, "agency action committed to agency discretion," might bar review. But, that suggestion also seems wrong. Of course, when agencies hire personnel, their discretion is broad and courts must be reluctant to interfere. American Federation of Government Employees v. Hoffman, 543 F.2d 930, 938 (D.C.Cir.1976), cert. denied, 430 U.S. 965, 97 S. Ct. 1645, 52 L. Ed. 2d 356 (1977); Bielec v. United States, 456 F.2d 690, 695, 197 Ct. Cl. 550 (1972); cf. Borsari v. Federal Aviation Administration, 699 F.2d 106, 109 (2d Cir.), cert. denied, --- U.S. ----, 104 S. Ct. 115, 78 L. Ed. 2d 115 (1983) ("[A] smooth functioning government requires that public officials be afforded broad discretion to make employment decisions based on their view of the agency's best interest."). But, the fact that an agency enjoys broad discretionary powers does not mean judicial review is forbidden; it simply means that the reviewing court is unlikely to find against the agency, for the agency is unlikely to have acted unlawfully. See generally Berger, Administrative Arbitrariness and Judicial Review, 65 Colum.L.Rev. 55 (1965); 4 K. Davis, Administrative Law Treatise, Sec. 28.16 (1958) and Administrative Law of the Seventies, Sec. 28.16 (1976). To have bite, the "committed to agency discretion," exemption must forbid the court to review even those decisions in which the agency may have acted un lawfully--where it may have exceeded its broad powers. Sometimes courts will find that Congress intended to preclude such review, typically when review involves foreign affairs, the military, or other areas in which the very act of reviewing may impede the agency's ability to carry out its functions. See, e.g., Chicago & Southern Air Lines, Inc. v. Waterman Steamship Corp., 333 U.S. 103, 68 S. Ct. 431, 92 L. Ed. 568 (1948); Braniff Airways, Inc. v. CAB, 581 F.2d 846 (D.C.Cir.1978); Langevin v. Chenango Court, Inc., 447 F.2d 296 (2d Cir.1971); Hahn v. Gottlieb, 430 F.2d 1243 (1st Cir.1970); Saferstein, Nonreviewability: A Functional Analysis of "Committed to Agency Discretion," 82 Harv.L.Rev. 367 (1968). But here, as previously mentioned, there is no reason to believe that Congress wished or needed to preclude review of unlawful action in order to carry out a statutory objective. To the contrary, Congress has specifically provided for judicial review of many related actions affecting not only employed civil servants but also "job applicants." 5 U.S.C. Sec. 7703. In sum, we believe that OPM's discretionary powers to accept or to reject job applications are broad. 5 U.S.C. Sec. 1302. But, court review of a "final agency decision," to determine whether the agency's action was lawful, is nonetheless appropriate.

In reviewing for arbitrariness, we recognize the broad authority that the relevant statutes grant OPM to develop, and to administer, ALJ hiring criteria. See 5 U.S.C. Secs. 1305, 3105, 5372. Moreover, the agency practice here challenged consists of an administrative interpretation of its own regulations. An agency's interpretation of its own regulations is rarely set aside. United States v. Larionoff, 431 U.S. 864, 872-73, 97 S. Ct. 2150, 2155-56, 53 L. Ed. 2d 48 (1977); Bowles v. Seminole Rock Co., 325 U.S. 410, 413-14, 65 S. Ct. 1215, 1217, 89 L. Ed. 1700 (1945). But see Jicarilla Apache Tribe v. FERC, 578 F.2d 289, 292-93 (10th Cir.1978). Still, we believe that the agency interpretation here at issue is so totally unreasonable as to be without support in the authorizing statute. Or to put the matter in APA terms, it is unreasonable to the point that it constitutes an "abuse" of the agency's discretionary powers. 5 U.S.C. Sec. 706(2)(A).

First, the distinction between a day of trial preparation for a case that reaches trial and one that settles prior to that trial is often a distinction without a difference. The legal work involved will often be identical. Indeed a trial lawyer who tends to settle cases may be a better trial lawyer than one who does not.

Second, the agency's refusal to take this fact of "preparation time equivalence" into account is not the minor sort of irrationality that typically accompanies any administrative effort to shape rules and practices. Rather, it is a serious irrationality that likely thwarts the very purpose of the basic regulation, namely, to allow lawyers with trial experience to qualify to become ALJ's. Dugan alleges and supports with affidavits from experienced practitioners the well known fact that between 75 percent and 90 percent of all cases filed in court eventually settle. Thus, if the agency insists upon 400 days of "trial practice" and counts only those days related to cases actually tried, the typical trial lawyer would have to engage in 1600 to 4000 days of trial case work to qualify. And 1600 to 4000 days amounts to more workdays than likely exist in the seven year period that OPM allows the applicant to canvass for relevant experience. Of course, these numbers may not be exactly right; and some lawyers will qualify regardless. But, the numbers do show that Dugan has pointed out something seriously, not trivially, wrong with OPM's system.

Third, OPM cannot claim that its interpretation is necessary to prevent applicants from resting upon experience that is all "preparation" and no "trial," for there is a perfectly obvious way to guarantee that the applicant has actual trial experience, namely, to require that the applicant have spent a certain number of days in court trying cases.

Fourth, OPM cannot easily claim that its practice is administratively necessary to help it segregate "real" trial preparation experience from work not truly related to trial preparation. All that is at issue here is whether OPM will proceed to its next stage of selection when it writes to opposing counsel and the trial judge to determine in some detail what happened in the particular case. OPM remains free to separate "ersatz" from "real" in light not only of the applicant's description but also of what it learns through its correspondence (which it routinely undertakes anyway once past the threshold). This is to say that OPM cannot rest upon a "need for mechanical rule" justification in an area where it makes subjective, qualitative judgments regardless.

Fifth, in any event, OPM did not even advance the justifications we have just mentioned. Rather, we might take its justifications to consist of those we have gleaned from the record, prior proceedings, and its briefs in this court. Cf. Motor Vehicle Manufacturers Association of the United States v. State Farm Mutual Automobile Insurance Co., --- U.S. ----, ----, 103 S. Ct. 2856, 2870, 77 L. Ed. 2d 443, 462 (1983) ("[A]n agency's action must be upheld, if at all, on the basis articulated by the agency itself."). They consist of the following:

a. OPM points to language in its qualification announcement stating that trial lawyers, relying on trial, rather than administrative law, experience, must show "actual participation in the preparation and trial" of cases. This, and other similar language in the announcement interpreted in a commonsense manner, however, says that the agency wants lawyers with (1) experience preparing cases for trial and (2) experience trying cases; it does not rigidly insist that all preparation experience must involve a case that actually goes to trial.

b. OPM points to language at page 11 of its announcement that says the applicant relying upon trial experience must prepare "a list ... of ... court cases which he or she has prepared and tried ... to demonstrate 2 full years (400 workdays)." We doubt that even this language has to be read literally, as OPM has done, to make irrelevant all preparatory experience that is not followed by at least a day of trial. But, even if it did say just that, the existence of the language would merely show that OPM has the practice; it would not justify it.

c. OPM states that a group of experts, including representatives of the American Bar Association and the federal regulatory agencies, decided on these selection criteria. There is every reason to believe that these experts wanted trial lawyers to have experience trying cases as well as preparing cases for trial. But the record reveals no reason to believe that these experts hoped for, intended, or knew about OPM's apparently unnecessarily rigid interpretation of this requirement.

d. OPM's appeals panel added that "it has always been necessary for an applicant to have prepared and presented a case in court in order to receive credit for it as qualifying experience." This argument is based upon administrative history. But, there is no evidence of a relevant longstanding administrative practice. Cf. NLRB v. Bell Aerospace Co., 416 U.S. 267, 274-75, 94 S. Ct. 1757, 1761-62, 40 L. Ed. 2d 134 (1975) ("a court may accord great weight to the longstanding interpretation placed on a statute by an agency charged with its administration"); Massachusetts Trustees of Eastern Gas & Fuel Associates v. United States, 377 U.S. 235, 241, 84 S. Ct. 1236, 1241, 12 L. Ed. 2d 268 (1964) ("Some weight is due to the consistent interpretation of the Maritime Commission: the agency entrusted with administration of the statute."). There is nothing in the record to indicate that OPM has hired many trial lawyers in the past or any suggestion that anyone previously challenged the practice here at issue.

e. A memorandum written by Administrative Law Judge Irving Sommer states that Dugan's allegation that the

requirements are arbitrary, whimsical, etc. does not merit serious concern. The record of the Office of Administrative Law Judges, Office of Personnel Management, fashioned throughout the years and acknowledged by all knowing members of the bar and bench as one that has seen the appointment of a group of ALJ's who are fair, intelligent and learned in all facets of the law and procedure is full testimony to the merits of the examination process.

Dugan, however, is not directly attacking the quality of the existing group of Administrative Law Judges.

f. We have searched the record for any additional justification for OPM's practice. We have found nothing except frequent references to the broad discretion that OPM enjoys in selecting ALJs. The government places particular weight upon two such cases, Friedman v. Devine, supra, and Bromberg v. Civil Service Commission, supra. The first of these upheld OPM's decision not to count as litigation experience the applicant's work in the Advice Division of the Office of General Counsel, where his contact with "formal cases" consisted primarily of preparing Memoranda advising NLRB regional offices to issue complaints or to dismiss cases. The second case upheld the Civil Service Commission's determination that an applicant who practiced as an Assistant Attorney General in Illinois where he dealt almost exclusively with state revenue matters should receive less credit for experience than a federal lawyer in grade GS15. The OPM decisions and practices in these two cases seem better grounded in reason that the practice at issue here.

Sixth, the record reveals that those OPM officials who specifically focused on the problem that Dugan raised were skeptical about the value of OPM's practice. At one point Senator John Chafee, to whom Dugan had sent a letter of complaint, wrote to OPM noting that many of Dugan's points "appear to be right on target." In reply, Mr. Arch Ramsay, the acting Director of OPM, stated

I am informed that OALJ, ..., is considering asking applicants to differentiate in describing trial experience as to whether cases that fell short of trial on the merits involved participation in pretrial procedures in court so as to demonstrate experience other than office practice.

In a similar vein, Edward T. Rhodes, the Chairman of the ALJ Appeals Panel wrote that Dugan

"raises an interesting point ..., that is his concern of how a practicing lawyer in the private arena ... might be eligible to become an Administrative Law Judge in view of their type of practice and settlement percentages.... [I]t would be an interesting statistic to determine how many active ALJ's qualify from the private sector as opposed to the government sector ...."

The record does not reveal the answer to Judge Rhodes' question.

In sum, OPM's practice of counting as relevant experience trial preparation work related to cases that are actually tried, but refusing to count as relevant likely identical experience related to cases that settle, strikes us as unreasonable. It threatens to hinder significantly, rather than to help, achieve the purpose of the regulation that it purports to interpret. OPM has provided no sensible justification for the practice. And, the practice, on its face, does not appear a reasonably necessary adjunct for practical administration of OPM's ALJ qualifications. Under these circumstances, it seems "arbitrary, capricious, an abuse of discretion" within the meaning of the APA. 5 U.S.C. Sec. 706(2)(A).

We add, lest this case be taken incorrectly as a precedent authorizing strict judicial scrutiny of agency decisions not to hire applicants, that such is not our intent. OPM has not yet evaluated Dugan's application; it has refused to do so on the basis of a "screening rule." We find only that the particular screening criterion that OPM used is, for the reasons set out, unlawful. OPM's evaluation of a candidate's qualifications is not here in issue.

We wish to mention one final matter. The record in this case gives us some reason to fear that OPM may not handle the matter expeditiously. In December 1980 as soon as OPM turned him down, Mr. Dugan wrote OPM asking which cases it had not accepted and why. It took OPM nine months and many letters before it purported to provide a serious answer to this question. That answer, when it finally came, still did not explain which cases were rejected for which reasons. It listed some reasons that are trivial (for example, it says that a major case that Dugan had listed as tried in the Rhode Island federal district court would not count because Dugan had not provided the address of Chief Judge Pettine). It also listed several reasons that are not at issue on this appeal. The lack of clarity in the administrative record, our consequent uncertainty as to whether there are other "threshold" matters that might lead OPM not to consider Dugan's application, and the history of delay, lead us to suggest that the district court determine the future course of these proceedings with an eye towards expeditious resolution of the case, keeping in mind its power to "compel agency action ... [that is] unreasonably delayed." 5 U.S.C. Sec. 706(1).

The district court's order of dismissal is reversed. The case is remanded to the district court for further proceedings consistent with this opinion.

In light of the delay appellant has suffered, the fact that his allegations and arguments on appeal proved correct and the fact that the government failed to provide an initial full briefing of the relevant issues, we award double costs to appellant.