Citizens Awareness Network, Inc., Petitioner, Andnational Whistleblower Center et al., Intervenors, v. United States of America and United States Nuclear Regulatory Commission, Respondents, Andnuclear Energy Institute, Inc., Intervenor.public Citizen Critical Mass Energy and Environment Program et al., Petitioners, v. United States of America and United States Nuclear Regulatory Commission, Respondents, 391 F.3d 338 (1st Cir. 2004)

Disenchanted with its existing procedural framework for the conduct of adjudicatory hearings, the Nuclear Regulatory Commission (NRC or Commission) promulgated new rules designed to make its hearing processes more efficient. These new rules greatly reduce the level of formality in reactor licensing proceedings but, at the same time, place certain unaccustomed restrictions upon the parties. The petitioners and petitioner-intervenors are public interest groups. Supported by the Attorneys General of five states (who have filed a helpful amicus brief), they claim that the new rules violate a statutory requirement that all reactor licensing hearings be conducted in accordance with sections 554, 556, and 557 of the Administrative Procedure Act (APA), 5 U.S.C. §§ 554, 556 & 557.¹  In the alternative, they claim that the Commission has not put forth an adequate justification for so substantial a departure from prior practice and that, therefore, the new rules must be set aside as arbitrary and capricious. Fully cognizant of the gravity of our task, we have studied the complex statutory and regulatory framework and scrutinized the plenitudinous administrative record. After completing that perscrutation and grappling with an antecedent jurisdictional question, we find that the new procedures in fact comply with the relevant provisions of the APA and that the Commission has furnished an adequate explanation for the changes. Consequently, we deny the petitions for review.

The NRC is the federal agency charged with regulating the use of nuclear energy, including the licensing of reactors used for power generation. See 42 U.S.C. § 2201. The Atomic Energy Act requires the Commission to hold a hearing "upon the request of any person whose interest may be affected," id. § 2239(a) (1) (A), before granting a new license, a license amendment, or a license renewal.

The NRC's predecessor agency, the Atomic Energy Commission (AEC), originally interpreted this provision as requiring on-the-record hearings in accordance with the APA. See Hearings Before the Subcommittee on Legislation, Joint Committee on Atomic Energy, 87th Cong. 60 (1962) (letter of AEC Commissioner Loren K. Olsen). These hearings closely resembled federal court trials, complete with a full panoply of discovery devices and direct and cross-examination of witnesses by advocates for the parties. Such hearings proved to be very lengthy; some lasted as long as seven years.

In 1982, the NRC relaxed its approach for certain types of licensing proceedings. See, e.g., In re Kerr-McGee Corp., 15 N.R.C. 232, 235, 1982 WL 31680 (1982) (determining that formal hearings are not necessary in materials licensing cases). Although the results were heartening, the Commission nevertheless retained the full range of trial-like procedures for reactor licensing cases. The passage of time brought further changes: faced with the prospect of hearings on many license renewal applications in the near future — a large number of reactors were initially licensed in the decade from 1960 to 1970 and the standard term for such licenses was forty years — the Commission began to reassess its adjudicatory processes, focusing particularly on the procedures used in reactor licensing cases. The NRC's issuance, in 1998, of a policy on the conduct of adjudicatory proceedings, 63 Fed.Reg. 41,872 (Aug. 5, 1998), marked the inception of this process. This policy statement reiterated the NRC's commitment to expeditious adjudication and urged hearing officers to employ a variety of innovative case-management techniques in order to improve hearing efficiency.

While encouraging better utilization of existing procedures, the Commission also began pondering possible procedural revisions. In January of 1999, the NRC's general counsel drafted a legal memorandum concluding that the Atomic Energy Act did not require reactor licensing hearings to be on the record and, accordingly, that the Commission had the option of replacing the existing format with a truncated regime. Later that year, the Commission held a widely attended workshop on hearing procedures. Building on this foundation, the Commission published a notice of proposed rulemaking on April 16, 2001, 66 Fed.Reg. 19,610, suggesting a major revision of its hearing procedures. In an accompanying statement, the Commission took the position that section 189 of the Atomic Energy Act, 42 U.S.C. § 2239, does not require reactor licensing proceedings to be on the record.

On January 14, 2004, the NRC published a final rule, along with a response to the comments that the proposed rule had generated. See 69 Fed.Reg. 2,182. With minor exceptions, the final rule replicated the proposed rule. The statement of considerations for the final rule reiterated the Commission's view that reactor licensing hearings may be informal.

The new rules took effect on February 13, 2004. Although they apply to all adjudications conducted by the NRC, the petitioners only challenge their application to reactor licensing proceedings. We therefore confine our ensuing discussion to that aspect of the new rules.

Under the old protocol, all reactor licensing hearings were conducted according to the procedures outlined in 10 C.F.R. part 2, subpart G. The subpart G rules resemble those associated with judicial proceedings.²  They include a complete armamentarium of traditional discovery devices (e.g., requests for document production, interrogatories, and depositions). 10 C.F.R. § 2.705. The parties may make motions for summary disposition (although the hearing officer is not required to entertain them). Id. § 2.710. There is an evidentiary hearing at which testimony is presented through direct and cross-examination of witnesses by the parties. Id. § 2.711.

Under the new rules, reactor licensing hearings are, for the most part, to be conducted according to a less elaborate set of procedures described in 10 C.F.R. pt. 2, subpart L.³  The new subpart — which differs materially from the old subpart L — limns a streamlined hearing procedure. Unlike subpart G, subpart L does not provide for traditional discovery. 10 C.F.R. § 2.1203. Instead, parties in hearings governed by subpart L are required to make certain mandatory disclosures (akin to "open file" discovery) anent expert witnesses, expert witness reports, relevant documents, data compilations, and claims of privilege. Id. § 2.336.

The hearings themselves also differ. Under subpart L, the presumption is that all interrogation of witnesses will be undertaken by the hearing officer, not the litigants. Id. § 2.1207. Parties are allowed to submit proposed questions in advance of the hearing, but the presiding officer is under no compulsion to pose them. Id. Parties are not allowed to submit proposed questions during the hearing unless requested to do so by the presiding officer. Id. Cross-examination is not available as of right, although a party may request permission to conduct cross-examination that it deems "necessary to ensure the development of an adequate record for decision." Id. § 2.1204. A party seeking leave to conduct cross-examination must submit a cross-examination plan, which will be included in the record of the proceeding regardless of whether the request is allowed. Id.

The petitioners — we use that phrase broadly to include the petitioner-intervenors — took umbrage at these changes and brought these petitions for judicial review. Their primary claim is that the Commission erred in its determination that reactor licensing proceedings do not have to be fully formal adjudications. In their view, the new rules do not comply with the APA's requirements for on-the-record adjudication and, therefore, cannot stand. As a fallback, the petitioners assert that even if the new rules are not ultra vires, they must be set aside as arbitrary and capricious.

The parties have operated on the assumption that this court has first-instance jurisdiction to hear and determine their petitions for judicial review. We are not so sanguine — and we are cognizant that, as a court of limited jurisdiction, subject-matter jurisdiction will not accrete to us either by the parties' acquiescence or by their consent. Espinal-Dominguez v. Puerto Rico, 352 F.3d 490, 495 (1st Cir. 2003). Consequently, we asked the parties to address what we perceived to be a thorny question relating to our authority to entertain these petitions. Before proceeding to the merits of the petitioners' asseverational array, we must resolve that question.

The facts are as follows. The petitioners premise jurisdiction on the Administrative Orders Review Act, 28 U.S.C. §§ 2341-2351, better known as the Hobbs Act. In pertinent part, that statute confers original jurisdiction on the courts of appeals to hear petitions for judicial review of "all final orders of the [NRC] made reviewable by section 2239 of title 42." Id. at § 2342(4). In turn, 42 U.S.C. § 2239(b) makes reviewable, inter alia, "[a]ny final order entered in any proceeding of the kind specified in subsection (a) of this section." The proceedings enumerated in that subsection include those for "the granting, suspending, revoking, or amending of any license or construction permit, or application to transfer control, and in any proceeding for the issuance or modification of rules and regulations dealing with the activities of licensees." Id. § 2239(a).

Read literally, these interlocking statutes would not seem to grant jurisdiction to this court. After all, the petitioners are challenging a rule, not an order. The APA, which is made applicable to the Commission by 42 U.S.C. § 2231, defines an order as "the whole or a part of a final disposition, whether affirmative, negative, injunctive, or declaratory in form, of an agency in a matter other than rule making...." 5 U.S.C. § 551(6) (emphasis supplied). Thus, the action at issue here — a rulemaking — would appear to fall outside the scope of review provided by the Hobbs Act.

Even if one were tempted to suppose that Congress simply misspoke in limiting Hobbs Act jurisdiction to the review of orders, other sections of the Act would seem to militate against a judicial reconstruction of the term "order" to encompass rulemaking. The Act explicitly provides for initial court of appeals review of "all rules, regulations, or final orders" of the Secretary of Transportation, the Federal Maritime Commission, and the Surface Transportation Board. 42 U.S.C. §§ 2342(3), 2342(5). The principle is clear that Congress's use of differential language in various sections of the same statute is presumed to be intentional and deserves interpretive weight. See Duncan v. Walker, 533 U.S. 167, 173, 121 S. Ct. 2120, 150 L. Ed. 2d 251 (2001); In re 229 Main St. Ltd. P'ship, 262 F.3d 1, 5-6 (1st Cir. 2001).

Were we writing on a pristine page, we would likely find this careful parsing persuasive and thus dismiss the case so that the petitioners could seek initial review in an appropriate district court. The page, however, is cluttered, not pristine. There is a substantial body of precedent elaborating the scope of the Hobbs Act with respect to both the NRC and other agencies to which it applies.

The key case is Florida Power & Light Co. v. Lorion, 470 U.S. 729, 105 S. Ct. 1598, 84 L. Ed. 2d 643 (1985), in which the Supreme Court determined that Hobbs Act jurisdiction existed in the courts of appeals for initial review of the NRC's denial of citizen petitions to suspend or revoke licenses. Id. at 746, 105 S. Ct. 1598. In making this determination, the Court declared that the language of section 2239 was ambiguous as to whether it limited judicial review to orders entered in proceedings under that section. Id. at 736, 105 S. Ct. 1598. The Court then declared that the Hobbs Act should be interpreted broadly, so as to maximize the availability of initial circuit court review of licensing proceedings. Id. at 745, 105 S. Ct. 1598.

The Court laid out two grounds in support of this reasoning. First, it cited efficiency concerns. In this regard, the Court deemed initial circuit court review the better use of judicial resources, observed that such a course eliminates one layer of review, and stressed that there is usually no need for the compilation of either a fresh or an augmented record in agency review proceedings. Id. at 744, 105 S. Ct. 1598. Second, the Court harangued against the evils of piecemeal review. In this regard, it warned that when Congress clearly places initial review of some agency actions in the courts of appeals, the jurisdictional provision should not be interpreted narrowly to shunt review of other agency actions to the district courts. Id. at 741-42, 105 S. Ct. 1598. For these reasons, the Court admonished that "[a]bsent a firm indication that Congress intended to locate initial APA review of agency action in the district courts, we will not presume that Congress intended to depart from the sound policy of placing initial APA review in the courts of appeals." Id. at 745, 105 S. Ct. 1598.

Lorion has displayed remarkable vitality. The Seventh Circuit applied its teachings in Commonwealth Edison Co. v. U.S. NRC, 830 F.2d 610 (7th Cir. 1987), finding jurisdiction to review the Commission's assessment of fees for the processing of a licensing application. Id. at 613. In the court's view, the assessment was sufficiently related to a licensing proceeding to ground circuit court jurisdiction. Id. at 612-13. Pertinently for present purposes, the court, in the exercise of its discerned jurisdiction, reviewed the underlying rules on which the Commission had based its assessment. Id. at 616.

Closer to home, this court has applied Lorion to find jurisdiction when a contrary reading of the applicable statute would, for no apparent reason, have divided judicial review between the district courts and the courts of appeals. See City of Boston v. U.S. HUD, 898 F.2d 828, 834-35 (1st Cir. 1990). The Third Circuit has gone even further, holding that Lorion creates a presumption of initial circuit court review "absent clear and convincing evidence of a contrary congressional intent." Conoco, Inc. v. Skinner, 970 F.2d 1206, 1214 (3d Cir. 1992).

Although the question is close, we conclude that appellate jurisdiction is proper in this case. In reaching this conclusion, we start with the premise that both the Hobbs Act and the Atomic Energy Act are ambiguous as to their reach. Furthermore, while the term "order" has a clear meaning for APA purposes, its placement in section 2239 of the Atomic Energy Act suggests that Congress might not have used it with the same precision in connection with the intersection of the Hobbs Act and the Atomic Energy Act. Cf. Hanover Ins. Co. v. United States, 880 F.2d 1503, 1504 (1st Cir. 1989) (noting that the same word may have different meanings in different statutory contexts). That premise is bolstered by the fact that the Atomic Energy Act uses the terms "order" and "rule" inconsistently. For example, section 2239(b) refers to "[a]ny final order entered in any proceeding of the kind specified in subsection (a)," but section 2239(a) includes proceedings "for the issuance or modification of rules and regulations dealing with the activities of licensees." This disharmony renders the meaning of "order" in this context uncertain.

Given these amphibolies, we believe that the policies announced by the Supreme Court in Lorion deserve special weight. We interpret Lorion as holding that original jurisdiction in the courts of appeals is proper to review any NRC action that could be cognizable in a petition for review from a proceeding under section 2239. This interpretation is consistent with the Lorion Court's instruction that jurisdictional statutes should be construed so that agency actions will always be subject to initial review in the same court, regardless of the procedural package in which they are wrapped. Lorion, 470 U.S. at 742, 105 S. Ct. 1598. By like token, an affirmation of jurisdiction in this case is consistent with the Lorion Court's conclusion that judicial efficiency is best served by limiting the layers of review. Id. at 744-45, 105 S. Ct. 1598. On this basis, and in conformity with our earlier decision in City of Boston, we conclude that we have jurisdiction to entertain these petitions.

We divide our discussion of the merits into three segments, corresponding with the petitioners' most lively bruited points.

The mainstay of the petitioners' challenge is the proposition that the new rules exceed the Commission's statutory authority. The petitioners start with the premise that 42 U.S.C. § 2239 requires the NRC to conduct licensing hearings on the record, that is, in strict accordance with the relevant provisions of the APA. See supra note 1. In their view, the new rules fail to satisfy that requirement and, therefore, must be pole-axed. In the pages that follow, we examine both the petitioners' premise and their conclusion.

Section 2239 requires the Commission, "upon the request of any person whose interest may be affected" by certain agency actions, to hold "a hearing." It does not explicitly require that the hearing be on the record. We have held, however, that the degree of formality that a hearing must afford does not necessarily turn on the presence or absence of an explicit statutory directive. If, even absent such a directive, the nature of the hearing that Congress intended to grant is clear, then that intention governs. Dantran, Inc. v. U.S. Dep't of Labor, 246 F.3d 36, 46 (1st Cir. 2001); Seacoast Anti-Pollution League v. Costle, 572 F.2d 872, 876 (1st Cir. 1978). We assume arguendo, favorably to the petitioners, that the Seacoast rule still obtains.⁴ 

The petitioners advance several arguments for holding that Congress, in enacting section 2239, purposed to require on-the-record hearings in reactor licensing cases. In addition to canvassing the legislative history and cataloging the relevant amendments to the statute, they point out that for approximately four decades the NRC and its predecessor agency, the AEC, interpreted the statute as requiring on-the-record hearings in reactor licensing proceedings. In response, the NRC highlights the ambiguity of the statute and attempts to situate the latest round of changes in a larger history of procedural experimentation. The Commission also notes that some courts have interpreted section 2239 to allow informal hearings in licensing proceedings not involving reactors. See, e.g., City of W. Chicago v. U.S. NRC, 701 F.2d 632, 645 (7th Cir. 1983) (licensing of nuclear materials). Last — but far from least — the Commission urges us to defer to its judgment that informal hearings are a suitable prophylactic for reactor licensing. Cf. Lattab v. Ashcroft, 384 F.3d 8, 19-20 (1st Cir. 2004) (deferring to the agency's judgment on the proper application of a procedural statute).

For years, the courts of appeals have avoided the question of whether section 2239 requires reactor licensing hearings to be on the record. See, e.g., Kelley v. Selin, 42 F.3d 1501, 1510-14 (6th Cir. 1995) (discussing, but not resolving, the issue while approving the use of informal hearings for materials storage issues); Nuclear Info. & Res. Serv. v. NRC, 969 F.2d 1169, 1180 (D.C. Cir. 1992) (en banc) (deeming the issue forfeited and declining to decide it); Union of Concerned Scientists v. U.S. NRC, 920 F.2d 50, 53 n. 3 (D.C. Cir. 1990) (finding the procedural rules at issue to comply with the APA and declining to decide whether formal hearings are required); City of W. Chicago, 701 F.2d at 642-43 (distinguishing reactor licensing from materials licensing and addressing only the latter). We too decline to resolve this issue. Because the new rules adopted by the Commission meet the requirements of the APA it does not matter what type of hearing the NRC is required to conduct in reactor licensing cases.

Before elaborating our reasoning on this point, we must dispense with a procedural theory advocated by the petitioner Public Citizen. It is a bedrock principle that a court may only uphold an administrative action on a rationale advanced by the agency in the administrative proceeding. SEC v. Chenery Corp., 318 U.S. 80, 95, 63 S. Ct. 454, 87 L. Ed. 626 (1943). Embracing this principle, Public Citizen asserts that the Commission has waived the argument that the new rules satisfy the APA's requirements because, in promulgating the new rules, it relied exclusively on its view that section 2239 does not mandate on-the-record hearings. Thus, it cannot now rely on a different rationale to defend the rules in court.

This assertion reads the record through rose-colored glasses. The Commission explicitly memorialized in the statement of considerations for the final rule the view that even if reactor licensing hearings were required to be on the record, the new rules would meet that requirement. 69 Fed.Reg. at 2,192 ("[T]he Commission believes that ... the hearing procedures in each of these subparts meets [sic] the requirements for an on-the-record hearing under the APA...."). No more was exigible to preserve the point. Accordingly, we turn to the merits of this rationale.

We exercise plenary review over the Commission's compliance with the APA. See Dantran, 246 F.3d at 48 (stating that agencies' interpretations of statutes they do not administer are not entitled to particular deference). The APA lays out only the most skeletal framework for conducting agency adjudications, leaving broad discretion to the affected agencies in formulating detailed procedural rules. See Am. Trucking Ass'ns, Inc. v. United States, 627 F.2d 1313, 1321 (D.C. Cir. 1980). In specific terms, the APA requires only that the agency provide a hearing before a neutral decisionmaker and allow each party an opportunity "to present his case or defense by oral or documentary evidence, to submit rebuttal evidence, and to conduct such cross-examination as may be required for a full and true disclosure of the facts." 5 U.S.C. § 556(d).⁵ 

The petitioners urge that the magnitude of the risks involved in reactor licensing proceedings warrant the imposition of a more elaborate set of safeguards. It is beyond cavil, however, that, short of constitutional constraints, a court may not impose procedural requirements in administrative cases above and beyond those mandated by statute (here, the APA). Vt. Yankee Nuclear Power Corp. v. Natural Res. Def. Council, Inc., 435 U.S. 519, 543-44, 98 S. Ct. 1197, 55 L. Ed. 2d 460 (1978); Union of Concerned Scientists, 920 F.2d at 53. Accordingly, we are not at liberty to impress on the Commission (or any other agency, for that matter) a procedural regime not mandated by Congress. The NRC's new rules will, therefore, succumb to the petitioners' first line of attack only if they fail to provide the minimal procedural safeguards actually demanded by the APA. See Nat'l Classif. Comm. v. United States, 765 F.2d 1146, 1151 (D.C. Cir. 1985).

We turn now from the general to the particular. The rulemaking at issue here effected several changes in the Commission's procedures. The petitioners focus their challenge on two aspects of the newly minted process. First, they object to the Commission's decision to eliminate discovery. Second, they complain about the Commission's decision to circumscribe the availability of cross-examination. Because these are the only issues on which the petitioners have offered developed argumentation, we confine our analysis to those portions of the new rules. Cf. United States v. Zannino, 895 F.2d 1, 17 (1st Cir. 1990) (holding that "a litigant has an obligation to spell out its arguments squarely and distinctly, or else forever hold its peace" (citations and internal quotation marks omitted)).

We begin with the question of whether the new rules fall below the APA's minimum requirements by eliminating discovery. The Commission points out, and the petitioners do not seriously contest, that the APA does not explicitly require the provision of any discovery devices in formal adjudications. See 5 U.S.C. § 556; see also Kelly v. U.S. EPA, 203 F.3d 519, 523 (7th Cir. 2000); Frilette v. Kimberlin, 508 F.2d 205, 208 (3d Cir. 1974). Thus, if the APA requires the Commission to provide any discovery to satisfy the standards for formal adjudications, that discovery must be necessary either to effectuate some other procedural right guaranteed by the APA or to ensure an adequate record for judicial review. Cf. U.S. Lines, Inc. v. Fed. Maritime Comm'n, 584 F.2d 519, 540 (D.C. Cir. 1978) (stating that an agency charged with holding a hearing to determine the public interest must provide adequate means of public participation); Seacoast, 572 F.2d at 876-77 (noting that in some cases procedural requirements may be implied to ensure adequate judicial review).

The petitioners suggest that discovery is necessary to realize the right of citizen-intervenors to present their case and submit an informed rebuttal. See 5 U.S.C. § 556. If discovery is unavailable, this thesis runs, citizen-intervenors will be unable to gather the evidence needed to support their contentions and, thus, will be shut out of meaningful participation in licensing hearings.

This thesis is composed of more cry than wool. The petitioners argue as if the new rules have eliminated all access to information from opposing parties — but that is a gross distortion. The new rules provide meaningful access to information from adverse parties in the form of a system of mandatory disclosure. See 10 C.F.R. § 2.336. Although there might well be less information available to citizen-intervenors under the new rules, the difference is one of degree. There is simply no principled way that we can say that the difference occasioned by replacing traditional discovery methods with mandatory disclosure is such that citizen-intervenors are left with no means of adequately presenting their case.

Nor do we think that full-dress discovery is essential to ensure a satisfactory record for judicial review. The Commission's final decision in any hearing must survive review based on the evidence adduced in the hearing. 5 U.S.C. § 556(e). The applicant bears the burden of proof in any licensing hearing, id. § 556(d), and it will have every incentive to proffer sufficient information to allow the agency to reach a reasoned decision. That same quantum of information should be adequate for a reviewing court to determine whether the agency's action is supportable.

To say more on this point would be to paint the lily. There is simply no discovery-linked conflict between the new rules and the APA's on-the-record adjudication requirement. The petitioners' first line of argument is, therefore, a dead end.

Turning to cross-examination, the petitioners' contentions fare no better: the new rules meet the APA's requirements. To explain this conclusion, we first must strip away the rhetorical flourishes in which the petitioners shroud their reasoning.

It is important to understand that, contrary to the petitioners' importunings, the new rules do not extirpate cross-examination. Rather, they restrict its use to situations in which it is "necessary to ensure an adequate record for decision." 10 C.F.R. § 2.1204. The legitimacy of this restriction must be weighed in light of the fact that the APA does not provide an absolute right of cross-examination in on-the-record hearings. Seacoast, 572 F.2d at 880. The APA affords a right only to such cross-examination as may be necessary for a full and fair adjudication of the facts. Id. Equally to the point, "[t]he party seeking to cross-examine bears the burden of showing that cross-examination is in fact necessary." Id. at 880 n. 16.

The Commission represents that, despite the difference in language, it interprets the standard for allowing cross-examination under the new rules to be equivalent to the APA standard. 69 Fed.Reg. at 2,195-96. When an agency provides a plausible interpretation of its own procedural rules and there is no record or pattern of contrary conduct a court has no right either to slough off that interpretation or to deem it disingenuous. Cf. Albathani v. INS, 318 F.3d 365, 378-79 (1st Cir. 2003) (accepting the agency's good faith in carrying out its procedures, while acknowledging that evidence to the contrary might warrant judicial intervention). Given the Commission's stated interpretation, the new rules on cross-examination cannot be termed inconsistent with the dictates of the APA. Nor do we see how cross-examination that is not "necessary to ensure an adequate record for decision" could be necessary to ensure appropriate judicial review.

Because we find that the new rules meet the APA requirements for on-the-record adjudications, we hold that their promulgation does not exceed the Commission's authority. Consequently, the petitioners' ultra vires argument founders.

Our labors are not yet done. Even though we hold that the new rules are within the ambit of the NRC's authority, we still must consider whether its decision to discard the old in favor of the new was arbitrary and capricious. See 5 U.S.C. § 706.

An agency's rules, once adopted, are not frozen in place. The opposite is true: an agency may alter its rules in light of its accumulated experience in administering them. Rust v. Sullivan, 500 U.S. 173, 186-87, 111 S. Ct. 1759, 114 L. Ed. 2d 233 (1991). An agency must, however, offer a reasoned explanation for the change. If the agency fails to furnish such an explanation, or if the proffered explanation fails to demonstrate that the agency fully considered its new course, the revised rules must be set aside. See Motor Vehicle Mfrs. Ass'n of the U.S. v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 42, 103 S. Ct. 2856, 77 L. Ed. 2d 443 (1983) ("If Congress established a presumption from which judicial review should start, that presumption ... is ... against changes in current policy that are not justified by the rulemaking record.").

The petitioners begin this branch of their assault with a global challenge to the rationality of departing from the NRC's forty-year-old procedural regime. They do not suggest that the Commission's goal of improving hearing efficiency is impermissible; rather, they maintain that the current procedures worked well enough and that the Commission has made no showing that they were the source of any past inefficiencies. The petitioners further note evidence in the rulemaking record suggesting that efficiency gains could be brought about through less drastic measures, such as more aggressive case management.

To resolve this point, we first must determine what an agency must show to justify modifying its procedural rules. As a general principle, agencies have broad authority to formulate their own procedures — and the NRC's authority in this respect has been termed particularly great. Union of Concerned Scientists, 920 F.2d at 54; BPI v. Atomic Energy Comm'n, 502 F.2d 424, 428 n. 3 (D.C. Cir. 1974). A necessary corollary of this authority is the freedom to experiment with different procedural formats. Consequently, tinkering with rules is by no means a forbidden activity.

Of course, there are limits on this prerogative. An agency may not act precipitously or in an irrational manner in revising its rules. But so long as these limits are observed, it is not the place of a reviewing court to second-guess the agency's decision as to when to make procedural changes. It is enough that the agency reasonably determines that existing processes are unsatisfactory and takes steps that are fairly targeted at improving the situation.

In this case, the NRC has determined that its existing rules of practice lead to hearings that are cumbersome, unnecessarily protracted, and wasteful of the resources of the parties and the Commission. This determination warrants a high degree of deference. See Vt. Yankee, 435 U.S. at 543-44, 98 S. Ct. 1197 (holding that agencies' evaluations of their procedural needs are entitled to great respect). Although the petitioners may disagree, we descry nothing in the record that would support setting aside the rule on the basis that the agency should have left well enough alone. It would unfairly handcuff administrators if agencies could not change their procedures simply because certain of their constituencies admired the status quo.

That leaves the how of the Commission's new rules; that is, the question whether the Commission acted arbitrarily in making a specific set of changes in order to achieve its goal of improved hearing efficiency. To clear this hurdle, the Commission must show both that its new rules constitute a rational means for achieving its stated objective and that it sensibly rejected other options considered in the rulemaking proceeding (including the option of maintaining the status quo). See State Farm, 463 U.S. at 43-44, 103 S. Ct. 2856. This calculus must fairly account for any benefits lost by modifying existing rules, as well as any advantages expected to be gained through the adoption of updated rules. See id.

Here, too, we address only the NRC's decisions to replace traditional discovery with mandatory disclosure and to restrict the availability of conventional cross-examination. With regard to discovery, the NRC explained in its statement of considerations for the final rule that the proposed substitute "has the potential to significantly reduce delays and resources expended by all parties in discovery." 69 Fed.Reg. at 2,194. The Commission also expressed the view that replacing traditional discovery with mandatory disclosure would eliminate a substantial amount of motion practice related to discovery matters. Id. Finally, the Commission determined that any prejudice to citizen-intervenors from eliminating traditional discovery would be offset by the mandatory disclosure requirements and the general public availability of NRC documents. Id.

The petitioners renew their exhortation that discovery is necessary for citizen-intervenors to participate effectively in reactor licensing hearings. They speculate that without discovery they will be without access to large amounts of detailed technical information (information that is available to the applicant and the Commission staff) and, thus, will be unable to respond adequately to technical and factual arguments.

We agree with the petitioners that the Commission's explanation for the change in discovery practice is thin. The Commission baldly states its belief that eliminating traditional discovery will shorten hearings and conserve resources, but it provides no empirical analysis of its experience with traditional discovery from which an outside observer can determine what benefits the Commission might reasonably expect. In a substantive rule, this omission might be fatal. See, e.g., Portland Cement Ass'n v. Ruckelshaus, 486 F.2d 375, 393 (D.C. Cir. 1973) ("It is not consonant with the purpose of a rulemaking proceeding to promulgate rules on the basis of inadequate data, or on data that, [in] critical degree, is known only to the agency.").

In the realm of procedure, however, agencies are presumed to have special competence and, accordingly, are held to less exacting standards of explication. See Union of Concerned Scientists, 920 F.2d at 54 (noting that procedural determinations "fall uniquely within the expertise of the agency"). To add to this leeway, we are not willing to ignore matters that are common knowledge to courts of law. Discovery, especially in complex matters, is both time-consuming and costly. We do not think it can reasonably be questioned that the replacement of discovery with mandatory disclosure will make reactor licensing hearings faster and less expensive.

The Commission also has explained that it believes any harm to citizen-intervenors will be minimal. Although the petitioners offer some hypothetical examples of information that may be unavailable under the new rules, they have not made a persuasive case that mandatory disclosure will undercompensate for the loss of traditional discovery. We thus find no basis for setting aside the new rules on discovery-related grounds.

The new rules' outlook on cross-examination presents a closer question. The Commission reasons that restricting cross-examination will reduce the amount of testimony taken and make hearings more efficient. 69 Fed.Reg. at 2,196. The Commission further observes that, in its experience, cross-examination is not always helpful to the resolution of scientific or technical issues. Id.

The petitioners retort that cross-examination is a vital component of a citizen-intervenor's case. They note that citizen-intervenors often lack the resources to present their own expert testimony and must rely on cross- questioning of the adverse party's experts to make their case. They also stress the value of cross-examination as a means for bolstering public confidence in licensing hearings. Tellingly, the Commission's own administrative judges agree that cross-examination is helpful for the resolution of issues raised in many licensing hearings. In addition to the reasons advanced by the petitioners, the administrative judges note that the prospect of cross-examination discourages exaggeration in direct testimony because witnesses are aware that they will have to defend their statements later.

Experience in the courts has left no doubt that cross-examination can be a useful tool. Had the new rules abolished cross-examination entirely, we might well find the Commission's action insupportable. Importantly, however, the new rules do not completely do away with cross-examination. Rather, they leave its availability to the discretion of the hearing officer. Just as we will not ignore the fact that discovery is resource-consuming, we will not presume that all — or, perhaps, even most — cross-examination is essential to the just resolution of issues. With this in mind, we find no fault with the Commission's decision to attempt to curtail unnecessary cross-examination. Cf. 5 U.S.C. § 556(d) ("[T]he agency as a matter of policy shall provide for the exclusion of irrelevant, immaterial, or unduly repetitious evidence."). Accordingly, we cannot say that it is arbitrary and capricious for the Commission to leave the determination of whether cross-examination will further the truth-seeking process in a particular proceeding to the discretion of the individual hearing officer.

We do, however, add a caveat. The APA does require that cross-examination be available when "required for a full and true disclosure of the facts." Id. If the new procedures are to comply in practice with the APA, cross-examination must be allowed in appropriate instances. Should the agency's administration of the new rules contradict its present representations or otherwise flout this principle, nothing in this opinion will inoculate the rules against future challenges.

One petitioner, Citizens Awareness Network (CAN), charges that the new rules are unconstitutional because they deprive citizen-intervenors of fundamental political rights and discriminate against them in violation of the Fifth Amendment. We explore these charges.

CAN's first charge implodes because there is no fundamental right to participate in administrative adjudications. Reactor licensing (unlike, say, voting) is not "preservative of other basic civil and political rights." Reynolds v. Sims, 377 U.S. 533, 562, 84 S. Ct. 1362, 12 L. Ed. 2d 506 (1964). Assuming, for argument's sake, that citizen-intervenors have a protected liberty interest in the outcome of reactor licensing proceedings — a proposition that we consider extremely dubious, see City of W. Chicago, 701 F.2d at 645 — the quantum of process required before the government may deprive citizen-intervenors of that interest would depend on the three-part analysis adumbrated in Mathews v. Eldridge, 424 U.S. 319, 335, 96 S. Ct. 893, 47 L. Ed. 2d 18 (1976). CAN makes no effort to apply the Mathews rubric to the rules at issue, and we will not do CAN's homework for it. The issue is, therefore, forfeit. See Zannino, 895 F.2d at 17.

As for equal protection, CAN claims that we should apply strict scrutiny because citizen-intervenors are a "discrete and insular minority." This claim is meritless. The Supreme Court has made it crystal clear that the criteria for deeming a class suspect are both rigorous and specific. Johnson v. Robison, 415 U.S. 361, 375 n. 14, 94 S. Ct. 1160, 39 L. Ed. 2d 389 (1974); San Antonio Ind. Sch. Dist. v. Rodriguez, 411 U.S. 1, 28, 93 S. Ct. 1278, 36 L. Ed. 2d 16 (1973). As a class, citizen-intervenors cannot begin to meet those criteria. They are not "saddled with such disabilities, or subjected to such a history of purposeful unequal treatment, or relegated to such a position of political powerlessness as to command extraordinary protection from the majoritarian political process," Rodriguez, 411 U.S. at 28, 93 S. Ct. 1278, nor do they share any "immutable characteristic determined solely by the accident of birth," Johnson, 415 U.S. at 375 n. 14, 94 S. Ct. 1160.

Belaboring this point would serve no useful purpose. Whatever legitimate grievances citizen-intervenors may have, it is absurd to equate discrimination against them with the historic discrimination against racial and other minorities that lies at the core of suspectedness. The bottom line is that citizen-intervenors are not a suspect class. We so hold.

This holding means, of course, that rational basis review applies, not strict scrutiny. See Boivin v. Black, 225 F.3d 36, 42 (1st Cir. 2000). Thus, CAN bears the burden of demonstrating that no plausible set of facts exists that could forge a rational relationship between the challenged rules and the government's legitimate goals. Id. at 44; Montalvo-Huertas v. Rivera-Cruz, 885 F.2d 971, 978-79 (1st Cir. 1989). For the reasons set forth above, see supra Part III(B), there can be no doubt that the Commission's action is rationally related to a legitimate government purpose. It follows inexorably, as night follows day, that CAN's constitutional argument is meritless.

We need go no further. Procedural flexibility is one of the great hallmarks of the administrative process-and it is a feature that courts must be reluctant to curtail. Though the Commission's new rules may approach the outer bounds of what is permissible under the APA, we find the statute sufficiently broad to accommodate them. Similarly, the Commission's judgments as to when its procedures need fine-tuning and how they should be retooled are ones to which we accord great respect. We cannot say that the Commission's desire for more expeditious adjudications is unreasonable, nor can we say that the changes embodied in the new rules are an eccentric or a plainly inadequate means for achieving the Commission's goals. Accordingly, both of the instant petitions must be denied.

The petitions for judicial review are denied and dismissed.

LIPEZ, Circuit Judge.

Although I concur fully in Judge Selya's thoughtful and comprehensive opinion, I write separately to describe some oddities about this case which should not go unnoticed. The basic proposition of Judge Selya's decision is indisputably correct: the new rules promulgated by the Nuclear Regulatory Commission (NRC) to reduce the level of formality in reactor licensing proceedings comply with the "on-the-record" requirements of the Administrative Procedure Act (APA). Yet that legal proposition was largely an afterthought of the NRC in the effort to justify its new rules. Instead, the NRC principally argued in the long run-up to this case that 42 U.S.C. § 2239, which simply requires the Commission to hold a hearing "upon the request of any person whose interest may be affected" before granting a new license, did not invoke the requirements for formal adjudication (commonly referred to as "on-the-record" hearings) under the APA.

It is striking that so many smart people at the NRC could be so wrong for so long about the requirements of the APA. Although this history does not affect the outcome of this case, it should be noted as a cautionary tale about the power of analogy and the endurance of unexamined legal theories. This history also serves to explain some of the legitimate frustrations of the petitioners, who felt that they were dealing with a moving target as the NRC tried to justify its new regulations. With so much at stake in these nuclear reactor licensing proceedings, the rulemaking process should have followed a steadier course. For reasons I shall explain, this was not the rulemaking process at its best.

The terminology for hearings under the APA can be imprecise and confusing. The everyday meaning of terms like "formal" and "informal" sometimes creeps into the discussion, although those terms have specific, functional definitions under the APA. As Judge Selya notes, the terms "formal" and "on-the-record" are generally used as shorthand for hearings that must be conducted pursuant to the requirements of 5 U.S.C. §§ 554, 556, and 557 of the APA. Other terms, too, are sometimes used to refer to such procedures — "trial-type" and "quasi-judicial." These vague and indefinite terms are particularly mischievous because they evoke images of courtroom trials, and they have contributed to the false impression that the APA's requirement of on-the-record hearings involves procedures more akin to civil trials than is actually the case.

To be specific, § 554 requires that, in cases of an "adjudication required by statute to be determined on the record after opportunity for an agency hearing," the agency must follow the procedures outlined in §§ 556 and 557. Although the statutory text at issue here is itself rather pithy, these procedures can be usefully condensed into the following ten points:

1. The agency must give notice of legal authority and matters of fact and law asserted. § 554(b).

2. The oral evidentiary hearing must be presided over by an officer who can be disqualified for bias. § 556(b).

3. Presiding officers cannot have ex parte communications. §§ 554(d), 557(d) (1).

4. Parties are entitled to be represented by attorneys. § 555(b).

5. The proponent of an order has the burden of proof. § 556(d).

6. A party is entitled to present oral or documentary evidence. § 556(d).

7. A party is entitled "to conduct such cross-examination as may be required for a full and true disclosure of the facts." § 556(d).

8. Orders can be issued only on consideration of the record of the hearing. § 556(d).

9. The transcript of testimony and exhibits is the exclusive record for decision and shall be made available to parties. § 556(e).

10. The decision must include "findings and conclusions, and the reasons or basis therefor, on all the material issues of fact, law, or discretion presented on the record." § 557(c) (3) (A).

See Richard J. Pierce, Jr., Administrative Law Treatise § 8.1 (4th ed.2002). Strikingly, there is no reference to discovery in these statutory provisions of the APA, and cross-examination is assured only if necessary "for a full and true disclosure of the facts." 5 U.S.C. § 556(d). Most of these provisions relate to the conduct and responsibilities of the presiding officer or the basis for agency orders (on the record). Only a few relate to the conduct of the hearing itself. These APA requirements leave agencies with a great deal of flexibility in tailoring on-the-record hearing procedures to suit their perceived needs.

If hearings are not required to be "on the record," the procedures of §§ 556 and 557 are not triggered; the only section of the APA applicable to the proceedings is § 555, titled "Ancillary matters." Section 555(b) entitles a party to be represented by a lawyer, § 555(c) entitles people who have submitted data or evidence to retain copies of their submissions, and § 555(e) requires agencies to give prompt notice when they deny a petition made in connection with a proceeding, and to give a brief statement of the grounds for denial. Additionally, subsections (c) and (d) require that process, subpoenas, and other investigative demands must be made in accordance with law. Of course, these "informal" hearings must also comply with basic due process requirements.

From the beginning of its proposed rulemaking, the NRC repeatedly referred to the procedures outlined in the new regulations as "informal," as opposed to the outmoded formal procedures of the past. The clear implication was that the new informal procedures would not meet the APA's requirements for formal, on-the-record hearings. Thus, the NRC believed that it first had to establish that its authorizing statute, the Atomic Energy Act (AEA), did not require it to hold on-the-record hearings for reactor licensing.

Judge Selya outlines some of the important history of the rulemaking in this case. However, there is more to this curious history that is worth telling. From the 1998 Policy Statement to the Notice of Proposed Rulemaking, the NRC barely contemplated the possibility that it could reform its hearing procedures to its liking and still comply with the APA after all. Indeed, at first, the NRC did not propose any dramatic changes to its hearing procedures. In its August 5, 1998 Policy Statement, the NRC stated that it hoped to encourage a renewed vigor in the enforcement of already-existing hearing procedures by the Atomic Safety and Licensing Boards (ASLBs).⁶  To expedite hearings, the NRC advocated greater adherence to schedules, more rigorous enforcement of time limits for filing (for example, allowing extensions of time only in "unavoidable and extreme circumstances"), more rigorous enforcement of contention requirements,⁷  and tighter management of discovery. It also announced that the NRC "may consider further changes to the Rules of Practice as appropriate to enable additional improvements to the adjudicatory process."

True to its word, the NRC issued a Staff Requirements Memorandum to its Office of General Counsel (OGC) on July 22, 1999, directing it to develop a proposed rulemaking. At the same time, the Commission noted that it would also pursue a legislative solution by lobbying Congress to confirm its authority to reform licensing hearings as it wished. Again, the Staff Requirements Memorandum never suggested that the NRC could reduce the formality of its hearing procedures while staying within the strictures of the APA's requirements for on-the-record hearings. Instead, the NRC apparently still believed that the more informal licensing procedures it sought would not comply with the APA, and that it had to establish its freedom from the APA's strictest requirements.

The Staff Requirements Memorandum also directed the NRC's OGC to solicit the views of interested parties on the proposed rulemaking. Consequently, the general counsel held a two-day meeting in October 1999, called a "hearing process workshop," with representatives from the nuclear industry, citizens' groups (including the petitioners in this case), other federal agencies, academia, and the NRC's Atomic Safety Board and Licensing Panel. Although the OGC encouraged a wide-ranging conversation, no one raised the possibility that the NRC licensing procedures could be more informal yet still comply with the APA.

In the Notice of Proposed Rulemaking itself, 66 Fed.Reg. 19,610 (April 16, 2001), which for the first time proposed specific changes to the NRC's hearing procedures, there was still no argument that the proposed procedures complied with the APA's requirements. Instead, the notice offered pages of legal analysis on the history of the AEA, all intended to justify the NRC's ability to promulgate new hearing procedures that are not subject to the APA's requirements for on-the-record hearings.

Other choices made by the NRC in its regulatory overhaul further emphasize the firmness of its conviction that the supposedly "informal" procedures it was proposing did not comply with the APA's requirements for on-the-record hearings. In the one instance where no one disputes that the NRC must hold on-the-record hearings — the licensing of construction and operation of uranium enrichment facilities, see 42 U.S.C. § 2243(b) — the Notice described this process as "requiring formal trial-type hearing procedures to be used." 66 Fed.Reg. at 19,623. Consequently, the NRC believed that it could not use new subpart L for these hearings but had to resort to the more formal procedures of subpart G.⁸ 

Strikingly, in the entire record of this rulemaking prior to the promulgation of the Final Rule, I can find only one footnote hinting that anyone at the NRC thought that it could reduce the formality of its procedures while at the same time complying with the strictures of the APA. In a footnote in its January 1999 memo, the OGC acknowledged the possibility of eliminating the "elements of Subpart G that go beyond the Administrative Procedure Act's requirements for `on-the-record' hearings. One immediate effect would be to eliminate formal discovery in NRC adjudications." The memo contains no further discussion of how far beyond the APA's requirements the OGC understood Subpart G to go.

Not until publication of the Final Rule itself did the NRC assert for the first time that the new procedures comply with the APA's requirements for an on-the-record hearing-and even here, the NRC devotes only a few sentences to the issues of cross-examination, discovery, and the presiding officer.⁹  Indeed, the NRC's few statements are easily reproduced in their entirety. In response to concerns about the reforms to cross-examination, the NRC stated: "The Commission believes that this approach strikes an appropriate balance in the use of cross-examination, and is consistent with the requirements of the Administrative Procedure Act (APA), which does not require cross-examination for on-the-record proceedings unless necessary for a `fair and true disclosure of the facts.'" 69 Fed.Reg. 2182, 2188 (Jan. 14, 2004). A little later, the NRC states: "The Commission's consideration of cross-examination in the hearing process begins with the observation that parties have no fundamental right to cross-examination, even in the most formal hearing procedures provided in Subpart G." 69 Fed.Reg. at 2195-96.

As for concerns about the availability of discovery, the NRC stated: "Thus, the mandatory disclosure requirement in subpart C, the hearing file provision in subparts G, L, and N[,] and the requirement for an LSN and `electronic docket' in subpart J, go well beyond the `discovery' provisions for full, on-the-record adjudicatory hearings under the APA."¹⁰  69 Fed.Reg. at 2189. As for presiding officers, the NRC described how the new regulations provided for either an ALJ or a three-member ASLB to preside over the hearing, and then stated: "The Commission has taken this step to ensure that all of these proceedings meet the requirements with regard to a presiding officer for an on-the-record hearing under the APA." 69 Fed.Reg. at 2191. However, despite these few references in the Final Rule to meeting the APA's requirements for on-the-record hearings, the NRC's primary rationale for its new procedures remained its long-standing position that reactor licensing hearings did not have to comply with the on-the-record requirements of the APA.¹¹ 

In considerable part, administrative agencies set the terms of the debate in the rulemaking process with the arguments they advance in support of their rulemaking initiatives. If certain arguments are unmistakably primary, those arguments will draw most of the attention during the administrative process, and during the judicial review that follows. Not surprisingly, opponents will believe that the primacy of the argument means that it is the most important argument to address. If, in the end, the dispositive issue on appeal is a different issue, addressed only glancingly in the administrative process, there has been enormous wasted effort, and the courts of appeal will be poorly served by appellate briefing that reflects the outdated emphases of the administrative process. That is precisely what happened here.

In their petition for review, petitioners never even mentioned the argument of the NRC that its new procedures complied with the APA's requirements for on-the-record hearings. Not one of the petitioners addressed the argument in their opening briefs.¹²  Indeed, the NRC tried to exploit this omission by arguing that petitioners had thereby waived the issue. In response, Public Citizen stated that the NRC argued from the beginning of the rulemaking that its freedom from the APA's on-the-record requirements was essential to its ability to fashion informal procedures, and "[t]here would have been no need for page after page of argument that Section 189 does not require on-the-record hearing procedures in reactor licensing cases if the NRC had believed that the new Subpart L procedures conformed to these requirements."¹³  Public Citizen added that the few references to APA- compliance in the Final Rule were nothing more than "stray remarks" when weighed against the overwhelming number of statements to the contrary made by the NRC throughout the rulemaking. Public Citizen then tried to turn this lament into a legal argument based upon SEC v. Chenery Corp., 318 U.S. 80, 63 S. Ct. 454, 87 L. Ed. 626 (1943). Although the lament is a fair one, the legal argument does not work.

In Chenery, the Supreme Court warned courts that they must not substitute their own policy judgments for those of the administrative agencies whose decisions they review. The Court explained that this rule did not disturb the settled rule that appellate courts can affirm trial court decisions that are right for the wrong reasons:

In confining our review to a judgment upon the validity of the grounds upon which the Commission itself based its action, we do not disturb the settled rule that, in reviewing the decision of a lower court, it must be affirmed if the result is correct although the lower court relied upon a wrong ground or gave a wrong reason. The reason for this rule is obvious. It would be wasteful to send a case back to a lower court to reinstate a decision which it had already made but which the appellate court concluded should properly be based on another ground within the power of the appellate court to formulate. But it is also familiar appellate procedure that where the correctness of the lower court's decision depends upon a determination of fact which only a jury could make but which has not been made, the appellate court cannot take the place of the jury. Like considerations govern review of administrative orders. If an order is valid only as a determination of policy or judgment which the agency alone is authorized to make and which it has not made, a judicial judgment cannot be made to do service for an administrative judgment. For purposes of affirming no less than reversing its orders, an appellate court cannot intrude upon the domain which Congress has exclusively entrusted to an administrative agency.

SEC v. Chenery Corp., 318 U.S. 80, 88, 63 S. Ct. 454, 87 L. Ed. 626 (1943) (citation and internal quotation marks omitted) (Chenery I).¹⁴ 

The dispute in Chenery came before the Court again in 1947, after the SEC had "reexamined the problem, recast its rationale and reached the same result." SEC v. Chenery Corp., 332 U.S. 194, 196, 67 S. Ct. 1575, 91 L. Ed. 1995 (1947) (Chenery II). Accepting the SEC's new decision as justified by an administrative determination that "is based upon substantial evidence and is consistent with the authority granted by Congress," id. at 207, 67 S. Ct. 1575, the Court noted the "important corollary" to Chenery's rule of judicial review of agency decisions:

If the administrative action is to be tested by the basis upon which it purports to rest, that basis must be set forth with such clarity as to be understandable. It will not do for a court to be compelled to guess at the theory underlying the agency's action; nor can a court be expected to chisel that which must be precise from what the agency has left vague and indecisive. In other words, `We must know what a decision means before the duty becomes ours to say whether it is right or wrong.'

Id. at 196, 67 S. Ct. 1575 (quoting United States v. Chicago, M., St. P. & P. R.R. Co., 294 U.S. 499, 511, 55 S. Ct. 462, 79 L. Ed. 1023 (1935)). The Chenery decisions have continuing vitality today. See, e.g., NLRB v. K. River Cmty. Care, Inc., 532 U.S. 706, 721, 121 S. Ct. 1861, 149 L. Ed. 2d 939 (2001); Nat'l Tower, LLC v. Plainville Zoning Bd. of Appeals, 297 F.3d 14, 21 (1st Cir. 2002); Rizek v. SEC, 215 F.3d 157, 161 (1st Cir. 2000) (citing the "well-established rule that agencies must sufficiently articulate the grounds of their decisions so that appellate courts are able to perform their function of judicial review meaningfully.")

Despite the NRC's disproportionate attention to its original premise that it could alter its procedures for the licensing of nuclear reactors free of the on-the-record requirements of the APA, and its scant attention to the APA-compliance argument that prevails here (found primarily in several sentences placed in the Final Rule), I agree with Judge Selya that the NRC preserved the rationale that has become the basis for affirming its rulemaking initiative. However, if the Chenery cases have a spirit, the NRC came perilously close to violating it here, with the unfortunate consequences for efficient administrative process and effective appellate review that I have already described. Although, to quote Chenery II, we did not have "to guess at the theory underlying the agency's action," 332 U.S. at 197, 67 S. Ct. 1575, we had to find it with too little help from the parties because of the NRC's failure to understand the APA's flexibility.

The NRC's belated recognition that the new licensing procedures might in fact comply with the on-the-record requirements of the APA is all the more surprising because sources contemporaneous with the APA's passage suggest that flexibility has always been a hallmark of the APA, and that agencies have always had considerable discretion to structure on-the-record hearings to suit their particular needs. This flexibility is nowhere more evident than in determining the role of cross-examination in on-the-record hearings.

The Attorney General's Manual on the Administrative Procedure Act (1947) is a "key document" for interpreting the APA, Am. Min. Cong. v. Mine Safety & Health Admin., 995 F.2d 1106, 1109 (D.C. Cir. 1993). We have described it as containing the "most authoritative" account of the history of the Act's passage, Warder v. Shalala, 149 F.3d 73, 79 (1st Cir. 1998). See also V. Yankee Nuclear Power Corp. v. Natural Res. Def. Council, Inc., 435 U.S. 519, 546, 98 S. Ct. 1197, 55 L. Ed. 2d 460 (1978) (the Attorney General's Manual is "a contemporaneous interpretation previously given some deference by this Court because of the role played by the Department of Justice in drafting the legislation"). The Attorney General's Manual offers a vision of cross-examination entirely consistent with that advanced by the NRC in this rulemaking.

The Manual begins by stressing the general importance of cross-examination in on-the-record hearings, cautioning that "it is clear that the `right to present his case or defense by oral or documentary evidence' does not extend to presenting evidence in affidavit or other written form so as to deprive the agency or opposing parties of opportunity for cross-examination." AG's Manual at 77. Technical evidence may be introduced in written form as long as its admission "would not prejudicially deprive other parties or the agency of opportunity for cross-examination. Thus, technical and statistical data may be introduced in convenient written form subject to adequate opportunity for cross-examination and rebuttal." Id.

The Attorney General's Manual goes on, however, to acknowledge that the general opportunity to cross-examine is subject to restrictions which become more salient as the complexity of the hearing's subject matter increases. On this point, the Manual quotes from the Report of the House Committee on the Judiciary on the APA. The Report cautions that the APA's provision for "such cross-examination as may be required for a full and true disclosure of the facts" does not

confer a right of so-called `unlimited' cross-examination. Presiding officers will have to make the necessary initial determination whether the cross-examination is pressed to unreasonable lengths by a party or whether it is required for the `full and true disclosure of the facts' stated in the provision. Nor is it the intention to eliminate the authority of agencies to confer sound discretion upon presiding officers in the matter of its extent. The test is — as the section states — whether it is required `for a full and true disclosure of the facts.' In many rule making proceedings where the subject matter and evidence are broadly economic or statistical in character and the parties or witnesses numerous, the direct or rebuttal evidence may be of such a nature that cross-examination adds nothing substantial to the record and unnecessarily prolongs the hearings.

H.R.Rep. No.1980, 79th Cong., 2d Sess., 37.

The Attorney General's Manual and the House Report serve as good indicators that Congress, when it passed the APA, understood that agencies needed a considerable amount of flexibility in fashioning hearing procedures for on-the-record hearings. Despite the frequent use of terms like "trial-type" and "quasi-judicial" over the years to refer to on-the-record hearings, agencies have always been able to adapt their procedures for on-the-record hearings under the APA. Today, this statute of general applicability governs the procedures for an enormous variety of hearings — everything from relatively simple claims for workers' compensation, to enforcement proceedings under the National Labor Relations Act or the Occupational Health and Safety Act, to complex rate-setting hearings before the Federal Energy Regulatory Commission. See Manual for Administrative Law Judges (ALJ Manual), 49-51.

This historical flexibility is confirmed by a modern-day guide to the conduct of on-the-record hearings, the Manual for Administrative Law Judges, which provides a thorough overview of the current state of on-the-record procedures.¹⁵  Although the Manual is primarily designed as a practical aid for ALJs, it also offers an analysis of the fundamental requirements of the APA, including a section on the special problems presented by complex hearings at both the agency and ALJ levels. See ALJ Manual at 49-70. Ultimately, if the ALJ Manual tells us anything about what a typical on-the-record hearing looks like, it is that there is no typical hearing.

In its section on complex hearings, the ALJ Manual begins by noting that the term "quasi-judicial" is most often used to refer to "relatively simple cases." Id. at 50. "Typically, these quasi-judicial proceedings are nearly identical to a formal adjudication without a jury," with the full panoply of pleadings, pre-hearing discovery, and witnesses who testify orally on direct and cross-examination. Id. at 49. Complex cases, however, are "another matter," bearing less resemblance to our traditional concept of a civil trial. Id. at 51. The ALJ Manual cites the NRC's use of ASLBs as an "innovative approach to complex cases" with highly technical subject matter. Id. at 52. In general, the ALJ Manual emphasizes the use of written submissions of direct and rebuttal evidence: "Typically, much of the testimony is highly technical and lengthy, and is submitted in written form prior to the hearing." Id. at 51. "Preparation and exchange of direct and rebuttal evidence is usually beneficial in complex cases." Id. at 56.¹⁶ 

Like the Attorney General's Manual, the ALJ Manual emphasizes the basic importance of the opportunity to cross-examine in on-the-record hearings, noting that "judges should be extremely cautious about denying parties an opportunity to cross-examine witnesses." Id. at 55 n. 149. In the end, however, and again like the Attorney General's Manual, the ALJ Manual also acknowledges the agencies' need to adapt hearing procedures to suit especially complex cases, observing that "[u]nless witness credibility is involved, cross-examination is frequently confined to clarifying the exhibits, determining the source of the material, and testing the basis for the witness's conclusions." Id. at 83. In fact, the ALJ Manual even notes a proposal by one legal commentator that "the major rebuttal of expert opinion testimony should take place not by cross-examination but by submission, prior to the hearing, of rebuttal testimony prepared by the opponent's experts." Id.

These sources, both contemporaneous with the APA's passage and modern, show that procedures in on-the-record hearings, despite sometimes being described as "trial-type" procedures, can in fact stray considerably from the procedures found in civil trials as the subjects of the administrative proceedings become more complex and more technical. This flexibility is inherent in the APA, and has been acknowledged by commentators and by courts. See Seacoast Anti-Pollution League v. Costle, 572 F.2d 872, 880 (1st Cir. 1978) (finding no basis to petitioners' argument that the APA required presiding officer to afford opportunity for cross-examination); Cellular Mobile Systems of Pa., Inc. v. FCC, 782 F.2d 182, 198 (D.C. Cir. 1985) ("Cross-examination is therefore not an automatic right conferred by the APA; instead, its necessity must be established under specific circumstances by the party seeking it.").¹⁷  With these abundant sources pointing the way, the NRC's belated recognition that the APA could, in fact, accommodate its procedural reforms is all the more puzzling.

For most of the history of this rulemaking, the NRC argued that it did not have to comply with the APA's on-the-record requirements in refashioning its procedures for reactor licensing hearings. Belatedly, and then only sketchily, the NRC advanced the alternative argument that its proposed procedures complied with those on-the-record requirements. The staying power of old theories and flawed analogies (the repeated references to trial-type proceedings) may account for some of this delay. Whatever the reasons, the deleterious effect of this late insight on the rulemaking process and our review of it is undeniable. Countless hours were wasted during the administrative process fighting over the tired issue of whether 42 U.S.C. § 2239 requires reactor licensing hearings to be on the record. This tired issue dogged judicial review as well. Although we have done what Chenery requires — affirming on a basis advanced by the agency itself during the administrative process — we got there with too little help from the parties. There is a victory here for the NRC, but it should be a cause for self-examination rather than jubilation.