concurring:
This case is a classic example of the dangers that can flow from an indiscriminate attempt to read legislative meaning into congressional tea leaves. I will therefore begin by highlighting the lessons to be drawn from the case before addressing an aspect of the court’s opinion with which I cannot agree.
When one undertakes to use legislative history as a tool of statutory construction, surely the first part of wisdom is to remember that Congress is a political as well as a legislative body, and that its members will put the privileges and facilities of their respective chambers to political as well as legislative uses. Thus not every utterance to be found in committee reports or the Congressional Record may be assumed to represent statutory gold.
In the case before us, a modest degree of caution would have spared the Board the errors identified in the court's opinion, as the political story that emerges from the record is clear enough. In 1973, key members of the Senate and House labor committees decided the time had come to eliminate the exemption of nonprofit health care institutions from coverage by the National Labor Relations Act. At the same time, they recognized that success required reconciling the strong differences existing between the two constituencies most directly involved; namely, the unions’ goal of an unrestricted right to organize health care employees, and hospital administrators’ concern over their continued ability to deliver health care services uninterrupted by labor disputes.
In the Senate, two bills were introduced and referred to the Committee on Labor *187and Public Welfare. The first, which was authored by Senators Cranston and Javits, would have brought nonprofit health care employees under NLRA coverage without the addition of any industry-specific restrictions. The second, introduced by Senator Taft, incorporated restraints on the right to strike, a ceiling on bargaining units, and other provisions sought by hospital officials.
Ultimately, in order to break a deadlock, the Senate and House committees worked out a compromise with representatives of labor and management that would meet the former’s desire to bring nonprofit health care employees under the NLRA with a minimum of restrictions while, at the same time, addressing the latter’s concerns over the potential disruption of hospital care. As part of the arrangement, the Cranston-Javits and Taft bills were withdrawn; Senator Taft introduced a substitute measure, S. 3203, that preserved his restrictions on strikes but dropped his cap on bargaining units; and the Senate and House committees agreed to include the admonitory language in their reports. As a result, the legislative baton passed to S. 3203, and the bargaining units issue moved from the legislative to the political agenda.
This fact should have been self-evident from Senator Dominick’s clear-cut statement as to why he had voted against reporting S. 3203 out of committee:
I am also concerned about the bill’s failure to treat bargaining units____ The committee has recognized the need for special treatment of the hospital industry. A proliferation of bargaining units, however, can pose a serious threat to uninterrupted health-care.
... A hospital should be protected by statute from being placed in a position of continual bargaining with different units in order to minimize the threats to patient care.
Individual views of Senator Peter H. Dominick, S.Rep. No. 766, 93d Cong., 2d Sess. 44-45 (1974), U.S.Code Cong. & Admin. News 1974, p. 3958, reprinted in Legislative History at 51-52. Senator Dominick was a careful legislator. As a member of the Senate Committee on Labor and Public Welfare, he was fully aware of the report language. Nevertheless he declined to support the compromise because it failed to provide the legal safeguards he considered necessary.
If there were still room for doubt, it should have been put to rest by the fact that when Senator Cranston described the key elements of the legislation in his capacity as floor manager for the 1974 Amendments, he made no mention of the committee admonition. See 120 Cong.Rec. 12,936-37 (1974). Nor did Senators Williams and Javits when, in their respective capacities as chairman and ranking member of the committee, they presented their own analyses of S. 3203. Id. at 12,937-40. The reason is obvious. The legislation before the Senate did not amend section 9 of the Act or otherwise address the question of bargaining units; therefore, the admonition was not relevant to its analysis.
This does not mean that the agreement to include the report language was an empty gesture. To the contrary, there can be little doubt that the two committees expected the Board to pay attention to their directive — not because it had the force of law, but because agencies are not given to ignoring the commands of potentates who control their budgets and oversee their operations. As counsel for one agency recently acknowledged in oral argument before this court, while an instruction in an oversight committee’s report did not bind his agency legally, it did so “as a practical matter.” To underscore his point, he added: “[W]e are not talking law school enforcement, legal textbook arguments; we’ré talking political reality here.” Meredith Corp. v. FCC, 809 F.2d 863 (D.C.Cir. 1987).
This political reality is well understood by professionals. See, e.g., J. Mashaw & R. Merrill, Administrative Law: The American Public Law System 74-83 (1985) (identifying “pervasive methods of legislative control” over agencies in general); Gross, Conflicting Statutory Purposes: Another Look at Fifty Years of NLRB Law Making, 39 Indus. & Lab.Rel.Rev. 7, 8-10 *188(1985) (detailing political control over the NLRB in particular). Thus the inclusion of the admonition in the committee reports could be expected to provide the margin of assurance required to secure the support of at least some of the legislators and lobbyists who were concerned over the prospect of proliferating bargaining units.
As the admonitory language served a political rather than a legal purpose, each of the key legislators no doubt felt free to “accentuate the positive” by providing his own distinct interpretation on its meaning and effect. This may explain the contrasting glosses they placed on the language, as noted in section II.A. of the court’s opinion. Senator Taft, for example, categorized the admonition as stressing “the necessity [for] the Board to reduce and limit the number of bargaining units in a health care institution,” 120 Cong.Rec. 12,944 (1974) (statement of Sen. Taft) (emphasis added), while Senator Williams asserted that “the committee clearly intends that the Board give due consideration to its admonition to avoid an undue proliferation of units____” 120 Cong.Rec. 22,575 (1974) (statement of Sen. Williams) (emphasis added). It is unlikely that either senator was upset by the other’s interpretation because no doubt each understood the other was engaged in reassuring his own constituencies in order to consolidate support for their common enterprise.
I do not claim my scenario is accurate, merely that it is plausible. As such, it suggests an endemic interplay, in Congress, of political and legislative considerations that makes it necessary for judges to exercise extreme caution before concluding that a statement made in floor debate, or at a hearing, or printed in a committee document may be taken as statutory gospel. Otherwise, they run the risk of reading authentic insight into remarks intended to serve quite different purposes. Furthermore, to the degree that judges are perceived as grasping at any fragment of legislative history for insights into congressional intent, to that degree will legislators be encouraged to salt the legislative record with unilateral interpretations of statutory provisions they were unable to persuade their colleagues to accept; a process made especially easy in the Senate, whose members are allowed to introduce into the printed record of a debate remarks that were never spoken on the floor, hence never subjected to the hazard of rebuttal.
An additional reason for the exercise of judicial caution is that judges confined to the printed page may not always be able to separate legislative wheat from political chaff, to identify and discard remarks that are designed not so much to inform colleagues as to reach beyond Capitol Hill to audiences ranging from lobbyists who may help a beleaguered legislator achieve his goals, to home state journalists on whose reports the next election may well depend. It is not always easy to distinguish which remarks will fit in which category, especially when read years after the event by judges who have neither the time to review the entire history of a particular bill nor the experience to filter out the political overtones.
The hazards of reconstructing legislative intent are unfortunately real; witness the remarkable fact that all ten circuits hitherto construing post-1974 nonprofit health care unit certifications have accorded legal weight to the committee admonition. See maj. op. at 704. Two circuits, the Ninth and Tenth, have even gone so far as to order the Board to adopt the disparity-of-interest test, thereby effectuating by judicial fiat what Senator Taft concluded he could not accomplish by statute. The balance of the circuits ruling on this issue — the Second, Third, Fourth, Fifth, Sixth, Seventh, Eighth, and Eleventh — fare better, but only in that they do not themselves dictate the precise weight to be accorded the admonition. See, e.g., maj. op. at 703-705. Indeed, to date only Judge Fairchild of the Seventh Circuit has stated, in a brief dissent, what we take to be the essential point: as the admonition and the remarks addressed to it were divorced from the legislation then before Congress, they are an illegitimate source of authority for the agency’s construction of the law. See maj. op. at 713.
*189Whatever may have persuaded our sister circuits to mistake legislative politics for law, we find the record unambiguous. Senator Taft sought to limit bargaining units in nonprofit health care institutions, but withdrew the proposal as part of a larger compromise. Given this undisputed legislative background, the legal conclusion we reach is inescapable. Agencies receive their power to impose decisions on private parties from statutory law, and not from the advisory instructions of particular committees. Had the court’s opinion stopped with the conclusion that the published admonition was a legal nullity, I would have been in full agreement. Unfortunately, the court abandons the logic of its own position by then according legal significance to a single committee’s failure to act.
First, the court states that because a Senate committee did not modify section 9, “Congress implicitly approved the Board’s forty-year construction of section 9 to embody community-of-interest criteria.” Maj. op. at 711 (emphasis added) (footnote omitted). See also maj. op. at 711 (“Congress ... did consider modifying ... [section 9]”), 711 (“Congress, however, decided against the modification of section 9 proposed by Senator Taft.”), id (“Congress’ rejection of the Taft proposal____”). Moreover, while declining to say whether the Board may in its discretion “switch to another unit standard,” id, the opinion observes, in ominous tones, that “the Board would have to explain its action adequately, particularly because the Board has always construed section 9 to embody community-of-interest criteria.” Id. at n. 65 (citations omitted) (emphasis in original). At best, in light of the manifest reality that Congress never contemplated a proposed modification to section 9, the court’s opinion engages in what is in effect an advisory opinion.
The notion of implicit approval is not so easily cabined as dicta, however, for it holds to the idea, rejected by the very rationale of the case, that authoritative congressional intent may be deduced from a committee’s failure to adopt a proposal. Formally, the question is the proper weight to be accorded legislative proposals that do not reach Congress. The answer should be self-evident.
A committee’s purposes cannot be imputed to Congress as a whole, and references to committee action (or inaction) on the Senate floor will not change this fact. Senator Taft’s original bill was never presented for debate, and neither the Senator nor anyone else offered floor amendments that would have given Congress the opportunity to accept or reject his proposed limitation on unit representation. Therefore, it cannot be said that Congress — which acts through a majority of its 535 members — expressed any opinion whatever on the merits of his proposal, let alone endorsed a forty-year-old interpretation of which most members were in all probability wholly unaware.
Certainly, one does not hear the court arguing that because the committee failed to adopt the Cranston-Javits bill, Congress implicitly disapproved the application of the community-of-interest test to nonprofit health care workers. To travel this path is to end up at the bizarre hypothesis, actually advanced by one commentator, that the withdrawal of both bills should be taken to mean that Congress intended to end up somewhere in the middle between a numerical cap and the traditional test. See Bum-pass, Appropriate Bargaining Units in Health Care Institutions: An Analysis of Congressional Intent and its Implementation by the National Labor Relations Board, 20 B.C.L.Rev. 867, 884-86 (1979). Rather, it seems to me this is a classic case in which to acknowledge that “[ijnaction by Congress is due generally to reasons too diffuse, ephemeral, and ultimately unaccountable ...” to provide a basis for statutory construction. Advanced Micro Devices v. CAB, 742 F.2d 1520, 1541 (D.C.Cir. 1984).
Wong Yang Sung v. McGrath, 339 U.S. 33, 70 S.Ct. 445, 94 L.Ed. 616 modified, 339 U.S. 908, 70 S.Ct. 564, 94 L.Ed. 1336 (1950), is also on point. The issue there was whether the newly passed Administrative Procedure Act applied to deportation hearings. The Immigration Department, after an adverse decision in district court, sought *190exempting legislation from Congress. Committees in both houses reported out favorable bills, but Congress adjourned and no further action was taken. The government and Sung each sought to read significance into this state of affairs. But the Supreme Court, in words wholly applicable here, declared the committee actions of no consequence whatsoever: “We draw, therefore, no inference in favor of either construction of the Act — from the Department’s request for legislative clarification, from the congressional committee’s willingness to consider it, or from Congress’ failure to enact it.” Id. at 47-48, 70 S.Ct. at 453.
The court makes an equally unwarranted attempt to entrench the community-of-interest standard by “placpng] significance on the Board’s contemporaneous construction of the 1974 Amendments.” Maj. op. at 714. Although the issue is not before us, the suggestion here is that our review of the Board’s 1984 St. Francis II decision must defer to the policy applied by the agency to bargaining unit questions in the period immediately following the enactment of the Health Care Amendments, as long as that policy does not clearly conflict with the intent of Congress. This approach contradicts the obvious necessity and indeed the very logic for deference to agency decisionmaking, which is to place policy choices (within the limits permitted by statute) in the hands of delegated agents of Congress, subject to their expertise and experience. See, e.g., Beth Israel Hospital v. NLRB, 437 U.S. 483, 508, 98 S.Ct. 2463, 2477, 57 L.Ed.2d 370 (1978) (“The authority of the Board to modify its construction of the Act in light of its cumulative experience is, of course, clear.” (citations omitted)).
Although it is true that contemporaneous agency decisionmaking may sometimes be accorded special deference, I believe the panel misapplies this rule in the present case. The rationale for special deference to a contemporaneous construction is to aid courts in construing ambiguous statutory language and does not operate when the NLRB, in light of eleven years’ experience applying the Act to nonprofit health care workers, overturns a prior construction of its organic statute in favor of a new permissible construction. See, e.g., NLRB v. J. Weingarten, Inc., 420 U.S. 251, 266, 95 S.Ct. 959, 968, 43 L.Ed.2d 171 (1975) (striking down court of appeals for “impermissibly encroachpng] upon the Board’s function” “to adapt the Act” “in light of changing industrial practice and the Board’s cumulative experience”); Columbia Broadcasting System, Inc. v. FCC, 454 F.2d 1018, 1026 (D.C.Cir.1971) (“We do not challenge [an agency’s] well established right to modify or even overrule an established precedent or approach, for an administrative agency concerned with furtherance of the public interest is not bound to rigid adherence of its prior rulings.” (footnote omitted)).
Rather, the appropriate legal standard is rooted in the deference due permissible NLRB constructions of its broadly delegated authority. The well-known, settled requirement is that the agency departure from precedent be based on reasoned decisionmaking. Id. (“[W]hen an agency decides to reverse its course, it must provide an opinion or analysis indicating that the standard is being changed and not ignored, and assuring that it is faithful and not indifferent to the rule of law.” (footnote omitted)); Oil, Chemical & Atomic Workers Int’l v. NLRB, 806 F.2d 269, 273-74 (D.C.Cir.1986). Had the Board, in St. Francis II, given a reasoned explanation for its shift in policy and not placed a legally untenable reliance on the committees’ admonition, we would have been required to determine whether its most recent construction met the standards outlined in Weingarten, 420 U.S. at 267, 95 S.Ct. at 968 (“fair and reasoned balance upon a question within its special competence, [the] newly arrived at construction of § [9] does not exceed the reach of that section, and the Board has adequately explicated the basis of its interpretation”). If we determined that the Board had met those standards, we would have been obliged to defer to it whatever our own assessment of the correctness of the Board’s original construction.
*191If it were to hold otherwise, an appellate court would be substituting its view of the statute, under the rubric of deference to an abandoned agency view, for the agency’s own updated judgment. Such an outcome would be contrary to the Court’s express direction that the Board’s reconciliation of conflicting interests of labor and management be “subject to limited judicial review.” Id. at 267, 95 S.Ct. at 968 (quoting NLRB v. Truck Drivers Local 449, Int’l Bhd. of Teamsters, 353 U.S. 87, 96, 77 S.Ct. 643, 648, 1 L.Ed.2d 676 (1957). This is true when the Board reverses ground, and even when its newest view is not required by the Act. It need only be a permissible construction. Id. at 266-67, 95 S.Ct. at 968-67. See also NLRB v. Local 103, Int’l Ass’n of Bridge, Structural & Ornamental Iron Workers, 434 U.S. 335, 351, 98 S.Ct. 651, 660, 54 L.Ed.2d 586 (1978) (“An administrative agency is not disqualified from changing its mind; and when it does, the courts still sit in review of the administrative decision and should not approach the statutory construction issue de novo and without regard to the administrative understanding of the statute[ ].”); Chevron U.S.A. Inc. v. Natural Resources Defense Council, Inc., 467 U.S. 837, 843 n. 11, 104 S.Ct. 2778, 2782 n. 11, 81 L.Ed.2d 694 (1984) (“The court need not conclude that the agency construction was the only one it permissibly could have adopted to uphold the construction, or even the reading the court would have reached if the question initially had arisen in a judicial proceeding.” (citations omitted)).
Against this clear authority, the doctrine of contemporaneous construction cannot be invoked here to trump or undermine the NLRB’s freedom to review its past policy in determining how section 9 should be applied to nonprofit health care workers in the future.
The court's emphasis on the settled weight of the community-of-interest standard seems particularly inappropriate given the Board’s obligation to construe section 9 in light of the whole statute, including the Health Care Amendments. Congress charged the Board, not circuit judges, to make “appropriate” unit determinations for employees in a complex, hitherto uncovered industry, given to “unique considerations that do not apply in the industrial settings with which the Board is more familiar.” Beth Israel, 437 U.S. at 508, 98 S.Ct. at 2477 (quoting NLRB v. Beth Israel Hospital, 554 F.2d 477, 481 (1st Cir.1977)). Our role is to stand and wait.
In remanding, we do not say how or whether experience with the nonprofit hospital setting might influence the standard adopted by the Board. We do say, and it is quite enough, that the Board may not attach legal significance to the committees’ admonition.