concurring in part and dissenting in part:
I am in full agreement with the majority’s treatment of Winter Park's § 307(b) claim and of the issues addressed in part II.C. of the opinion. I dissent from the majority’s constitutional analysis. The critical factor inducing the Commission to award the license to Rainbow was that its 90 percent credit for Hispanic ownership outweighed Metro’s 19.8 percent credit for black ownership. Although in West Michigan Broadcasting Co. v. FCC, 735 F.2d 601 (D.C.Cir.1984), our circuit found it constitutional for the Commission to make race pivotal in the allocation of a radio channel, I believe that case has been largely undermined by the Supreme Court’s more recent decisions in Wygant v. Jackson Board of Education, 476 U.S. 267, 106 S.Ct. 1842, 90 L.Ed.2d 260 (1986), and City of Richmond v. J.A. Croson Co., — U.S. -, 109 S.Ct. 706, 102 L.Ed.2d 854 (1989).
The Commission and intervenors on its side assert two possible constitutional justifications for the FCC’s minority preference policy. First, the FCC claims that its policy fosters diversity in broadcast programming, a “non-remedial” purpose. See FCC Brief at 29-31. I do not believe the justification can survive the Croson Court’s analysis, particularly its indictment of racial stereotyping. Second, Rainbow claims that the FCC’s policy is also designed as a remedy for the effects of “historical under-representation of minorities in broadcasting,” independently of any link to broadcasting diversity. See Rainbow Brief at 22 and generally at 21-26.1 As the Commission itself never articulated this remedial theory, the theory cannot sustain the decision here unless Congress mandated Commission use of racial preferences even in the event of the collapse of the Commission’s own justification. See Continuing Resolution of December 22, 1987, Pub.L. No. 100-202, 101 Stat. 1329; Departments of Commerce, Justice and State, the Judiciary, and Related Agencies Appropriations Act, 1989, Pub.L. No. 100-459, 102 Stat. 2186 (1988). Particularly in light of the doubts created by Wygant and Croson as to whether this alternative theory could shore up the constitutionality of the Commission’s policy, I believe the case should be remanded to the Commission for it to interpret Congress’s directions.
I. The Program Diversity Purpose
The word “diversity” comprises a multitude of goals and methods. The Commission has adopted, and the Supreme Court approved, “cross-ownership” rules intended by the Commission to deconcentrate ownership and thereby to foster diversity of viewpoints. FCC v. National Citizens Committee for Broadcasting, 436 U.S. 775, 98 S.Ct. 2096, 56 L.Ed.2d 697 (1978). But despite the Commission’s using the word “diversity” to label the goal of its racial preferences, the approaches have al*144most nothing in common. Deconcentrating ownership bears no resemblance to allocating licenses on the basis of the applicants’ race, for the purpose of generating programs “targeted” at specific ethnic groups.
The FCC notes in its brief that “the purpose of the [minority preference] policy has been clearly explained by the Commission and Congress as being to enhance program diversity by increasing ownership of stations by significant groups that are substantially underrepresented in station ownership.” FCC Brief at 47 n. 47. If the Commission’s purpose were ethnically proportional representation of programmers for its own sake, its action would be clearly unconstitutional in the absence of a remedial justification. As Justice Powell said in Bakke, “[Preferring members of any one group for no reason other than race or ethnic origin is discrimination for its own sake. This the Constitution forbids.” Regents of the University of California v. Bakke, 438 U.S. 265, 307, 98 S.Ct. 2733, 2757, 57 L.Ed.2d 750 (1978); see also Loving v. Virginia, 388 U.S. 1, 11, 87 S.Ct. 1817, 1823, 18 L.Ed.2d 1010 (1967) (“this Court has consistently repudiated ‘[distinctions between citizens solely because of their ancestry’ as being ‘odious to a free people whose institutions are founded upon the doctrine of equality.’ ”) (quoting Hirabayashi v. United States, 320 U.S. 81, 100, 63 S.Ct. 1375, 1385, 87 L.Ed. 1774 (1943)).
At several points the Commission’s arguments suggest that the stated program diversity goal has in fact dissolved into a simple quest for racial or ethnic proportionality in license ownership. The Commission has never developed any data suggesting that the listening or viewing tastes of any ethnic group are either distinctive or undersupplied. Indeed, responding to the argument that its present scheme is of unlimited duration (because there is no measureable goal whose attainment would call a halt), its brief simply referred the reader to data on minority underrepresentation in license ownership. See FCC Brief at 47 n. 47, referring to id. at 30 n. 33. Nonetheless, I take the Commission at its word — or better — and examine the program-diversity rationale on the premise that it is genuine.
Under the view adopted by a majority of the Supreme Court for the first time in Croson, race-based decision criteria must withstand “strict scrutiny” to survive an equal protection attack. See Croson, — U.S. at-, 109 S.Ct. at 721 (Part III-A)2; id. at-, 109 S.Ct. at 735 (Scalia, J., concurring); see also id. at-, 109 S.Ct. at 752 (Marshall, J., dissenting) (noting it as novel aspect of majority decision). More specifically, the government’s goal must be a “compelling” one and the means selected must be “narrowly tailored” to fulfillment of that goal.
In asserting the compelling character of the goal of program diversity, the Commission seeks to ride on Justice Powell’s endorsement of a “diversity” rationale in the context of educational admissions policy. See Bakke, 438 U.S. at 311-12, 98 S.Ct. at 2759-60. But the diversity purportedly sought by the Commission is altogether different from what Justice Powell would have sanctioned. Ethnic diversity in the classroom enables those present to see individual members of ethnic groups as they are. Far from depending on some link between race and conduct, it is a potent device against ethnic stereotyping. In his dissent in Wygant v. Jackson Board of Education, 476 U.S. 267, 106 S.Ct. 1842, 90 L.Ed.2d 260 (1986), Justice Stevens made just this point — that the “inclusion of minority teachers in the educational process inevitably tends to dispel [the] illusion” that skin color is a proper basis for inferring character or conduct. Id. at 316, 106 S.Ct. at 1869 (emphasis added).
Observers of programming from a minority-owned station, of course, cannot see *145the owners and thereby learn the irrelevance of race to conduct. Indeed, the Commission believes that viewers who could do so would learn what it apparently believes already — that skin color is a sound basis for predicting behavior. For the Commission’s theory to make sense, in fact, skin color or ethnicity must be an excellent clue not only to what owners want to provide but also to what viewers and listeners want to receive. Its race preference plan is based at the core on precisely the racial stereotyping that the Croson majority identified as the central evil in the use of race-based decision criteria. See — U.S. at-, 109 S.Ct. at 721 (Part III — A); id. at ---, 109 S.Ct. at 730-34 (Stevens, J., concurring); id. at-,-, 109 S.Ct. at 735, 739 (Scalia, J., concurring).
Sustaining the program diversity theory would require the court to have some basis for believing that minority ownership would lead to “minority programming” in some sense that is both intelligible and permissible, and, second, that broadcast offerings are not currently “diverse.” There is no basis in the record for either. No party here has offered a definition of minority programming that is empirically verifiable and can serve as a goal that government may constitutionally pursue. Thus it is not surprising that the data do not support any claim that the Commission’s racial decision criteria are even reasonably (much less “narrowly”) tailored to fulfill any constitutional goal. Moreover, the only data to which our attention has been called, taken at face value and assuming arguendo that they are meaningful in terms other than pure racial stereotyping, suggest that the goal posited by the Commission has been achieved quite independently of its racial decision criteria.
As Judge Silberman has shown in Shurberg Broadcasting of Hartford, Inc. v. FCC, 876 F.2d 902 (D.C.Cir.1989), Congress made no findings which support the theory of a nexus between race and diverse programming when it adopted a lottery system for allocating broadcast licenses. See Shurberg, Silberman op. at 923 (“Nowhere in the House report [H.R.Rep. No. 765, 97th Cong., 2d Sess. (1982) U.S.Code Cong. & Admin.News 1982 p. 2237]3 is there evidence presented of a nexus between program diversity and minority ownership, nor is there reference to evidence upon which the committee drew.”). Nor was this evidentiary void filled by the Continuing Resolution in 1987. See Pub.L. No. 100-202, 101 Stat. at 1329-31; Shurberg, Silberman op. at 926 (S.Rep. No. 182, 100th Cong., 1st Sess. 76 (1987), added only an assertion that “[diversity of ownership results in diversity of programming and improved service to minority and women audiences.”).
Several parties in this case urge that a Congressional Research Service Report, Minority Broadcast Station Ownership and Broadcast Programming: Is There A Nexus? (June 29, 1988), proves a link between minority ownership and diverse programming. See FCC Brief at 46-47; Rainbow Brief at 24. Examination of the CRS Report is valuable not so much for what it proves — which as I will try to show is very little — as for what it suggests about the inevitable collapse of any evidence purporting to link racial criteria for station allocation with a goal of program diversity.
In preparing the study, the Service examined data from nearly 9000 responses to an FCC survey of about 11,000 American radio and television stations. See CRS Report at 2. The survey first asked station owners what percentage of the ownership interest in their station was held by women or members of specific minority groups.4 It then sought data on programming conduct. Both radio and TV station owners were asked whether their programming was directed to or provided special services for any of nine listed groups, including specific minorities, women, children and *146senior citizens. Respondents were asked to state whether the targeting or format was for 1-19 hours per week or for more than 20 hours per week.
The CRS found some correlation between minority ownership and “audience-targeted” programming. Whereas only 20 percent of stations without any black ownership responded that they engage in “black programming,” 65 percent of stations with black ownership said that they did so. Id. at 13. For Hispanic ownership and “Hispanic programming,” the corresponding figures are 10 percent and 59 percent. Id. at 13, 15. Without employing any of the conventional paraphernalia of modem statistics, the CRS hesitantly inferred a causal connection:
[Tjhere is a strong indication that minority and women station ownership results in a greater degree of minority programming [than does nonminority ownership]. Therefore, an argument can be made that FCC policies that enhanced minority and women station ownership may have resulted in more minority and other audience targeted programming. To the degree that increasing minority programming across audience markets is considered adding to programming diversity, then, based on the FCC survey data, an argument can be made that the FCC preference policies contributed, in turn, to programming diversity.
Id. at cover page.
First I will note a rather plain methodological flaw in the survey and then turn to its basic conceptual defect. The designers of the survey made no effort to adjust for possible differences in the racial composition of the various station owners’ potential audiences. This makes it impossible to examine the role of market demand as a force behind programming decisions. “Hispanic-targeting”5 is obviously more likely to be profitable in Miami than in Minneapolis. Thus, if specific minorities are more likely to own stations in areas where they are numerous (as seems likely), the difference in “targeting” that the Report hesitantly attributes to the owners’ racial characteristics may be due simply to their rational responses to demand. The Report itself acknowledged that it “would be difficult to determine to what degree station programming strategies are market driven rather than the results of minority ownership interests.” CRS Report at 3.
Far more critical than this methodological flaw, however, is the survey’s failure to define “minority-targeted programming.” As the survey omitted that vital step, it and the Report prove only that black owners are more likely to say that they target black audiences, Hispanic owners that they target Hispanic audiences, and so on.
One can easily understand why the Commission refrained from defining black-targeted or other “minority-targeted” programming. The entire exercise would almost necessarily have constituted the crudest type of racial stereotyping. I say “almost” advisedly. It is conceivable that survey data might support the notion that blacks (say) had a greater taste for some specific programming feature. If that were true as an empirical matter, it would not be racist to say so. But if it were, and if it is permissible for the Commission to pursue program diversity by means other than general dispersion of ownership, it would simply expose the availability of a non-race-specific solution for the Commission: it could ascertain whether that sort of programming was underrepresented in relation to demand and take such steps as might be constitutionally available to stimulate its supply. If the answer is that there are no constitutionally permissible steps that it might take, then it is hard to see why the Commission has greater authority to stimulate the supply by its present means — which are not only roundabout but race-based.
The Commission avoided obvious racial stereotyping in its survey design by the blunt device of refusing to define the various types of “minority” programming at issue. The omission renders its data meaningless. It also tends to undermine any *147claim by the Commission that its goal was “compelling”: no one can seriously argue that there is a compelling governmental interest in increasing the proportion of station owners who say they engage in something they call minority programming.
The Commission’s goal also appears indeterminate, so that neither the Commission nor any observer will ever be able to discern that it has been attained. But all members of the Court appear to agree that a racial preference cannot be considered a “narrowly tailored” means if it can never reach its stated end. See, e.g., United States v. Paradise, 480 U.S. 149, 171, 107 S.Ct. 1053, 1067, 94 L.Ed.2d 203 (1987) (Brennan, J., for himself and Justices Marshall, Blackmun and Powell) (among factors determining whether race-conscious remedies are appropriate are “the flexibility and duration of the relief’). In Croson four members of the Court severely criticized theories that “could be used to ‘justify’ race-based decisionmaking essentially limitless in scope and duration.” — U.S. at-, 109 S.Ct. at 723 (Part III-A). They rejected outright remedies “that are ageless in their reach into the past, and timeless in their ability to affect the future.” Id. (quoting Wygant, 476 U.S. at 276, 106 S.Ct. at 1848 (plurality op.)). Although the other two concurring Justices did not join that portion of the Court’s opinion, their opinions show no greater tolerance of racial preferences that have no clearly specified goal whose attainment will bring the preference to an end. See — U.S. at-, 109 S.Ct. at 730 (Stevens, J., concurring) (“When [government] creates a special preference, or a special disability, for a class of persons, it should identify the characteristic that justifies the special treatment. When the classification is defined in racial terms, I believe that such particular identification is imperative.”) (quoting Fullilove, 448 U.S. at 552-54, 100 S.Ct. at 2813-14); — U.S. at-, 109 S.Ct. at 738 (Scalia, J., concurring) (“[T]he remedial power extends no further than the scope of the continuing constitutional violation____ And it is implicit in our cases that after the [violative] school system has been completely disestablished, the States may no longer assign students by race.”).
The stated goal here — some sort of optimal or proportional relation of programming to ethnic taste — is undefined and seemingly not definable except in racial terms. Plainly there can be no assurance of an end to the racial preference if there is no way — except the Commission’s conclusory say-so — of ascertaining when the goal is reached.
The Commission gives some clue as to how in fact it would measure the attainment of its goals. Its response in this very case to the argument that the minority preference was “open-ended as to duration” was to point to the figures comparing the percent of minorities in the population with the percent of stations owned and controlled by minorities. See FCC Brief at 47 n. 47, referring to id. at 30 n. 33. Thus the Commission apparently intends to stop only when proportionality in ownership is attained. Whatever the Commission may think it is doing, its stated criteria indicate that its supposedly nonremedial program-diversity goal is only a stalking horse for making license ownership replicate the racial proportions of the population at large.
The Commission might escape the challenge of defining its programming goal by positing the existence of some ineffable attribute of programming by minorities, arising out of such elements as may be common to the experience of the specific minority group. To the extent that such an attribute exists, one could say that, by definition, programming by minority-owned stations will reflect the ineffable characteristic. There would be no empirical problem.
This response would prove too much. If government may simply assume some inherently non-measurable race-conduct link and pursue diversity of conduct by allocating government benefits on a basis of racial proportions, then obviously there is no constitutional obstacle to race-based decision criteria. The Commission would be free to orchestrate an ethnically defined array of radio and TV programs, each designed to match “black tastes,” “white *148tastes,” “Hispanic tastes” or “Asian-American tastes” — or, indeed, as its ethnic analysts advanced in sophistication, “Jewish tastes,” “WASP tastes” or “Italian tastes.” Neither Croson nor any of the Court’s decisions in the field can stand for this.
Thus the CRS study not only fails to supply the Commission’s factual premises but suggests that it is asking questions that are unanswerable without racial stereotyping.
For a reader willing to ignore these flaws, however, the Report leads to the conclusion that the Commission has already achieved its purported goal. For purposes of testing the relation between “minority-targeted” programming and minority populations, the Commission considered data from ten cities grouped into two categories, one with five large Standard Metropolitan Statistical Areas (“SMSA”) and the other with five smaller cities. See id. at 349 & n. 1. The study’s data show that nonminority-owned programmers produce minority-targeted formats in percentages higher than the percentages of the various minorities in the general population for every minority group in both types of cities. For the five large cities the data are as follows:
Percent of Nonminority 6 Percent of Stations Broadcasting Population to Target Minority
Black 17.6 22
Hispanic 14.9 20
Asian/Pac 3.3 5
Ind/Alask 1.7 3
CRS Report at 25, 61-64.
Similarly, in the five smaller cities, non-minority broadcasters also appear to “target” minority audiences in greater percentages than would be called for by their raw proportions of the population:
Percent of Nonminority Percent of Stations Broadcasting Population to Target Minority
Black 9.8 24.2
Hispanic 3.4 26.4
Asian/Pac .44 27.2
Ind/Alask 1.4 26.4
Id. at 26, 73-76. If black-targeted programming as the term is used by the survey respondents has any meaning that is related to some permissible goal, the Commission should declare victory.
This court’s 1984 decision in West Michigan, sustaining the Commission’s racial diversity rationale, cannot survive Croson. There a majority of the Supreme Court for the first time insisted on “strict scrutiny” of racial preferences. See above at 349-350. Moreover, that opinion gave unprecedented weight to precisely the criteria under which the Commission’s rationale is vulnerable and that West Michigan disregarded. First, while West Michigan makes no mention of the hazards of racial stereotyping, the prevailing opinions in Croson all express the justices’ concern that in the absence of strict judicial review, the political branches’ use of racial criteria may reflect and reinforce racial stigmatization. See above at 357-58. While West Michigan casually asserted that the “FCC comparative evaluation process generally conforms to Justice Powell’s model,” 735 F.2d at 615, the Croson Court’s evident concern over stereotyping requires us to probe seriously the FCC’s assumptions about race as a predictor of conduct — an empirical matter irrelevant to Justice Powell’s diversity theory.
Second, the justices’ statements of concern that racial preferences may drag on interminably, see above at 352, have no parallel in West Michigan — an omission that is hardly surprising as those statements appear to have no roots in prevailing opinions issued before West Michigan. (Justice Stevens’s statement is taken from his dissent in Fullilove.) Indeed, West Michigan never so much as mentioned the requirement of “narrow tailoring.”
I therefore believe that Croson fatally undermines West Michigan’s conclusory validation of the Commission’s racial diversity rationale.
II. The Remedial Purpose
If the Commission's racial preference cannot withstand the equal protection challenge on the basis of its program diversity *149purpose, the next question is whether the intervenor ’s remedial theory fills the gap. One difficulty, obviously, is the theory’s possible constitutional infirmity. But the court need reach that issue only if the preference is otherwise a proper exercise of the Commission’s authority. It can be such an exercise only if (a) the Commission has properly adopted a remedial program in the discharge of its general authority to advance the public interest in the allocation of licenses or (b) if Congress has mandated such a remedial program. The first possible basis is clearly absent here; the Commission has never sought to explain its racial preference as an attempt to remedy discrimination. The second basis is highly doubtful; Congress’s instructions in the Continuing Resolution of December 22, 1987, Pub.L. No. 100-202, 101 Stat. 1329, 1329-31 (1987), and the Departments of Commerce, Justice and State, the Judiciary, and Related Agencies Appropriations Act, 1989, Pub.L. No. 100-459, 102 Stat. 2186, 2216-17 (1988), are at best ambiguous. The legislative histories do not support a broad reading of the language used by Congress, and in light of what are at least serious doubts about the constitutionality of any such remedial program on the present record, I believe the appropriations acts do not clearly mandate more than continuation of the Commission’s preexisting program of racial preferences for purposes of program diversity.
A. The Commission’s Stated Purpose
At no time has the Commission sought to explain or justify the racial preference now before us as a remedy for racial discrimination. Under well-established principles of administrative law, we cannot uphold a Commission action on the basis of reasoning that the Commission has not adopted. SEC v. Chenery Corp., 318 U.S. 80, 88, 63 S.Ct. 454, 459, 87 L.Ed. 626 (1943) (“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.”).
The background of the Commission’s policy is no secret. In TV 9, Inc. v. FCC, 495 F.2d 929 (D.C.Cir.1973), this court overturned a Commission decision that had refused to award a license applicant credit for minority ownership. Noting that the Commission had rested its efforts to diffuse ownership on the belief that such diffusion would bring about diverse program content, the court declared that the goal of maximum diversification of ownership entailed “favorable consideration to an applicant who ... gives a local minority group media entrepreneurship.” Id. at 937. The court appeared to make the credit contingent upon a showing that the minority ownership would be likely to enhance diversity of content:
We hold only that when minority ownership is likely to increase diversity of content, especially of opinion and viewpoint, merit should be awarded.
Id. at 938 (emphasis added) (citations omitted).
In Garrett v. FCC, 513 F.2d 1056 (D.C. Cir.1975), the court reviewed another Commission decision declining to award credit for minority ownership, where the applicant had offered no evidence either of dissatisfaction among minorities with existing programs or of any plan to cure it. Despite having appeared in TV 9 to make credit turn on specific circumstances indicating a likelihood of program diversity, the court now stated that “black ownership and participation together are themselves likely to bring about programming that is responsive to the needs of the black citizenry.” Id. at 1063.
The Commission responded to these developments. See Statement of Policy on Minority Ownership of Broadcasting Facilities, 68 F.C.C.2d 979 (1978). In the Minority Ownership Report of its Minority Ownership Task Force, it stated:
Despite the fact that minorities constitute approximately 20 percent of the population, they control fewer than one percent of the 8500 commercial radio and television stations currently operating in this country.
*150Acute underrepresentation of minorities among the owners of broadcast properties is troublesome because it is the licensee who is ultimately responsible for identifying and serving the needs and interests of his or her audience. Unless minorities are encouraged to enter the mainstream of the commercial broadcasting business, a substantial proportion of our citizenry will remain underserved and the larger, non-minority audience will be deprived of the views of minorities.
Federal Communications Commission, Minority Ownership Taskforce Report, Minority Ownership in Broadcasting 1 (May 17, 1978) (emphasis in original).
The references to “underrepresentation” are not linked, of course, to any claim of discrimination. The Commission’s rationale remained the one consistently articulated by this court — an extension of the idea that ownership diversity would entail program diversity.
The Commission has, indeed, explicitly abjured any remedial purpose. As it says in its brief here, “[T]he FCC's goal in implementing the preference policy ... has not been to remedy prior discrimination against minorities or to provide remedial benefits.” FCC Brief at 30. Further, in its Brief on Rehearing En Banc at 18, Steele v. FCC, 806 F.2d 1126 (D.C.Cir.1986), it declared the absence of the sort of factual underpinnings best suited to support any such remedial purpose: “There has never been a finding, nor so far as we know even an allegation, that the FCC engaged in prior discrimination against racial minorities or women in its licensing process.” Because of its lack of any remedial purpose, it has, of course, never sought to analyse whether past discrimination affords any basis for such a program.
If we assess the Commission’s racial preference as an exercise of its mandate to pursue the public interest, and if I am correct in finding that Croson and Wygant have toppled West Michigan's constitutional validation of the program diversity theory, then the Commission’s lack of any effort to support racial preferences as a remedy for discrimination would normally require a remand to the Commission for such further consideration as it found appropriate. Such a remand would be inappropriate, however, if Congress had mandated a remedial program. I therefore turn to the relevant appropriations acts for 1988 and 1989.
B. The Appropriations Acts
In appropriating funds for fiscal year 1988, Congress provided:
That none of the funds appropriated by this Act shall be used to repeal, to retroactively apply changes in, or to continue a reexamination of, the policies of the Federal Communications Commission with respect to comparative licensing, distress sales and tax certificates granted under 26 U.S.C. 1071, to expand minority and women ownership of broadcasting licenses, including those established in Statement of Policy on Minority Ownership of Broadcast Facilities, 68 F.C.C.2d 979 and 69 F.C.C.2d 1591, as amended 52 R.R.2d [1301] (1982) and Mid-Florida Television Corp., [69] F.C. C.2d 607 Rev.Bd. (1978), which were effective prior to September 12, 1986, other than to close MM Docket No. 86-484 with a reinstatement of prior policy and a lifting of suspension of any sales, licenses, applications, or proceedings, which were suspended pending the conclusion of the injury____
Pub.L. No. 100-202, 101 Stat. 1329-31 (1987). Congress renewed this spending prohibition in the relevant appropriations measure for fiscal year 1989. See Pub.L. No. 100-459, 102 Stat. 2186, 2216-17 (1988) (same language). Accordingly, the Commission proceeded to resolve the present licensing dispute by reference to the preexisting policy. See Metro Broadcasting, Inc., 3 F.C.C. Red 866 (1988).
If Congress had enacted the racial preference here at issue into law, of course, that preference would be a part of the law governing the Commission’s license allocation activities; the Commission would be bound to apply it, and we to uphold the application in the absence of constitutional *151defect. But that is not what Congress did. It ordered a kind of mental standstill, prohibiting the use of funds appropriated under the acts to “repeal, to retroactively apply changes in, or to continue a reexamination of, the [specified] policies,” “other than to close” the docket in which the Commission was reexamining its various racial and gender preferences “with a reinstatement of prior policy.” In this context “policy” is ambiguous. So far as comparative licensing is concerned, the laws do not state whether the permissible policy, i.e., the one which the Commission may spend the appropriated funds to reinstate, is the Commission’s pre-existing pursuit of program diversity through racial preferences, or outright racial preferences for some other (unspecified) purpose. Thus they do not address what the Commission should do if it applies its prior policy (defined in the only terms that the Commission has ever articulated), and the policy as so defined and rationalized is found constitutionally defective.
In at least two respects the context of the appropriations acts militate against reading them as mandating the use of racial preferences even if the Commission’s diversity-based ones are unconstitutional. First, it would seem anomalous for Congress to lock the Commission into a policy broader than it had ever before applied without specifically articulating the new policy and its bases. Second, as the restriction is limited in time to the period of the Commission’s reliance on the appropriated funds, the Commission would be free to abandon the policy thereafter, a result that seems especially curious for a new policy.
The legislative history of the Continuing Resolution does not entirely clarify Congress’s intent, but it contains no hint that Congress contemplated an enlarged reading of the FCC’s “policy.” The restriction was added to H.R. 2763 during Senate committee hearings, with the statement that the FCC’s reexamination of the minority preference policies was “unwarranted.” S.Rep. No. 182, 100th Cong., 1st Sess. 76 (1987). The Report went on to note that “Congress has expressed its support for such policies in the past and has found that promoting diversity of ownership of broadcast properties satisfies important public policy goals. Diversity of ownership results in diversity of programming and improved service to minority and women audiences.” Id. The only other specific references to the minority preference policy that I uncovered in the legislative history were brief remarks by Senator Lautenberg during the Senate discussions of H.R. 2763. In supporting the amendment, which he had proposed, Senator Lautenberg echoed the Report’s focus on program diversity:
Important goals of our communications policy should be the promotion of diverse viewpoints, and service to diverse audiences. To meet these goals, it is important that we promote diversity of ownership of broadcast licenses____ Diversity of ownership results in diversity of programming and improved service to minority and women audiences.
133 Cong.Rec. S14395 (daily ed. October 15, 1987) (statement of Sen. Lautenberg). Obviously, neither of these expressions refers to any remedial purpose.
Similarly, the legislative history of the 1989 appropriations restriction gives no reason to believe that Congress was imposing a remedial rationale on the agency. The spending prohibition in the relevant act for 1989 was introduced in the Senate committee report, S.Rep. No. 100-388, 100th Cong., 2d Sess. 78-79 (1988). The sole remarks concerning this addition were that “[t]he Committee has also continued language from previous appropriations acts with regard to VHF-UHF swaps and minority and women ownership of broadcasting licenses.” Id. at 79. Senator Hollings was the only legislator who spoke on the minority ownership provision during either the House or Senate débates on the bill. He discussed the CRS Report, which had been recently issued, and said:
The report clearly demonstrates that minority ownership of broadcast stations does increase the diversity of viewpoints presented over the airwaves.
The FCC’s rationale for questioning the constitutionality of the minority/female ownership policies is factually and *152legally invalid. I am a strong supporter of diversity in ownership and programming____ The language in this year’s appropriations bill simply preserves long-standing policies of both the Congress and the FCC____
134 Cong.Rec. S10021 (daily ed. July 27, 1988) (statement of Sen. Hollings) (emphasis added). Thus the legislative history of the 1989 appropriation, like that for 1988, appears to evidence only a congressional wish to continue the policy of trying to enhance program diversity.
Serious doubts about the constitutionality of a remedial rationale for the Commission’s racial preference rule also call for a narrow reading of the appropriations acts. Crowell v. Benson, 285 U.S. 22, 62, 52 S.Ct. 285, 296, 76 L.Ed. 598 (1932).7 It is true that in West Michigan this court relied on a remedial purpose to support its finding of the preference’s constitutionality. See 735 F.2d at 614-16. But as the FCC’s purpose was exclusively to promote programming diversity, only that purpose was properly under review. The portion of West Michigan addressing remedial concerns is therefore dictum.
Even were West Michigan a holding on the remedial purpose, its reasoning appears largely undermined by Wygant and Croson. First, it accepted findings of past discrimination that are likely insufficient under present standards. Second, it completely neglected the question whether the “remedy” was “narrowly tailored,” an issue now likely to be critical.
Discrimination. The West Michigan court noted that the allocation there under review “came on the heels” of Congress's passage of § 115 of the Communications Amendments Act of 1982, Pub.L. No. 97-259, 96 Stat. 1087, 1094-95 (codified at 47 U.S.C. § 309(i)(3)(A) and (C)(ii)), authorizing Commission use of lotteries for license allocation but subjecting such lotteries to special minority preference requirements. 735 F.2d at 613, 615-16. The court noted that Congress had based the preference requirements on findings of “underrepresentation” of minorities in the broadcast media and had characterized this underrepresentation as flowing from “ ‘the effects of past inequities stemming from racial and ethnic discrimination.' ” Id. at 614 (quoting H.R. Conf.Rep. No. 97-765; 97th Cong., 2d Sess. 43 (1982), U.S.Code Cong. & Admin.News 1982, p. 2287). The court read the enactment of the lottery provisions under these circumstances “as giving congressional confirmation to the factual bases of [the] remedial nature [of the FCC’s various minority preference policies].” 735 F.2d at 616.
I agree with Judge Silberman’s conclusion in Skurberg that these “findings” are inadequately supported to show prior discrimination by the FCC or the broadcast industry. See Skurberg, Silberman op. 876 F.2d at 914 (“Neither Congress nor the FCC ever found any evidence to link minority ‘under-representation’ to discrimination by the FCC or to particular discriminatory practices in the broadcasting industry.”) (citation omitted); see generally id. at 914-16. In the absence of findings of any discrimination by the agency or in the industry, the congressional statements can sustain the preference only if general societal discrimination suffices to support a race-based remedial program.
Three Supreme Court decisions are relevant to an assessment of such general discrimination: Fullilove v. Klutznick, 448 U.S. 448, 100 S.Ct. 2758, 65 L.Ed.2d 902 (1980), Wygant, and Croson. In Fullilove the Court upheld a 10% minority set-aside program established by Congress for federal grants to state and local governments for use in public works projects. Both Chief Justice Burger’s plurality opinion and Justice Powell’s concurrence stressed Con*153gress’s unique role in ensuring equal protection under the Fourteenth Amendment. See 448 U.S. at 472-78, 100 S.Ct. at 2771-75, id. at 503-06, 100 S.Ct. at 2787-89 (Powell, J., concurring). Both opinions, however, left obscure the extent to which Congress’s special role left it free to mandate racial preferences solely on the basis of general societal discrimination. Justice Powell read the record as showing that Congress had made a particularized finding of discrimination by the governmental unit involved: “Congress reasonably concluded that private and governmental discrimination had contributed to the negligible percentage of public contracts awarded minority contractors.” Id. at 503, 100 S.Ct. at 2787 (citation omitted). The plurality was less clear on whether the identified discrimination was specifically in the contracting business, or merely in society at large. The opinion declared that Congress “had before it ... evidence of a long history of marked disparity in the percentage of public contracts awarded to minority business enterprises. This disparity was considered to result not from any lack of capable and qualified minority businesses, but from the existence and maintenance of barriers to competitive access which had their roots in racial and ethnic discrimination____” Id. at 478, 100 S.Ct. at 2774.
In Wygant, where the Court invalidated a purely local racial preference, the plurality’s holding and rationale appeared to reject generalized findings of discrimination as a basis for race-based remedial measures. In his plurality opinion, Justice Powell observed that the Court had never held that societal discrimination alone sufficed to justify a racial classification. “Rather, the Court has insisted upon some showing of prior discrimination by the governmental unit involved before allowing limited use of racial classifications in order to remedy such discrimination.” Wygant, 476 U.S. at 274, 106 S.Ct. at 1847 (emphasis added). While the Court held that there need not be “findings” of prior unlawful discrimination, it did require that a “firm basis” exist for believing that the public entity had engaged in discrimination. Id. at 293, 106 S.Ct. at 1856 (O’Connor, J., concurring). Justice O’Connor suggested as an example of such a “basis” the evidence needed to support a prima facie Title VII pattern or practice claim. Id. at 292, 106 S.Ct. at 1856.
Wygant, of course, does not directly consider societal discrimination as a legitimate factual predicate for race-based policies adopted by Congress. And the ambiguity of Fullilove carries forward to Croson. Although the majority nowhere states that Congress has the right to base racially-conscious programs solely on evidence of societal discrimination, neither does it explicitly reject this proposition. See — U.S. at -, 109 S.Ct. at 719 (Part II) (“That Congress may identify and redress the effects of society-wide discrimination does not mean that, a fortiori, the States and their political subdivisions are free to decide that such remedies are appropriate.’’).
The Court’s recent disposition of H.K. Porter Co. v. Metropolitan Dade County indicates that at least some aspects of Croson will apply to congressionally authorized racial preferences. The court of appeals had upheld a local racial preference program adopted under the authority of the Surface Transportation Assistance Act of 1978, Pub.L. No. 95-599, 92 Stat. 2689 (1978), and the Supreme Court remanded “for further consideration in light of” Croson. — U.S. -, 109 S.Ct. 1333, 103 L.Ed.2d 804 (1989), vacating 825 F.2d 324 (11th Cir.1987); see also Croson, — U.S. at-, 109 S.Ct. at 740 (Marshall, J., dissenting) (commenting that the Croson majority opinion represented a “deliberate and giant step” away from Fullilove).
The unresolved ambiguity of Fullilove and Croson leaves it impossible to reach a firm opinion as to the evidence of discrimination needed to sustain a congressional mandate of racial preferences. But I pause to note how little there is here. There is no finding by a court or anyone of discrimination in the allocation of broadcasting channels. Nor has any evidence been presented to this court that would amount to prima facie evidence of a discriminatory pattern or practice. See Wygant, 476 U.S. at 292-94, 106 S.Ct. at 1856-*15457 (O’Connor, J., concurring) (suggesting that such evidence would suffice). Apart from the conclusory references by congressional committees discussed above, there is simply the Commission’s observation that only 2.1 percent of radio and TV stations are minority owned and controlled, as against a 23.5 percent minority share of the general population. FCC Brief at 30 n. 33.
Discrepancies between the ethnic proportions among persons receiving a government benefit and among the population at large cannot be evidence of discrimination where special qualifications are needed for receipt of the benefit. See, e.g., Croson, — U.S. at-, 109 S.Ct. at 725 (“[w]here special qualifications are necessary, the relevant statistical pool for purposes of demonstrating discriminatory exclusion must be the number of minorities qualified to undertake the particular task.”) (citations omitted). Justice O’Connor went on to point out that there was no evidence of the number of minority business enterprises in the relevant market, and indeed noted evidence that black-owned businesses were not homogeneously distributed among industries. Id. As the FCC’s “integration” requirements compel the synthesis of ownership and day-to-day management in the same hands, the relevant pool consists of potential owner-managers of radio and TV stations. The figures discussed by the FCC — including those in the CRS Report— nowhere reveal the number or proportion of minorities in this group. In the case at hand, the Rainbow principals are of Hispanic origin, yet we have no data on the relative eligibility and participation rates of this group. See Croson, — U.S. at-, 109 S.Ct. at 728 (highlighting complete absence of evidence of discrimination against Spanish-speaking and other non-black beneficiaries of Richmond’s set-aside program, program’s “random inclusion of racial groups that ... may never have suffered from discrimination in the construction industry,” and its “gross overinclusiveness”).
Nonetheless, because of the persisting uncertainty over the nature of the discrimination necessary for congressionally mandated remedial preferences, I turn to the issue of whether the preference is “narrowly tailored” or properly “fits” the evil to be cured.
Narrow tailoring. The Croson majority took two ameliorating features of the program upheld in Fullilove and made their absence critical to the rejection of the Richmond plan. First, in Fullilove three justices upholding the preference noted that before adopting the set-asides Congress had considered and attempted alternatives which were not based on racial criteria. See Fullilove, 448 U.S. at 463-67, 100 S.Ct. at 2767-69; id. at 511, 100 S.Ct. at 2791 (Powell, J., concurring). In Croson those justices of the majority whose constitutional views were most tolerant of racial preferences found the Richmond council’s failure to consider such alternatives a critical defect. See — U.S. at-, 109 S.Ct. at 728 (“The principal opinion in Fullilove found that Congress had carefully examined and rejected race-neutral alternatives before enacting the MBE [Minority Business Enterprise] set-aside.”). As the Commission adopted the racial preference on the basis of its and this court's program diversity theory, and as it is most unclear whether Congress has ever endorsed a remedial theory for the program (the point indeed at issue here), it can hardly be said that the program rests on anyone's findings as to the inadequacy of non-race-based alternatives.
Second, the three justices of the Fullilove plurality had noted as “significant” that the congressional program at issue provided for waiver of the 10% minority participation requirement “to avoid dealing with an MBE who is attempting to exploit the remedial aspects of the program by charging an unreasonable price, i.e., a price not attributable to the present effects of past discrimination.” Fullilove, 448 U.S. at 488, 100 S.Ct. at 728. In Croson, the majority opinion emphasized the absence of any such built-in limit. “Unlike the program upheld in Fullilove, the ... system focuses solely on the availability of MBEs; there is no inquiry into whether or not the particular MBE seeking a racial preference has suffered from the effects of past discrimination by the city or prime contrac*155tors.” Croson, — U.S. at-, 109 S.Ct. at 729. This reading of Fullilove — whereby the set-aside seemingly applies only to victims of prior discrimination by the agency or industry involved — further undercuts the Commission’s claim that its policy is sufficiently narrow in scope.8 Indeed, it is hard to see how a program can be “narrowly tailored” as a remedy for societal discrimination if competitors have no opportunity to show that individual beneficiaries have suffered no impairment of their license-securing ability attributable to that discrimination.
Moreover, in Croson (for the first time) the Court made the point that where an agency is making decisions “on a case-by-case basis,” it was hard “to see the need for a rigid numerical quota.” — U.S. at -, 109 S.Ct. at 730. For such programs, the Court plainly believed, it was practical to ascertain whether discrimination had in fact impaired the proposed beneficiary's competitive position. Although the FCC’s preference is not in form a “rigid numerical quota” (a point explored immediately below), the Court’s analysis seems equally applicable. Where the government is proceeding on a case-by-case basis, there is no obvious need to employ a mechanical racial preference and forego inquiry into the beneficiaries’ actual circumstances. No FCC preference for successful Hispanic entrepreneurs can remedy the years spent by other Hispanics picking lettuce as a result of racial discrimination. At least it cannot if we think of people as individuals.
In upholding the FCC’s racial preference policy, West Michigan relied heavily on the FCC’s treatment of race as merely one of many “factors.” 735 F.2d at 613. I am unpersuaded that the multi-factored approach in this case — if that is what it really is — saves the minority preference scheme. As Judge Silberman explained in his opinion in Shurberg at 919, “[T]he constitutional distinction between a tolerable and intolerable burden on the innocent turns on the uniqueness and value of the opportunity lost.” Here, as in Shurberg, Metro was denied a comparative hearing on the only new license currently being offered in the Orlando area. Thus, Rainbow’s victory, in which the minority preference was dispositive, deprived the other competitors of their only chance for a new license for the foreseeable future.
Moreover, on the present facts the line between a simple quota and a multifactored allocation is painfully thin. The Commission has made clear that it proposes to keep giving race weight until license ownership matches the racial composition of the population at large. See above at 352. That looks like a quota. The only refinement is that the Commission is working to fulfill it gradually — as it must, given the limited number of stations available to allocate each year. The Commission’s candid briefing thus appears to remove a distinction that the West Michigan court found critical. See West Michigan, 735 F.2d at 613.
Even if one assumes, however, along with West Michigan, that there are adequate findings of discrimination, that the FCC’s minority preference program is genuinely a multi-factored test rather than pursuit of a quota, and that the burden on nonminority applicants is sufficiently light to pass constitutional scrutiny, I believe that the plan is undermined by the two defects that Croson made critical — the absence of consideration of alternative nonracial solutions and the omission of a waiver provision that effectively confines the benefits to individuals who have suffered identifiable discrimination.
It is plain that the FCC has not offered a remedial rationale for its policy. Nor could it possibly be said that the appropriations acts clearly establish a remedial policy, in light of the textual ambiguity concerning the word “policy” in the appropriations *156acts, their legislative histones, and the strong possibility that even supported as a remedy the FCC’s racial preference program would violate the equal protection clause. Whether we could uphold a Commission finding that the acts established such a policy, given the deference it enjoys in matters of law under Chevron U.S.A. v. NRDC, 467 U.S. 837, 842-45, 104 S.Ct. 2778, 2781-83, 81 L.Ed.2d 694 (1984), is not before us, in the absence of any FCC consideration of the point. See Phillips Petroleum Co. v. FERC, 792 F.2d 1165, 1172 (D.C.Cir.1986) (deference must be based on agency’s exercise of its interpretive judgment).
CONCLUSION
Finding the program diversity theory unconstitutional because plainly based on impermissible racial stereotypes, and finding no adoption of any remedial justification by the Commission or Congress, I would remand to the Commission for further proceedings. The FCC’s first task on remand would be to interpret the relevant appropriations act for 1989 to determine whether it mandates a remedial preference policy. If the FCC determines that it does not, it would be free to reassess the preference policy in light of the unconstitutionality of the program diversity rationale.
. The FCC’s counsel asserts that the Commission has not adopted this latter, so-called "remedial” rationale for its minority preference policy. See FCC Brief at 30 ("The FCC’s goal in implementing the preference policy, however, has not been to remedy prior discrimination against minorities or to provide remedial benefits.”).
. Justice O’Connor wrote a six-part opinion in Croson. A majority of the Court agreed to three of those parts — I, III-B and IV. All citations to these majority statements will be in the usual form, without parenthetical reference to authorship. Part II of Justice O’Connor’s opinion was joined by Justices Rehnquist and White. Parts III-A and V were joined by Justices Rehnquist, White and Kennedy. All citations to these plurality portions of the opinion will include the relevant part number, but will not specify authorship.
. The report explains the House-Senate conference’s approval of § 115 of the Communications Amendments Act of 1982, Pub.L. No. 97-259, 96 Stat. 1087, 1094-95 (codified at 47 U.S.C. § 309(i)(3)(A) and (C)(ii)).
. The minority groups were Black (not of Hispanic origin), Hispanic, American Indian or Alaskan Native, and Asian or Pacific Islander.
. I assume for the sake of argument that "Hispanic targeting” has an intelligible meaning. If it is a fancy phrase for Spanish-language programming, then indeed it would.
. "Nonminority" as used here and in the following table means not belonging to the specific minority “targeted.”
. The principle that statutes should be read so as to avoid constitutional questions where fairly possible appears to extend to situations like this one, in which the event casting doubt on a statute’s constitutionality (here, the decision in Croson) occurs after Congress enacted the law. See, for example, United States v. Thirty-Seven Photographs, 402 U.S. 363, 367-75, 91 S.Ct. 1400, 1403-08, 28 L.Ed.2d 822 (1971) (federal statute written in 1930 interpreted to incorporate 14-day time limit on commencement of forfeiture proceedings, and 60-day limit on their completion, both inferred from 1965 decision of Supreme Court).
. It also sheds light on the question of whether after Croson Congress can find remedying general societal discrimination to be a compelling state interest under the Fourteenth Amendment. If the only permissible (narrowly tailored) remedies are those which extend solely to victims of past discrimination perpetrated by the governmental unit or industry at hand, then the only discrimination that even Congress may remedy is discrimination that can be shown to account for the specific beneficiary’s plight.