[659]*659OPINION
KOZINSKI, Circuit Judge.
We must decide whether the Reindeer Industry Act of 1937, 25 U.S.C. §§ 500 et seq., prohibits reindeer herding by non-natives in Alaska.
I
Contrary to popular belief, reindeer are neither native to Alaska nor part of the Alaskan native way of life. Around the turn of the century, white settlers had just about exhausted the natural food supply of Alaskan natives, who traditionally relied on fishing and hunting. Dr. Sheldon Jackson, the famous missionary, realized that raising livestock could provide natives a stable food supply, so he arranged for the importation of reindeer from Russia.1 See Alaska Legislation: Hearing on Bills Relating to the Territory of Alaska Before the House Comm, on the Territories, 75th Cong., 1st Sess. 4-5 (1937) (statement of Rep. Dimond). But white settlers also muscled into the reindeer business, prompting Congress to pass the Reindeer Act of 1937. The Act appropriated $2 million to buy all reindeer owned by nonnatives in Alaska and give them to the natives. The Act also strictly limits the sale of reindeer to non-natives.
In 1986, appellant Williams informed the Bureau of Indian Affairs Area Director in Alaska that he intended to import reindeer from Canada for commercial purposes; he asked whether his plan would run afoul of the Reindeer Act. See Reindeer Herders Ass’n v. Juneau Area Dir., 23 IBIA 28, 35 (1992). The Area Director referred the inquiry to the Regional Solicitor’s Office, which held that it would not. Nothing in the Act, the Solicitor noted, specifically prohibits nonnatives from owning and selling reindeer. The Solicitor further noted that the Act’s prohibitions on selling reindeer to non-natives apply only to two categories of reindeer: those owned by the government and those owned by natives, see 25 U.S.C. § 500i (“Live reindeer in Alaska, and the increase thereof, acquired by the Secretary of the Interior ..., and live reindeer in Alaska, and the increase thereof, owned by the said natives ... shall not be sold or transferred ... to anyone other than ... natives of Alaska-”). The Regional Solicitor, therefore, held that these restrictions didn’t apply to appellants’ imported Canadian reindeer and that appellants were free to sell their reindeer to anyone. See Reindeer Herders Ass’n, 23 IBIA at 35-37 (quoting the Regional Solicitor’s opinion). In 1989, the Regional Solicitor reconsidered the issue and reached the same conclusion. See id. at 39-41. The Area Director officially adopted the Regional Solicitor’s interpretation, and native reindeer herders appealed to the Interior Board of Indian Appeals (IBIA).
The IBIA acknowledged that the Act says nothing about non-native ownership of reindeer. Nevertheless, it held that, based on the Act’s policy, structure and legislative history, the statute “must be construed to prohibit non-Native entry into the reindeer industry in Alaska, regardless of the source of the reindeer involved.” Id. at 69. The district court upheld the IBIA’s interpretation and rejected appellants’ claim that the Act violates equal protection. See E.R. at tab 9. The non-native reindeer herders appeal.
II
We start, as did the Regional Solicitor, by noting that the Reindeer Act does not by its terms guarantee Alaskan natives a monopoly in the reindeer business. To be sure, a lot there suggests that Congress meant to give natives a big leg up in the business. The Act’s declared purpose is to “preserv[e] the native character” of the reindeer industry, 25 U.S.C. § 500, and it authorizes the Secretary of Interior to organize and manage the reindeer industry so that natives have “responsibility ... in all branches of said industry,” id. § 500f. To that end, the Act directs the Secretary to buy all reindeer and equipment [660]*660needed to effectuate the Act’s purpose, id. § 500a, and imposes a continuing duty on any non-natives who come into possession of reindeer to register them with the Interior Department, id. § 500b, presumably so the government can buy the reindeer for the natives. Most significantly, the Act prohibits the sale of government- and native-owned reindeer to non-natives unless the reindeer are to be slaughtered within 30 days of sale or exported from Alaska “and never brought back alive.” Id. § 500i.
Taken as a whole, these provisions do place significant obstacles in the way of any non-native who would operate a reindeer business in Alaska. To begin with, as of 1937, the Act ensured that only natives would be established in the industry; a non-native who wanted to get into the business would have had to overcome the advantages of incumbency enjoyed by natives, and suffered the disadvantage of not being allowed to buy local reindeer or any of their offspring. See 25 U.S.C. § 500i. At the time the Act was passed, this meant that non-natives would have had to obtain the reindeer from Scandinavia or Russia where they are indigenous. Moreover, non-natives intending to start a reindeer business would have had to buy grazing land, as the government allows only natives to graze reindeer on federal land. Id. § 500m; 43 C.F.R. § 4310.2 (1995). This too would have been a huge obstacle because 99% of land in Alaska, until statehood in 1950, was owned by the federal government. See Richard O. Stern, et. al, Eskimos, Reindeer, and Land 187 (1980). Congress may well have thought that, by passing the Act, it was effectively precluding anyone other than natives from entering the reindeer business. Nevertheless, nothing in the Act actually prohibits non-natives from entering the Reindeer business if they can overcome all these obstacles.2
Were we interpreting the statute free from any constraint, therefore, we would conclude that the Reindeer Act does not proscribe appellants’ activities. But, as often happens, we must construe the statute subject to a number of countervailing considerations. In the first place, we owe the IBIA’s interpretation substantial deference. See Chevron U.S.A. v. Natural Res. Def. Council, 467 U.S. 837, 842-43, 104 S.Ct. 2778, 2781-82, 81 L.Ed.2d 694 (1984).3 Then too, we are required to construe statutes favoring Native Americans liberally in their favor. See, e.g., County of Yakima v. Confederated Tribes and Bands, 502 U.S. 251, 269, 112 S.Ct. 683, 116 L.Ed.2d 687 (1992). Weighing on the other side is appellants’ claim that the Reindeer Act, as interpreted by the IBIA, violates the constitutional guarantee of equal protection of the laws.
Ill
It is a close question whether-even giving the agency the full measure of deference to which it is entitled under Chevron and adding in the special solicitude to which that interpretation is entitled because it favors [661]*661natives-we could uphold the agency’s interpretation. The simple fact is that the Aet-which is so explicit on so many points-does not say the one thing the agency reads it to say: that non-natives may not herd reindeer.4 This is not a minor detail-a mere interstice-that the agency is entitled to fill in at its discretion. Rather, the total and perpetual exclusion of a majority of the population of Alaska from a particular enterprise is the kind of significant feature we would normally expect Congress to spell out if that were its intent. We are aware of no other provision in the Code of Federal Regulations that effectuates quite so sweeping a policy based on such ambiguous statutory language.
At the same time, we recognize that for the past 60 years natives have enjoyed a de facto monopoly in the Alaskan reindeer business. This monopoly has developed under the watchful supervision of the agency charged with enforcing the Reindeer Act and is, therefore, entitled the added force of a long-standing construction. See Reindeer Herders Ass’n, 23 IBIA at 34 (noting that, since 1939, the Interior Department and general public “assumed” the Act precluded nonnative ownership). Under these circumstances, we conclude that the agency’s interpretation of the Reindeer Act, favoring natives as it does, is not unreasonable.
IV
This only brings us to a far more complex question: To what extent is a court bound to defer to an agency’s interpretation where that interpretation raises difficult constitutional questions? The Supreme Court has spoken twice on this subject. In DeBartolo Corp. v. Florida Gulf Coast Trades Council, 485 U.S. 568, 108 S.Ct. 1392, 99 L.Ed.2d 645 (1988), the Court rejected the NLRB’s interpretation of section 8(b)(4) of the National Labor Relations Act, 29 U.S.C. § 158(b)(4), that would have prohibited certain handbill-ing by unions. See 485 U.S. at 588, 108 S.Ct. at 1404. Although the NLRB was entitled to Chevron deference, id. at 574, 108 S.Ct. at 1397, the Court refused to defer to the NLRB because its interpretation raised serious First Amendment questions.
The Court reached a different conclusion in Rust v. Sullivan, 500 U.S. 173, 111 S.Ct. 1759, 114 L.Ed.2d 233 (1991). At issue there were Department of Health and Human Services regulations issued pursuant to Title X of the Public Health Service Act, 42 U.S.C. §§ 300 to 300a-6, that prohibited federally funded family planning clinics from providing abortion counseling. Although the regulations were challenged on constitutional grounds, the Court applied Chevron, upholding the regulations as a reasonable interpretation of an ambiguous statute, and proceeded to address the constitutional question on the merits. 500 U.S. at 191-92, 111 S.Ct. at 1771-72.
We do not read Rust as overruling DeBar-tolo. Rather, the Rust majority simply did not find the constitutional doubts raised by the regulations to be “grave and doubtful” enough to defeat the presumption of Chevron deference. 500 U.S. at 191, 111 S.Ct. at 1771. As Professor Davis has noted, “All nine Justices [in Rust ] agreed that ‘a statute must be construed, if fairly possible, to avoid not only the conclusion that it is unconstitutional but also grave doubts on that score.’ The Justices divided 5 to 4, however, on the issue of whether the agency’s construction ... raised ‘the sort of grave and doubtful constitutional questions ... that would lead us to assume that Congress did not intend to authorize’ the construction announced in the [662]*662agency rule.” 1 Davis, Administrative Law § 3.6, at 129 (quoting Rust, 500 U.S. at 190-91, 111 S.Ct. at 1770-71) (citations omitted). Equally important, the Rust majority noted that applying DeBartolo would effectively eliminate Chevron deference for all regulations touching on abortion. See 500 U.S. at 191, 111 S.Ct. at 1771 (“The extensive litigation regarding governmental restrictions on abortion since our decision in Roe v. Wade, 410 U.S. 113, 93 S.Ct. 705, 35 L.Ed.2d 147 (1973), suggests that it was likely that any set of regulations promulgated by the Secretary ... would be challenged on constitutional grounds.”).
Rust and DeBartolo, read together, require courts to scrutinize constitutional objections to a particular agency interpretation skeptically. Only if the agency’s proffered interpretation raises serious constitutional concerns may a court refuse to defer under Chevron. This is the approach taken by one of our sister circuits. Compare Chamber of Commerce v. Federal Election Com’n, 69 F.3d 600, 605 (D.C.Cir.1995) (declining to accord Chevron deference to a Federal Election Commission regulation because it would raise troubling constitutional questions) with Republican Nat’l Comm. v. Federal Election Comm’n, 76 F.3d 400, 409 (D.C.Cir.1996) (according Chevron deference because “we can easily resolve the [constitutional questions] through the application of controlling precedent”), cert. denied, — U.S. -, 117 S.Ct. 682, 136 L.Ed.2d 607 (1997).
This reading of DeBartolo and Rust does not offend Chevron’s underlying principles and preserves the balance of power between the judicial and the political branches. First, an agency is entitled to Chevron deference largely because Congress has delegated interpretive authority to it. See Pauley v. BethEnergy Mines, 501 U.S. 680, 696, 111 S.Ct. 2524, 2533, 115 L.Ed.2d 604 (1991). But, just as we will not infer from an ambiguous statute that Congress meant to encroach on constitutional boundaries, we will not presume from ambiguous language that Congress intended to authorize an agency to do so. At the core of DeBartolo lies the presumption that, if Congress means to push the constitutional envelope, it must do so explicitly. See 485 U.S. at 575, 108 S.Ct. at 1397-98.
Second, an agency’s interpretation is generally accorded Chevron deference because the agency has superior expertise in the particular area. When agencies adopt a constitutionally troubling interpretation, however, we can be confident that they not only lacked the expertise to evaluate the constitutional problems, but probably didn’t consider them at all. See, e.g., Gilbert v. NTSB, 80 F.3d 364, 366-67 (9th Cir.1996) (“Generally, challenges to the constitutionality of a statute or a regulation promulgated by an agency are beyond the power or jurisdiction of an agency.”). That was the case here, as the IBIA declared itself unable to address the constitutional arguments raised against its interpretation. Reindeer Herders Assn., 23 IBIA at 48. Judges, by contrast, specialize in “say[ing] what the law is.” Marbury v. Madison, 5 U.S. (1 Cranch) 137, 176, 2 L.Ed. 60 (1803). The balance of expertise therefore shifts against judicial deference to agency interpretations when a constitutional line is about to be crossed. See The Supreme Court, 1990 Term-Leading Cases, 105 Harv. L.Rev. 177, 398 (1991) (“When constitutional rights are implicated ..., the balance of values clearly shifts against agency deference.”).
Third, the DeBartolo rule doesn’t usurp the policymaking powers of the political branches. When courts attempt to give meaning to a hopelessly ambiguous statute using tools of statutory interpretation, they often do engage in camouflaged policymak-ing. See 1 Davis, Administrative Law § 3.6, at 130 (“It is the very indeterminacy of the ‘traditional tools’ that gives judges the discretion to make policy decisions through the process of statutory construction.”). But when a court construes a statute so as to avoid a difficult constitutional question, it is not making a policy choice; rather, it is practicing constitutional narrowing. See United States v. X-Citement Video, 982 F.2d 1285, 1295 n. 6 (9th Cir.1992) (Kozinski, J., dissenting in part) (“Statutory interpretation is an attempt to divine the meaning of the statute as [enacted]. Constitutional narrowing seeks to add a constraint to the statute that its drafters plainly had not meant to put [663]*663there.”). The agency remains free to adopt any interpretation that doesn’t come perilously close to the constitutional boundary. And, of course, Congress may still force the issue by removing the statutory ambiguity and making it clear that it chooses the constitutionally doubtful interpretation-in which event, the courts will have to confront the constitutional question squarely.
To be sure, DeBartolo does constrain agency power by precluding some policy options because they raise serious constitutional questions, even though they may ultimately turn out to be constitutional. Cf William N. Eskridge, Jr. & Phillip P. Frickey, The Supreme Court 1993 Term, Forword: Law As Equilibrium, 108 Harv.L.Rev. 26, 85 (1994) (constitutional narrowing creates the possibility of “stealth constitutionalism” by which judges enforce constitutional norms without having to actually articulate them); Richard A. Posner, Statutory Interpretation-in the Classroom and in the Courtroom, 50 U.Chi.L.Rev. 800, 816 (1983) (the canon allows the creation of “a judge-made constitutional ‘penumbra’”). Rust, however, limits this intrusion on agency power to situations where it’s absolutely necessary. According to Rust, constitutional narrowing should displace Chevron only when the constitutional problems are truly “grave” and never when it would effectively preclude all policy options because all possible interpretations raise constitutional problems.
Whether the IBIA’s interpretation here is entitled to Chevron deference therefore turns on the seriousness of the constitutional doubts it raises.5
Y
Appellants claim that the Reindeer Act, as construed by the IBIA, violates equal protection; they make two arguments. First, they contend that the IBIA’s interpretation can’t be upheld even under the pro-native standard of Morton v. Mancari, 417 U.S. 535, 94 S.Ct. 2474, 41 L.Ed.2d 290 (1974). In any event, they argue, Mancari was overruled by Adarand Constructors, Inc. v. Pena, 515 U.S. 200, 115 S.Ct. 2097, 132 L.Ed.2d 158 (1995), and the IBIA’s interpretation cannot possibly survive Adarand's strict scrutiny.
In Mancari, the Supreme Court rejected an equal protection challenge to a statutory hiring preference for Indians in the Bureau of Indian Affairs (BIA). The Court held that statutory preferences for Indians were “political” not “racial” classifications, 417 U.S. at 553 n. 24, 94 S.Ct. at 2484 n. 24, and would be upheld “[a]s long as the special treatment can be tied rationally to the fulfillment of Congress’ unique obligation toward the Indians....” Id. at 555, 94 S.Ct. at 2485. Although the Court has never overturned a statute or treaty affecting Indians or natives since Mancari, see Stuart Minor Benjamin, Equal Protection and the Special Relationship: The Case of Native Hawaiians, 106 Yale L.J. 537, 548 (1996); Felix S. Cohen, Handbook of Federal Indian Law 657 (2d ed. 1982), the IBIA’s interpretation makes the Reindeer Act very different from a lot of other legislation pertaining to Native Americans.
The preference at issue in Mancari only applied to the BIA, an agency created for the purpose of serving Indians. In approving the preference, the Court noted that “a blanket exemption for Indians from all civil service examinations” would pose an “obviously more difficult question.” See 417 U.S. at 554, [664]*66494 S.Ct. at 2484. While Mancari did not have to confront the question of a naked preference for Indians unrelated to unique Indian concerns, the IBIA’s interpretation of the Reindeer Act would force us to confront that very issue. According to the IBIA, the Reindeer Act provides a preference in an industry that is not uniquely native, whether the beneficiaries live in a remote native village on the Seward Peninsula or in downtown Anchorage. The Act in no way relates to native land, tribal or communal status, or culture.
The government, intervenors and amici have been unable to cite another law that gives natives so broad a preference.6 They point out that Congress has exempted Alaskan natives from certain prohibitions against hunting and fishing.7 Unlike raising livestock, however, hunting and fishing wild game is an integral and time-honored part of native subsistence culture, see David S. Case, Alaska Natives and American Laws 276 (1984) (subsistence hunting and fishing “are intricately woven into the fabric of [native] social, psychological, and religious life”), and all of these exemptions only permit taking of animals for subsistence or cultural uses. See 16 U.S.C. §§ 712(1), 1153, 1371(b), 1539(e).
Legislation that relates to Indian land, tribal status, self-government or culture passes Mancari’s rational relation test because “such regulation is rooted in the unique status of Indians as ‘a separate people’ with their own political institutions.” United States v. Antelope, 430 U.S. 641, 646, 97 S.Ct. 1395, 1399, 51 L.Ed.2d 701 (1977). As “a separate people,” Indians have a right to expect some special protection for their land, political institutions (whether tribes or native villages), and culture. We therefore have upheld statutes that provide Indians with special fishing rights that were promised tó them by treaty, United States v. Decker, 600 F.2d 733, 740-41 (9th Cir.1979), and subsidies for “low-income housing in remote Alaskan villages,” Alaska Chapter, Assoc. Gen. Contr. v. Pierce, 694 F.2d 1162, 1163 (9th Cir.1982).
As in Alaska Chapter, id. at 1167, we can discern Mancari’s scope by looking to the [665]*665cases it cited as examples of permissible special treatment for Indians, see 417 U.S. at 555, 94 S.Ct. at 2485: Each case dealt with life in the immediate vicinity of Indian land. E.g., Morton v. Ruiz, 415 U.S. 199, 94 S.Ct. 1055, 39 L.Ed.2d 270 (1974) (providing welfare benefits to Indians, but only those who live “on or near” reservations); McClanahan v. Arizona State Tax Comm’n, 411 U.S. 164, 93 S.Ct. 1257, 36 L.Ed.2d 129 (1973) (tax exemption for income derived wholly from reservation sources); Simmons v. Eagle Seelatsee, 384 U.S. 209, 86 S.Ct. 1459, 16 L.Ed.2d 480 (1966), aff'g 244 F.Supp. 808 (E.D.Wash.1965) (limiting the right to inherit reservation land only to Indians); Williams v. Lee, 358 U.S. 217, 79 S.Ct. 269, 3 L.Ed.2d 251 (1959) (tribal courts and their jurisdiction over reservation affairs).
While Mancari is not necessarily limited to statutes that give special treatment to Indians on Indian land, we do read it as shielding only those statutes that affect uniquely Indian interests. See, e.g., Antelope, 430 U.S. at 646, 97 S.Ct. at 1398 (“Both Mancari and Fisher [ v. District Court, 424 U.S. 382, 96 S.Ct. 943, 47 L.Ed.2d 106 (1976),] involved preferences or disabilities directly promoting Indian interests in self-government, whereas in the present case we are dealing, not with matters of tribal self-regulation, but with federal regulation of criminal conduct within Indian country implicating Indian interests.'” (emphasis added)). For example, we seriously doubt that Congress could give Indians a complete monopoly on the casino industry or on Space Shuttle contracts. At oral argument, counsel for the government conceded that granting natives a monopoly on all Space Shuttle contracts would not pass Mancari’s rational-relation test. Counsel could only distinguish the Space Shuttle preference from a reindeer preference by noting that, in 1937, natives were heavily involved in the reindeer business whereas they aren’t involved in the Space Program. The casino example defies this distinction, but is equally unrelated to “Congress’ unique obligation toward the Indians.” Mancari, 417 U.S. at 555, 94 S.Ct. at 2485.
The Supreme Court’s recent decision in Adarand only adds to our constitutional doubts about the IBIA’s interpretation of the Reindeer Act. In Adarand, the Court ruled that racial classifications by the federal government are subject to strict scrutiny. The Court overruled Metro Broadcasting, Inc. v. FCC, 497 U.S. 547, 110 S.Ct. 2997, 111 L.Ed.2d 445 (1990), in part because its application of intermediate scrutiny to federal racial classifications was inconsistent with the strict scrutiny applied to state classifications. See Adarand, 515 U.S. at 225-27, 115 S.Ct. at 2112. Although the majority emphasized that it was only overruling Metro Broadcasting, id. at 233-35, 115 S.Ct. at 2116, Justice Stevens in dissent argued that the majority’s “concept of ‘consistency [in equal protection jurisprudence] ... would view the special preferences that the National Government has provided to Native Americans since 1834 as comparable to the official discrimination against African Americans that was prevalent for much of our history.” Id. at 244, 115 S.Ct. at 2121 (Stevens, J., dissenting) (footnote citing Mancari omitted). If Justice Stevens is right about the logical implications of Adarand, Mancari’s days are numbered. See Benjamin, 106 Yale L.J. at 567.
Under strict scrutiny, the IBIA’s interpretation would almost certainly render the Reindeer Act unconstitutional. Assuming that Congress has a compelling interest in assisting natives economically,8 the IBIA’s interpretation precludes any chance of defending the Reindeer Act as narrowly tailored to serve that interest. A complete race-based ban is the broadest possible remedy. Unlike a subsidy, set-aside or even a quota, an absolute ban deprives the disfa[666]*666vored racial group of all opportunity to participate, thus placing a tremendous burden on innocent third parties. This makes the remedy almost by definition not narrowly tailored. See, e.g., United States v. Paradise, 480 U.S. 149, 182, 107 S.Ct. 1053, 1072, 94 L.Ed.2d 203 (1987) (plurality opinion) (noting that a narrowly tailored employment preference was not “an ‘absolute bar’ to white advancement”); Sheet Metal Workers v. EEOC, 478 U.S. 421, 481, 106 S.Ct. 3019, 3053, 92 L.Ed.2d 344 (1986) (“While white applicants for union membership may be denied certain benefits available to their nonwhite counterparts, the court’s orders do not stand as an absolute bar-”). Moreover, non-natives have been effectively excluded from the reindeer industry for almost 60 years. See pp. 662-63 supra. While a ban on non-native participation in the reindeer industry may have been narrowly tailored in 1937, it may no longer be today. See Adarand, 515 U.S. at 238, 115 S.Ct. at 2118 (a race-conscious program must be “appropriately limited such that it “will not last longer than the discriminatory effects it is designed to eliminate’”) (quoting Fullilove v. Klutznick, 448 U.S. 448, 513, 100 S.Ct. 2758, 2792-93, 65 L.Ed.2d 902 (1980) (Powell, J., concurring)). This is especially true where the economic and subsistence needs of natives have changed dramatically in the interim-most notably, by passage of the Alaska Native Claims Settlement Act in 1971, which gave natives 44 million acres of land, almost $1 billion, and shares of stock in newly-created native corporations. See H.R.Rep. No. 92-523 (1971), reprinted in 1971 U.S.C.C.A.N. 2192, 2193; 43 U.S.C. § 1601 et seq. (ANSCA); Bay View, Inc. v. Ahtna, Inc., 105 F.3d 1281, 1283 (9th Cir.1997).
Finally, a race-conscious remedy will not be deemed narrowly tailored until less sweeping alternatives-partieularly race neutral ones-have been considered and tried. See Associated Gen. Contractors v. City & County of S.F., 813 F.2d 922, 939 (9th Cir.1987); see also Adarand, 515 U.S. at 237, 115 S.Ct. at 2118. We note, for example, that Congress recently preserved native subsistence hunting and fishing rights by enacting a race-neutral preference for “rural” Alaskans, regardless of race. See Alaska National Interest Lands Conservation Act of 1980 (ANILCA), 16 U.S.C. § 3111 et seq.; Village of Gambell v. Clark, 746 F.2d 572, 581 (9th Cir.1984).
The constitutional questions raised by the IBIA’s interpretation are grave and, as intervenors and amici point out, implicate an entire title of the United States Code. We see no reason to unnecessarily resolve them when a less constitutionally troubling construction is readily available. We therefore interpret the Reindeer Act as not precluding non-natives in Alaska from owning and importing reindeer.9
The Reindeer Act erects a number of barriers to non-native participation in the reindeer business, but it does not prohibit nonnatives in Alaska from owning, importing or selling reindeer. Congress, of course, is free to adopt the IBIA’s interpretation of the Reindeer Act, in which case we would be required to confront explicitly the difficult constitutional questions outlined above. Meanwhile, the decision of the district court is REVERSED and the case is remanded for proceedings consistent with this opinion.