ON PETITION FOR REHEARING
Before BAUER, Chief Judge, CUMMINGS, CUDAHY, POSNER, COFFEY, FLAUM, EASTERBROOK, RIPPLE, MANION, and KANNE, Circuit Judges, and ESCHBACH, Senior Circuit Judge.On January 30, 1987, the defendants-appellants filed a petition for rehearing with suggestion of rehearing en bane. A majority of the panel voted to deny the petition for rehearing. Judge ESCHBACH voted to grant the petition. The petition is accordingly denied. A judge in regular active service requested a vote on the suggestion of rehearing en banc. In light of the amended opinion of the panel filed today, the suggestion of rehearing en banc did not secure a majority.
POSNER, COFFEY, EASTERBROOK, and MANION, Circuit Judges, voted to grant rehearing en banc. WOOD, Circuit Judge, did not participate in the consideration or decision of this case. EASTERBROOK, Circuit Judge, with whom POSNER, COFFEY, and MANION, Circuit Judges, join, dissenting from the denial of rehearing en banc.*The panel’s opinion, as amended, holds that Ill.Rev.Stat. ch. 8 § 37-9(c) does not authorize the Illinois Racing Board to conduct administrative searches of living cubicles at race tracks. Then it declares Thoroughbred Rule 322 and Harness Racing Rule 25.19 unconstitutional on two grounds: warrantless searches unauthorized by statute bear a special burden of justification, and the regulations do not contain standards to guide the discretion of the administrative officials. All of the track’s backstretchers consented to the searches, but the court says that the consents are invalid because the state did not have the authority to search over objection. Finally, the court deals with searches of the backstretchers’ persons at the track. Having made so much of its conclusion that § 37-9(c) authorizes the search of business premises but not living cubicles, the panel nonetheless holds that “the same rules apply” to searches conducted on the business premises.
If the panel had said: “Searches of living quarters and persons require either a warrant or some criteria limiting the discretion of the officers, criteria Illinois does not supply”, this would be a plausible though problematic disposition. New York v. Burger, — U.S. -, 107 S.Ct. 2636, 96 L.Ed.2d 601 (1987), holds that police may search regulated businesses (in Burger auto junkyards) without a warrant, a regular pattern, or any announced criteria. The Court rejected a claim that the searches must be predictable, pointing out that to the extent people can tell when the police will arrive, they can use that knowledge to hide evidence of wrongdoing. — U.S. at -n. 2,---& nn. 21, 22, 107 S.Ct. at 2639 n. 2, 2648 & nn. 21-22. Burger does not use the approach of Bionic Auto Parts & Sales, Inc. v. Fahner, 721 F.2d 1072 (7th Cir.1983), another junkyard case on which the panel heavily relied. See 808 F.2d 601, 604-06. Bionic said that each inspection must be justified and conducted to “minimize the dangers” of random searches (721 F.2d at 1079); Burger held that only the program of searches requires justification, and that particular searches do not need additional support. But Burger does not deal with searches of persons and living quarters located on business *35premises, and sooner or later the Court will have to do so. The Board can amend its rules and avoid the difficulty, or press on to the only Court that can resolve the issue. We could add little by trying to apply Burger and similar cases to the searches of living quarters and backstretchers.
But the panel did not stop with the observation that searches of persons and their living quarters are different from searches of the rest of the business premises; it did not even start there. It started by making an independent decision on a question of state law, as if the state were just another litigant. It used the conclusion about state law as a basis of its constitutional decision. This approach is highly questionable. So is the panel’s treatment of consent. The panel’s approach to the interpretation of state laws could govern many cases, as would its handling of consent. Long after Ill.Rev. Stat. ch. 8 § 37-9(c), Thoroughbred Rule 322, and Harness Racing Rule 25.19 have been amended or forgotten, we will have to live with the principles the majority used. These principles deserve more attention than they have received.
1. If the backstretchers had filed a suit under the diversity jurisdiction seeking judicial review of the rules on the ground that they are unauthorized by statute, the suit would have been dismissed because the eleventh amendment deprives the district court of authority to adjudicate such suits. If the backstretchers had filed a suit under 42 U.S.C. § 1983 and added a pendent claim under state law, they still would have lost. Pennhurst State School & Hospital v. Halderman, 465 U.S. 89, 104 S.Ct. 900, 79 L.Ed.2d 67 (1984) (Pennhurst II), holds that a federal court may not award relief against a state on the basis of state law. As the Court said, “it is difficult to think of a greater intrusion on state sovereignty than when a federal court instructs state officials on how to conform their conduct to state law.” 465 U.S. at 106, 104 S.Ct. at 911. Yet that is exactly what the panel has done. It has concluded that the Board does not understand state law and used that as the springboard of its constitutional holding.
The panel’s award of relief was not based directly on state law. So the panel may take comfort from Ex parte Young, 209 U.S. 123, 28 S.Ct. 441, 52 L.Ed. 714 (1908). But Pennhurst II depends not on the eleventh amendment but on principles of immunity that have developed in the shadow of that amendment. It establishes the proper role of federal courts in telling state governments the meaning of state law. That a federal court possesses constitutional power to revise a state’s view of the state’s law does not imply that the court should do so. Four justices, dissenting in Pennhurst II, argued that a state, with its choice of poison, would prefer to lose on state rather than constitutional grounds, and that a federal court does well to avoid the constitutional issue. 465 U.S. at 159-63, 104 S.Ct. at 939-41 (Stevens, J., dissenting). The Court nonetheless rejected this position. No justice in Pennhurst II suggested that a court should use state statutory grounds to create a constitutional problem on which the state then would lose. In Pennhurst II resort to state grounds would have obviated a federal issue; in this case, the panel’s holding on state law set the state up for a fall on a federal issue.
There are at least five ways to find out what state law means in a case like ours. One is to accept the view of the executive branch of the state. A second is to certify the question to the supreme court of the state. A third is to abstain, as Judge Eschbach urged in dissent from the panel’s opinion. A fourth is to review the issue of state law with at least the deference given the statutory interpretations of federal agencies. A fifth is to decide the meaning of the law de novo, as if this were a dispute between private parties. The panel ignored three of these methods and brushed aside abstention, proceeding to give its views on the meaning of state law unencumbered by deference to the state’s interpretation. This is no way to treat state governments.
I suggested in Huggins v. Isenbarger, 798 F.2d 203, 207-09 (7th Cir.1986) (concurring opinion), that the best approach is the first: to accept the position of the State of *36Illinois on the meaning of the state’s law. The Illinois Racing Board interpreted Ill. Rev.Stat. ch. 8 § 37-9(c) when it issued its rules. The Attorney General of Illinois filed a brief in this court representing that the Board’s construction of state law is correct. No state court has questioned this. The State of Illinois thus has presented us with an authoritative construction of its law. If the state had spoken through the Supreme Court of Illinois, we would treat the court’s interpretation as beyond our ken; we would accept it merely because the court had said it. Why should we listen only to state courts and ignore the views of other officials of the state? Many constructions of law come from courts, but courts pronounce only when necessary to decide cases; executive officials construe laws for other purposes, cf. Carson v. Block, 790 F.2d 562, 565 (7th Cir.1986); Mother Goose Nursery Schools, Inc. v. Sendak, 770 F.2d 668 (7th Cir.1985). When executive officials are authorized to construe the law, they speak for the state as authoritatively as courts do. Unless a federal court may choose who speaks for the state, the court ought to respect the views of whoever has spoken for the state. Cf. Hilton v. Braunskill, — U.S. —, 107 S.Ct. 2113, 2120, 95 L.Ed.2d 724 (1987); Ohio Bureau of Employment Services v. Hodory, 431 U.S. 471, 477-80, 97 S.Ct. 1898, 1902-04, 52 L.Ed.2d 513 (1977); Barrera v. Young, 794 F.2d 1264, 1269 (7th Cir.1986). How Illinois apportions governmental powers, including the power to construe statutes, is none of our concern. Whalen v. United States, 445 U.S. 684, 689 n. 4, 100 S.Ct. 1432, 1436 n. 4, 63 L.Ed.2d 715 (1980); Mayor of Philadelphia v. Educational Equality League, 415 U.S. 605, 615 n. 13, 94 S.Ct. 1323, 1330 n. 13, 39 L.Ed.2d 630 (1974); Highland Farms Dairy, Inc. v. Agnew, 300 U.S. 608, 612, 57 S.Ct. 549, 551, 81 L.Ed. 835 (1937); Prentis v. Atlantic Coast Line Co., 211 U.S. 210, 225, 29 S.Ct. 67, 69, 53 L.Ed. 150 (1908); Dreyer v. Illinois, 187 U.S. 71, 84, 23 S.Ct. 28, 32, 47 L.Ed. 79 (1902); United Beverage Co. v. Indiana Alcoholic Beverage Commission, 760 F.2d 155 (7th Cir.1985). Cf. City of Newport v. Iacobucci, — U.S. —, 107 S.Ct. 383, 385-86, 93 L.Ed.2d 334 (1986). Illinois has supplied us with an interpretation of its statute, which we should accept unless set aside by processes sufficient under the law of Illinois.
Perhaps the Attorney General is not authorized by state law to speak for Illinois. We should give the views of the executive branch the sort of respect provided by state law. National Surety Corp. v. Midland Bank, 551 F.2d 21, 26 (3d Cir.1977). Thus if the Attorney General’s status as an advocate diminishes the force of his views, or if the Attorney General’s view is trumped by that of a court, we should respect that allocation of governmental powers. Huggins, 798 F.2d at 208-09. But it is commonplace for courts to defer to the views of federal agencies and of the Solicitor General of the United States— even when those views are advanced during litigation. E.g., Japan Whaling Ass’n v. American Cetacean Society, — U.S. —, 106 S.Ct. 2860, 2867-68, 92 L.Ed.2d 166 (1986); Haig v. Agee, 453 U.S. 280, 291, 101 S.Ct. 2766, 2773, 69 L.Ed.2d 640 (1981). States, which need not adopt the same separation of powers found within the federal government, may give greater force to statutory interpretations of executive officials. The parties have not addressed the extent to which the Board’s, and the Attorney General’s, construction of § 37-9(c) is authoritative under state law. We certainly ought not assume, as the panel does, that it is worthless. In Pennhurst II the plaintiffs relied on an opinion of the Supreme Court of Pennsylvania, which the court of appeals held showed that the legal position of the executive branch of Pennsylvania was untenable. The Supreme Court held that even the views of the highest court of Pennsylvania did not allow a federal court to override the executive branch’s construction of Pennsylvania law. Why may we disagree with the executive branch of Illinois when no state court has spoken?
Certification of the state law question would be one way to avoid this difficulty— at least when the eleventh amendment is not in play, see Citizens for John W. *37Moore Party v. Board of Election Commissioners, 781 F.2d 581, 584-86 (7th Cir. 1986) (dissenting opinion) — but the panel did not certify the question. Abstention is another way to obtain the views of the state courts. The panel in our case declined to abstain, pointing out that abstention disrupts the progress of the case and is not a sound way to proceed when raised belatedly. Yet the high costs of abstention do not explain why the panel did not certify the question. Certification entails neither the costs nor the delay associated with abstention and is appropriate when the statute is susceptible of multiple interpretations. Houston v. Hill, — U.S.—,— —, 107 S.Ct. 2502, 2512-15, 96 L.Ed.2d 398 (1987). Surely either is preferable to holding an entire administrative scheme unconstitutional because unsupported by state law. Even belated abstention is attractive if the alternative is the federal court’s substitution of its judgment for the state officials’.
There is one more option: deference to the state agency’s construction of state law, as we would defer to a federal agency’s construction of federal law. If the Board’s rules had been issued by a federal agency, we would have asked not whether the construction is right but whether it is reasonable. See Chevron U.S.A. Inc. v. Natural Resources Defense Council, Inc., 467 U.S. 837, 842-45, 862-66, 104 S.Ct. 2778, 2781-83, 2791-93, 81 L.Ed.2d 694 (1984); Watkins v. Blinzinger, 789 F.2d 474, 478 (7th Cir.1986). The regulations were promulgated under a law giving the Board authority to inspect the “race track, facilities and other places of business” of the licensees. The panel says that the Board may not inspect the living cubicles at the track because “other places of business” “effectively defines the earlier listed places as places of business.” This is the eiusdem generis approach. A reasonable person might deny the applicability or force of this saw, see Harrison v. PPG Industries, Inc., 446 U.S. 578, 587-89, 100 S.Ct. 1889,1895-96, 64 L.Ed.2d 525 (1980); United States v. Turkette, 452 U.S. 576, 581-82, 101 S.Ct. 2524, 2527-28, 69 L.Ed.2d 246 (1981), but even if applicable this canon does not answer all questions. The cubicles are part o/the “race track [and] facilities” of the racing licensee, and they promote the operation of the racing business. Licensees make quarters available at the track for their convenience, not because they want to operate apartment houses. The backstretchers may regard the cubicles as residences, but the licensees regard them as part of the track complex. The statute authorizes the Board to inspect the facilities, track, and business of the licensees; that the licensee’s place of business is someone else’s hotel room does not necessarily confine the Board’s statutory power. Cf. United States v. Cerri, 753 F.2d 61 (7th Cir.1985) (agents may search a gun dealer’s home without a warrant, if he chooses to do business at home). As in Chevron, the legislature does not appear to have considered, and therefore it has not settled, the problem at hand.
There is, moreover, a different way to read the statute: it authorizes inspection of the track proper, and of “other places of business” away from the track. Racing licensees do not use the tracks year ’round; sometimes more than one licensee uses a track; every licensee has some place of business away from the track. The function of the “other places of business” language then is not to close some portions of race tracks to the Board but to ensure that the Board can follow the business wherever the licensee goes. This reading is consistent with the reasons the Board is authorized to inspect. Papers suggesting improprieties may be hidden anywhere the licensee may be found; drugs given to horses may be hidden at the track and elsewhere. It would be most surprising if the Board may not inspect at least every comer of the race track proper for drugs, forbidden implements, and suspicious papers. Yet under the panel’s decision, the licensee can put part of the track off limits by the expedient of inviting an employee to sleep there. That is not an inevitable reading of the statute, one so compelling that we would say that a federal agency exceeded its power in reading the statute to embrace the whole track. The panel treated the *38meaning of “race track [and] facilities” as a pure question of law on which it could take a clear shot. It is not appropriate for a federal court to give less deference to a state agency’s interpretation of a state statute than to a federal agency’s interpretation of a federal statute; the distinction cuts the other way.
2. The panel scrutinized the statutory authority for the Board’s rules because it believed that searches are more readily sustained if conducted on statutory authority. Maybe so for searches by federal officials, because a federal court should respect Congress’ decision that a category of searches is “reasonable”. United States v. Watson, 428 U.S. 411, 416, 96 S.Ct. 820, 824, 46 L.Ed.2d 598 (1976). But because states need not observe the separation of functions that prevails within the federal government, they may entrust to executive officials the task of deciding for the political branches what is reasonable. No case I have found even hints that a search by a state official, under color of state regulations, is any different for constitutional purposes from a search under color of a state statute; indeed the distinction between “statute” and “regulation” presupposes a separation of functions that states are free to modify. Many cases sustain state searches, conducted on administrative, regulatory, or no authorization, without suggesting that a statute would have supplied a firmer base. E.g., O’Connor v. Ortega, — U.S. —, 107 S.Ct. 1492, 94 L.Ed.2d 714 (1987); Colorado v. Bertine, — U.S.—, 107 S.Ct. 738, 93 L.Ed.2d 739 (1987). See also McDonell v. Hunter, 809 F.2d 1302 (8th Cir.1987), and Shoemaker v. Handel, 795 F.2d 1136 (3d Cir.1986), both sustaining administrative searches that were supported entirely by regulations. The fourth amendment does not allow administrative officials to issue warrants but is otherwise silent on who makes policy for Illinois concerning administrative searches. Illinois v. Krull, — U.S. —, 107 S.Ct. 1160, 94 L.Ed.2d 364 (1987), holds that evidence gathered under authority of a state statute later declared unconstitutional may be used in a criminal case, but this is based on considerations peculiar to the exclusionary rule rather than on a belief that legislation is better than a regulation at declaring the policy of Illinois about the propriety of administrative searches. State regulations and state statutes should have equal weight when the question is: “what does society believe is a ‘reasonable’ administrative search?”.
At all events, the searches of the back-stretchers’ persons at the track are authorized by both statute and regulation; the track (outside the backstretchers’ cubicles) is a place of business of the licensee, so the regulation is authorized even on the panel’s treatment of § 37-9(c). This search has all the support the State of Illinois as a whole can furnish. The court has necessarily declared § 37-9(c) unconstitutional as applied to personal searches at the track.
3. The backstretchers were required to consent to searches as a condition of their employment. The panel’s approach to these consents is to say that because the state cannot search the backstretchers’ cubicles or persons against their will, the state cannot require consent either. This has the curious consequence that consent is valid whenever it is not needed (because the state may conduct the search without consent) and invalid whenever it is necessary (because the state is forbidden to search over objection). This eliminates consent as a ground for search.
Although the panel does not articulate its rationale, it must be making an “unconstitutional conditions” argument. The state did not ask for consent, as in Schneckloth v. Bustamonte, 412 U.S. 218, 93 S.Ct. 2041, 36 L.Ed.2d 854 (1973). It made consent a condition of employment. “Consent” extracted by threat of a violation of one’s constitutional rights is not effective; it is no different from the proposal “your money or your life”, because either option makes the person worse off. Another panel of the court recently held that “consent” extracted by a threat of disbarment is valid, see Lewis v. Lane, 816 F.2d 1165, 1169 (7th Cir.1987), and I wonder how these decisions may be reconciled, but that is not my principal concern.
*39Ours is not a simple “unconstitutional condition”. The panel did not hold that administrative searches of backstretchers’ quarters and persons always violate the fourth amendment. It has held only that the searches are unauthorized by statute and that the regulations are (so far) insufficiently detailed. There has never been a doctrine of “unstatutory conditions” or “insufficiently circumscribed regulatory conditions”. Why can’t people be asked to consent to a kind of search that the statute has not yet authorized?
Moreover, the state demands consent only from those who live or work at the track. Employees are free to live elsewhere and avoid searches of their quarters. This is one of the grounds on which courts sustain airport searches: you can’t board a plane without consenting to a search, but you can travel by car or train if you like. So too at the track. The demand is not unconditional; the employee controls the security of his quarters by his choice of abode. The panel’s decision casts a pall over all consents in which the choice is genuine because the person has a right to say no by choosing another line of work, another place to live, a different mode of travel, and so on.
Some recent cases call the unconstitutional conditions doctrine itself into question. Snepp v. United States, 444 U.S. 507, 509 n. 3, 100 S.Ct. 763, 765 n. 3, 62 L.Ed.2d 704 (1980), holds that an employee may surrender by contract his first amendment right tó speak; the Court did not think it important that the consent was required as a condition of employment. Buckley v. Valeo, 424 U.S. 1, 54-58 & n. 62, 96 S.Ct. 612, 651-53 & n. 62, 46 L.Ed.2d 659 (1976), holds that the government may condition monetary support for political campaigns on a surrender of the constitutional right to spend unlimited sums for speech. Posadas de Puerto Rico Associates v. Tourism Co. of Puerto Rico, — U.S. —, 106 S.Ct. 2968, 2979, 92 L.Ed.2d 266 (1986), uses the principle that the greater power (to ban gambling) includes the lesser power (to condition a gambling permit on surrender of some first amendment rights). “The greater power includes the lesser” is the traditional antagonist of the “unconstitutional conditions” principle. An inferior federal court may not proceed as if the doctrine of unconstitutional conditions were in perfect health.
Neither the panel’s proposition that consent is ineffectual when the government lacks the power to impose its will over objection, nor the contrasting view of Lewis that knuckling under to a show of authority is voluntary, is very attractive. To determine whether acquiescence in the face of a demand is significant, we must evaluate the nature and strength of the reasons for the demand (as the Supreme Court did in Snepp), the options open to the person faced with the demand (here to obtain quarters off the track’s premises or to change jobs), the extent to which the scope of any consent is reasonable in light of the purposes to be served, and so on. Many cases say that the government may demand consent when it has strong reasons. The airport search cases are good examples. Searches at race tracks also vindicate important interests. The panel did not deny that the government has a substantial interest in keeping drugs away from horses (and jockeys, see Shoemaker, 795 F.2d at 1141-43); it apparently did not believe that the strength of the state’s interest is relevant. When there are strong reasons for conducting a search, when the demand affects only a tiny portion of the jobs available in the state (so that saying no and changing jobs is a real option), when the consent approves a search that fits the need like a glove — in short, in this case— the state may use the consent even if it may not act over objection.
If the backstretchers’ consents are valid, then the state may carry out its searches even if the Board’s regulations do not sufficiently confine the agents’ discretion. And if these regulations are inconsistent with the fourth amendment, our court ought to give the right reasons for that conclusion. The panel’s opinion does not give Illinois the deference in the interpretation of state law that is its due, and the panel’s preference for legislation over regulation requires the state to conform its governance *40to the panel’s views of how states ought to be organized. The questions of principle glossed over by the panel’s opinion are far more important than the outcome of this case, and they are worth the extra judicial time necessary to get them right.