In 1999 the Chicago Board of Education brought this suit for copyright infringement against a Chicago public school teacher named Schmidt and a corporation that owns a local newspaper called Substance, aimed at such teachers, which Schmidt edits. (To simplify the opinion, we'll ignore the corporation.) In August 2002 the magistrate judge issued an injunction mysteriously captioned "Permanent Restraining Order," from which no appeal was taken. He issued a "Final Judgment Order" in January of the following year, awarding the school board $500 in damages but not referring to the injunction. Schmidt filed a timely notice of appeal from that order. But was that too late to challenge the "Permanent Restraining Order"? One possibility is that the order was a temporary injunction that became permanent when the "Final Judgment Order" was issued. If so, then since an appeal from a final judgment brings up for review any interlocutory order that has not become moot, Schmidt could defer his appellate challenge until the final judgment was entered. And if, as other orders in the case suggest, including the clerk's minute order recording the issuance of the "Permanent Restraining Order," the magistrate judge meant by that term "Permanent Injunction," the failure to appeal from it when it was entered was still not fatal. "[I]nterlocutory appeal of a permanent injunction is not mandatory, as the order remains part of the case and merges with the final judgment, from which an appeal can then be taken." Prudential Securities, Inc. v. Yingling, 226 F.3d 668, 671 (6th Cir.2000); see also Victor Talking Machine Co. v. George, 105 F.2d 697 (3d Cir.1939). To speak of "merger" in this context is to state a conclusion rather than give a reason. But it is the right conclusion because judicial economy is served by the consolidation of as many issues in a litigation as possible in a single appeal. That is why it is almost never mandatory to file an interlocutory appeal in order to preserve an issue for appellate review. 16 Charles Alan Wright et al., Federal Practice and Procedure § 3921, pp. 19-20 (2d ed.1996); see, e.g., Clark v. Merrill Lynch, Pierce, Fenner & Smith, Inc., 924 F.2d 550, 553 (4th Cir.1991).

The following facts are either undisputed or indisputable. At some considerable expense (more than $1 million, according to Schmidt, though the actual figure is not in the record), the school board created and copyrighted a series of standardized tests that it called the "Chicago Academic Standards Exams" (CASE). These were, in copyright lingo, "secure tests." "A secure test is a nonmarketed test administered under supervision at specified centers on specific dates, all copies of which are accounted for and either destroyed or returned to restricted locked storage following each administration. For these purposes a test is not marketed if copies are not sold but it is distributed and used in such a manner that ownership and control of copies remain with the test sponsor or publisher." 37 C.F.R. § 202.20(b)(4). To maintain secrecy, the Library of Congress does not retain a copy of such a test. § 202.20(c)(2)(vi).

The first step in making the CASE tests was to create a pool of questions — the record does not indicate how many — that would be drawn on to create the individual exams, each consisting of between three and 30 questions. The teachers who administered the exams were instructed not to make copies of them and to collect the test papers at the end of each exam so that the tests could be reused. Reuse of questions in standardized testing is not a sign of laziness but a way of validating a test, since if performance on the same questions is inconsistent from year to year this may indicate that the questions are not well designed and are therefore eliciting random answers. Educational Testing Service v. Simon, 95 F. Supp. 2d 1081, 1084 (C.D.Cal.1999); Association of American Medical Colleges v. Mikaelian, 571 F. Supp. 144, 147 (E.D.Pa.1983), aff'd by unpublished order, 734 F.2d 3 (3d Cir. 1984); Association of American Medical Colleges v. Carey, 482 F. Supp. 1358, 1367 (N.D.N.Y.1980); Robert A. Kreiss, "Copyright Fair Use of Standardized Tests," 48 Rutgers L.Rev. 1043, 1049 (1996). Such validation is of particular importance for a new battery of standardized tests, such as CASE, the subject of a three-year pilot program. And publication of standardized tests would not only prevent validation by precluding reuse of any of the questions in them, but also require the school board to create many new questions, at additional expense; and they might not be as good as the original questions, in which event there would be diminished quality as well as added cost. Hence the copyright category "secure tests."

It may seem paradoxical to allow copyright to be obtained in secret documents, but it is not. See National Conference of Bar Examiners v. Multistate Legal Studies, Inc., 692 F.2d 478, 483-87 (7th Cir.1982). For one thing, the tests are not secret from the students taking them. For another, federal copyright is now available for unpublished works that the author intends never to see the light of day. Salinger v. Random House, Inc., 811 F.2d 90, 94-97 (2d Cir.1987); see 17 U.S.C. §§ 104(a), 301(a); Harper & Row, Publishers, Inc. v. Nation Enterprises, 471 U.S. 539, 548, 105 S. Ct. 2218, 85 L. Ed. 2d 588 (1985). Most important, tests are expressive works that are not costlessly created, and the costs are greater and so the incentive to create the tests diminished if the tests cannot be reused. Educational Testing Services v. Katzman, 793 F.2d 533, 544 (3d Cir.1986); Emily Campbell, Comment, "`Testing' the Copyright Clause: Copyright Protection for Educational and Psychological Tests," 69 Neb. L.Rev. 791, 796-805 (1990). There is no analytical difference between destroying the market for a copyrighted work by producing and selling cheap copies and destroying the subsequent years' market for a standardized test by blowing its cover.

In the newspaper that he edits, Schmidt published six of the tests given in January 1999. He did this because he thought them bad tests and that he could best demonstrate this by publishing them in full. His answer to the school board's complaint asserted that the unauthorized copying and publication of the tests were a fair use and therefore not a copyright infringement. The district court dismissed the fair use defense on the pleadings. This was irregular, since no evidence had yet been presented and the defense was not meritless on its face. The judge said that Schmidt had offered "no indication that [the defendants] plan to present facts, arguments, or analyses in support of a fair use." The judge was mistaken. He had overlooked a memorandum submitted by Schmidt that attempted to do just that.

To convert the school board's motion to dismiss into a motion for summary judgment, the judge would have had to notify Schmidt of his intention and give him an opportunity to present evidence that would create a triable issue. Fed. R.Civ.P. 12(b); R.J. Corman Derailment Services, LLC v. International Union of Operating Engineers, Local Union 150, 335 F.3d 643, 649 (7th Cir.2003); Edward Gray Corp. v. National Union Fire Ins. Co., 94 F.3d 363, 366-67 (7th Cir.1996); 5A Wright et al., supra, § 1366, pp. 501-06 (1990). The judge didn't do that. But the requirement of notice has a purpose, which is to give the opponent a chance to show that he can present evidence that will defeat summary judgment. If Schmidt cannot point to any pertinent evidence that he might be able to obtain and present in opposition to summary judgment, the judge's pratfall was harmless and a remand to enable the formal conversion of the school board's motion to dismiss to a motion for summary judgment would be pointless. Loeb Industries, Inc. v. Sumitomo Corp., 306 F.3d 469, 479-80 (7th Cir. 2002); see also Ribando v. United Airlines, Inc., 200 F.3d 507, 509-10 (7th Cir. 1999); cf. Walker v. National Recovery, Inc., 200 F.3d 500, 504 (7th Cir.1999); Kingman Park Civic Ass'n v. Williams, 348 F.3d 1033, 1036, 1041-43 (D.C.Cir. 2003); Taylor v. FDIC, 132 F.3d 753, 762-63, 765-66 (D.C.Cir.1997). What is pertinent evidence depends on the applicable substantive law, here the scope of the privilege of fair use in relation to secure tests.

Each side of this case has a one-sided view of the law. The school board contends that the case is governed by Harper & Row Publishers, Inc. v. Nation Enterprises, supra, a case in which a magazine published unauthorized excerpts from Gerald Ford's copyrighted memoirs before the memoirs were published, thus jumping the gun and by doing so reducing the value of the copyright. The school board points out that Schmidt impaired the value of the exams by publishing them prematurely. (Presumably the school board will not complain if they are published in Substance a century from now.) But the board overlooks the fact that the Nation was not publishing the excerpts from Ford's memoirs in order to be able to criticize Ford, or anyone or anything else, and the further fact that Schmidt was not trying to steal the school board's market, because unlike Harper & Row the school board did not intend ever to publish the copyrighted work that the alleged infringer published first. The second point has little force, as we'll see; but the first is important. As Schmidt points out, one office of the fair use defense is to facilitate criticism of copyrighted works by enabling the critic to quote enough of the criticized work to make his criticisms intelligible. Copyright should not be a means by which criticism is stifled with the backing of the courts. And since doubts that fair use could ever be a defense to infringement of a copyright on an unpublished work (see, e.g., Salinger v. Random House, Inc., supra, 811 F.2d at 97) have now been stilled (see 17 U.S.C. § 107; Sundeman v. Seajay Society, Inc., 142 F.3d 194, 204-05 (4th Cir.1998); Wright v. Warner Books, Inc., 953 F.2d 731, 740 (2d Cir.1991)), the fact that the CASE tests were quasi-secret does not exclude the possibility of a fair use defense.

But Schmidt overreads our decision in Ty, Inc. v. Publications Int'l Ltd., 292 F.3d 512 (7th Cir.2002), to hold that a purpose of criticizing creates a privilege of unlimited copying regardless of the harm to the copyright owner's legitimate interests. Ty was trying to enjoin the publication of Beanie Babies collectors' guides that contained criticisms of some of the Beanie Babies, with accompanying photographs that constituted derivative works of the soft-sculpture Beanie Babies and hence prima facie infringements of Ty's copyrights. There was no danger that if Ty failed to enjoin the publication of the collectors' guides the value of its copyrights would be impaired, because collectors' guides are not substitutes for the Beanie Babies themselves. More precisely, the only harm that could come to Ty from the unauthorized publication of the guides would be caused by the criticisms of particular Beanie Babies, whereas a harm to the school board from Schmidt's copying and publication that is independent of any criticisms is the cost of creating substitutes for the questions that Schmidt published and of extending the period of the pilot program because his publishing the six tests prevented the board from asking any of the questions in those tests in subsequent years.

The board thus is correct that Schmidt did it harm going beyond the force of his criticisms. Yet he was entitled to criticize the tests and to do that effectively he had to be able to quote from them, just as a parodist has to be able to quote, sometimes very extensively, from the parodied work in order to make the criticism of it that is implicit in parodying it comprehensible. Campbell v. Acuff-Rose Music, Inc., 510 U.S. 569, 588, 114 S. Ct. 1164, 127 L. Ed. 2d 500 (1994); Ty, Inc. v. Publications Int'l Ltd., supra, 292 F.3d at 518; Suntrust Bank v. Houghton Mifflin Co., 268 F.3d 1257, 1271-74 (11th Cir.2001). Indeed, there is no per se rule against copying in the name of fair use an entire copyrighted work if necessary. Sony Corp. of America v. Universal City Studios, Inc., 464 U.S. 417, 449-50, 104 S. Ct. 774, 78 L. Ed. 2d 574 (1984); Ty, Inc. v. Publications Int'l Ltd., supra, 292 F.3d at 521; Sega Enterprises Ltd. v. Accolade, Inc., 977 F.2d 1510, 1526 (9th Cir.1992). This would be obvious if the school board had copyrighted each question separately, as it might well have done, since it wants to reuse questions without necessarily reusing an entire test.

So where to draw the line? The question cannot be answered precisely. The fair use defense defies codification. As we said in Ty, the four factors that Congress listed when it wrote a fair use defense (a judicial creation) into the Copyright Act in 1976 are not exhaustive and do not constitute an algorithm that enables decisions to be ground out mechanically. Ty, Inc. v. Publications Int'l Ltd., supra, 292 F.3d at 522; see also Campbell v. Acuff-Rose Music, Inc., supra, 510 U.S. at 577-78, 114 S. Ct. 1164; Harper & Row, Publishers, Inc. v. Nation Enterprises, supra, 471 U.S. at 560, 105 S. Ct. 2218. The general standard, however, is clear enough: the fair use copier must copy no more than is reasonably necessary (not strictly necessary — room must be allowed for judgment, and judges must not police criticism with a heavy hand) to enable him to pursue an aim that the law recognizes as proper, in this case the aim of criticizing the copyrighted work effectively. Ty, Inc. v. Publications Int'l Ltd., supra, 292 F.3d at 521; Kelly v. Arriba Soft Corp., 336 F.3d 811, 820-21 (9th Cir.2003); Sundeman v. Seajay Society, Inc., supra, 142 F.3d at 206.

The burden of proof is on the copier because fair use is an affirmative defense, Campbell v. Acuff-Rose Music, Inc., supra, 510 U.S. at 590, 114 S. Ct. 1164; Harper & Row Publishers, Inc. v. Nation Enterprises, supra, 471 U.S. at 561, 105 S. Ct. 2218; Video Pipeline, Inc. v. Buena Vista Home Entertainment, Inc., 342 F.3d 191, 197 (3d Cir.2003), and Schmidt has presented no evidence sufficient to withstand summary judgment. But because of the bizarre circumstance that the district court decided the issue of fair use and then permitted discovery, we would be forgiving had Schmidt indicated in his appeal briefs or at oral argument what evidence he would like to obtain or present, though he didn't obtain any in the district court proceedings, to prove that he copied no more of the tests than he had to do in order to make his criticisms of them intelligible. He insists that he had to copy six tests out of the 22 to 44 tests given in January 1999 (it is unclear from the record which number is correct, so we will give Schmidt the benefit of the doubt and assume that it is the higher number) in order to drive his criticisms home. But he does not explain why, or indicate witnesses or documents that might support such an argument. Granted that he had to quote some of the test questions in order to substantiate his criticisms, why entire tests? Does he think all the questions in all six tests bad? He does not say. What purpose is served by quoting the good questions? Again, no answer.

Schmidt's opening brief cites the memorandum that he had filed in the district court — the memorandum the judge had overlooked in rejecting the fair use defense on the pleadings. In the memorandum, Schmidt did offer answers to these questions. Only the answers are no good, so that even if incorporation by reference were a valid method of bringing facts and arguments to the attention of an appellate court — which it is not, DeSilva v. DiLeonardi, 181 F.3d 865, 867 (7th Cir.1999) ("a brief must make all arguments accessible to the judges, rather than ask them to play archaeologist with the record"); Hunter v. Allis-Chalmers Corp., 797 F.2d 1417, 1430 (7th Cir.1986) ("our rules do not permit issues to be preserved by references to documents filed in the district court"); Northland Ins. Co. v. Stewart Title Guaranty Co., 327 F.3d 448, 452-53 (6th Cir. 2003) — treating the memorandum as if it were a brief in this court would merely confirm that Schmidt has no fair use defense.

The memorandum argues that the school board does not intend to sell the tests, and so Schmidt isn't eating into their market by publishing the tests. This is true, but irrelevant, because he is destroying the value of the tests and the fact that it's not a market value has no significance once the right to copyright unpublished works is conceded, as it must be. The memorandum argues that expert testimony would establish that there is no educational value in publishing the exact same exam year after year. No one supposes there is; the argument rather is that some questions must be carried over to future years in order to validate the exam. The memorandum argues that if Schmidt quoted only a few of the questions, the school board would respond that he was cherry-picking the worst. But if the board did that without a solid grounding, it would open the door for him to quote additional questions; it would be a pro tanto waiver of confidentiality. The memorandum's remaining argument is that expert witnesses can be found to testify that the CASE tests are "dramatically inadequate." No doubt. But in so arguing the memorandum misunderstands the fair use privilege of criticism. It is not a privilege to criticize just bad works, and there is no right to copy copyrighted works promiscuously merely upon a showing that they are bad.

There is more than a suspicion that Schmidt simply does not like standardized tests. That is his right. But he does not have the right, as he believes he does (he claims a right to copy any test that an expert will testify is no good), to destroy the tests by publishing them indiscriminately, any more than a person who dislikes Michelangelo's statue of David has a right to take a sledgehammer to it. From the amicus curiae briefs filed in this case, moreover, it is apparent that many other teachers share Schmidt's unfavorable opinion of standardized tests. (A cynic might say that this is because such tests can make teachers look bad if their students don't do well on them.) So if Schmidt can publish six tests, other dissenters can each publish six other tests, and in no time all 44 will be published. The board will never be able to use the same question twice, and after a few years of Schmidtian tactics there will be such difficulty in inventing new questions without restructuring the curriculum that the board will have to abandon standardized testing. Which is Schmidt's goal.

If ever a "floodgates" argument had persuasive force, therefore, it is in this case. And this suggests another fair use factor that supports the school board: the aspect of academic freedom that consists of the autonomy of educational institutions, see, e.g., Grutter v. Bollinger, 539 U.S. 306, ___, 123 S. Ct. 2325, 2339, 156 L. Ed. 2d 304 (2003); Piarowski v. Illinois Community College Dist. 515, 759 F.2d 625, 629 (7th Cir.1985) (cases involving a university and a college respectively, but the point is general), including their authority here gravely threatened to employ standardized tests in support of their conception of their educational mission. If Schmidt wins this case, it is goodbye to standardized tests in the Chicago public school system; Schmidt, his allies, and the federal courts will have wrested control of educational policy from the Chicago public school authorities.

The appeal raises other issues, but they require little discussion. Despite the Supreme Court's recent statement that "copyright law contains built-in First Amendment accommodations" and that in any event the First Amendment "bears less heavily when speakers assert the right to make other people's speeches," Eldred v. Ashcroft, 537 U.S. 186, 219, 221, 123 S. Ct. 769, 154 L. Ed. 2d 683 (2003), and our repetition of the point in In re Aimster Copyright Litigation, 334 F.3d 643, 656 (7th Cir.2003), Schmidt devotes most of the argument portion of his briefs to attempting to persuade us that the First Amendment authorizes what he did. (He fails to cite either case.) The First Amendment adds nothing to the fair use defense. The defense, so far as relates to this case, is the point of balance between the right to criticize, which is one aspect of freedom of expression, and the incentive to create expressive works, which is another aspect of the same freedom, the aspect that undergirds the Constitution's copyright clause. Wasting time on the First Amendment, Schmidt failed to show that he may have a viable fair use defense.

His further argument that the school board lacks a valid copyright registration on which to found this suit is frivolous. Although a copyright no longer need be registered with the Copyright Office to be valid, an application for registration must be filed before the copyright can be sued upon. 17 U.S.C. § 411(a); 2 Melville B. Nimmer & David Nimmer, Nimmer on Copyright § 7.16[B][1][a] (2003). In its application for registration the school board claimed to have a copyright in the entire contents of the tests. Had the claim been false, the registration should not have issued and maybe therefore the copyright could not have been sued upon. See Whimsicality, Inc. v. Rubie's Costume Co., 891 F.2d 452, 453, 456 (2d Cir.1989). (Or maybe yes, because the copyright would have been registered, and because the statute requires only a refused registration, which might be the equivalent of an improper registration, not an actual registration, as the premise for the suit. We need not decide.) Schmidt argues that the school board did not have a copyright in the entire contents of the tests. He bases this argument on a squabble that erupted between the school board and a research center at UCLA called CRESST that had been hired to review and evaluate the questions and help the board analyze the results of the tests. CRESST may have devised a few of the questions, though this is unclear and in any event they would probably have been work for hire, 17 U.S.C. § 101, so that the school board would have owned the copyright anyway. § 201(b). Unsurprisingly, CRESST did not claim copyright in any part of the tests, and to resolve the squabble had assigned any copyright it might have in its analyses to the board.

All that remains to consider is the injunction. Remarkably, Schmidt does not question its scope or application. But because an injunction frequently affects third parties and consumes judicial resources, a court has an independent duty (Marseilles Hydro Power, LLC v. Marseilles Land & Water Co., 299 F.3d 643, 647 (7th Cir. 2002)) to assure that the injunctions it issues comply with the directive of Fed. R.Civ.P. 65(d) that injunctions "shall be specific in terms" and "shall describe in reasonable detail ... the act or acts sought to be restrained." See, e.g., Schmidt v. Lessard, 414 U.S. 473, 476-77, 94 S. Ct. 713, 38 L. Ed. 2d 661 (1974) (per curiam); International Longshoremen's Ass'n, Local 1291 v. Philadelphia Marine Trade Ass'n, 389 U.S. 64, 73-76, 88 S. Ct. 201, 19 L. Ed. 2d 236 (1967).

The injunction flunks this test resoundingly. Cf. Educational Testing Services v. Katzman, supra, 793 F.2d at 544-45. The defendants are enjoined from "copying distribution of copies, making derivative copies, displaying copies an performing copies of the Board's examinations, including but not limited to the CASE examinations, whenever created by the Board." The first phrase — "copying distribution of copies" — is gibberish; probably there was meant to be an "or" between "copying" and "distribution." The reference to "displaying copies an[d] performing copies" is opaque since what is at issue is the copying itself and the publication of the copies, not display or performance. The inclusion in the injunction of all other examinations that the school board may someday create (there is no sunset provision), an inclusion brought about by the phrase "including but not limited to the CASE examinations, whenever created by the Board," is overbroad. The school board may not own the copyrights in all its tests, even those it creates (work-for-hire copyright rights can be waived); not all its test questions may be intended to be reused; not all its tests may be secure tests.

This is an appallingly bad injunction. It must be modified to enjoin the defendants, and those in privity with them or with notice of the injunction, just from copying or publishing or otherwise distributing copies of secure Chicago public school tests, in whole or substantial part, in which the school board has valid and subsisting copyright, without the board's authorization. No evidentiary basis has been laid for a broader injunction.

The judgment is affirmed in part, and vacated in part, and the case is remanded for further proceedings consistent with this opinion, the parties to bear their own costs in this court.