Part 1 · Chapter 5
Idea-Expression, Functionality and Other Exclusions—Part II
Merger and Scènes à Faire
The Merger Doctrine
Very occasionally a literary work will fail to qualify for copyright protection because it lacks the modicum of creativity required by Feist. Telephone directories and works based on the public domain with no recognizable authorial changes are not eligible for copyright protection and may be copied at will. However, sometimes even works that display that modicum of creativity will also be ineligible for copyright protection because potentially copyrightable expression has merged with uncopyrightable facts, ideas, or functions. The uncopyrightability of the forms in Baker v. Selden is an early example of the merger doctrine at work, although the Court did not expressly invoke the doctrine by name.
Every copyrightable work contains unprotectable elements. In very rare cases, there is complete overlap between original expression and unprotectable facts, ideas, or methods. In such cases, the merger doctrine holds that the freedom of facts, ideas, and methods wins out. Under the merger doctrine, when “there is only one feasible way of expressing an idea, so that if the expression were copyrightable it would mean that the idea was copyrightable,” the expression is not protected. See Bucklew v. Hawkins, Ash, Baptie & Co., LLP, 329 F.3d 923, 928 (7th Cir. 2003). Or to put in another way, the “merger doctrine” holds that, in certain circumstances, where authors have only one or very few ways of expressing an idea or function, the expression and the idea are merged, and the expression is treated as unprotectable. See Pamela Samuelson, Reconceptualizing Copyright’s Merger Doctrine, 63 Journal of the Copyright Society of the U.S.A. 417 (2016).
In Morrissey v. Procter & Gamble Co., 379 F.2d 675 (1st Cir. 1967), for example, the plaintiff was the copyright owner of a set of rules for a sales promotional contest of the ‘sweepstakes’ type involving the social security numbers of the participants. Morrissey alleged that the defendant, Procter & Gamble, had copied and thus infringed Rule 1 of its rules.
Plaintiff’s Rule
1. Entrants should print name, address and social security number on a boxtop, or a plain paper. Entries must be accompanied by * * * boxtop or by plain paper on which the name * * * is copied from any source. Official rules are explained on * * * packages or leaflets obtained from dealer. If you do not have a social security number you may use the name and number of any member of your immediate family living with you. Only the person named on the entry will be deemed an entrant and may qualify for prize.
‘Use the correct social security number belonging to the person named on entry * * * wrong number will be disqualified.’
Defendant’s Rule
1. Entrants should print name, address and Social Security number on a Tide boxtop, or on (a) plain paper. Entries must be accompanied by Tide boxtop (any size) or by plain paper on which the name ‘Tide’ is copied from any source. Official rules are available on Tide Sweepstakes packages, or on leaflets at Tide dealers, or you can send a stamped, self-addressed envelope to: Tide ‘Shopping Fling’ Sweepstakes, P.O. Box 4459, Chicago 77, Illinois.
‘If you do not have a Social Security number, you may use the name and number of any member of your immediate family living with you. Only the person named on the entry will be deemed an entrant and may qualify for a prize.
‘Use the correct Social Security number, belonging to the person named on the entry - wrong numbers will be disqualified.’
Holding for the defendant, the First Circuit said that:
When the uncopyrightable subject matter is very narrow, so that ‘the topic necessarily requires,’ …, if not only one form of expression, at best only a limited number, to permit copyrighting would mean that a party or parties, by copyrighting a mere handful of forms, could exhaust all possibilities of future use of the substance.
In such circumstances it does not seem accurate to say that any particular form of expression comes from the subject matter. However, it is necessary to say that the subject matter would be appropriated by permitting the copyrighting of its expression. We cannot recognize copyright as a game of chess in which the public can be checkmated.
Three points are worth keeping in mind in relation to merger. First, the merger is not a sui generis or ad hoc principle, it is one manifestation of the broader principle that copyright protection does not extend to the protection of facts, ideas, functions, or processes. Second, the merger doctrine operates to exclude what might have otherwise been protectable expression from the ambit of copyright protection. Third, the merger doctrine can deny copyright to a work in total, but merger analysis is also used by courts to identify unprotectable elements within protectable expression that must be filtered in any infringement analysis.
Scènes à Faire
Stock themes and conventions belong in the public domain.
Like the merger doctrine, the scènes à faire doctrine also polices the idea-expression distinction. Under the scènes à faire doctrine stock or standard literary devices are not considered protectable, even if they are expressive. The doctrine’s name is borrowed from the French who “use a very expressive phrase in dramatic literature: ‘scènes à faire’ that is, scenes which ‘must’ be done.” Schwarz v. Universal Pictures Co., 85 F. Supp. 270, 275 (S.D. Cal. 1945).
As the Seventh Circuit explained in Bucklew v. Hawkins, Ash, Baptie & Co., 329 F.3d 923, 929 (7th Cir. 2003):
… copyright owner can’t prove infringement by pointing to features of his work that are found in the defendant’s work as well but that are so rudimentary, commonplace, standard, or unavoidable that they do not serve to distinguish one work within a class of works from another.
… Every expressive work can be decomposed into elements not themselves copyrightable—the cars in a car chase, the kiss in a love scene, the dive bombers in a movie about Pearl Harbor, or for that matter the letters of the alphabet in any written work. The presence of such elements obviously does not forfeit copyright protection of the work as a whole, but infringement cannot be found on the basis of such elements alone; it is the combination of elements, or particular novel twists given to them, that supply the minimal originality required for copyright protection.
As the Second Circuit explained in Hoehling v. Universal City Studios, Inc., 618 F.2d 972, 979 (2d Cir. 1980), the scènes à faire doctrine is justified on the basis that “it is virtually impossible to write about a particular historical era or fictional theme without employing certain ‘stock’ or standard literary devices.”
How stock-standard does a stock character have to be to constitute scènes à faire? Some courts put this in terms of whether expression is “indispensable,” but that is an overstatement. An element can be “standard” or “customary” without being indispensable. See, e.g., Hoehling v. Universal: “incidents, characters or settings which are as a practical matter indispensable, or at least standard, in the treatment of a given topic.”
Scènes à faire will not usually make a work uncopyrightable, but it will make aspects of the work irrelevant for infringement analysis. Like an unprotectable idea, any aspect of the work that is scènes à faire is unprotectable as such. The fact that work A and work B both contain the same scènes à faire elements does not go very far to prove one was copied from the other in the usual case, and it certainly does not establish that any such copying was wrongful, even if proven or admitted.
Lexmark International, Inc. v. Static Control Components, Inc., 387 F.3d 522 (6th Cir. 2004)
At the time of this litigation, Lexmark made and sold printers and printer cartridges. Lexmark print cartridges contained a simple computer program (the Toner Loading Program) and an authentication code stored on an embedded microchip. For a Lexmark printer to accept a print cartridge, it would have to receive an authentication code from the chip inside the cartridge. The printers were designed this way so that Lexmark could sell toner cartridges as single-use items, or that could only be refurbished by Lexmark. Static Control supplied its own chips that would defeat the Lexmark authentication sequence and thus allow third parties to sell remanufactured Lexmark toner cartridges. Each of Static Control’s SMARTEK chips also contained an exact copy of Lexmark’s Toner Loading Program, which was necessary to make its product compatible with Lexmark’s printers.
Lexmark alleged that Static Control violated its copyright in the Toner Loading Program and that it violated the DMCA by circumventing Lexmark’s access controls. This extract focuses on the copyright issues in the case. DMCA anti-circumvention issues are addressed in a later chapter.
Circuit Judge Sutton
… As this case comes to the court, the parties agree that computer programs may be entitled to copyright protection as “literary works” under 17 U.S.C. § 101 and may be protected from infringement under 17 U.S.C. § 106. And that is true with respect to a computer program’s object code (the binary code — a series of zeros and ones — that computers can read) and its source code (the spelled-out program commands that humans can read).
The parties also agree that Lexmark has registered the Toner Loading Program with the Copyright Office, which is an infringement suit prerequisite, see 17 U.S.C. § 411(a), and which constitutes prima facie evidence of the copyright’s validity, see id. § 410(c). And the parties agree that SCC shoulders the burden of rebutting the presumptive validity of Lexmark’s copyright.
The parties also share common ground when it comes to most of the general principles of copyright infringement applicable to this case. A plaintiff may establish a claim of copyright infringement by showing (1) ownership of a valid copyright in the computer program at issue (here, the Toner Loading Program) and (2) that the defendant copied protectable elements of the work. See Feist Publications, Inc. v. Rural Tel. Serv. Co., 499 U.S. 340, 361 (1991). The first prong tests the originality and non-functionality of the work, both of which are presumptively established by the copyright registration. The second prong tests whether any copying occurred (a factual matter) and whether the portions of the work copied were entitled to copyright protection (a legal matter). If no direct evidence of copying is available, a claimant may establish this element by showing that the defendant had access to the copyrighted work and that the copyrighted work and the allegedly copied work are substantially similar.
As to the first prong, the Supreme Court has instructed that “original ... means only that the work was independently created by the author (as opposed to copied from other works), and that it possesses at least some minimal degree of creativity,” even if the work is not a “novel” one. Feist, 499 U.S. at 345-46 (originality requires both “independent creation plus a modicum of creativity”). And although constitutionally mandated, the threshold showing of originality is not a demanding one. Id. at 345 (“To be sure, the requisite level of creativity is extremely low; even a slight amount will suffice.”).
But even if a work is in some sense “original” under § 102(a), it still may not be copyrightable because § 102(b) provides that “[i]n no case does copyright protection for an original work of authorship extend to any idea, procedure, process, system, method of operation, concept, principle, or discovery, regardless of [its] form.” This provision embodies the common-law idea-expression dichotomy that distinguishes the spheres of copyright and patent law. “Unlike a patent, a copyright gives no exclusive right to the art disclosed; protection is given only to the expression of the idea — not the idea itself.” Mazer v. Stein, 347 U.S. 201, 217 (1954); see also Baker v. Selden, 101 U.S. 99, 101-02 (1879) (explaining that while a book describing a bookkeeping system is worthy of copyright protection, the underlying method described is not); Computer Assocs. Int’l, Inc. v. Altai, Inc., 982 F.2d 693, 703 (2d Cir.1992) (“It is a fundamental principle of copyright law that a copyright does not protect an idea, but only the expression of the idea.”). While this general principle applies equally to computer programs, id.; see also House Report at 5667 (extending copyright protection to computer programs only “to the extent that they incorporate authorship in programmer’s expression of original ideas, as distinguished from ideas themselves”), the task of separating expression from idea in this setting is a vexing one, see Altai, 982 F.2d at 704 (“The essentially utilitarian nature of a computer program further complicates the task of distilling its idea from its expression.”); Sega Enters., Ltd. v. Accolade, Inc., 977 F.2d 1510, 1524 (9th Cir.1992); see also Lotus Dev. Corp. v. Borland Int’l, Inc., 49 F.3d 807, 819-20 (1st Cir.1995) (Boudin, J., concurring). “Compared to aesthetic works, computer programs hover even more closely to the elusive boundary line described in § 102(b).” Altai, 982 F.2d at 704.
In ascertaining this “elusive boundary line” between idea and expression, between process and non-functional expression, courts have looked to two other staples of copyright law — the doctrines of merger and scènes à faire. Where the “expression is essential to the statement of the idea,” CCC Info. Servs., Inc. v. Maclean Hunter Mkt. Reports, Inc., 44 F.3d 61, 68 (2d Cir.1994); see also Lotus Dev., 49 F.3d at 816 (“If specific words are essential to operating something, then they are part of a ‘method of operation’ and, as such, are unprotectable.”), or where there is only one way or very few ways of expressing the idea, Warren Publ’g, Inc. v. Microdos Data Corp., 115 F.3d 1509, 1519 n. 27 (11th Cir.1997), the idea and expression are said to have “merged.” In these instances, copyright protection does not exist because granting protection to the expressive component of the work necessarily would extend protection to the work’s uncopyrightable ideas as well. See Gates Rubber Co. v. Bando Chem. Indus., Ltd., 9 F.3d 823, 838 (10th Cir.1993); see also Murray Hill Publ’ns, Inc. v. Twentieth Century Fox Film Corp., 361 F.3d 312, 319 n. 2 (6th Cir.2004) (noting that where idea and expression are intertwined and where non-protectable ideas predominate, expression is not protected). For computer programs, “if the patentable process is embodied inextricably in the line-by-line instructions of the computer program, [] then the process merges with the expression and precludes copyright protection.” Atari I, 975 F.2d at 839-40.
For similar reasons, when external factors constrain the choice of expressive vehicle, the doctrine of “scènes à faire” — “scenes,” in other words, “that must be done” — precludes copyright protection. See Twentieth Century Fox Film, 361 F.3d at 319-20. In the literary context, the doctrine means that certain phrases that are “standard, stock, ... or that necessarily follow from a common theme or setting” may not obtain copyright protection. Gates Rubber, 9 F.3d at 838. In the computer-software context, the doctrine means that the elements of a program dictated by practical realities — e.g., by hardware standards and mechanical specifications, software standards and compatibility requirements, computer manufacturer design standards, target industry practices, and standard computer programming practices — may not obtain protection. Id. (citing case examples); see Sega Enters., 977 F.2d at 1524 (“To the extent that a work is functional or factual, it may be copied.”); Brown Bag Software v. Symantec Corp., 960 F.2d 1465, 1473 (9th Cir.1992) (affirming district court’s finding that “plaintiffs may not claim copyright protection of an expression that is, if not standard, then commonplace in the computer software industry”). As “an industry-wide goal,” programming “efficiency” represents an external constraint that figures prominently in the copyrightability of computer programs. Altai, 982 F.2d at 708.
Generally speaking, “lock-out” codes fall on the functional-idea rather than the original-expression side of the copyright line. Manufacturers of interoperable devices such as computers and software, game consoles and video games, printers and toner cartridges, or automobiles and replacement parts may employ a security system to bar the use of unauthorized components. To “unlock” and permit operation of the primary device (i.e., the computer, the game console, the printer, the car), the component must contain either a certain code sequence or be able to respond appropriately to an authentication process. To the extent compatibility requires that a particular code sequence be included in the component device to permit its use, the merger and scènes à faire doctrines generally preclude the code sequence from obtaining copyright protection. See Sega Enters., 977 F.2d at 1524 (“When specific instructions, even though previously copyrighted, are the only and essential means of accomplishing a given task, their later use by another will not amount to infringement.”) (quoting National Commission on New Technological Uses of Copyrighted Works, Final Report 20 (1979)) (emphasis added).
In trying to discern whether these doctrines apply, courts tend to “focus on whether the idea is capable of various modes of expression.” Mason v. Montgomery Data, Inc., 967 F.2d 135, 138 (5th Cir.1992) (quoting Franklin Computer, 714 F.2d at 1253); Atari I, 975 F.2d at 840 (“The unique arrangement of computer program expression which generates the data stream does not merge with the process so long as alternate expressions are available.”). The question, however, is not whether any alternatives theoretically exist; it is whether other options practically exist under the circumstances. See Altai, 982 F.2d at 708 (“While, hypothetically, there might be a myriad of ways in which a programmer may effectuate certain functions within a program ... efficiency concerns may so narrow the practical range of choice as to make only one or two forms of expression workable options.”); Atari I, 975 F.2d at 840 (noting that “no external factor dictated the bulk of the program” and finding the program copyrightable). In order to characterize a choice between alleged programming alternatives as expressive, in short, the alternatives must be feasible within real-world constraints.
The Supreme Court’s decision in Feist helps to illustrate the point. In Feist, the alleged infringer had included 1,309 of the plaintiff’s alphabetically-organized telephone book listings in its own telephone directory. The facts comprising these listings, it was clear, theoretically could have been organized in other ways — for instance, by street address or phone number, or by the age or height of the individual. But by virtue of tradition and settled expectations, the familiar alphabetical structure copied by the defendant amounted to the only organizational option available to the defendant. For these reasons, the Supreme Court determined that alphabetical phone listings did not satisfy the low threshold of originality for copyright protection.
* * *
In applying these requirements to this case, it helps to clarify the terms of debate between the parties. Lexmark claims copyright protection in, and infringement of, the code that composes its Toner Loading Program. It has not alleged that SCC copied any other portion of its chip, including any of the data on which the SHA-1 algorithm — the authentication sequence or “secret handshake” — appear. Presumably that is because SCC replaced Lexmark’s SHA-1 function with a different publicly available encryption program to enable interoperability of its chip with Lexmark’s printers. When it comes to the merits of the infringement claim, the parties primarily debate whether the Toner Loading Program satisfies the originality requirement (prong one), as distinct from whether any copying by SCC is substantially similar to the Lexmark chip (prong two). That is because the parties agree that SCC’s SMARTEK chip copied all aspects of the Toner Loading Program.
In our view, the district court committed three related legal errors in determining that Lexmark had a likelihood of prevailing on its copyright claim with respect to the Toner Loading Program. First, the district court concluded that, because the Toner Loading Program “could be written in a number of different ways,” it was entitled to copyright protection. In refusing to consider whether “external factors such as compatibility requirements, industry standards, and efficiency” circumscribed the number of forms that the Toner Loading Program could take, the district court believed that the idea-expression divide and accompanying principles of merger and scènes à faire play a role only in the “substantial similarity” analysis and do not apply when the first prong of the infringement test (copyrightability) is primarily at issue. In taking this path, the district court relied on cases invoking the pronouncement [in Nimmer on Copyright] that the idea-expression divide “constitutes not so much a limitation on the copyrightability of works, as it is a measure of the degree of similarity that must exist between a copyrightable work and an unauthorized copy.” And in concluding more generally that the copyrightability of a computer program turns solely on the availability of other options for writing the program, the court relied on several cases from other circuits.
This reasoning, to start with, conflicts with Feist. As the Supreme Court’s recent decision suggests, one does not satisfy the originality requirement for copyright protection merely by showing that the work could have been put together in different ways. Just as it failed to suffice in Feist that the author of the competing telephone book could have organized the listings in some manner other than the individual’s last name, so it does not suffice here that SCC could have written the Toner Loading Program in some other way. As in Feist, the court must ask whether the alternative ways of putting together the competing work are feasible in that setting.
Nor does Nimmer support the district court’s “a number of different ways” reasoning. As a matter of practice, Nimmer is correct that courts most commonly discuss the idea-expression dichotomy in considering whether an original work and a partial copy of that work are “substantially similar” (as part of prong two of the infringement test), since the copyrightability of a work as a whole (prong one) is less frequently contested. But the idea-expression divide figures into the substantial similarity test not as a measure of “similarity”; it distinguishes the original work’s protectable elements from its unprotectable ones, a distinction that allows courts to determine whether any of the former have been copied in substantial enough part to constitute infringement. Both prongs of the infringement test, in other words, consider “copyrightability,” which at its heart turns on the principle that copyright protection extends to expression, not to ideas. See Brown Bag Software, 960 F.2d at 1475-76 (noting that copyrightability is an aspect of both parts of the infringement test); see also Bateman v. Mnemonics, Inc., 79 F.3d 1532, 1545 (11th Cir.1996) (deciding that district court erred by not permitting consideration of merger, scènes à faire doctrines in instance of literal copying); Lotus Dev., 49 F.3d at 815 (assessing whether the literally copied menu command structure is copyrightable in light of § 102(b)’s limitation). When a work itself constitutes merely an idea, process or method of operation, or when any discernible expression is inseparable from the idea itself, or when external factors dictate the form of expression, copyright protection does not extend to the work. See Bateman, 79 F.3d at 1546 n. 28 (“Compatibility and other functionality challenges to originality ... are applied so as to deny copyright protection to a particular work or portion of a work.”) (emphasis added); Mason, 967 F.2d at 138 n. 5 (rejecting argument that merger doctrine applies only to question of infringement and noting that “this court has applied the merger doctrine to the question of copyrightability”).
Neither do the cited cases support the district court’s initial frame of reference. Franklin Computer, 714 F.2d 1240, and Formula International, 725 F.2d 521, involved copies of Apple’s operating system program — a program whose size and complexity is to the Toner Loading Program what the Sears Tower is to a lamppost. Given the nature of the Apple program, it would have been exceedingly difficult to say that practical alternative means of expression did not exist and indeed no defendant in the cases advanced that argument. [The court distinguished Franklin Computer, Whelan Associates and E.F. Johnson.]
Second, given the district court’s mistaken view of the legal standard for distinguishing protectable expression from unprotectable ideas, the constraints on the Toner Loading Program established by the evidence need to be reconsidered. To discern whether “originality” exists in the work, the court should ask whether the ideas, methods of operation and facts of the program could have been expressed in any form other than that chosen by the programmer, taking into consideration the functionality, compatibility and efficiency demanded of the program.
In presenting evidence in support of its motion for a preliminary injunction, Lexmark focused on establishing that the Toner Loading Program could have been written in other ways. Dr. Maggs, Lexmark’s expert, described several possible alternatives in his declaration: (1) different constants and equations could be used; (2) a lookup table could be used in lieu of equations; (3) some measure other than torque could be used to approximate toner level (e.g., the number of pages printed); or (4) the same equations could be used in a different sequence. He concluded that over 50 different programs could be written to substitute for the Toner Loading Program.
Dr. Goldberg, SCC’s expert, acknowledged that certain changes could be made to the program, for example, by changing the sequence of elements in the program, or by writing the Toner Loading Program in a different programming language altogether. But Dr. Goldberg conceded this point only as a theoretical matter, as he concluded that functionality and efficiency considerations precluded any material changes to the Toner Loading Program.
Dr. Goldberg concluded that several external constraints limit the options available in designing the Toner Loading Program. For one, the Printer Engine Program that downloads and executes the program understands only a single programming language composed of eight simple commands. For another, the program must consist of only 55 bytes because the printer downloads only these particular bytes. Efficiency considerations and the physical realities of the printer and toner cartridge also restrict the forms that the Toner Loading Program could take. As a result, Dr. Goldberg concluded, these external factors together “dictate the way that the simple toner loading program looks,” and the resulting program is a “no-thought translation of the formulas to the language that the internal loading program must be written in, and the programmer doesn’t have much choice.” Dr. Goldberg responded to Dr. Maggs’ testimony that the Toner Loading Program could take alternative forms by noting that Dr. Maggs’ proposed changes were trivial — that they did not make any “substantial difference to the nature of the program” — or that they were so inefficient and repetitive as to be “ridiculous.” Instead, Dr. Goldberg concluded, the Toner Loading Program as it is written is the most “straightforward, efficient, natural way to express the program.” By contrast, Dr. Maggs’ testimony did not reference any of these functional considerations discussed by Goldberg, meaning that the record fails to establish any affirmative support for the contention that Dr. Maggs’ proposed alternatives satisfy the memory restrictions of the program.
Even aside from Dr. Goldberg’s testimony that the Toner Loading Program is the most efficient means of calculating toner levels, the alternatives suggested by Dr. Maggs do not appear to support the district court’s initial conclusion that the program is expressive. Dr. Maggs’ first and third suggestions — that different equations and values or a different means of measuring toner level altogether could have been used — do not appear to represent alternative means of expressing the ideas or methods of operations embodied in the Toner Loading Program; they appear to be different ideas or methods of operation altogether. Selection from among competing ideas or methods of operation generally does not result in copyright-protectable expression. See Bateman, 79 F.3d at 1546 n.29 (“Generally there is more than one method of operation or process that can be used to perform a particular computer program function. However, methods of operation and processes are not copyrightable.”) Nor would the use of a “lookup table” appear to differ meaningfully from the use of other equations directly. Instead of executing a mathematical formula on a given input, this program merely “looks up” in a data table the output of that same formula for the given input value. Finally, Dr. Maggs’ fourth suggestion — that the same equations could be reordered — does not appear to show originality because such alterations may be too trivial to support a finding of creative expression. See M.M. Bus. Forms, 472 F.2d at 1140 (holding that paraphrasing of words on legal forms does not contain the requisite originality for copyright protection).
To the extent these alternatives suggest any originality in the Toner Loading Program, at any rate, the quantum of originality may well be de minimis and accordingly insufficient to support the validity of Lexmark’s copyright in the work. See Mitel, Inc. v. Iqtel, Inc., 124 F.3d 1366, 1373-74 (10th Cir.1997) (determining that plaintiff’s “arbitrary selection” of several numbers required only “de minimis creative effort” and did not “evince enough originality to distinguish authorship”); cf. Sega Enters., 977 F.2d at 1524 n. 7 (noting that 20-byte code is of de minimis length and therefore likely a “word” or “short phrase” that is not protected by copyright law); Murray Hill Publ’ns, Inc. v. ABC Comm’ns, Inc., 264 F.3d 622, 633 (6th Cir.2001) (noting that short movie line was “a phrase or slogan not worthy of copyright protection in its own right”). Because the district court initially looked at these issues and this evidence through the wrong frame of reference, its conclusion that the Toner Loading Program had sufficient originality to obtain copyright protection does not support the preliminary injunction. At the permanent injunction stage of this dispute, we leave it to the district court in the first instance to decide whether the Toner Loading Program has sufficient originality to warrant copyright protection.
Third, and perhaps most significantly, the district court erred in assessing whether the Toner Loading Program functions as a lock-out code. Even if the constraints described by Dr. Goldberg — the programming language, the program size, efficiency concerns — did not dictate the content of the Toner Loading Program, the fact that it also functions as a lock-out code undermines the conclusion that Lexmark had a probability of success on its infringement claim.
The Toner Loading Program, recall, serves as input to the checksum operation that is performed each time the printer is powered on or the printer door is opened and closed (i.e., for toner cartridge replacement). After downloading a copy of the Toner Loading Program to calculate toner levels, the Printer Engine Program runs a calculation — the checksum — using every data byte of the Toner Loading Program as input. The program then compares the result of that calculation with a “checksum value” that is located elsewhere on Lexmark’s toner cartridge chip. If any single byte of the Toner Loading Program is altered, the checksum value will not match the checksum calculation result. Only if the checksum value is correspondingly changed to accommodate any alterations in the data bytes will the printer function.
In addition to its general conclusion that external constraints on the program were not relevant, the district court concluded that the checksum operation did not operate as a strict constraint on the content of the Toner Loading Program because “SCC’s identical copying of Lexmark’s Toner Loading Programs went beyond that which was necessary for compatibility.” According to the district court, the program could be altered rather simply, even in view of the checksum operation, because reasonable trial and error of no more than 256 different data combinations would have enabled SCC to discover and encode the correct checksum value on its chip. In reaching this conclusion, the court downplayed the significance of Dr. Goldberg’s testimony regarding the importance of contextual information to the ease or difficulty of guessing the correct checksum value, saying that Dr. Goldberg called the task “extraordinarily difficult,” then dismissing his testimony without further explanation. Dr. Goldberg, however, did not describe this endeavor as “extraordinarily difficult” but as “computationally impossible,” (emphasis added) — a point that Lexmark did not then, and does not now, refute. Contrary to Judge Feikens’ suggestion, moreover, Lexmark offered no evidence to show that the task of “turning off” the checksum operation altogether (without contextual information) would be any different from, or any less arduous than, the task of altering the checksum value to accommodate another program.
The difficulty of deriving the proper checksum value and the corresponding degree to which the checksum operation acts as a constraint on the content of the bytes comprising the Toner Loading Program may be an open question at the permanent injunction phase. But for purposes of the preliminary injunction, Dr. Goldberg’s unchallenged testimony that it would be “computationally impossible” to modify the checksum value without contextual information suffices to establish that the checksum operation imposes a compatibility constraint in the most literal sense possible: if any single byte of the Toner Loading Program is altered, the printer will not function. On this record, pure compatibility requirements justified SCC’s copying of the Toner Loading Program.
* * *
In defense of the district court’s decision, Lexmark raises several other arguments, all unavailing. First, Lexmark notes that it “creatively inserted” in the Toner Loading Program a computer code representation of its stock ticker symbol, “LXK.” Lexmark describes this segment as “non-functional” because it does not translate into source code contributing to the toner-calculating program. It is not clear whether these three letters would support a finding of creative expression in the work as a whole. See Sega Enters., 977 F.2d at 1524 n. 7 (noting that Sega’s 20-byte initialization code is of de minimis length and therefore likely a “word” or “short phrase” that is not protected by copyright law as described in 37 C.F.R. § 202.1(a)); see also Feist, 499 U.S. at 344 (concluding that phone book listings are not copyrightable even though the listings contained four fictitious listings that the plaintiff included for infringement detection purposes). What is clear is that the bytes containing the “LXK” reference are functional in the sense that they, like the rest of the Toner Loading Program, also serve as input to the checksum operation and as a result amount to a lock-out code that the merger and scènes à faire doctrines preclude from obtaining protection.
Second, Lexmark argues that if the Toner Loading Program is not copyrightable, then “most computer programs would not be copyrightable.” But the slope of this decision is neither as slippery nor as steep as Lexmark suggests. Most computer programs do not simultaneously operate as a lock-out code that is “computationally impossible” to alter without input from the programmer; and most programs are not as brief as this one.
In reaching this conclusion, we do not mean to say that brief computer programs are ineligible for copyright protection. Short programs may reveal high levels of creativity and may present simple, yet unique, solutions to programming quandaries. Just as a mathematician may develop an elegant proof, or an author may express ideas in a spare, simple, but creative manner, see, e.g., e.e. cummings, Selected Poems (Richard S. Kennedy ed., 1994), so a computer programmer may develop a program that is brief and eligible for protection. But unless a creative flair is shown, a very brief program is less likely to be copyrightable because it affords fewer opportunities for original expression.
Third, invoking the Federal Circuit’s decision in Atari I, 975 F.2d at 840, Lexmark argues that even if the Toner Loading Program amounts to a lock-out code, it still may be eligible for protection. In Atari I, Nintendo developed a program (known as “10NES”) that blocked its game console from accepting unauthorized game cartridges. Relying on this program, Nintendo sold game cartridges that generated a data stream that “unlocked” the game console, allowing it to load and run the game. The Federal Circuit determined that the 10NES program was copyrightable despite arguments that the program constituted unprotectable ideas rather than expression and that the merger doctrine precluded copyright protection.
The Federal Circuit’s rationale for accepting copyright protection for the 10NES program does not undermine our conclusion because the 10NES program was not a “lock out” code in the same sense that the Toner Loading Program is. In Atari, the data bytes of the 10NES program did not themselves do the “unlocking” of the game console; the program, when executed, generated an arbitrary stream of data that in turn enabled the console to function. That same data stream, the court concluded, could have been produced by a number of alternative programs; for this reason, the expression contained in the computer program did not “merge” with the process. Id. (“The unique arrangement of computer program expression which generates that data stream does not merge with the process so long as alternate expressions are available. In this case, Nintendo has produced expert testimony showing a multitude of different ways to generate a data stream which unlocks the NES console.”) id. (“[N]o external factor dictated the bulk of the [10NES] program.”). Here, by contrast, the data bytes comprising the Toner Loading Program themselves act as the input to the checksum operation that must be successfully completed for the printer to operate. None of these bytes of the program can be altered without impeding printer functionality given the compatibility requirements created by the checksum operation. See Atari III, 1993 WL 207548, at *1 (distinguishing current compatibility requirements from future ones, and excluding program code “that is strictly necessary” to comply with the former from copyright protection under the merger concept). Compatibility requirements in Atari, in short, did not preclude the possibility of substituting other programs for the 10NES, while they do here.
For like reasons, Judge Feikens is correct that a poem in the abstract could be copyrightable. But that does not mean that the poem receives copyright protection when it is used in the context of a lock-out code. Similarly, a computer program may be protectable in the abstract but not generally entitled to protection when used necessarily as a lock-out device.
Notes and questions
(1) In Lexmark Int’l, Inc. v. Static Control Components, Inc., 387 F.3d 522 (6th Cir. 2004), the Sixth Circuit held that a simple computer program used as a lock-out device was not copyrightable because of the merger and scènes à faire doctrines. In a part of the decision not extracted here, the court also indicated that even if it were copyrightable the defendant’s use to enable compatible printer cartridges to work in Lexmark printers would have been fair use. Do you agree with the court’s application of the merger doctrine in this case? Could the court have applied this doctrine differently?
(2) How does the merger doctrine balance the need to protect creative works with the need to keep ideas freely available? What are the broader policy implications of denying copyright protection to functional elements like lock-out codes? How might this impact innovation and competition in the technology sector?
(3) The Sixth Circuit treats merger and scènes à faire as a matter of whether feasible alternatives exist within “real-world constraints” and not simply whether alternatives exist in theory. The court of appeals faulted the district court for “refusing to consider whether external factors such as compatibility requirements, industry standards, and efficiency circumscribed the number of forms that the Toner Loading Program could take.”
(4) Notice that the Sixth Circuit treats merger and scènes à faire as limitations on copyright scope that can arise after the work has been created. The court says that the question is not simply “whether any alternatives theoretically exist” but rather “whether other options practically exist under the circumstances” and of course, circumstances change over time. Can you frame an argument based on Baker v. Selden that this must be the correct view? For a different view, see Oracle America, Inc. v. Google Inc., 750 F.3d at 1363-1364.
(5) The Lexmark court rejects the position of the Nimmer treatise [at that time, it may have changed] that merger and scènes à faire are only applicable to assessing substantial similarity. The court held that an analysis of copyrightability informed by the doctrines of merger and scènes à faire is required to establish the existence of a valid copyright and to determine whether a particular act of reproduction amounts to infringement.
(6) Does the court’s conclusion in Lexmark that “when external factors dictate the form of expression, copyright protection does not extend to the work” make sense in absolute terms? What if Lexmark had a poem that functioned as a lock-out code, would the poem be uncopyrightable for all purposes, or simply when used in that one functionally constrained manner?
The Law Itself is Not Copyrightable
Georgia v. Public.Resource.Org, Inc. 140 S.Ct. 1498 (2020)
Chief Justice Roberts delivered the opinion of the Court.
The Copyright Act grants potent, decades-long monopoly protection for “original works of authorship.” 17 U.S.C. § 102(a). The question in this case is whether that protection extends to the annotations contained in Georgia’s official annotated code.
We hold that it does not. Over a century ago, we recognized a limitation on copyright protection for certain government work product, rooted in the Copyright Act’s “authorship” requirement. Under what has been dubbed the government edicts doctrine, officials empowered to speak with the force of law cannot be the authors of—and therefore cannot copyright—the works they create in the course of their official duties.
We have previously applied that doctrine to hold that non-binding, explanatory legal materials are not copyrightable when created by judges who possess the authority to make and interpret the law. See Banks v. Manchester, 128 U.S. 244 (1888). We now recognize that the same logic applies to non-binding, explanatory legal materials created by a legislative body vested with the authority to make law. Because Georgia’s annotations are authored by an arm of the legislature in the course of its legislative duties, the government edicts doctrine puts them outside the reach of copyright protection.
I
A
The State of Georgia has one official code—the “Official Code of Georgia Annotated,” or OCGA. The first page of each volume of the OCGA boasts the State’s official seal and announces to readers that it is “Published Under Authority of the State.”
The OCGA includes the text of every Georgia statute currently in force, as well as various non-binding supplementary materials. At issue in this case is a set of annotations that appear beneath each statutory provision. The annotations generally include summaries of judicial decisions applying a given provision, summaries of any pertinent opinions of the state attorney general, and a list of related law review articles and similar reference materials. In addition, the annotations often include editor’s notes that provide information about the origins of the statutory text, such as whether it derives from a particular judicial decision or resembles an older provision that has been construed by Georgia courts.
The OCGA is assembled by a state entity called the Code Revision Commission. In 1977, the Georgia Legislature established the Commission to recodify Georgia law for the first time in decades. The Commission was (and remains) tasked with consolidating disparate bills into a single Code for reenactment by the legislature and contracting with a third party to produce the annotations. A majority of the Commission’s 15 members must be members of the Georgia Senate or House of Representatives. The Commission receives funding through appropriations “provided for the legislative branch of state government.” OCGA § 28-9-2(c) (2018). And it is staffed by the Office of Legislative Counsel, which is obligated by statute to provide services “for the legislative branch of government.” §§ 28-4-3(c)(4), 28-9-4. Under the Georgia Constitution, the Commission’s role in compiling the statutory text and accompanying annotations falls “within the sphere of legislative authority.” Harrison Co. v. Code Revision Commission, 244 Ga. 325, 330 (1979).
Each year, the Commission submits its proposed statutory text and accompanying annotations to the legislature for approval. The legislature then votes to do three things: (1) “enact[]” the “statutory portion of the codification of Georgia laws”; (2) “merge[]” the statutory portion “with [the] annotations”; and (3) “publish[]” the final merged product “by authority of the state” as “the ‘Official Code of Georgia Annotated.’” OCGA § 1-1-1 (2019).
The annotations in the current OCGA were prepared in the first instance by Matthew Bender & Co., Inc., a division of the LexisNexis Group, pursuant to a work-for-hire agreement with the Commission. The agreement between Lexis and the Commission states that any copyright in the OCGA vests exclusively in “the State of Georgia, acting through the Commission.” Lexis and its army of researchers perform the lion’s share of the work in drafting the annotations, but the Commission supervises that work and specifies what the annotations must include in exacting detail. Under the agreement, Lexis enjoys the exclusive right to publish, distribute, and sell the OCGA. In exchange, Lexis has agreed to limit the price it may charge for the OCGA and to make an unannotated version of the statutory text available to the public online for free. A hard copy of the complete OCGA currently retails for $412.00.
B
Public.Resource.Org (PRO) is a nonprofit organization that aims to facilitate public access to government records and legal materials. Without permission, PRO posted a digital version of the OCGA on various websites, where it could be downloaded by the public without charge. PRO also distributed copies of the OCGA to various organizations and Georgia officials.
In response, the Commission sent PRO several cease-and-desist letters asserting that PRO’s actions constituted unlawful copyright infringement. When PRO refused to halt its distribution activities, the Commission sued PRO on behalf of the Georgia Legislature and the State of Georgia for copyright infringement. The Commission limited its assertion of copyright to the annotations described above; it did not claim copyright in the statutory text or numbering. PRO counterclaimed, seeking a declaratory judgment that the entire OCGA, including the annotations, fell in the public domain. …
II
We hold that the annotations in Georgia’s Official Code are ineligible for copyright protection, though for reasons distinct from those relied on by the Court of Appeals. A careful examination of our government edicts precedents reveals a straightforward rule based on the identity of the author. Under the government edicts doctrine, judges—and, we now confirm, legislators—may not be considered the “authors” of the works they produce in the course of their official duties as judges and legislators. That rule applies regardless of whether a given material carries the force of law. And it applies to the annotations here because they are authored by an arm of the legislature in the course of its official duties.
A
We begin with precedent. The government edicts doctrine traces back to a trio of cases decided in the 19th century. In this Court’s first copyright case, Wheaton v. Peters, 8 Pet. 591 (1834), the Court’s third Reporter of Decisions, Wheaton, sued the fourth, Peters, unsuccessfully asserting a copyright interest in the Justices’ opinions. In Wheaton’s view, the opinions “must have belonged to someone” because “they were new, original,” and much more “elaborate” than law or custom required. Id., at 615 (argument). Wheaton argued that the Justices were the authors and had assigned their ownership interests to him through a tacit “gift.” Id., at 614. The Court unanimously rejected that argument, concluding that “no reporter has or can have any copyright in the written opinions delivered by this court” and that “the judges thereof cannot confer on any reporter any such right.” Id., at 668 (opinion).
That conclusion apparently seemed too obvious to adorn with further explanation, but the Court provided one a half century later in Banks v. Manchester, 128 U.S. 244 (1888). That case concerned whether Wheaton’s state-court counterpart, the official reporter of the Ohio Supreme Court, held a copyright in the judges’ opinions and several non-binding explanatory materials prepared by the judges. Id., at 249-251. The Court concluded that he did not, explaining that “the judge who, in his judicial capacity, prepares the opinion or decision, the statement of the case and the syllabus or head note” cannot “be regarded as their author or their proprietor, in the sense of [the Copyright Act].” Id., at 253. Pursuant to “a judicial consensus” dating back to Wheaton, judges could not assert copyright in “whatever work they perform in their capacity as judges.” Banks, 128 U.S. at 253 (emphasis in original). Rather, “the whole work done by the judges constitutes the authentic exposition and interpretation of the law, which, binding every citizen, is free for publication to all.” Ibid. (citing Nash v. Lathrop, 142 Mass. 29 (1886)).
In a companion case decided later that Term, Callaghan v. Myers, 128 U.S. 617 (1888), the Court identified an important limiting principle. As in Wheaton and Banks, the Court rejected the claim that an official reporter held a copyright interest in the judges’ opinions. But, resolving an issue not addressed in Wheaton and Banks, the Court upheld the reporter’s copyright interest in several explanatory materials that the reporter had created himself: headnotes, syllabi, tables of contents, and the like. Callaghan, 128 U.S. at 645. Although these works mirrored the judge-made materials rejected in Banks, they came from an author who had no authority to speak with the force of law. Because the reporter was not a judge, he was free to “obtain[] a copyright” for the materials that were “the result of his [own] intellectual labor.” 128 U.S. at 647.
These cases establish a straightforward rule: Because judges are vested with the authority to make and interpret the law, they cannot be the “author” of the works they prepare “in the discharge of their judicial duties.” Banks, 128 U.S. at 253. This rule applies both to binding works (such as opinions) and to non-binding works (such as headnotes and syllabi). It does not apply, however, to works created by government officials (or private parties) who lack the authority to make or interpret the law, such as court reporters.
The animating principle behind this rule is that no one can own the law. “Every citizen is presumed to know the law,” and “it needs no argument to show... that all should have free access” to its contents. Nash, 142 Mass. at 35 (cited by Banks, 128 U.S. at 253-254). Our cases give effect to that principle in the copyright context through construction of the statutory term “author.” Id., at 253. Rather than attempting to catalog the materials that constitute “the law,” the doctrine bars the officials responsible for creating the law from being considered the “author[s]” of “whatever work they perform in their capacity” as lawmakers. Ibid. (emphasis added). Because these officials are generally empowered to make and interpret law, their “whole work” is deemed part of the “authentic exposition and interpretation of the law” and must be “free for publication to all.” Ibid.
If judges, acting as judges, cannot be “authors” because of their authority to make and interpret the law, it follows that legislators, acting as legislators, cannot be either. Courts have thus long understood the government edicts doctrine to apply to legislative materials. See, e.g., Nash, 142 Mass. at 35, 6 N.E. at 560 (judicial opinions and statutes stand “on substantially the same footing” for purposes of the government edicts doctrine); Howell v. Miller, 91 F. 129, 130-131, 137-138 (CA6 1898) (Harlan, J., Circuit Justice, joined by then-Circuit Judge Taft) (analyzing statutes and supplementary materials under Banks and Callaghan and concluding that the materials were copyrightable because they were prepared by a private compiler).
Moreover, just as the doctrine applies to “whatever work [judges] perform in their capacity as judges,” Banks, 128 U.S., at 253, it applies to whatever work legislators perform in their capacity as legislators. That of course includes final legislation, but it also includes explanatory and procedural materials legislators create in the discharge of their legislative duties. In the same way that judges cannot be the authors of their headnotes and syllabi, legislators cannot be the authors of (for example) their floor statements, committee reports, and proposed bills. These materials are part of the “whole work done by [legislators],” so they must be “free for publication to all.” Ibid.
Under our precedents, therefore, copyright does not vest in works that are (1) created by judges and legislators (2) in the course of their judicial and legislative duties.
B
1
Applying that framework, Georgia’s annotations are not copyrightable. The first step is to examine whether their purported author qualifies as a legislator.
As we have explained, the annotations were prepared in the first instance by a private company (Lexis) pursuant to a work-for-hire agreement with Georgia’s Code Revision Commission. The Copyright Act therefore deems the Commission the sole “author” of the work. 17 U.S.C. § 201(b). Although Lexis expends considerable effort preparing the annotations, for purposes of copyright that labor redounds to the Commission as the statutory author. Georgia agrees that the author is the Commission.
The Commission is not identical to the Georgia Legislature, but functions as an arm of it for the purpose of producing the annotations. The Commission is created by the legislature, for the legislature, and consists largely of legislators. The Commission receives funding and staff designated by law for the legislative branch. Significantly, the annotations the Commission creates are approved by the legislature before being “merged” with the statutory text and published in the official code alongside that text at the legislature’s direction. …
2
The second step is to determine whether the Commission creates the annotations in the “discharge” of its legislative “duties.” Banks, 128 U.S. at 253. It does. Although the annotations are not enacted into law through bicameralism and presentment, the Commission’s preparation of the annotations is under Georgia law an act of legislative authority, and the annotations provide commentary and resources that the legislature has deemed relevant to understanding its laws. Georgia and Justice GINSBURG emphasize that the annotations do not purport to provide authoritative explanations of the law and largely summarize other materials, such as judicial decisions and law review articles. But that does not take them outside the exercise of legislative duty by the Commission and legislature. Just as we have held that the “statement of the case and the syllabus or head note” prepared by judges fall within the “work they perform in their capacity as judges,” Banks, 128 U.S. at 253, so too annotations published by legislators alongside the statutory text fall within the work legislators perform in their capacity as legislators.
In light of the Commission’s role as an adjunct to the legislature and the fact that the Commission authors the annotations in the course of its legislative responsibilities, the annotations in Georgia’s Official Code fall within the government edicts doctrine and are not copyrightable.
III
Georgia resists this conclusion on several grounds. … Georgia suggests that we should resist applying our government edicts precedents to the OCGA annotations because our 19th-century forebears interpreted the statutory term author by reference to “public policy”—an approach that Georgia believes is incongruous with the “modern era” of statutory interpretation. But we are particularly reluctant to disrupt precedents interpreting language that Congress has since reenacted. As we explained last Term in Helsinn Healthcare S. A. v. Teva Pharmaceuticals USA, Inc., 139 S.Ct. 628 (2019), when Congress adopts the language used in an earlier act, we presume that Congress adopted also the construction given by this Court to such language, and made it a part of the enactment. A century of cases have rooted the government edicts doctrine in the word “author,” and Congress has repeatedly reused that term without abrogating the doctrine. The term now carries this settled meaning, and “critics of our ruling can take their objections across the street, [where] Congress can correct any mistake it sees.” Kimble v. Marvel Entertainment, LLC, 576 U.S. 446, 456 (2015).
…
Turning to our government edicts precedents, Georgia insists that they can and should be read to focus exclusively on whether a particular work has “the force of law.” Justice Thomas appears to endorse the same view. But that framing has multiple flaws.
Most obviously, it cannot be squared with the reasoning or results of our cases—especially Banks. Banks, following Wheaton and the “judicial consensus” it inspired, denied copyright protection to judicial opinions without excepting concurrences and dissents that carry no legal force. 128 U.S. at 253 (emphasis deleted). As every judge learns the hard way, “comments in [a] dissenting opinion” about legal principles and precedents “are just that: comments in a dissenting opinion.” Railroad Retirement Bd. v. Fritz, 449 U.S. 166, 177, n. 10 (1980). Yet such comments are covered by the government edicts doctrine because they come from an official with authority to make and interpret the law.
Indeed, Banks went even further and withheld copyright protection from headnotes and syllabi produced by judges. 128 U.S. at 253. Surely these supplementary materials do not have the force of law, yet they are covered by the doctrine. The simplest explanation is the one Banks provided: These non-binding works are not copyrightable because of who creates them—judges acting in their judicial capacity. See ibid.
The same goes for non-binding legislative materials produced by legislative bodies acting in a legislative capacity. There is a broad array of such works ranging from floor statements to proposed bills to committee reports. Under the logic of Georgia’s “force of law” test, States would own such materials and could charge the public for access to them. …
Georgia minimizes the OCGA annotations as non-binding and non-authoritative, but that description undersells their practical significance. Imagine a Georgia citizen interested in learning his legal rights and duties. If he reads the economy-class version of the Georgia Code available online, he will see laws requiring political candidates to pay hefty qualification fees (with no indigency exception), criminalizing broad categories of consensual sexual conduct, and exempting certain key evidence in criminal trials from standard evidentiary limitations—with no hint that important aspects of those laws have been held unconstitutional by the Georgia Supreme Court. Meanwhile, first-class readers with access to the annotations will be assured that these laws are, in crucial respects, unenforceable relics that the legislature has not bothered to narrow or repeal.
If everything short of statutes and opinions were copyrightable, then States would be free to offer a whole range of premium legal works for those who can afford the extra benefit. A State could monetize its entire suite of legislative history. With today’s digital tools, States might even launch a subscription or pay-per-law service.
There is no need to assume inventive or nefarious behavior for these concerns to become a reality. Unlike other forms of intellectual property, copyright protection is both instant and automatic. It vests as soon as a work is captured in a tangible form, triggering a panoply of exclusive rights that can last over a century. 17 U.S.C. §§ 102, 106, 302. If Georgia were correct, then unless a State took the affirmative step of transferring its copyrights to the public domain, all of its judges’ and legislators’ non-binding legal works would be copyrighted. And citizens, attorneys, nonprofits, and private research companies would have to cease all copying, distribution, and display of those works or risk severe and potentially criminal penalties. §§ 501-506. Some affected parties might be willing to roll the dice with a potential fair use defense. But that defense, designed to accommodate First Amendment concerns, is notoriously fact sensitive and often cannot be resolved without a trial. Cf. Harper & Row, Publishers, Inc. v. Nation Enterprises, 471 U.S. 539, 552 (1985). The less bold among us would have to think twice before using official legal works that illuminate the law we are all presumed to know and understand.
Thankfully, there is a clear path forward that avoids these concerns—the one we are already on. Instead of examining whether given material carries “the force of law,” we ask only whether the author of the work is a judge or a legislator. If so, then whatever work that judge or legislator produces in the course of his judicial or legislative duties is not copyrightable. That is the framework our precedents long ago established, and we adhere to those precedents today.
*
Justice GINSBURG, with whom Justice BREYER joins, dissenting.
Beyond doubt, state laws are not copyrightable. Nor are other materials created by state legislators in the course of performing their lawmaking responsibilities, e.g., legislative committee reports, floor statements, unenacted bills. Not all that legislators do, however, is ineligible for copyright protection; the government edicts doctrine shields only “works that are (1) created by judges and legislators (2) in the course of their judicial and legislative duties.” Ante, at 1508 (emphasis added). The core question this case presents, as I see it: Are the annotations in the Official Code of Georgia Annotated (OCGA) done in a legislative capacity? The answer, I am persuaded, should be no.
To explain why, I proceed from common ground. All agree that headnotes and syllabi for judicial opinions—both a kind of annotation—are copyrightable when created by a reporter of decisions, Callaghan v. Myers, 128 U.S. 617 (1888), but are not copyrightable when created by judges, Banks v. Manchester, 128 U.S. 244, 253 (1888). That is so because “[t]he whole work done by ... judges,” ibid., including dissenting and concurring opinions, ranks as work performed in their judicial capacity. Judges do not outsource their writings to arms or adjuncts, to be composed in their stead. Accordingly, the judicial opinion-drafting process in its entirety—including the drafting of headnotes and syllabi, in jurisdictions where that is done by judges—falls outside the reach of copyright protection.
One might ask: If a judge’s annotations are not copyrightable, why are those created by legislators? The answer lies in the difference between the role of a judge and the role of a legislator. “[T]o the judiciary” we assign “the duty of interpreting and applying” the law, Massachusetts v. Mellon, 262 U.S. 447, 488 (1923), and sometimes making the applicable law, see Friendly, In Praise of Erie—and of the New Federal Common Law, 39 N.Y.U. L. Rev. 383 (1964). See also Marbury v. Madison, 1 Cranch 137, 177 (1803) (“It is emphatically the province and duty of the judicial department to say what the law is.”). In contrast, the role of the legislature encompasses the process of “making laws”—not construing statutes after their enactment. Mellon, 262 U.S. at 488; see Patchak v. Zinke, 138 S.Ct. 897, 905 (2018) (plurality opinion) (“The legislative power is the power to make law.”). The OCGA annotations, in my appraisal, do not rank as part of the Georgia Legislature’s lawmaking process for three reasons.
First, the annotations are not created contemporaneously with the statutes to which they pertain; instead, the annotations comment on statutes already enacted. … In short, annotating begins only after lawmaking ends. This sets the OCGA annotations apart from uncopyrightable legislative materials like committee reports, generated before a law’s enactment, and tied tightly to the task of law-formulation.
Second, the OCGA annotations are descriptive rather than prescriptive. Instead of stating the legislature’s perception of what a law conveys, the annotations summarize writings in which others express their views on a given statute. For example, the OCGA contains “case annotations” for “all decisions of the Supreme Court of Georgia and the Court of Appeals of Georgia and all decisions of the federal courts in cases which arose in Georgia construing any portion of the general statutory law of the state.” Id., at 403. Per the Code Revision Commission’s instructions, each annotation should “accurately reflect the facts, holding, and statutory construction” adopted by the court. Id., at 404. The annotations are neutrally cast; they do not opine on whether the summarized case was correctly decided. See, e.g., OCGA § 17-7-50 (2013) (case annotation summarizing facts and holdings of nine cases construing right to grand jury hearing). This characteristic of the annotations distinguishes them from preenactment legislative materials that touch or concern the correct interpretation of the legislature’s work.
Third, and of prime importance, the OCGA annotations are “given for the purpose of convenient reference” by the public, § 1-1-7 (2019); they aim to inform the citizenry at large, they do not address, particularly, those seated in legislative chambers.2
Footnote 2: Suppose a committee of Georgia’s legislature, to inform the public, instructs a staffer to write a guide titled “The Workways of the Georgia Legislature.” The final text describing how the legislature operates is circulated to members of the legislature and approved by a majority. Contrary to the Court’s decision, I take it that such a work, which entails no lawmaking, would be copyrightable.
Annotations are thus unlike, for example, surveys, work commissioned by a legislature to aid in determining whether existing law should be amended.
The requirement that the statutory portions of the OCGA “shall be merged with annotations,” § 1-1-1, does not render the annotations anything other than explanatory, referential, or commentarial material. Annotations aid the legal researcher, and that aid is enhanced when annotations are printed beneath or alongside the relevant statutory text. But the placement of annotations in the OCGA does not alter their auxiliary, nonlegislative character.
* * *
Because summarizing judicial decisions and commentary bearing on enacted statutes, in contrast to, for example, drafting a committee report to accompany proposed legislation, is not done in a legislator’s law-shaping capacity, I would hold the OCGA annotations copyrightable and therefore reverse the judgment of the Court of Appeals for the Eleventh Circuit.
[Justice THOMAS, with whom Justice ALITO joined, and with whom Justice BREYER joined in part, also dissented.]
Notes and questions
(1) In Wheaton v. Peters, 33 U.S. (8 Pet.) 591 (1834) the Supreme Court held that “no reporter has or can have any copyright in the written opinions delivered by this Court; and that the judges thereof cannot confer on any reporter any such right.” In Banks v. Manchester, 128 U.S. 244 (1888) the Court addressed the same issue with respect to the opinions of state court judges and held that “no copyright could under the statutes passed by Congress, be secured in the products of the labor done by judicial officers in the discharge of their judicial duties.” But in Callaghan v. Myers, 128 U.S. 617 (1888) the Court held that although “there can be no copyright in the opinions of the judges, or in the work done by them in their official capacity as judges,” the additional intellectual contribution of the court reporter—things such as organizing the cases; writing annotations such as headnotes and syllabi to appear alongside the opinions in the reports—may be entitled to copyright protection in the hands of the reporter as a private individual. What does Georgia v. Public.Resource.Org add to these cases?
In Georgia v. Public.Resource.Org, Inc. 140 S.Ct. 1498 (2020) the Supreme Court held that the Official Code of Georgia Annotated, a document that was published under the authority of the state and yet contained non-binding annotations on and summaries of Georgia law, was ineligible for copyright protection under the “government edicts doctrine.” Writing for the majority, Chief Justice Roberts explained that “officials empowered to speak with the force of law cannot be the authors of—and therefore cannot copyright—the works they create in the course of their official duties.”
(2) Section 8 of the Copyright Act of 1909 provided that “no copyright shall subsist in the original text of any work which is in the public domain ... or in any publication of the United States Government, or any reprint, in whole or in part, thereof.” A similar provision persists in section 105 of the current Act.
17 U.S. Code § 105. Subject matter of copyright: United States Government works
Copyright protection under this title is not available for any work of the United States Government, but the United States Government is not precluded from receiving and holding copyrights transferred to it by assignment, bequest, or otherwise.
Why wasn’t Georgia v. Public.Resource.Org decided on the basis of Section 105?
(3) In Veeck v. Southern Building Code Congress International, Inc., 293 F.3d 791 (5th Cir. 2002) (en banc), the Fifth Circuit held that privately authored building codes lost their copyright status once they were incorporated into law. However, various other courts have declined to extend the rule in other similar contexts. See, e.g., CCC Info. Servs., Inc. v. Maclean Hunter Mkt. Reports, Inc., 44 F.3d 61 (2d Cir. 1994); Practice Mgmt. Info. Corp. v. Am. Med. Ass’n, 121 F.3d 516 (9th Cir. 1997), amended, 133 F.3d 1140 (9th Cir. 1998).
(4) A recent Fifth Circuit opinion extends the logic of Veeck to codes incorporated into foreign law. In Canadian Standards Association v. P.S. Knight Company, 112 F.4th 298 (5th Cir. 2024), cert. denied, No. 24-537 (U.S. Jan. 21, 2025), a majority held that copying Canadian Standards Association’s codes was not actionable under the United States’ Copyright Act, because those codes had become “the law” of Canada, being fully incorporated by reference into at least one Canadian statute or regulation.
Fair use as an alternative to copyrightability for codes that become law?
Should issues relating to copyright in privately authored building codes and the like be addressed in terms of copyrightability or fair use? The court of appeals for the DC Circuit turned to this issue in American Society For Testing And Materials v. Public.Resource.Org, Inc. 896 F.3d 437 (D.C. Cir. 2018). The case involves our old friend Public.Resource.Org, again. As part of its self-proclaimed mission “to make the law and other government materials more widely available,” Public.Resource.Org (“PRO”) posted copies of a number of incorporated standards to its public website. Between 2012 and 2014, PRO uploaded hundreds of technical standards, which, collectively, were downloaded tens of thousands of times. In mid-2013, several standards organizations, including the American Society for Testing & Materials sued PRO, asserting claims of copyright infringement, among other things.
Tens of thousands of technical standards developed by private organizations are incorporated into or referenced by law in different ways and with varying consequences. As the court explained:
When agencies or legislatures incorporate private standards into law, they often do so by reference—that is, instead of spelling out the requirements of a standard within legislative or regulatory text, they reference the standard being incorporated and direct interested parties to consult that standard in order to understand their obligations. … Just as the incorporation process varies, so too—and this is central to the issues before us—do the legal consequences of any given incorporation. This is hardly surprising, given that federal, state, and local legislatures and agencies have incorporated by reference thousands of technical standards. Indeed, by ASTM’s own count, the Code of Federal Regulations alone has incorporated by reference over 1,200 of its standards. This appeal, which concerns ten standards incorporated by reference into law, reflects just a sliver of that diversity.
One way in which the incorporated standards vary is how readily they resemble ordinary, binding law. At one end of this spectrum lie incorporated standards that define one’s legal obligations just as much as, say, a local building code—except that the specific legal requirements are found outside the two covers of the codebook. The NFPA 70 tank-barge plug specification discussed above, which the relevant regulation mentions by name in making compliance mandatory, is one such example. Another is the incorporation of ASTM D975-07, the “Standard Specification for Diesel Fuel Oils,” into the U.S. Code. It provides that a retailer of certain biofuels need not affix any special labels to its fuel so long as the fuel “meets ASTM D975 diesel specifications.” 42 U.S.C. § 17021(b)(1). These laws impose legally binding requirements indistinguishable from, for example, a cigarette-labeling obligation, see 15 U.S.C. § 1333(a), except that the federal law imposing that obligation expressly specifies, without reference to an external standard, exactly what qualifies as a cigarette, see id. § 1332(1).
At the other end of the spectrum lie standards that serve as mere references but have no direct legal effect on any private party’s conduct. One example is the incorporation of ASTM D86-07, the “Standard Test Method for Distillation of Petroleum Products and Liquid Fuels at Atmospheric Pressure,” which a federal regulation describes as a “reference procedure” used by the Environmental Protection Agency and regulated motor-vehicle manufacturers to determine whether the boiling point for certain gasoline used for “exhaust and evaporative emission testing” falls within a permissible range. 40 C.F.R. § 86.113-04(a)(1). The regulation creates only one relevant legal obligation: the regulated entity, in testing vehicular emissions, must use gasoline that meets specifications expressly laid out within the regulation itself. The incorporation of an external standard merely tells the regulated entity how it can ensure that the gasoline it uses in fact satisfies the codified requirements.
Judge Tatel’s majority opinion in the ASTM case acknowledged that PRO had raised “a serious constitutional concern with permitting private ownership of standards essential to understanding legal obligations.” However, the court thought “it best at this juncture to address only the statutory fair use issue—which may provide a full defense to some, if not all, of the [Standards Developing Organizations’] infringement claims in this case—and leave for another day the question of whether the Constitution permits copyright to persist in works incorporated by reference into law.”
In terms of fair use, the majority noted that PRO had distributed the standards for the purpose of educating the public about the specifics of governing law and it generally credited the argument that paraphrases, summaries, and descriptions, would not adequately “capture the precision that is necessary to understand the legal obligations that governments impose and enforce.” But the court regarded this as a highly fact specific inquiry. Citing Banks, the court reiterated that “the express text of the law falls plainly outside the realm of copyright protection” and held further that “standards incorporated by reference into law are, at best, at the outer edge of copyright’s protective purposes.” Judge Tatel further explained that:
Where the consequence of the incorporation by reference is virtually indistinguishable from a situation in which the standard had been expressly copied into law, this factor weighs heavily in favor of fair use. But where the incorporation does not lend to such easy substitution, fair use is harder to justify.
The court of appeals instructed the district court to take these principles into account on remand and also to conduct a more nuanced analysis of the potential market effect of PRO’s activities. On remand the district court entered summary judgment for Public Resource as to the 184 standards that had been directly incorporated by reference, 597 F. Supp. 3d 213 (D.D.C. 2022), and the D.C. Circuit affirmed, holding that the non-commercial dissemination of standards incorporated into law is a fair use. American Society for Testing & Materials v. Public.Resource.Org, Inc., 82 F.4th 1262 (D.C. Cir. 2023). Notably, the 2023 decision qualified the suggestion in the 2018 opinion that what weighs most heavily in favor of fair use is the copying of material of “legal import.” The most important question, the court said, is what material counts as law, and even explanatory or background material within an incorporated standard may aid in understanding the legal duties it imposes. Copying portions of a standard that are not themselves mandatory is therefore not fatal to a fair use defense. Judge Katsas’ concurring opinion in the 2018 decision is extracted below.
American Society for Testing & Materials v. Pub.Resource.Org, 896 F.3d 437 (D.C. Cir. 2018)
Circuit Judge Katsas concurring:
The plaintiffs here claim a copyright over binding legal texts, which would enable them to prevent anyone from gaining access to that law or copying it for the public. See 17 U.S.C. § 106. Moreover, saying what that law is, without plaintiffs’ permission, would expose an individual to injunctive relief, impoundment, damages, attorneys’ fees, and potentially even criminal liability. See id. §§ 502-506. As a matter of common-sense, this cannot be right: access to the law cannot be conditioned on the consent of a private party, just as it cannot be conditioned on the ability to read fine print posted on high walls. See Suetonius, Gaius Caligula ¶ XLI, in The Lives of the Caesars (J.C. Rolfe trans., Macmillan Co. 1914) (“he . . . had the law posted up, but in a very narrow place and in excessively small letters, to prevent the making of a copy”).
Not surprisingly, precedent confirms this instinct. In Banks v. Manchester, 128 U.S. 244 (1888), the Supreme Court held that judges cannot copyright their opinions, in part because their work “constitutes the authentic expression and interpretation of the law, which, binding every citizen, is free for publication to all.” Id. at 253. Moreover, two courts of appeals have confirmed that Banks remains good law under the modern Copyright Act of 1976. In Building Officials & Code Administrators v. Code Technology, Inc., 628 F.2d 730 (1st Cir. 1980), the First Circuit vacated a preliminary injunction that would have enforced the copyright of a model building code as enacted into Massachusetts law. While not definitively deciding the question, the court reasoned that enforcement of the copyright could not be “squared with the right of the public to know the law to which it is subject.” Id. at 735. Similarly, in Veeck v. Southern Building Code Congress International, 293 F.3d 791 (5th Cir. 2002) (en banc), the Fifth Circuit held that “as law,” model rules adopted by a legislative body “enter the public domain and are not subject to the copyright holder’s exclusive prerogatives.” Id. at 793.
Today, the Banks rule might rest on at least four possible grounds: the First Amendment; the Due Process Clause of the Fifth Amendment; Section 102(b) of the Copyright Act, which denies copyright protection to “any idea, procedure, process, system, method of operation, concept, principle, or discovery,” 17 U.S.C. § 102(b); or Section 107 of the Act, which sets forth the fair-use doctrine, id. § 107. The Court today reasonably avoids what it correctly regards as “a serious constitutional concern” under the First and Fifth Amendments. And it expressly reserves, in substance though not by name, the question whether Section 102(b) extends protection to private standards as enacted into law.
The Court’s fair-use analysis faithfully recites the governing four-factor balancing test, yet, in conducting the balancing, it puts a heavy thumb on the scale in favor of an unrestrained ability to say what the law is. Thus, when an incorporated standard sets forth binding legal obligations, and when the defendant does no more and no less than disseminate an exact copy of it, three of the four relevant factors—purpose and character of the use, nature of the copyrighted work, and amount and substantiality of the copying—are said to weigh “heavily” or “strongly” in favor of fair use. This analysis closely parallels Banks, which the Court explicitly invokes in its discussion of factor two. The Court acknowledges the thinness of the record in this case, and it appropriately flags potentially complicating questions about how particular standards may be incorporated into law, and whether such standards, as so incorporated, actually constitute “the law.” But, where a particular standard is incorporated as a binding legal obligation, and where the defendant has done nothing more than disseminate it, the Court leaves little doubt that the dissemination amounts to fair use.
With that understanding, and recognizing that the Section 102(b) and constitutional issues remain open in the unlikely event that disseminating “the law” might be held not to be fair use, I join the Court’s opinion.
Notes and questions
(1) In his concurring opinion in ASTM, Judge Katsas seems intent on leaving us in no doubt that publishing copyrighted works incorporated as binding legal obligations will be fair use. For another recent case on this issue, see International Code Council, Inc. v. UpCodes, Inc., 2020 WL 2750636 (S.D.N.Y. May 26, 2020).
(2) Judge Katsas’ instinct now looks like the mainstream. In American Society for Testing & Materials v. UpCodes, Inc., No. 24-2965 (3d Cir. Apr. 7, 2026), the Third Circuit affirmed the denial of a preliminary injunction against a for-profit platform that reproduced ASTM standards incorporated into the building codes of Philadelphia and other jurisdictions. Following the D.C. Circuit, the court held UpCodes likely to succeed on fair use because its use “achieves the distinct objective of making the law freely accessible.” Two things about that holding are worth noticing. It is a preliminary posture, so the court held only that UpCodes was likely to succeed, not that it had won. And it is a fair use holding and nothing more: the court did not decide whether the standards are copyrightable, and it did not decide the government edicts question.
That reticence is the pattern rather than the exception. Courts confronted with privately drafted standards that have been absorbed into law appear increasingly willing to resolve the dispute through fair use rather than by holding that the material has ceased to be copyrightable at all. There is a reason for the caution. Georgia v. Public.Resource.Org is in some respects a sweeping decision, but follow the logic of the government edicts doctrine and it arguably does not reach this situation: the doctrine rests on the proposition that the people are the constructive authors of the law, and the material at issue here was not written by the government in the first place. Fair use lets a court reach a sensible result without having to say that a standards body’s work stops being its own the moment a legislature points at it.
(3) Is transformative use the right frame for these cases? The Third Circuit reasoned that making the law freely accessible is a distinct purpose from the one ASTM had in publishing the standard. That is a recognizable transformative use argument, but it may be doing work that a more direct argument would do better. What is really carrying these decisions is not that the second use communicates a different message; it is that the public has a powerful interest in reading the rules it is bound by. Is it clarifying or obscuring to route that interest through the first fair use factor? Would the analysis be more honest if courts said plainly that the public interest in access to law justifies the copying, whether or not the use is transformative in the sense Campbell and Warhol contemplate?
(4) Congress has been asked to reverse the trend. In March 2026 Senators Coons, Cornyn, Hirono and Tillis reintroduced the Pro Codes Act, which would preserve copyright in standards incorporated by reference into law so long as the standard is made available to read free of charge on a publicly accessible website. It is a direct legislative response to Veeck, P.S. Knight and the ASTM line, and it has not been enacted. Should it pass, it would raise a question the courts have so far avoided: whether Congress may confer copyright in the text of binding legal obligations at all. If the government edicts doctrine is merely a rule of statutory construction, Congress can displace it. If it reflects something in the constitutional grant, or in due process notions of what it means for law to be knowable, a statute that keeps the law under copyright would be vulnerable.
Other Exclusions from Copyright Protection
We already addressed the non-copyrightability of words, titles, and short phrases in an earlier chapter under the heading of “Locating the Threshold of Copyrightability.”
Taxonomies
There is some case law to the contrary, but the better view is that taxonomies of arbitrary naming conventions are not copyrightable. For example in Southco, Inc. v. Kanebridge Corp., 390 F.3d 276 (3d Cir. 2004) (en banc), the Third Circuit held that a list of serial numbers used to identify hardware parts was not copyrightable.
After discussing the originality requirement and the implications of the Feist decision, the court held:
In this case, the Southco product numbers are not “original” because each number is rigidly dictated by the rules of the Southco system. Because ideas may not be copyrighted, Southco does not assert any claim of copyright in its numbering system, but instead focuses on the part numbers themselves. The numbers, however, do not reflect any creativity.
The court also held that the part numbers were uncopyrightable on a separate ground, the uncopyrightability of short phrases and titles. The court explained that “since at least 1899, it has been the practice of the Copyright Office to deny registration to ‘words and phrases.’” It summarized and accepted the policy rationale for the doctrine put forward by the government in an amicus brief as follows:
The government suggests that this practice serves at least two purposes. First, the government notes that a short phrase such as a part number typically lacks any creativity whatsoever. Second, the government suggests that extending copyright protection to part numbers would unduly interfere with the legitimate use of the numbers in question. Because the owner of a copyright has the exclusive rights to reproduce the copyrighted work, if a part number (say, XXXXXXXXX, to take the example discussed above) were copyrighted, any use of the number would potentially infringe the copyright. Moreover, if Southco’s nine-digit numbers are protected, would there be a principled basis for denying protection to a number with, say, seven or five digits? Could a company or person thereby obtain the exclusive right to use the number 4,710,202 or 47,102? In light of the huge number of part and product numbers (and other analogous numbers) that now exist, this prospect gives reason for concern. Although the fair use defense would presumably protect the use of such numbers in most situations, fair use is an affirmative defense and may impose an undue burden.
The law in the Sixth Circuit is the same. In ATC Distribution Group v. Whatever It Takes Transmissions & Parts, Inc., 402 F.3d 700, 706 (6th Cir. 2005), the court of appeals held that a taxonomy for assigning unique identifiers to auto transmission parts by sorting them into categories and sub-categories was not copyrightable. Just as in Southco, the court recognized the system that generated the unique identifiers as an uncopyrightable idea and the numbers so generated as either unoriginal products of that system or as excluded by merger. The court also invoked the short works doctrine following a similar reasoning to Southco.
For a contrary view, see American Dental Association v. Delta Dental Plans Association, 126 F.3d 977, 979 (7th Cir. 1997), arguing that:
Facts do not supply their own principles of organization. Classification is a creative endeavor. Butterflies may be grouped by their color, or the shape of their wings, or their feeding or breeding habits, or their habitats, or the attributes of their caterpillars, or the sequence of their DNA; each scheme of classification could be expressed in multiple ways. Dental procedures could be classified by complexity, or by the tools necessary to perform them, or by the parts of the mouth involved, or by the anesthesia employed, or in any of a dozen different ways. The Code’s descriptions don’t “merge with the facts” any more than a scientific description of butterfly attributes is part of a butterfly.
The Seventh Circuit’s opinion in American Dental appears to confuse originality in the system with originality in the codes that are the product of that system.