dbasepaper

CAN A COMPUTER LANGUAGE BE COPYRIGHTED? THE STATE OF CONFUSION IN COMPUTER COPYRIGHT LAW by Steve C. Posner

From the legal and business standpoint, this paper is a bit dated. Nobody has tried to copyright a computer language since Ashton-Tate's initial attempt. Ashton-Tate no longer exists as a company. New paradigms of computer copyright law, such as the abstraction-filtration-comparison analysis, have replaced some the models examined in this paper. However, I believe this paper -- which won Nathan Burkan and Computer Law Writing Competition awards and has been both published and anthologized -- stands as a worthwhile analysis of the application of essential copyright principles to computer languages, as well as a useful analysis for lawyers of what makes up a computer language. It is reprinted with Computer/Law Journal permission.

If a particular section listed in the Table of Contents interests you, click on that line in the Table and you will be brought right to the desired section. Click on the superscripted footnote numbers to be brought to the chosen footnote.

Table of Contents

1. The Attempt to Copyright a Computer Language: a Focus for Inquiry

2. Potential Ramifications for the Computer Industry

3. The Purpose of Copyright: to Promote the Progress of Science and Useful Arts

4. The Idea/Expression Dichotomy and Computer Languages

5. The Theoretical Nature of a Computer Language: Idea

6. The Context of a Computer Language: Compilers and Other Expressions

7. Over-broad Definition of a Program's Idea: the Worm in the Apple Cases

8. Copyright of Operating Systems: Blurring the Distinction Between Copyright and Patent
Coverage

9. "Structure, Sequence and Design": the Whelan v.Jastrow Non-breakthrough

10. Application of "Structure, Sequence and Design" to Computer Languages

11. "Structure, Sequence and Design" v. "Look and Feel": the Present State of Confusion. A
Model for Use in Accord with Traditional Literary/Audio-visual Copyright Distinctions.

12. Broderbund v. Unison World: "Structure, Sequence and Design" Run Amok

13. "Look and Feel" Applied to Computer Languages

14. The Copyright Policy of 1988: Creature of Neither Statute nor Case Law

15. Do We Need New Copyright Legislation for Computers?

16. Summary and Conclusion

1. The Attempt to Copyright a Computer Language: A Focus for Inquiry

Thousands of jobs and the future of the computer industry will be affected when a court decides whether a computer language can be copyrighted. The issue, never before litigated, arises in the context of a suit now pending in the United States District Court, Central District of California, Ashton-Tate Corporation v. Fox Software, Inc.

Hugely popular business database programs, dBASEII and its successors, dBASEIII and dBASEIV, were created and copyrighted by Ashton-Tate in 1984, 1986 and 1988, respectively. Foxbase+, marketed as a dBASEIII "workalike," began to be marketed by Fox Software in 1986. Although the issue of computer language copyright is not mentioned in the complaint,¹ Ashton-Tate's acquiescence to public use of the dBASE language is posited as an affirmative defense in Fox's answer,² and attorneys for both Plaintiff³and Defendant⁴ say the issue will be litigated. Edward Esber, Chairman and Chief Executive Officer of Ashton-Tate, wrote in an open letter to the software industry, "Ashton-Tate believes that the structure and organization of the dBASE programs are protectible under copyright law, including the screens, menus, basic file structures and dBASE language embodied in these products."⁵

This paper focuses on the copyrightability of computer languages for two reasons: First, because it may well happen that a language is held copyrightable, with results the author believes would be disastrous. Second, because a language serves as a metaphor around which can be organized an inquiry into the confusion which is now the general state of computer copyright law. This paper seeks to develop some rational and functional analytical framework consistent with existing copyright law. The call of Sholkoff,⁶ among others, for new computer copyright legislation is also briefly examined.

2. Potential Ramifications for the Computer Industry

At present, approximately 100,000 programmers⁷write professionally in the dBASE dialects.⁸If the court finds that Ashton-Tate owns a copyright in the dBASE language, these programmers and the users of a vast installed base of their programs will be liable for copyright infringement. All future programs written in the dBASE language will be subject to licensing fees and/or royalties, which will force up the price of development. Companies that have developed dBASE variants may find themselves enjoined from further sale of their products, should those products eat into Ashton-Tate's market and should Ashton-Tate deny their creators licenses. Should they be so enjoined, those features that exist in the enjoined products and do not exist in Ashton-Tate's, will disappear from the marketplace.

For purposes of this discussion, laches, acquiescence and any legal other issues beyond the nature of what can be copyrighted will be ignored. We are not concerned with the resolution of the particular case, Ashton-Tate v. Fox, because the issue goes beyond the dBASE language. If one language is copyrightable, are not all languages which are not unquestionably within the public domain also subject to copyright? Such possessory interests would run contrary to a major trend in the computer industry: the movement toward software compatibility. Companies throughout the industry have moved to the UNIX operating system because it can function on machines made by many different hardware vendors. Software developers have made "C" one of the industry's favorite languages because programs written in it can run in many different operating system environments. C was created by Dennis M. Ritchie of AT & T. AT & T also created UNIX, the operating system under which C originally ran. C grew and developed because, as Ritchie himself wrote, "C has proven to be a pleasant, expressive and versatile language for a wide variety of programs."⁹ Ray Duncan, a columnist for Dr. Dobb's Journal, a popular programmer's trade magazine, attributes the growth of C to "the penetration into academia that was made possible by Bell Laboratory's broadminded licensing plans for universities [with regard to UNIX --SP]."¹⁰ The result of easy access to C has been the development of an industry-wide pool of programmers expert in the language. The growth of the dBASE language has been similar; as dBASEIII and its variants became ubiquitous, programmers flocked to it and an industry-wide skill-pool developed. This paper argues that, should access to either of these languages have been restricted, programs written in these languages would probably have been restricted to the environments approved by the language creators, marketability of skills developed in these languages would also have been limited, and evolution of the pools of programming knowledge, and thus of the languages themselves, would have been slowed. If Ashton-Tate succeeds in copyrighting the dBASE language, it and new languages will be protected, less accessible and less desirable to programmers. The development and acceptance of new industry standard languages will be less likely.

A further effect could be a general slowdown in applications development, since the ability of corporate data processing managers to acquire or create new applications that would be compatible (that is, would share data and/or development environments) with their pre-existing programs and data would be severely limited. Despite the decision of the Third Circuit in Apple Computer, Inc. v. Franklin Computer Corp.,¹¹ which dismissed the importance of compatibility, it is a vital issue in the computer industry. Inmon writes:

The internal development is usually done in a widely
used language (COBOL, PL-1, FORTRAN, etc.). Selecting the
language(s) that is used parallels the same reasoning as
selection of operating systems and DBMS -- an available
talent pool and the existence of a wide variety of software
written in the selected language.¹² Inmon notes that while it is not always easy to move data and programs from one machine computer to another even when they use the same language, it's a lot easier than converting from one language to another.¹³

It seems undeniable that the progress of competitive business practices and technology are dependent on both software advances and the ability to efficiently handle data. If computer language copyright limitations were imposed only in America, the competitiveness of American companies with those of other technological nations would suffer.

The relationship between technology and the marketplace is a symbiotic one. On one hand technology depends on growth and acceptance in the marketplace for sustenance and the opportunity to become established. At the same time, many marketplaces depend on advances in technology for existence and the opportunity to grow. Without technological innovation entire marketplaces would never exist.¹⁴ One must ask whether retarding technological growth is the purpose of copyright law.

3. The Purpose of Copyright: to Promote Science and Useful Arts

Article I, Section 8 of the United States Constitution states that the purpose of copyright is "To promote the progress of science and useful arts, by securing for a limited time to authors and inventors the exclusive right to their respective writings and discoveries." A question for the Ashton-Tate court and, prospectively, for Congress, is whether copyright should be granted where it is clear that securing such rights to authors will retard the progress of science rather than promote it. The clear weight of authority is in the negative.

"The sole interest of the United States and the primary object in conferring the monopoly lie in the general benefits derived from the public labor of authors,"¹⁵ wrote the U.S. Supreme Court. Nimmer observes that:

The primary purpose of copyright is not to reward the author, but is rather to secure 'the general benefits derived by the public from the labors of authors....[T]he public benefits from the creative activities of authors, and....the copyright monopoly is a necessary condition to the full realization of such creative activities.¹⁶ Moreover, "when technological change has rendered its literal terms ambiguous, the Copyright Act must be construed in light of this basic purpose." ¹⁷

Copyright law has provided no stimuli to language development; this is obvious because languages have developed without being copyrighted. It is arguable that languages are developed and gain general acceptance and usefulness precisely because they are not protected.¹⁸

Undoubtedly, the creator of a language does invest considerable effort in his creation. Conley and Bryan write, "Regardless of the source of the elements, the author may legitimately claim that synthesizing the entire system required substantial creativity and ingenuity."¹⁹

However, not everything that requires originality and effort to create is copyrightable.

In no case does copyright protection of an original work of authorship extend to any idea, procedure, process, system, method for operation, concept, principal or discovery, regardless of the form in which it is described, explained, illustrated, or embodied in such work.²⁰ Policy issues aside, does the nature of a computer language qualify it for copyright protection?

4. The Idea/Expression Dichotomy and Computer Languages

The idea/expression dichotomy had its genesis in Baker v. Selden,²¹ which held that held that "copyright of a work on mathematical science cannot give to the author an exclusive right to the methods of operation which he propounds."²² The focus of this case was a bookkeeping system that included blank forms with ruled lines and headings. "The mere copyright of Selden's book did not confer upon him the exclusive right to make and use account-books, ruled and arranged as designated by him and described and illustrated in said book."²³

Mazer v. Stein ²⁴ held that "protection is given only to the expression of the idea -- not the idea itself." ²⁵

What, in a particular case, is "idea" and what is "expression" is considered one of the most difficult distinctions to make in copyright law. The traditional test was put forward in Dymow v. Bolton:²⁶ "if the idea can be expressed in a plurality of totally different manners, a plurality of copyrights may result, and no infringement will exist." In the more modern language of Apple v. Franklin, the issue is "whether particular ideas and expressions have merged."²⁷

However, whether such plurality or merger exist in a particular case is easier asked than determined. Learned Hand wrote, "Nobody has ever been able to fix that boundary, [between idea and expression] and nobody can." ²⁸ He also noted that "Decisions must therefore inevitably be ad hoc."²⁹

One of the major problems lies in discovering exactly what the "idea" is. A useful metaphor for the analysis described in case law³⁰ is that of a tree, in which the root is the "idea" and the multiple branches are the "expressions." However, at least in the computer world, the root seems particularly difficult to identify.

This paper argues that a computer language is just such an idea. But in order to fit a computer language into the analytical framework of the idea/expression dichotomy, we must understand precisely what a computer language is.

5. The Theoretical Nature of a Computer Language: Idea

Webster's New Collegiate Dictionary defines a language as "(5) a formal system of signs and symbols (as FORTRAN or a calculus in logic) including rules for the formation and transformation of admissible expressions."³¹ De Remer particularizes the elements of a computer language: "A 'grammar' is a formal device for specifying a potentially infinite 'language' (set of strings) in a finite way."³²

A "device" as used by De Remer and other theorists is not a tangible object that performs a task, rather, it is the theoretical construct of such a device. Part of such a grammatical device can be expressed as follows:

Figure 1

What is described here is a set of rules. The rule of line 40 dictates that any group of characters beginning and ending with quotation marks will be treated as a string of characters that can be manipulated -- for example, a client's name. Line 51 shows that the same group of characters preceded by a sharp sign (#) will be treated as a comment. (A comment is placed into a program's source code for the edification of any programmer reading it, but is ignored by the compiler that translates the code into machine language, and thus by the computer itself.)

Thus a computer language is a set of rules, an example of which, in the dBASE language, is this: If a programmer writes, "SET DEVICE TO PRINT," the output of the program will be directed to the printer. If the programmer breaks the rule by writing, "DEVICE SET TO PRINT," the program will not work.

Different languages have different sets of rules for accomplishing tasks. For example, in MAI/Basic IV's Business Basic language, the programmer might designate the printer device as "P1." He would assign that printer to one of several "channels" representing different devices; we'll use Channel (1). By writing the statement, "Print (1) X," he would direct the output of the program, the variable X, to the printer.

Having chosen a language, if the programmer writes a statement following one rule, the corresponding algorithm or process will be executed. Certain rules, such as those governing repetitive processing loops, require multiple statements. For example, from the dBASE language:

Do while .not. eof()
...a series of commands...
Enddo Furthermore, the use of some rules requires the use, at some prior place in the program, of others. A command to print a variable cannot work unless the value of the variable has previously been established.

Beyond this, however, the set of rules is not sequenced. In a programmer's manual, the creator of the language may list the various rules in some order, for example, alphabetical. But the language itself is a jumble of these rules, given order only when the programmer uses the language to write an ordered series of statements that becomes a program. The requirement of Rule B that rule A first have been satisfied can be seen as an attribute of rule B.

Each rule thus exists essentially in isolation.

What are the prospects of such a set of grammatical rules being copyrighted?

First, the law is, if anything, couched in terms of languages not being copyrightable. "Just as one cannot copyright the arabic number, so one cannot copyright the Latin alphabet or the English language."³⁴

Second, rules of a game have been held not to be copyrightable. "The rules of the game are perforce in the public domain as well as the game itself."³⁵ It is arguable that grammatical rules are analogous.

Third, although it was held by Learned Hand in Hartfield v. Peterson,³⁶ that a compilation of 75,000 code words and phrases was copyrightable both as to the compilation as a whole and as to the individual phrases themselves,³⁷ a computer language is distinguishable. It is not a compilation because it does not necessarily appear in any set order or arrangement. Language commands and functions can be ordered in a wide variety of ways by a programmer. The arrangement of a code book is in the order given it by the original author and no other. The potential copyrightability of individual phrases is more problematical. Conley and Bryan dismiss Hand's statement in Hartfield, writing:

Although any statement by Learned Hand in a copyright decision is not to be dismissed lightly, it can only be inferred that the remark is aberrant dictum. The statement was unnecessary to the decision since the defendant had copied portions of the compilation.³⁸ That the Hartfield opinion was written by Augustus N. Hand, not Learned Hand, may make this dictum easier to ignore. But this is not altogether satisfactory. It may be more useful to note that many of the elements of a computer languages are not phrases so much as phrase structures. An example in the dBASE language is LIST [<scope>] [<expression list>] [FOR<condition>]
[WHILE <condition>] [OFF] [TO PRINT]³⁹ This is the rule for writing a command that will list certain items of data residing in a database either to the screen or to a printer. The phrase itself can be written many different ways, each undoubtedly an expression. Even the descriptor of the rule, as typed above, is an expression. But the rule itself is only a rule, as the language as a whole is a set of rules. The closest analogue to such a single linguistic rule is an equation (which is not copyrightable). The closest analogue to the set of rules that comprises a language is a catalogue schema (which is copyrightable insofar as it indexes a set arrangement of information⁴⁰); a language schema, which describes only itself, seems distinguishable.

The dBASE language does contain certain elements that can be considered true "phrases"; however, these are few and generally not the sort of expressions that ought to be considered original. Examples are "SET DEVICE TO PRINT," "CLEAR MEMORY," "APPEND BLANK," etc. It seems doubtful that these rise even to the minimal level of originality required by copyright law.

Finally, clear proof that a language is an idea rather than expression is that it can, itself, be expressed in various ways. Manuals, compilers, and written grammatical schema such as Figure 1, above, are written or coded literary works and clearly copyrightable, but the language itself exists, conceptually, apart from these. Since it can expressed in several different ways, it cannot itself be an expression, and under the rule of Baker v. Selden and that case's descendants, cannot be considered copyrightable.

6. The Context of a Language - Compilers and Other Expressions

While the above is a theoretical description of a human-usable computer language, this language is only useful in the context of a compiler and in the ability of the compiled program to handle data.

The compiler translates the commands entered by the user into a form to which the computer can respond. Typically, compilers are commercially sold programs, and come bundled with certain features that turn the program into a development environment for that particular language. For example, Borland's Turbo C compiler is sold with its own editor and debugger. Microsoft's C compiler also comes with an editor and debugger,⁴¹ although they offer differences in form, feature and performance sufficient to provide a meaningful choice between the two products.

The compiler itself consists, conceptually, of a set of modules that translate individual user entered commands. The translation modules must exist in a set relation with one another in order that the user entered commands may work together as expected.

However, from compiler to compiler the modules need not work exactly the same way. Their functions are substantially similar, otherwise there would be no consistency to the language. But there are very significant differences in the exact procedures executed by the translation modules, which are themselves programs. Thus, in trade magazines it is common to see such comparisons as the following: "The most surprising run-time result is Turbo C's poor performance on the GRAPH test. The ellipse() and floodfill() functions perform more than twice as slowly as their Microsoft counterparts."⁴²

It is worth noting that the fundamental reason that Foxbase+ has been able to make inroads into Ashton-Tate's market is that the Foxbase+ compiler executes programs more than fifteen times faster than Ashton-Tate's dBASEIII+.⁴³ Additionally, while both compilers execute a core set of language rules that are recognized industry wide, Foxbase+ includes certain commands not included in dBASEIII+. Such additions are called, in the trade, "extensions." One reason that the Nantucket's Clipper compiler has made inroads into the dBASEIII+ market is that it is not only much faster than dBASEIII, but it offers many more functions.⁴⁴

Many types of compilers exist. Some compile an entire program and then run it. Some compile and run the program one step at a time, and are called "interpreters." Many languages require another process beyond compiling, called "linking." This occurs when the program is not complete in itself, but must be "linked" with pre-existing modules, already compiled, in order to run. The two ellipse() and floodfill() functions noted above are examples. The compiler comes with a "library" of such functions which, in writing the program, the user calls into operation simply by mentioning the function name. Third-party libraries are also sold, which provide further extensions to languages.

The second contextual area of concern is with file structures.

Most languages are designed with underlying schemes for creation of files that can then be customized to a user's or programmer's specifications. Thus, all the data required for efficient running of, say, a dental office can be organized and stored in a dBASE or Foxbase program.

The dBASE language was designed to create files of certain types: data, memo and index. Each type has a certain design. For example, a data file can consist of a number of fields, each field being a piece of information. These fields can be of type Character, Number, Date or Memo. Control characters surrounding each field determine the size and type of each field. In every database file, no matter what purpose it is used for or how it is organized by a user or programmer, every field of type "Number" will be surrounded by the same control characters. Every database file will have the same control characters indicating the beginning of that file, and the same characters designating the end. Dialects of the dBASE language vary in how flexibly they can utilize these basic file structures. For example, while Ashton-Tate's dBASEIII+ limited the user to files of no more than 128 fields, Nantucket's Clipper at least doubles that number.

Since part of the basic purpose of all dBASE dialects is to access files of the specific type described above, the language is tied to that basic file structure. If the basic file structure is copyrightable, then even if the language itself is in the public domain, it might as well be copyrighted because it can only be used with in conjunction with those file structures.

Such a basic file structure, prior to the entry of any user data, is merely an organization of control characters designating the beginning and end of the file, the beginning and end of each record, as well as the beginning, end and field-type of each field. Basic file structures can take various forms. Thus, the file created by a C-language compiler is different from the file created by a COBOL-language or dBASE-language compiler, although the user may enter identical information into each, through different programs. In this sense, a basic file structure can be seen as an expression of the idea of holding information. But except for whatever information a user derives from the standard file structure limits (you cannot, for example, enter a graphic image into a standard dBASE file, or create a file structure of more than 128 fields), the basic file structure is neither "original" nor "informative" in ways that case law requires for copyright,⁴⁵ any more than are the lines separating the columns of an accounting page.

Furthermore, both case law⁴⁶and Nimmer,⁴⁷ in treating compilations and forms, look to whether the basic structure presented suggests to the user a methodology or rationale for ordering his own information. The format of a basic file structure, by contrast, is altogether invisible to the user. It suggests no ordering of user information, nor does it offer any useful information except, again, for knowledge of what it cannot do. The basic schema of a legal reporter is <VOLUME> <SERIES NAME> <PAGE> and does not, in itself, rise to the slight originality requirements of copyright law. The schema of a basic file structure goes even less far; it doesn't even specify the order in which the elements appear.

We see, then, that a usable computer language, has three aspects: the theoretical set of rules that comprises the language's grammar, as well the compiler and the basic file structures, both of which are necessary for the language's practical use. The rules are clearly non-copyrightable "idea." The compiler is expression and copyrightable. The basic file layouts require so little originality and are so uninformative as not to be copyrightable either.

To determine whether a language is copyrightable, it is important to separate the theoretical construct -- the idea of the language and its rules -- from its practical expressions: manuals, descriptive writings, compilers, etc.

7. Over-broad Definition of a Program's Idea: the Worm in the Apple Cases

This paper argues that "idea" and "expression" with regard to languages can be most usefully analyzed if one views the grammar as the idea and the available compilers as the expressions.

It is possible, however, to define the idea far more broadly -- that is, in a much less rigorous way. For example, a computer language can be viewed as a instruction set that will enable the user to control screens, create files and file structures, and executable programs that use those screens and structures. In that case, nearly every language becomes an expression of the idea, since nearly all have those capabilities.

Now, there is certainly good reason to avoid too tight a definition of an "idea": One court, at least, has cautioned that "the 'idea' of any work could always be defined in such detail that the description of the expression would add nothing to the 'idea', thus allowing a defendant to engage in all but verbatim copying."⁴⁸

This paper argues that it is equally possible to sin in the other direction and define an idea in so little detail that almost any attempt to develop a new work based on that idea will run afoul of someone else's earlier copyright.

A practical example of failure to appreciate this problem exists in Apple v. Franklin:

Franklin claims that whether or not the programs can be rewritten, there are a "limited number of ways to arrange operating systems to enable a computer to run the vast body of Apple-compatible software," Brief of Appellee at 20. This claim has no pertinence to either the idea/expression dichotomy or merger. The idea which may merge with the expression, thus making the copyright unavailable, is the idea which is the subject of the expression. The idea of one of the operating system programs is, for example, how to translate source code into object code. If other methods of expressing that idea are not foreclosed as a practical matter, then there is no merger. Franklin may wish to achieve total compatibility with independently developed application programs written for the Apple II, but that as a commercial and competitive objective does not enter into the somewhat metaphysical issue of whether particular ideas and expressions have merged.[emphasis added]⁴⁹ The idea of that operating system program could just as easily, and perhaps more accurately, been stated as: a method of translating source code programs written for the Apple II computer into machine language executable by that computer.

The "idea" of the operating system program is further distorted by judicial refusal to recognize compatibility as an element of a program's "idea." No reason whatever is given for this refusal. Yet compatibility is an idea of profound consequence in the computer industry. Entire sectors of the industry have grown out of the need to make data housed in basic file structures created by different companies, and programs written by those companies, able to function with one another. The Third Circuit's statement in Franklin seems plainly inconsistent with its own later dictum in Whelan Assoc. v. Jaslow Dental Laboratory, that: "the purpose of a utilitarian work would be the work's idea, and everything that is not necessary to that purpose or function would be part of the expression of the idea."⁵⁰ This dictum was based on the court's reading of Baker v. Selden, supra, as focused on "the end sought to be achieved."⁵¹ If compatibility is part of a program's purpose -- whether the reason for that purpose is "commercial and competitive" or not -- how can it be anything but a part of the program's idea?

If the idea/expression dichotomy is to form the dividing line between what is copyrightable and what is not, a reasonable and consistent analytical framework for defining the "idea" of a computer program must be devised in light of the primary purpose of copyright law, which is to promote the progress of science and the useful arts. To discourage compatibility is to retard progress.

This paper argues that, at least with regard to computer languages, the method most likely to promote progress and least likely to disrupt the progress of the industry, is to regard the grammar/compiler distinction described in the previous section as equivalent to that between idea and expression. Since the creator of a language must also create a compiler in order to demonstrate the language's usefulness, and since the compiler is a copyrightable expression, the author's rights in his creation would be secured for a limited time. This would motivate compiler authors to create, since it would take some time for someone to come up with a superior compiler in that language. However, the creator of that next compiler would also be motivated, since he could copyright his work, at least, insofar as he did not steal any code from the language creator's compiler. This, too, is consistent with the purpose of the copyright statutes.

8. Copyright of Operating Systems: Blurring the Distinction Between Copyright and Patent Coverage

Both Apple v. Franklin and Apple Computer, Inc. v. Formula Int'l. Inc. have held operating systems to be copyrightable.⁵³

This section is written, not because of any disagreement with what the Third and Ninth Circuits have held with regard to the copyrightability of operating systems, but because of potential damage that could occur should these holdings be improperly read. In particular, of course, the potential effect on the copyrightability of computer languages is at issue. The more general threat is an incursion of copyright into realms historically ruled by patent law -- an event foreseen by Nimmer in his concurring opinion to the Final Report of the National Commission on New Technological Uses of Copyrighted Works (CONTU):

What is most troubling about the Commissions recommendation of open-ended copyright protection for all computer software is its failure to articulate any rationale which would not equally justify copyright protection for the tangible expression of any and all original ideas (whether of computer technology, business, or otherwise). If literary works are to be so broadly construed, the Copyright Act becomes a general misappropriation law, applicable as well in what has traditionally been regarded as the patent arena. . .⁵⁴ The Apple decisions would seem, at first glance, clearly contrary to the language of the statute, which prohibits copyright of "systems." However, both courts noted two aspects of an operating system program -- the processes carried out, and the instructions that define these processes and make possible their execution.⁵⁵ It is these instructions that the courts ruled copyrightable as "literary works" under § 102(a).⁵⁶ Focusing on the literary aspects of an operating system program, the Third Circuit simply held the systemic aspects of operating systems irrelevant: Since it is only the instructions which are protected, a 'process' is no more involved because the instructions in an operating system program may be used to activate the operation of the computer than it would be if instructions were written in ordinary English in a manual which described the necessary steps to activate an intricate complicated machine. ⁵⁷ ⁵⁸ Webster's defines a system as, inter alia: "a group of devices or artificial objects or an organization forming a network esp. for distributing something or serving a common purpose. . . <[other examples] >. . .<a data processing system>."⁵⁹

As worded, Apple v. Franklin,⁶⁰ in particular, can be read as dismissing the systemic aspect of an operating system program as irrelevant to a decision as to whether a copyright should be granted.

It must be remembered that the code of an operating system is itself merely an expression of the underlying system. The system itself -- the entity defined by Webster's -- is similar to a language grammar: the idea that underlies the expression. There are, as the losing defendant wrote in Apple v. Franklin, only a limited "number of ways to arrange operating systems to enable a computer to run the vast body of....compatible software."⁶¹ Defendant Franklin was not wrong in this. Defendant lost the case because it copied Apple's code line by line. Franklin could have written its own code to duplicate the Apple operating system's processes, and not infringed. Such copying could only be illegal if Apple's system processes were held to be copyrightable expression, in which case the decision would have been inconsistent with the Copyright Act -- just the sort of error Nimmer feared. In this, at least, the Franklin court ruled correctly:

If other programs can be written or created which perform the same function as an Apple's operating system program, then that program is an expression of the idea and hence copyrightable. In essence, this inquiry is no different than that made to determine whether the expression and idea have merged. . .⁶² The court's comment regarding compatibility⁶³ is logically inconsistent with this statement. If a process is not copyrightable, then clearly, independent creation of a program that duplicates that process is not an infringement. Therefore, compatibility must be an element of a program's idea, not its expression. If the only way to duplicate the process is to copy the code, then the expression and idea have merged and there is no copyright. In this case, the process should be patented.

Similarly, in the case of computer languages, there may also be only a limited number of ways for language compilers to be arranged in order to handle the set of user-entered commands and functions consistently. Even if the code per translation module is different, there will still be certain substantial similarities in the sequences in which they are called, from compiler to compiler for the same language.

It must be reaffirmed that systems -- including linguistic grammar systems -- are not copyrightable. First, because 17 U.S.C. 102(b) forbids it.⁶⁴ And secondly, because any attempt to duplicate a system's processes may require copying so substantial that the expression merges with the idea.

9. "Structure, Sequence and Design": the Whelan v. Jastrow Non-breakthrough

Whelan v. Jaslow is the most important of the computer copyright cases because it holds, for the first time, that not just the literal elements of a program can be copyrighted, but also something altogether intangible: the "structure, sequence and design" of the program. In fact, as will be shown, this is not really an extension of copyright principles. But it has been interpreted as a breakthrough, and subsequent decisions are using it is a platform on which to base dangerous and confusing decisions. By not limiting the scope of its holding, it seems to raise the possibility that things never before considered copyrightable will become so. The Apple cases falsely hint that a system program may be copyrighted without regard to its systemic aspects. The description in Whelan v. Jastrow of what may be copyrighted -- "structure, sequence and design" -- also, at first reading, may seem to speak to a program's systemic aspects. The author of this paper is not alone in his apprehensions. Berkowitz has observed:

The Third Circuit's extension of copyright protection of computer programs in Whelan Associates v. Jaslow Dental Laboratory, Inc., has potentially placed the copyright protection of computer programs in conflict with the patent system's protection of functional processes. The court's proposal to protect the intangible "look and feel" of a program is a difficult standard for other courts to interpret properly. At present, for lack of a better standard which would protect the interests of all parties, the courts should adhere to the well grounded standard of comparing evidence of only limited copying as proof of substantial similarity in a computer copyright infringement action.⁶⁵ In Ashton-Tate v. Fox, in which it is claimed a language is copyrightable, this twisting of the purpose of copyright law may come to full fruition.

However, this paper disagrees with Berkowitz in perhaps seeing a way out of this dilemma. This is to view the Whelan decision as simply an application of traditional copyright principles to the literary expression that is a computer program.

Although computer programs have long been copyrightable as literary works, programs -- unlike dramas, in which a sufficiently unique and developed plot can be copyrighted -- could, until Whelan, be copyrighted only as to their literal elements. Only the actual program code instructions were eligible for protection as literary works.

Consider Synercom Technology, Inc. v. University Computing Co.,⁶⁶ which held specifically that a program's structure and sequence are not copyrightable, that they are, in fact, elements of the program's idea. The court asked, "if sequencing and ordering is [sic] expression, what separable idea is expressed?"⁶⁷ This question makes eminent sense if the reader identifies "sequence and order" with the ordering of processes that is the "idea" of the system. However, it becomes irrelevant if one looks analyzes the organization of a computer program as one would look at the plot of a drama or novel.

The Third Circuit, in Whelan, declined to follow Synercom. "To the extent that Synercom rested on the premise that there was a difference between the copyrightability of sequence and form in the computer context and in any other context, we think that it is incorrect."⁶⁸ Whelan referred specifically to dramatic plot cases in holding that the "structure, sequence and organization" of a program is just as protectible as that of a literary or dramatic work.⁶⁹ Thus we see that Whelan is not a breakthrough at all, but merely a recognition that, if a computer program is a literary work, then all the protections available to novels and the literary aspects of dramatic works apply equally to computer programs.

However, the Whelan court made two important errors, not of analysis or decision, but of definition and omission.

First, by holding that the "idea" of the program was simply to aid "the efficient organization of a dental laboratory" -- just the sort of destructively broad definition discussed earlier in this paper -- it defined the "idea" far more broadly than the idea of a literary or dramatic work would be defined. The court's definition of the program's idea -- "the efficient organization of a dental office" -- was adequate for purposes of the decision at bar, but a poor descriptor for purposes of precedent.

Since, in its opinion, the Whelan court refers several times to the holdings and dicta of Learned Hand,⁷⁰ it may be useful to look at the method of analysis posited by Judge Hand in Nichols v. Universal Pictures Corp.:

But when the plagiarist does not take out a block in situ, but an abstract of the whole, decision is more troublesome. Upon any work, and especially upon a play, a great number of patterns, of increasing generality will fit equally well, as more and more of the incident is left out. The last may perhaps be no more than the most general statement of what the play is about, and at times might consist only of its title; but there is a point in this series of abstractions where they are no longer protected, since otherwise the playwright could prevent the use of his "ideas,", to which, apart from their expression, his property is never extended. . .We did not in Dymow v. Bolton, 11 F.(2d)690, hold that a plagiarist was never liable for stealing a plot. . .We found the plot of the second play too different to infringe, because the most detailed pattern, common to both, eliminated so much from each that its content went into the public domain.[Emphasis supplied].⁷¹ Viewed from this perspective, we can analyze the idea/expression dichotomy of a computer program as a "series of abstractions," seeking to find at which level of abstraction the most the most detailed pattern of organization is so unique as to be copyrightable. In theory, any program duplicating the original to at least this level would infringe. Any duplication of the pattern not reaching this level of detail would be in the public domain.

Let us illustrate again from the analysis in Nichols. The dispute in this case was as to alleged copying of incident and character.

There are but four characters common to both plays, the lovers and the fathers. The lovers are so faintly indicated as to be no more than stage properties. They are loving and fertile; that is really all that can be said of them, and anyone else is quite within his rights if he puts loving and fertile lovers in a play of his own, wherever he gets the cue. The plaintiff's Jew is quite unlike the defendant's. His obsession is his religion, on which depends such racial animosity as he has. He is affectionate, warm and patriarchal. None of these fit the defendant's Jew, who shows affection for his daughter only once. . .Both are grotesque extravagant and quarrelsome; both are fond of display; but these common qualities make up only a small part of their simple pictures, no more than anyone might lift if he chose. . .A comedy based upon conflicts between Irish and Jews, into which the marriage of their children enters, is no more susceptible of copyright than the outline of Romeo and Juliet.⁷² The levels of abstraction involved in this analysis may be charted as follows, from highest level of detail to the lowest. (For the sake of clarity, we will use "Romeo and Juliet," since the reader may be somewhat more familiar with it than with "Abie's Irish Rose," the play at issue in Nichols.):

"Romeo and Juliet," by Shakespeare, with all its poetry, character-based speeches, etc.
Detailed outline, laid out scene by scene, encounter by encounter, with detailed summaries of each character and that character's fate.
Somewhat detailed outline: boy and girl from feuding Italian families fall in love, after boy secretly crashes a party given by girl's family. Arrogant fop from girl's family kills boy's funny close friend and kinsman. Boy kills fop. Family battle-lines are drawn. Unable to be together, boy and girl mutually suicide. The character types are chosen because of the actors who are available to the author, among others, a two romantic leads, an athletic comic and a perfect fop.
Rough outline: Boy and girl from feuding families fall in love. They try to bring the families together but only make matters worse. Unable to be together, boy and girl do something desperate and both die.

Clearly the outline in item 4 is not copyrightable. Nor is the somewhat detailed outline sufficiently unique. Professor Chafee has written: "No doubt the line does lie somewhere between the author's idea and the precise form in which he wrote it down. I like to say that the protection covers the 'pattern' of the work * * * the sequence of events, and the development of the interplay⁷³of characters." By this analysis, the detailed outline of item 2 is sufficient for copyright, as of course, the play itself would be, were it not already in the public domain.

Now let us apply this mode of analysis to the program in Whelan and, more importantly, to the language issue of Ashton-Tate.
The program itself, as coded (and from which the actual copy, in Whelan, was made⁷⁴).
The technical and functional specifications of the program: flow charts, data flow diagrams, file structures and other detailed documentation from which the program could be coded in any language. These are all developed from the basic requirements of the system as posited by the ultimate user.
The basic requirements or "idea" of the system. These requirements, in Whelan, certainly included the purpose of "the efficient organization of a dental office" but also, had the dental office in question already had a pre-existing computer system, might have also included the requirement of compatibility pre-existing file formats and structures, as well as structures and programming languages, including compatibility, since"the purpose of a utilitarian work would be the work's idea, and everything that is not necessary to that purpose or function would be part of the expression of the idea.⁷⁵
The Whelan court description: "the efficient organization of a dental office."

Clearly items 1 and 2 are copyrightable. Clearly item 4 is not. The question is with regard to item 3. If the languages and file structures used are in the public domain -- as are the broad character types used in our "Romeo and Juliet" analysis, as well as Hand's "Abie's Irish Rose" analysis, then it seems there is no problem with regarding item 3 as "idea" rather than "expression." If they are not in the public domain, then item 3 ought only to be "idea" insofar as it does not consist of the copyrighted "expressions." Based on our earlier analysis, we find that neither language nor basic file structure⁷⁶are "expressions."

10. Application of "Structure, Sequence and Design" to Computer Languages

One might ask, however, whether, since a language has "structure, sequence and organization" -- which, in the case of the dBASE language, is part of the larger "structure, sequence and organization" that represents the dBASE III+ program -- the language can be copyrighted as part of that larger program. One can also ask whether such a copyright would protect the language if that language is used in a substantially different environment. Walt Disney Productions v. Air Pirates⁷⁷held that a copyrightable element of a work -- comic book characters -- could be held protected even though plaintiff had copyrighted only the comic books in which the characters appeared, not the characters themselves.

To answer: First, it is arguable that a language does not have an underlying "structure, sequence or design" because it is merely a methodology by which a programmer can create such structures of his own. Second, key to the Disney holding was that the element extracted from the protected work was itself copyrightable. A language would seem to be distinguishable in that, as an idea, it is not in itself copyrightable.

11. "Structure, Sequence and Design" v. "Look and Feel": the Present State of Confusion. A Model for Use in Accord with Traditional Literary/Audio-visual Copyright Distinctions.

The second sin of the Whelan decision was the Court's failure to explicitly limit or define the meaning of "structure, sequence and organization" to the facts of the case -- that is, to a situation in which an original program and the infringing copy both accept the same data and process it in almost precisely the same way. We have shown this to be a violation of literary copyright similar to the appropriation of the detailed plot of a play. But one court, at least, has extended the idea of "structure, sequence and design" to audio-visual works.⁷⁸ There is no reason to so extend the doctrine, since audio-visual copyright has been developing an analogous, new doctrine of its own, "look and feel."

The confusion of these two new doctrines, about which the law is anything but settled, threatens to break down the traditional distinctions between types of copyright. This, combined with the possible breakdown of the line between copyright and patent due to problems courts have had distinguishing idea from expression, leads one to fear that in the chaos, anything requiring originality may be copyrighted, including a language.

In the conviction that the confusion need not exist, the following section suggests a distinction between "structure, sequence and design" and "look and feel" along the line dividing literary from audio-visual copyrights.

"Look and feel" is a new legal vernacular which, along with "structure, sequence and organization," seems on its way to becoming a term of art. The phrase is derived from "total concept and feel" first used in Roth Greeting Cards v. United Card Co.⁷⁹ Although Sid & Marty Krofft Television Productions, Inc.,⁸⁰ held that similarity between characters could be established by the total concept and feel of two productions, as yet no case has been decided solely on the basis of "look and feel."

Initially, "look and feel" applied to audio-visual works. However, it is sometimes used more broadly, to cover the entire program. Sholkoff writes "It is the 'look and feel' or totality of all aspects of the software, when viewed as a whole, that forms the most valuable part of a computer program."⁸¹

Both "structure, sequence and design" and "look and feel" are outgrowths of the long line of cases that have held "substantial similarity" to be a test of whether one work infringes upon another.⁸²

This paper suggests that these two analytical criteria need to more clearly defined than they presently are, and suggests that they be defined and used as follows:

"Structure, sequence and design" should be confined to the literary copyright area. (Literary copyrights protect all code and documentation, which in turn are the sources of the program's organization and flow.) To determine whether "structure, sequence and design" of an work accused of infringement is "substantially similar" to a plaintiff's work, the "series of abstractions" test should be applied. The test, and very concept of "structure, sequence and design" is largely, if not completely, objective. It should be applied by a computer-literate master, as ordinary lay juries will rarely be qualified to make such an analysis.

On the other hand, "look and feel" ought to be viewed as a subjective criterion that concerns art more than science, ease and elegance more than raw efficiency. "The creative choices and artistic authorship of the screen display designers can be identified and should be evaluated wholly apart from the computer code used to implement the display."⁸³ "Look and feel" ought to be viewed as the work's overall visual concept, embracing all of a program's display screens and other visual outputs. One may conceptualize "look and feel" by imagining an art gallery in which many paintings are hung. It is not just the paintings themselves, but their arrangement which gives rise to the viewing experience of the gallery-visitor. As such, "look and feel" is readily identifiable by the lay observer.

This paper submits that the confusion of these two terms in both case law⁸⁴and literature is an outgrowth of the flaws of Whelan. One of the most egregious subsequent errors is Broderbund Software v. Unison World.⁸⁵

12. Broderbund v. Unison World: "Structure, Sequence and Design" Run Amok

Broderbund Software, Inc. v. Unison World interpreted Whelan to stand for: "the proposition that copyright protection is not limited to the literal aspects of a computer program, but rather that it extends to the overall structure of a program, including its audiovisual displays."⁸⁶ This is not what Whelan held and at least two commentators recognize this.⁸⁷ ⁸⁸ The Broderbund court held the program's audiovisual displays copyrightable despite the failure of plaintiff to register any claim to copyright of those displays. The court's logic seems to have been that since the displays exist as elements of the program's structure, sequence and organization, any copyright of the program covers all those elements. Therefore audio-visual displays are covered by a literary copyright.

Tito observes that "insofar as the [Broderbund] court's holding extends the Whelan rule regarding the protected portions of a computer program to screen displays, without distinguishing the separate copyright for the audiovisual works, the case was wrongly decided."⁸⁹

Yet, the reason that Broderbund occurred is that Whelan was seen as a breakthrough. If Whelan is viewed as it really is -- a mere application of the full scope of copyright protection to an area already protected under traditional doctrine -- we see that traditional copyright law holds good, that works that have historically been in the public domain ought to remain in the public domain, and that computer languages ought not to be subject to copyright.

13. "Look and Feel" Applied to Computer Languages

We have already looked at the structural aspects of a computer language. It remains to be asked whether it has a "look and feel" that rises to the level of copyrightable expression.

Generally, computer languages are communicated visibly; people don't speak COBOL, they write it. And the programs written in a language do have a characteristic appearance. For example, COBOL programs begin with statements organized into Environment and Data divisions. Programs written in C usually begin with pre-processor instructions that are prefixed with the character "#"; C program statements nearly all end with the character ";" and C procedures are all designated "<PROCEDURE NAME ()>."

This paper submits that these are no more expression than the characteristic look of English as opposed to German. It is not art that gives a language its appearance; it is convention. The medium is NOT the expression, as evinced by the many cases -- including Whelan v. Jastrow -- that have held mere translation from one language to another to be infringement.

Thus "look and feel" is not an issue that applies to computer languages.

14. The Copyright Office Policy of 1988: Creature of Neither Statute nor Case Law

Two years after the 1986 Broderbund decision, the Copyright Office announced that it would allow only one copyright registration per computer program.⁹⁰ The Copyright Office's new policy is to assess which of the copyright classes predominates in a particular program -- literary, audio-visual, etc. -- and to register the whole under that one class. This extension of the Broderbund erosion of traditional copyright class boundaries can only cause confusion, as it flies in the face of a number of decisions that have held computer audio-visual displays to be separately copyrightable as audio-visual works.⁹¹ Copyright Office policy is by no means the last word, however, courts will rarely substitute their own judgment for that of an administrative agency, unless the agency's policy is inconsistent with the statutory language or Congressional intent.⁹² It is equally clear that the administrative bulletin is not binding on the court.⁹³

Statutory language regarding the manner in which computer programs should be copyrighted is unclear. But however one interprets the statute, Copyright Office policy logically fails. Since 17 U.S.C. 101 clearly evinces Congressional intent that literary, audio-visual and other works be copyrighted separately, the Copyright Office policy should be overturned. On the other hand, Sec. 101 also defines computer programs as literary works, making no mention of other aspects of a program that may be copyrighted, such as audio-visual works. In either event, a policy of copyrighting all of a program's elements under the rubric of whichever aspect predominates cannot be seen as consistent with statutory language or Congressional intent, and should be struck down.

15. Do We Need New Copyright Legislation for Computers?

The current trend in decisions regarding computer copyright is clearly cause for concern. Sholkoff observes that "because the current standard is unclear, judicial determinations risk being arbitrary."⁹⁴

Logically, two alternative approaches toward resolution of the program exist.

The first is for the judicial system to garner the knowledge to interpret current law sensibly -- since courts will have to do so, whatever the current statutes are. This might be done in either of two ways: a pool of computer-expert masters might be established under the judicial system to keep current with developing technology, keep judges informed and act as expert witnesses. A more a traditional and perhaps better solution would be to legislate into existence a small administrative agency charged with remaining current and with holding adjudicatory hearings under computer-expert hearing officers.

The second, propounded by Sholkoff, among others, is that Congress pass laws to create a special statutory framework within which computer copyright issues can be analyzed. This approach is based on the premise that current copyright law is obsolete.

Current copyright protection for computer software remains unclear, inconsistent, and potentially under- protective and damaging to an industry which thrives on innovation. With standards that are questionable and confusing, and that fail to protect the financially important aspects of the work, current copyright law is not the most efficient approach.⁹⁵ The approach of legislating currency into copyright law has history and stare decisis behind it. "Repeatedly, as new developments have occurred in this country, it has been the Congress that has fashioned the new rules that new technology made necessary...it [is] settled that the protection given to Copyright is wholly statutory."⁹⁶ This approach may be useful in getting rid of some of the dead wood of current case law, but it shares with judge-made computer law one fundamental flaw: neither can possibly evolve as fast as the technology that must be governed, unless some group in the legislature or judiciary is charged with staying abreast of technical developments. Rather than rush ahead with new legislation that will quickly become obsolete, this paper suggests some ways of utilizing current statute and case law in a rational manner.

16. Summary and Conclusion:

Recent decisions in the Third and Ninth Circuits have raised questions regarding the potential application of copyright law to computer languages. Apple v. Franklin and Apple v. Formula Intern. called up the specter of copyright invading the realm of patent, by holding that operating systems could be copyrighted without explaining fully that it is only the literal code which could be copyrighted as an "expression" of the systemic "idea." The Apple courts made the problem worse by defining the "idea" of a system too broadly and excluding compatibility -- a concept vital to the computer industry -- from the possible components of an "idea." Whelan v. Jaslow repeated Franklin's error by defining the idea of a program too broadly, and added to the confusion by not defining the limits of its holding that a program's "structure, sequence and design" are subject to copyright. Lack of firm definitions for "structure, sequence and design" with "look and feel" exacerbates matters still more. Broderbund v. Unison World further fanned the flames by holding that a literary copyright could cover an audiovisual display, thus breaking down the traditional categories of copyright law. Lack of clear statutory language stating a Congressionally-intended method for copyrighting computer software, followed by a 1988 Copyright Office policy decision that seems wrong by any statutory interpretation, has added to the chaos.

In this uncertain environment, there is some danger that things never before deemed copyrightable will be copyrighted. This is precisely the danger that now looms in the Northern District of California, as Ashton-Tate attempts to copyright the dBASE computer language. The effects of this effort upon the programming community and businesses that depend on data processing could be disastrous.

The "idea/expression" danger can be avoided by defining the "idea" of a program with more specificity, by regarding compatibility as one of the potential elements of an idea, and by interpreting the Apple and Whelan decisions properly as cases that concerned literary copyright. The Apple decisions focused on the copyrightability of operating system code, not the conceptual system itself. Whelan analogized the organization of a system to the plot of a drama or novel, a mode of analysis which can be useful if one applies the "series of abstractions" analysis drawn from Nichols v. Universal Pictures.

By continuing to allow separate registration of the disparate elements of computer programs -- literary, audio-visual, and possibly phono-record -- triers of fact and law can better organize the issues before them and keep analyses relatively simple. Furthermore, both courts and people in the industry will have a clear idea of just what is protected by each copyright. Use of the recently developed tests of "structure, sequence and organization" and "look and feel" should apply respectively to the literary and audio-visual areas of copyright. The goal of each test is to determine whether there is "substantial similarity" and thus infringement. The former test should be applied on an objective basis by computer-trained masters utilizing Learned Hand's "sequence of abstractions" technique. "Look and feel," however, should be a subjective test exercisable by lay juries.

Nothing suggested here requires substantial revision of existing copyright law. The approaches put forward in this paper require only clarity in the application of present law and the disposal of such misguided precedents as have recently emerged as generalist judges grope to deal with an area requiring specialized knowledge. Such substantial new legislation as advocated by Sholkoff is probably not required, although it may be useful as an interim clarification. More useful legislation might establish a small administrative agency under the Copyright Office for the purpose of keeping abreast with technology and adjudication through administrative hearings under computer-expert hearing officers of technically difficult issues. These opinions could then be incorporated into the complete cases before the court (which may also contain issues of fact that can reasonably be settled by lay juries).

Traditional copyright law can function well in the computer field, if applied with understanding of the field's technology and industry. The temptation to proliferate new statutes because we don't understand how to apply the old, should be avoided.

THE END