When you picked up this collection, there was a name on the cover. That was my name, Paul Swartz, but when it is on the cover it does not function as a normal proper noun; it functions as the name of an author. Now names of authors are also proper nouns, but as names of creators of works they have additional responsibilities. Their names represent "rules that formed a certain number of concepts and theoretical relationships in their works."[1] The author's name represents not only a person, but ideas and relationships in and between works as well. However, the terms work and author themselves and the relationship between them are problematic than that. The modern view of the relationship is that the author creates a work from scratch. The work created is a stand-alone piece of text that requires only a reader to bring it to life. However, the relationship is more complicated than this. There is a text and a work, which consists of the language used: "the work can be seen...the text is a process of demonstration, speaks according to certain rules (or against certain rules); the work can be held in the hand, the text is held in language..."[2] The work, according to Barthes, is a physical object; the text is an abstract symbolic object. Barthes sees the text as a space of possibility constructed of language, a "methodological field...a process of demonstration..."[3] The text is an abstract object that is the site of the power we give to works of literature. Foucault, working from the side of the author, includes more into the concept of the work, complicating both the work itself and its relationship to the text. There are so many things that are connected to a work outside of its text, and it is unclear where the line should be drawn. Surely notes, drafts, and outlines, should be included, but what about "a reminder of an appointment, an address, or a laundry bill"[4]? These other related objects may not be read by every reader, but they can sometimes bring to light meanings in the text. Here, I use the term text to mean something farther from Barthes and closer to Foucault. Barthes sees a work and a text as two different kinds of objects, one physical, the other symbolic, but neither being a container for the other. A piece of writing can be a work or a text, but not both. I see the difference between work and text as that the work includes everything that can bring meaning to a text, but the text is the words, the linguistic core. The text is an abstract part of the work, and could theoretically be separated from the work, or brought into a new work. Citation, for example draws part of one text into another; it does not draw from the work because there is no physical transfer. This kind of link between texts is what Barthes has in mind when he sees the text as "a multi-dimensional space in which a variety of writings, none of them original, blend and clash."[5] He denies that even non-cited textual elements are not original, that they are always derived from something else; it's texts all the way down. The concept of a work in programming (hereafter program) is no less problematic. The program surely consists of the statements that the computer executes; that is the program equivalent of the text. But the statements that the computer executes are almost never those written by a programmer. The computer executes machine code, long strings of binary digits, and these are created by programs: either a compiler (which turns source code into machine code) or an interpreter (which runs the higher-level code directly). The compiler or interpreter is a necessary part of the program, but is not typically thought as being part of the program. The same is true of the libraries, external pieces of code, that the program uses. A program may use libraries for graphics, or for networks, or for higher-level math functions. These libraries are rarely written by the same programmer who is writing the program. And yet, the program will not run without access to those libraries. Although these other features are also linked to the program, they are different from the minutiae of an author's writing like notes or lists. The minutiae are not necessary for understanding of the work, but the libraries and other parts of the program are necessary for execution of the program. The minutiae are more related to the author than to the work specifically: a reminder of an appointment may have given an author an idea for a work, but the physical reminder is not likely to be included in the work itself. The libraries, however, are sometimes included directly with the program, say, when the program is being sold or packaged. Inline comments are the closest thing to the minutiae of programs. They are not required to run the program, but they can bring additional meaning for programmers who are reading the code. Comments describe in a general way what the code is doing, or pitfalls or problems the programmer encountered while writing the code. They can be longer than the code itself, if the code is particularly short or particularly complicated. The biggest difference between comments and the other minutiae of a literary work is that the comments are included with the source code. Rarely do authors include their notes with the published work; the comments are a part of the source code, included by not read by the computer. There is one other element, one which many theorists are actively working against including as part of the text: the intention of the author. What the author meant to say is irrelevant; only what the author actually wrote influences the meaning. Theories that require knowledge of the author's intention are said to fall prey to the 'intentional fallacy.' Roland Barthes sums this up most concisely as "a text's unity lies not in its origin but in its destination."[6] The unity that makes a text comes not from where it starts (as an idea in an author's mind) but from where it goes. For Barthes, this destination is a reader, the one person for whom all of the work that created the text is a single creation. As readers, we do not have access to the mind of the writer. The writer cannot tell us what he was thinking or how to interpret his work. All we have is the work in front of us, making any assumption of the writer's intentions a fallacy. This assumption is especially false with computer programs. What the programmer meant to write is irrelevant to how the program functions when it is executed. The programmer assumes that the program works in one way, and the computer interprets it. Ideally, the programmer and the computer interpret the program the same way. If they do not, it is a bug; the programmer's intention has not been conveyed to the computer, and so the programmer must try again. And if the concept of a work is confusing at best, what of the author? The person who writes a work is related to the author, but they are not the same. "The name of an author poses all the problems related to the category of the proper name,"[7] but it also poses new problems. The writer has physical features (e.g. height, weight, eye color) which do not influence how we as readers interpret the works she creates. The works which are linked to an author's name do influence the meaning we find in these works. We can tell what constitutes the function of the author's name with a simple thought experiment. "If I discover that Shakespeare was not born in the house that we visit today, this is a modification which, obviously, will not alter the functioning of the author's name. But if we proved that Shakespeare did not write those sonnets which pass for his, that would constitute a significant change and affect the manner in which the author's name functions."[8] The qualities that are relevant to the author's function are those that, if they were to change, would also change the meaning of that author's works. The name is relevant to the extent that it links works under that name. However, this is not the only relevance of the author's name. The name also delimits the works that are linked to it. The author's name creates a field of works that belong (for lack of a better term) to that name. The works draw meaning from their relationships with the author and with the other works. Adding or removing works to a field, or merging fields or splitting a field distorts the function of the name and the meaning of the works in that field. The use of the author's name as a link to the writer is also problematic because there is not a one-to-one correspondence between names and people. Some works are authored under a pseudonym, hiding the writer's true name. Multiple works under the same pseudonym can be linked, giving the pseudonym the power of the author's name. Some authors publish work under a pseudonym or pseudonyms as well as under their own names. This has the effect of separating the work of the pseudonyms from works under the author's true name. However, it does link the works: because they are written by the same physical person, discriminating readers can learn this fact. Other works have multiple authors. This confuses the issue of authorship by creating a question of which of the authors wrote which elements of the text. Scientific papers have this problem: often times the author listed first had little to do with the actual writing of the paper, but is the advisor or most senior member of the team. The relation between the author and the work is what Foucault calls the author function. This function exists in the relationship between a discourse and the society that created it. The author is a construction of society, a product of our collective relationship with the works. We as readers have created the persona of author, not to refer to a specific person, but to represent the function of the creator of a work. For literary works the author function is very important; the works draw meaning and status from having an author: "The meaning ascribed to [a work], and the status or value accorded it"[9] depend on their having an author. Not all works have this function to the same degree: notes, contracts, and lists, for example, do not have additional meaning from knowledge about their author; only their content is important. The author function is also less important in theater and film, due to their collaborative nature. We may go to a play because of the writer, but we are equally likely to go to a movie because of a director or a big-name actor or actress. Regardless of who draws viewers, the actual production is the work of many different people: writers, directors, producers, actors and actresses. What Foucault is try to get at here is that it is not a person who we put in the place of the author, but rather a set of responsibilities that are represented by the author's name. What about computer programs? Where do they fall on the spectrum of authored-ness? Software works the same regardless of what we know about the programmer. This means that programs do not derive additional meaning from knowledge about the author. Just as our society created the author to represent the person we as a society do not concentrate on the individuals writing our software. We are still drawn in to a degree because of the name attributed to it, but mostly we choose software because of it's functionality: what it can do for us. Commercial software nearly always has the name of the corporation attached to it. This is the first and most obvious authorship attribution commercial software receives. Finding the names of the actual programmers is difficult, if it is possible at all. Commercial programmers are paid, and so being rewarded with their name on the software is less important to the corporation. Open source software has the authors listed in a file that comes along with the program. Open source software is released as source code, the instructions that are translated by the compiler/interpreter into machine code. This makes it easier for other programmers to understand how the program works and modify its functionality. Commercial software is released as machine code. This means that modification is made much more difficult. Programmers want a reward for their work, and in open source software the reward is being recognized by the community. The AUTHORS file gives that recognition in a way that all can see. However, open source does typically recognize a single person as being at the head of a project. Eric Raymond has been one of open sources most prominent figures, and wrote a series of essays about his experiences and collected them in a book titled The Cathedral and the Bazaar. "Homesteading the Noosphere" discusses Raymond's experience with the open source culture. Despite having many people work on projects, open source programs have a very clear notion of ownership of programs: only the owner of a program can release modified versions of a program. Although it is technically and legally possible for other people to release updated versions of software, there is a strong community taboo against it. Violation of the taboo would likely result in the violator being expelled from the community. The reason for this taboo is to prevent the splitting of work in the community. If one did want to start releasing modified versions of a program (called 'forking'), they would have to release the new program under a different name to distinguish it from the original program. The reason for this importance is that the open source environment is a kind of gift culture. People get status by the gifts they give to the community; releasing someone else's project as your own is seen as gift theft.[10] Gift cultures (or gift economies) date back to the Stone Age, and have been a fruitful source of analysis in fields from anthropology to economics to political science. The individuals involved give goods without the expectation of an immediate return; the recipient is expected to give a gift sometime in the future. This contrasts a gift economy with a barter economy in which goods are traded at the same time. There is no assignment of value to a gift, like there is in a market economy. Perhaps the allure of such a culture derives from how far if differs from our own economy. The free-market economy, with it's giant corporations, propertization of everything, and reliance on commodities for trade, is a powerful force in our culture. Analysis of gift cultures could possibly bring their aspects (simplicity, altruism) into our own culture. The avoidance of gift theft is similar to the academic prohibitions against plagiarism. Plagiarism is the appropriation of someone else's work and then passing that work off as one's own. It is generally taken to also cover the ideas in a work, so that minor rephrasings are still considered plagiarized. Students are cautioned against using the work of others without citing it. Academia is also a kind of gift culture. The papers published by professors are like the programmers' gifts to the open source community. Those gifts give the giver certain benefits: name recognition, conference presence, tenure, etc. Appropriating another's work is gift theft, but with the added snub of passing that gift off as one's own. Plagiarism also disrupts the author function. Plagiarism interrupts the link between author and work, by giving multiple authors for the same work when there is truly only one author. It is difficult enough to attribute works when the authorship is clear. There is no true interaction between the work and the culture in the case of plagiarism; a copy under a new name brings nothing new to the table, only confusion. There is some link between a plagiarized work and the name of the person who copied it. Names like Kaavya Viswanathan and Stephen Ambrose are notable now for their link to plagiarism, and their names will forever be linked to that theft. However, the different ways that these two authors are accused of plagiarism can shed some light on what exactly plagiarism is. Ambrose was accused of using a few passages from other historians in his work without quotation marks. He did cite the works in footnotes, but not the specific passages that were found to be copied directly or very similar. The whole of Ambrose's work is still believed to be his own work. Viswanathan, on the other had, was accused of using passages from no fewer than five other authors throughout her entire novel. Investigations have been launched into her writing at a previous job as a newspaper intern. Her plagiarism calls into question the whole of her work as a writer. Clearly plagiarism can differ in scale, but what makes both of these cases plagiarism? It is not intention; both authors denied consciously included the work of others without citation. Nor is it a lack of general attribution; Ambrose did cite the works he plagiarized. Surely it requires the copying of another's work, but how much is plagiarism? In an essay for Harper's Magazine, Jonathan Lethem argues convincingly against the current copyright and intellectual property system, and in favor of the open source gift culture. He argues that all work is in a sense plagiarism, since to create new works authors invariably draw from the old: "consider the remarkable series of 'plagiarisms' that links Ovid's 'Pyramus and Thisbe' with Shakespeare's Romeo and Juliet and Leonard Bernstein's West Side Story, or Shakespeare's description of Cleopatra, copied nearly verbatim from Plutarch's life of Mark Antony and also later nicked by T. S. Eliot for The Waste Land."[11] Those varied works of art are depended on the plagiarism of their authors. Indeed, even Lethem's essay is dependent on plagiarism: nearly every word in this remarkable essay is copied from another source. As a reader, we are unaware of this until the last half of the essay, when he lists all ("except, alas, those sources I forgot along the way"[12]) of the sources from whom he used text. Even though only a few of the approximately 8000 words are Lethem's own, the essay is undoubtedly his, in his voice. His essay is like a remix of a song, a new creation that still draws from older works. But intentional copying can also be plagiarism; if either Ambrose or Viswanathan had claimed that they had copied intentionally, they would certainly be plagiarizing. In the end, one must judge plagiarism for oneself. There is no one standard that can be applied to all acts to make a judgment. All art is dependent on the work before it, and it is up to the author to make the decision how they would like to use the work of others. What they must keep in mind, however, is how society will decide if their work is a fair use of past work, or if it is plagiarized. Raymond also posits an interesting thesis at the end of the essay: "The verdict of history seems to be that free-market capitalism is the globally optimal way to cooperate for economic efficiency; perhaps, in a similar way, the reputation-game gift culture is the globally optimal way to cooperate for generating (and checking!) high-quality creative work." Capitalism has been great at providing lots of cheap goods, but it has not been as successful at creating art. Just looking at the number of actors, actresses, painters, photographers, and writers who have to wait tables to pay the rent makes this clear. They love their art, but cannot succeed in the capitalist system by producing it. The film and theater industries may disagree with this though, since they make money with their movies. But judging by the recent crop of multi-million dollar movies, money doesn't make high-quality creative work. It does seem to work for the open source community, however. Linux is a better operating system than ever; it's running some of the most popular websites on the internet today as well as desktops across world. Open source programming languages like Perl, Python, and Ruby are powering the newest applications and games. This gift culture encourages an economy of works based around the name of the person giving the gifts. This is exactly what the author function is about; the function is created in the relationship between the culture and the authors producing the gifts. The author function does several things. It gives ownership to a work, by attributing a person to the work. But the author function also introduces an element of danger. If an author can be attributed to a work, then that author is responsible for that work, both legally and morally. Giving this attribution is typically easy for a novel or other literary work. We take the entire work as having been created by that single author, and she is responsible for that work. Movies and plays get into murkier territory due to the number of people who work on a production. There is a scriptwriter, who provides the text that the actors speak; a director, who decides how that text should be realized; a producer, who manages the relationships between all of the players; and of course the actors and actresses who do the performing. Above the level of the direct influences, there are the corporate interests who fund the productions. When they fund a project, they also want to oversee what their money is producing. With all these influences, who is responsible for a performance? Auteur theory, popular since the 1950s, states that the value of a film comes from its director, since the director is responsible for the film. Auteur theory was popularized by François Truffaut, who argued that directors, not films, are what should be judged by film critics. The director, by controlling the actors and how he wants the script to be filmed, puts his personal vision into the films he creates. Because the director has such control over the film, his vision should be judged with regard to the film. The other people involved in the production are irrelevant. Auteur theory, however, have been criticized for failing to account for the number of people who are involved in a film. It tries to assign ultimate responsibility for a film to a single person when the film is obviously a collaborative work. But it is exactly this application of responsibility that gives this theory their power; giving responsibility allows theorists to focus their analysis on the relationship between a single person and the work instead of a group. Like films, computer software is also a collaborative effort between many people. Is auteur theory fruitful when applied to software? Commercial software is attributed in much the same way as the auteur theory expects of feature films. The name attached to the program gives that corporation the credit and responsibility for that software, as the director receives credit for a film. However, there is no reason that only the corporation should receive that amount of credit. Unlike the director, a corporation is generally responsible for funding the development of a work; it is still individuals who organize and write the code. The corporation is more like the producer or a corporate sponsor; it manages people and money, and oversee what is being produced. But the corporation is an abstract entity and can't write software. One possible explanation why corporations receive all the recognition in commercial software is the lack of concern users of software have for the actual programmers. Generally users only care who they complain to when something does not work, and that is rarely the original programmers. This goes back to the author being the site of responsibility for a work; the corporation's name is held responsible. However corporations, with a fear of lawsuits, try to escape this responsibility. Many commercial programs come with an End-User License Agreement (EULA), which is a contract that the user agrees to before being allowed to install the software. The EULA sets out the rights the user has with regard to the software, and the responsibilities the corporation has. Generally, this means that the user has no rights or guarantees to anything, and the corporation absolves itself of all responsibility. Software companies know that software is unreliable; even the best software does not function the same on every computer or even repeatedly on the same computer. The EULA serves to limit the liability of the company if the software functions incorrectly. What form 'incorrectly' takes can be everything from not performing the functions the user asked of the program to actively harming the user's computer. In the EULA for Microsoft Windows Vista, Microsoft's legal responsibility only extends to "direct damages up to the amount you paid for the software. You cannot recover any other damages, including consequential, lost profits, special, indirect, or incidental damages."[13] These restrictions mean that no matter what happens with the software, the most they will do is refund the cost of Windows. However, incorrectly is always incorrect to the user. Corporations need no protection from programs which do what the user expects. EULAs also restrict what the user may do with the software. Again, the Microsoft Windows Vista EULA denies the user the abilities to:
It prevents users from figuring out how the software functions, or working around bugs. Open source software is the exact opposite. By providing the source code, users are explicitly allowed to fix bugs and are encouraged to contribute their own work to the project. Another problem with giving credit to all the programmers is the difficulty in publicly acknowledging all the people involved in producing a large-scale work of software. Large projects can have upwards of 100 programmers, and no consistent place in the released software (such as the credits at the end of a movie) to place all those names. However, assigning responsibility internally can be easier with software than with other works. Unlike plays or films, software developed using a version control system can maintain very exact records of who is responsible for individual lines of code. However, other difficulties arise with assigning responsibility for software. Even with exact knowledge of who wrote individual lines, software is more than the sum of its lines. When a program does something that is unexpected, rarely is an error in a single line responsible. Bugs occur in combinations of lines, in the interaction between subroutines, between programs, between the software and the operating system, or between the software and the hardware. With all these levels of abstraction, pinning responsibility to an individual or even a suitably small group is rarely possible. There is also a question of what an error is. Computers only follow the instructions that they are given. We would hardly fault a factory worker for making bad cars if the instructions he was given were faulty. Often, the error is merely a difference of opinion: the programmer told the computer to do one action, and the user of the program expected another.[15] The author is a problematic concept, and no less problematic in computer software. However, the concept of authorship as a function of society, advocated by Foucault, and the traditional notions of authorship are not as helpful in resolving this problem. Neither concept adequately handles the multiplicity of people and code responsible for a program running correctly. Even film theory, with a focus on an inherently collaborative medium, fails to be a fruitful area of analysis because it too focuses on one person. Authorship of software is a collaborative act between many people, more people than are typically involved in a film or a play. A full analysis of the author in computer software would require much more space than I have here; this is only an initial attempt. Any theory of the author in programming would have to deal with issues of responsibility between blocks of code and between programmers. Responsibility is difficult to assign when multiple people influence the same work, but it is an important component of authorship. A theory would also need to handle the differences between commercial and open source software. The programmers who write in these (for lack of a better word) paradigms have different motivations, different values, and different attitudes towards authorship which need to be taken into account. Even with all this work to do in the future, I hope that this essay is a useful first step towards addressing these issues. Notes
|