Submit to Digest

Should Lawyers Dream of Electric Sheep? Digital Art: A Dynamic Misfit in a Static System

Copyright Commentary Notes

I. Introduction

The current copyright framework is becoming obsolete as we try to make a digital world run on an analog legal system. The Copyright Act covers “original works of authorship fixed in any tangible medium of expression, now known or later developed, from which they can be visually perceived, either directly or with the aid of a machine or device.” 17 U.S.C. § 401. The ability of the digital medium to interact dynamically with a user or generate unique visualizations along predefined parameters has enabled many novel forms of art which could not have been foreseen by the drafters of the Copyright Act of 1976.[1] Terms of art such as “fixed” or “copy” have lost much of their meaning, and the law’s notion of a tangible medium is becoming less relevant.[2] The digital arts are a misfit medium within current copyright categories as the underlying software and possible Internet platform add dynamic and collaborative elements which were previously absent from intellectual property. This Comment will address the disconnect between traditional copyright law and the realities of the digital age, using emerging forms of digital art as archetypal examples. 

II. The Digital Arts

Digital art is defined as a work of creative expression which has been non-trivially modified with digital technology as part of its creation or presentation. Works are often interactive, collaborative, or algorithmically generated along parameters set by the initial producer. The digital arts utilize software to produce different forms of creative expression, ranging from a static digital painting or photomontage to dynamic and unpredictable visualizations generated by a web program which uses constant input from the Internet.

The Copyright Act differentiates between “pictorial, graphic, and sculptural works” and audiovisual works, yet in regard to the digital arts it is much more relevant whether the work is static (fixed, lacking in change) or dynamic. In the past, copyrightable works were by their very nature static unless a human author effected change. Hence, requiring that a work be “fixed in any tangible medium of expression” made sense — fixation was the point at which legal copyrights attached to the completed artwork. In contrast, a digital artwork is never “fixed” in the traditional sense, as it is inseparable from the software and is regenerated as many times as the computer refreshes the image from its memory.

The digital arts often experiment with possibilities on the edge of technological innovation. A quick introduction to some of the terminology common within the genre is essential to understanding the following legal discussion. Digital art is often interactive — the software will generate a unique audiovisual display based upon user input. Works are often collaborative — many individuals make creative contributions, generally over the Internet. [3] In addition, digital artists often create generative algorithmic art by encoding algorithms that define the parameters of the artwork into the underlying software that guides its generation either autonomously or in response to input.  Computer art is a broader term than digital art and would include artworks such as unaltered (or minimally retouched) digital photography. One popular form of digital art depicts fractals – a common style of abstract algorithmic art where software calculates an image or animation from preset parameters. Information art, or informatism, is an emerging art form where the code generates artistic content by processing large amounts of information.

All of these factors are present in the famous “Electric Sheep” distributed computing project. Developer Scott Draves programmed a screensaver which displayed animated, evolving fractal flames. Once downloaded, the user could program and upload their own unique fractal “sheep,” download and modify existing sheep, or combine (mate) existing sheep to create hybrid fractals. The program was designed to upload and incorporate new sheep into the “flock” existing on the main server. These dynamically generated fractals could be voted up or down and a sorting algorithm took the votes into consideration when defining the parameters along which the fractals evolved. The Electric Sheep project was simultaneously generative, interactive, and collaborative. The original artist, downstream users, and server administrators all influenced the dynamically generated fractal flames. While a single snapshot of the resulting image could be fixed, the artwork itself consisted of a web of inputs dynamically generating a visualization through the underlying software algorithms, and the entire project was based over the internet.

A primary consideration when differentiating between the digital arts is whether the software was used as a tool in creating the work, or as the medium of expression. When software is merely used as a tool, there is a readily apparent difference between the code of the artwork and the imaging program. For example, digital photographs and paintings (“painted” on computer imaging software) are types of visual artwork in which electronic technology is used solely as a tool of production. When an artist uses Photoshop to create a digital painting, the underlying generative software is intuitively distinct from the code of the JPG file which records the image. Digital photographs and paintings are two of the few forms of digitally encoded artworks that comfortably fit within the current copyright categories, as there is a definable point where the work can be considered completed, static, and fixed. Contrast these to generative artworks, which actively use software algorithms as essential, dynamic parts of the resultant work. Software art utilizes the code as a medium of expression, and the artist generates aesthetically pleasing designs from dynamic user input, distributed systems, or real-time data sets – similar to the way a kaleidoscope generates unique patterns when a user manually turns the object box. Dynamic paintings are an intermediate example of software as a medium of expression.[4] The artist digitally paints an image and programs functions into the underlying code to control how the image will transform over time. The exact images rendered over time are unpredictable – the computer platform randomizes along parameters set by the artist.

A “fixed” animated audiovisual work is read in much the same way as a static digital image, with the self-contained file read by a compatible program and projected onto the user’s screen. For digital paintings, information visualization generators,[5] and even videogames, the underlying code has been fixed - software is used as a tool and does not change. However, in collaborative artworks, such as Electric Sheep, both the audiovisual display and the code are continually changing – software becomes a medium of expression. This type of work cannot be reconciled with traditional copyright law, which assumes a degree of fixation which may never be attained.

III. Current Legal Categorizations

Digital art is a legal misfit. It spans three statutorily defined authorship categories, so multiple and perhaps mutually exclusive copyright interests attach to the different aspects of a creative work. Under 17 U.S.C. §102(a)(1), the underlying software can be copyrighted as a literary work, or under §102(a)(6), transient animated displays may be copyrighted as an audiovisual work. See ex., Midway Manufacturing Co. v. Artic International, Inc., 547 F. Supp. 999 (N.D. Ill. 1982). In addition, under §102(a)(5), a work of visual art may be categorized as a “pictorial, graphic, and sculptural work.” Typically works covered by §102(a)(5) receive increased protection under the 1990 Visual Artists’ Rights Act (“VARA”). However, VARA limits its scope to limited-edition paintings, drawings, and photographs – audiovisual works are specifically excluded. Insofar as digital artworks are infinitely reproducible and often dynamic, they are excluded from most fine arts legislation. Problems also arise in determining when digital artworks qualify as “original works of authorship fixed in any tangible medium of expression” as required by the Copyright Act. Precise legal definitions of “original,” “author,” “fixed,” and “tangible” have proved elusive.

A work of digital art may not have an “author” in the traditional sense of the word. Legitimate authorship is undermined because “works produced by mechanical processes or random selection without any contribution by a human author are not registrable.”[6] When the artist is primarily a programmer who works directly with dynamic software, the legal status of the resulting visualization is uncertain – the computer is responsible for much of the content generated. Creative intervention by the artist avoids this pitfall, yet work that utilizes software as a medium of expression might run afoul of this limitation on authorship.[7] The requirement of “original works of authorship” also raises issues when the work is collaborative, or automates visualizations based on user input.

Legally, copyright is created when a work is “fixed” in a tangible medium, yet digital works are by nature transient. Numerous copies are generated as the coding is transferred and read and the audiovisual displays may be constantly changing. The Uniform Commercial Code makes an individualized exception to tangibility requirements for software to fit it within the older legal regime, yet the text specifically excludes the digital arts.[8] Circuits differ in how they deal with the disparity between legal definitions and digital realities. The Seventh Circuit affirmed in 1983 that transient images generated by an underlying program could be considered sufficiently fixed to receive copyright protection. Midway, Mfg. Co. v. Artic Int’l, Inc.,704 F.2d 1009, 1011 (7th Cir. 1983) (protecting video game displays as audiovisual works); see also Stern Electronics v. Kaufman, 669 F.2d 852 (2d Cir. 1982) (finding that, in addition to the source code, the images and sounds in a game could be copyrighted).[9] Similarly, the Ninth Circuit Court of Appeals held in 1993 that a computer technician fixed an infringing copy when he loaded a software program into RAM. MAI Systems Corp. v. Peak Computer, Inc., 991 F.2d 511 (9th Cir. 1993). Conversely, the Second Circuit held in 2008 that to be fixed, a work must be embodied in a medium for a period of more than transitory duration—momentary copying while streaming content did not constitute infringing copies. Cartoon Network, LP v. CSC Holdings, Inc., 536 F.3d 121 (2d Cir. 2008). In many ways, the courts must engage in technical analysis to which they are ill-suited and draw an arbitrary line. In a work—such as such as Electric Sheep—where both the software and the audiovisual displays are integral components of the creative expression, certain aspects would fall outside the law and remain unprotected.

IV. What Needs to be Done

Digital art occupies an uneasy position at the intersection of fine art, the applied arts, and software. Copyright protection extends equally to any work fixed in a tangible medium. Yet classification problems arise because works in a digital medium are inextricably linked to the underlying software and depend upon computers or electronic displays in order to be physically perceptible. Moreover, issues concerning more finely-tuned classifications and protections arise due to the diverse properties of creative works which fall under the umbrella of “digital art,” which do not conform to the neat categories laid out in copyright law. Insofar as digital images exist as information on a computer, they evade traditional fixation.[10]

Delineating the boundaries between the code and the visual artwork or disentangling individually copyrightable component parts may be a futile task.[11] Within a single digital artwork, copyright law allows for separate copyrights in the software, in the audiovisual display, or as a pictorial/graphic image. Looking back at the Electric Sheep example, the need for a rewrite of copyright law becomes painfully apparent. Scott Dravis initiated the project in 1999, yet users and server administrators are constantly uploading new algorithmic parameters for the fractal flames or recombining them in new ways. The computations are spread over thousands of user computers (when otherwise idle) which dynamically generate the finished animations and redistribute them over the user network.  The “sheep” in the main server number merely 100 or so at any given time. The different iterations show more than enough creativity to qualify for copyright protection, and after ten years very little may be left of Davis’ original work. All contributors may have a copyright claim, but no one could claim the amalgamated visual display in its entirety.[12]

The law creates exceptions and allowances to plaster over this cleft between traditional copyright concepts and digitized works, yet problems in statutory interpretation arise because the Act has not been rewritten to account for a basic shift in the nature of copyrightable works. Software allows works to be dynamically generative, and the Internet allows the easy exchange of information for collaboration. It is not possible to “fix” many works in a digital medium in a static, completed form. Authorship must be redefined to incorporate different levels of collaboration and contribution. Software must have a legal identity that accounts for its unique properties and the relationship it currently holds as a building block or language in which the other arts are written. We would not attempt to run cutting-edge software on a computer from 1976; we should be similarly sensible with our law.

[1] But see WGN Continental Broadcasting Co. v. United Video, Inc., 693 F.2d 622, 627 (7th Cir. 1982) (noting legislative history suggests Congress wanted courts to interpret definitions contained within the Copyright liberally, to avoid the need to periodically update the act).
[2] For an extreme example, developments in biotechnology and genetics have resulted in a new art form called “bioart,” wherein artists may modify genetic codes to form messages or grow sculpture from muscle or nerve cells.
[3] In legal arguments the digital arts are often differentiated from other types of creative work because that they are interactive. Yet “interactive” is inherently vague when used as a term of art – any experience of artwork or literature may be considered interactive to some extent. See Brown v. EMA, 131 S. Ct. 2729,  2737–2739 (2011)
[4] See, e.g.,
[5] See, e.g., (generates a visualization based on the frequency of word usage in a given sample of text).
[6] U.S. Copyright Office 503.03(a) Compendium II of Copyright Office Practices.
[7] Mechanically generated creative works blur the line between copyrightable “original works of authorship” and those to which no copyright can attach. Digital art pushes the limits, and has since its inception. Desmond Paul Henry, one of the first computer artists, constructed a succession of three drawing machines from analog bombsight computers salvaged from WWII bombers. The machines could not be programmed or store information, and relied on slight imperfections in construction to create abstract curvilinear line drawings. The machine would produce unique and unpredictable drawings, and the artist could interfere or not. Conceivably the “human author” contributed by inserting the paper and removing it when finished. Yet with automated digital artworks, even this small contribution is no longer necessary.
[8] See Nancy S. Kim, Expanding the Scope of the Principles of the Law of Software Contracts to Include Digital Content, 84 Tul. L. Rev. 1595, 1597 (2010).
[9] After Midway, the U.S. Copyright Office stopped allowing videogames to be registered as both an audiovisual work and a literary work. Instead, producers must choose whether to copyright the display or the underlying code.
[10] Gibbons, supra note 11 at 545.
[11] Dan L. Burk, Muddy Rules for Cyberspace, 21 Cardozo L. Rev. 121, 122–123 (1999).
[12] Informative cases have arisen within the video game industry. When user-generated content is solicited, game developers define terms in their licensing agreements. Typically the developers claim a copyright interest in the derivative work, deeming it to be fair use so long as it remains noncommercial. See Micro Star v. FormGen Inc. 154 F.3d 1107 (9th Cir. 1998); but see Galoob v. Nintendo, 964 F.2d 967 (9th Cir. 1984). Similar to the Electric Sheep project, the virtual world Second Life exemplifies a new kind of user relationship which provides the servers and platform, but much of the content is coded and uploaded by users. Contributors to Second Life then receive copyright interest in their work. The Electric Sheep project utilizes Creative Commons licenses to maintain an open legal platform for the collaborative work. <>