Broadly speaking, I’m in agreement with Grimmelmann’s argument, which is pretty well summed up in his opening sentences: “Copyright law doesn’t recognize computer programs as authors, and it shouldn’t. Some day it might make sense to, but if that day ever comes, copyright will be the least of our concerns.” The problem is, artists can and do very deliberately attempt to give computers creative agency, challenging the very notion of human creativity—and Grimmelmann's examples don't fully reckon with this. I'll quibble with a few things along the way, but the important thing is to consider a more challenging case: George Lewis’s computer program Voyager.
Briefly, Grimmelmann considers (and rejects) four possible circumstances under which computers might plausibly be considered to “create” a work: digital copies, digital works, algorithmic creation, sequential creation, and random creation.” Three of these (digital copies, digital works, and sequential creation) he correctly treats as trivial use of computers as tools for human creativity. It is in his consideration of algorithmic creation and random creation that his analysis does not cut deep enough.
First, it’s necessary to collapse a few distinctions and clarify a few understandings: an algorithm is a set of instructions, but it is not (necessarily) “a process whose steps are completely explicit.” That’s not a bad description of a computer program, but an algorithm can be much more flexible. It’s trite but instructive to think of algorithms as recipes: in ingredients (“5 apples”) algorithms can allow for a variety of inputs (imagine different varieties of apples, for example). This, crucially, allows for unexpected results from algorithmic execution.
Secondly, “randomness” is not an accurate category for discussion of computer-created works. Randomness indicates either a deterministic process occurring at a scale too fine to measure, or a provably random process like radioactive decay, where the distribution of discrete events can be probabilistically predicted over time, but within any given timeframe their occurrence or non-occurrence is unpredictable (cite: I asked a physicist!). As computer scientists know, although we’re working on randomness generators, basically the best we can get at this point is pseudorandomness. It’s important to note that randomness is “unknowable in advance ... uncorrelated with anything in the universe”—in a word, unpredictable. Unpredictability combined with utility or delight edges awfully close to a “spark of creativity,” especially when a system is designed specifically to maximize the freedom in which an algorithm operates.
As a partial aside, how randomness is programmed is essential to considering the role of creative agency. Sonic Pi, a popular environment for programming music both for recording and live-coding, will by default deliver the same set of “random” numbers each time a programmer calls a script with the command “rrand” the same pseudorandom sequence will be delivered (to be processed by other commands into notes, rhythms, or other attributes). As a performer or composer, one can learn to “play” the randomness—gaining control over its idiosyncracies like any other attribute of one’s instrument or medium.
It may also be worth noting that a sonic pi script is “in fixed form” as code or as a sound recording. This script will, unless it is modified, always produce this music, despite containing calls for randomness (in this case “choose” commands). Obviously this is an extremely trivial view of “randomness”—it’s only surprising the first time it’s produced—and the computer’s pseudorandomness should be viewed as an adjunct to the creativity of the user.
John Cage’s aleatory music (which is treated as a test-case for random creation) is closer to “randomly generated” but Cage aimed to reduce the role of the composer, not to increase the creativity of a process. As he wrote:
And what is the purpose of writing music? One is, of course, not dealing with purposes but dealing with sounds. Or the answer must take the form of a paradox: a purposeful purposelessness or purposeless play. This play, however, is an affirmation of life—not an attempt to bring order out of chaos nor to suggest improvements in creation, but simply a way of waking up to the very life we’re living, which is so excellent when one gets one’s mind and one’s desires out of the way and lets it act of its own accord. (John Cage, Silence, Hanover, NH: Wesleyan University Press, 1973, p. 12)
Cage’s copyrights are full of contradictions and ironies (and Cage himself relished these ironies more than the law does), but there is a more apposite model for Grimmelmann to probe: George Lewis’s Voyager. In Lewis's own words:
My analysis of Voyager as an interactive computer music system uses Robert Rowe's taxonomy of "player" and "instrument" paradigms, although these two models of role construction in interactive systems should be viewed as on a continuum along which a particular system's model of computer-human interaction can be located. In Rowe's terms, Voyager functions as an extreme example of a "player" program, where the computer system does not function as an instrument to be controlled by a performer.Lewis’s explicit aim in creating Voyager was to create an equal agent in composition. In doing so, he problematizes the notion of creativity when it comes to copyright. This radical re-envisioning of agency itself is key to considering computer authorship. If the creators of a computer program intend for the program itself to be creative, that must be reckoned with.
I conceive a performance of Voyager as multiple parallel streams of music generation, emanating from both the computers and the humans--a nonhierarchical, improvisational, subject-subject model of discourse, rather than a stimulus/response setup.
But law aims to create understandable categories, and art aims (among other things) to challenge our categories. It's an eternal (if usually friendly) struggle.