Response to Noah Weber’s Comments on Emily Howell
I want to thank Noah Weber for his thoughtful comments on Emily Howell’s music. My thanks to Weber may seem strange given the overall negative tone of his review. However, I greatly appreciate people who carefully listen and study the music they evaluate regardless of their ultimate findings. Rather than take each of Weber’s ideas in turn, I here group them according to common themes. I hope this does not present problems for readers in making connections between my clarifications and his points.
Weber defines language as syntax and grammar which I find redundant. Language is a combination of syntax (nouns, verbs, etc. in their proper order) and semantics (the meanings of these words). That music has syntax is undeniable. We have hundreds of years of proof in both classical and popular genres. That music contains semantics is less clear. In language we have dictionaries with clear meanings of words. In music, not so. Certainly the use of musical allusions such as “Dies Irae” and other quotations indicate vague meanings. However, I know of only a few attempts (see Cooke’s The Language of Music or the synthetic language Sol-re-sol) at creating dictionaries of meanings of musical ideas. And these musical dictionaries are necessarily incomplete, not to mention that no one I am aware of consults them. Therefore, music is half a language, the syntax half, with communication—the sending and receiving of an intended message—minimal at best. Words seem inadequate enough without resorting to melodies and harmonies as alternatives. Further, much of the magic of music, for me at least, rests in its ephemeral qualities. That we can hear the same work and each derive a different message and pleasure from it is part of music’s charm. Thus, intent on the composer’s part rests in getting the notes right, not in conveying some hidden message.
I have never claimed that Emmy has passed any kind of Turing test as Weber claims. In fact, while I greatly respect Alan Turing and his work, I don’t believe his test truly identifies intelligence. For example, what credentials does the individual judging have? Can he or she really tell the difference between intelligence and non-intelligence? As far as I know, no one has ever defined intelligence in a way that we can all agree upon. Without a complete notion of “I”, how can we have “A.I.”? What I have claimed is that Emily Howell has passed a test for artificial creativity, and I devote quite a number of pages to this in my book, Computer Models of Musical Creativity (MIT Press, 2006).
In reference to the story Weber finds problematic, his source (a popular journal) got it wrong. Here’s the correct version. After a performance of Emily Howell’s Opus 1 for two pianos, a non-music professor—who has a musical background—mentioned to my wife (one of the performers) that Howell’s piece was incredibly moving and the best work on the program. My wife, not me, had ‘advertently’ left out any information in the program notes about Emily being a computer program. Roughly six months later, this same audience member attended a lecture of mine on computer-composed music where I played the same Opus 1 via CD. On this occasion, not having made the connection between the two events, this professor claimed he could immediately tell what he’d heard was computer composed because it was cold and emotionless.
I will resist discussing either Weber’s prostitute metaphor or his references to eHarmony, iTunes, and so on. I simply don’t know enough about how any of these relate to Emily Howell to comment. However, the question of who owns Emily’s intellectual property rights should be clear. I do. I created the program. I interact with her as audience during the composition of new works. I decide if a new work is interesting enough to perform or record. The same is true of Emmy’s output, all the scores of which state “By David Cope with Experiments in Musical Intelligence.” I might also note that the reason for my programs only creating works by living composers in Cope style results from a reluctance to face the possible litigation that might follow.
Regarding the problems which Weber has detected in comparing the score of Land of Stone to the CD of same, recording is an extremely complex process. Between composer and listener are several layers of individuals that many may not know much about. Performers, conductors, producers, recording engineers, and editors come to mind. Each plays an important role in making a recording. The producer, for example, listens during the recording sessions to determine when the music is “in the can,” that is when all the notes are covered correctly in at least one of the “takes.” The recording engineer controls the mixing board and decides the balance between instruments. The editor decides which take works best and how to edit it into the takes of the music on either side of it. For Land of Stone, I had limited funds for the entire project. This meant limited time for each aspect. In the end, nothing works perfectly. Mistakes creep in during recording, especially when certain performers just don’t get it. Thus, substitutions can occur at the last minute. Often certain measures aren’t really in the can and a composer or editor may have to make a choice between releasing a recording with a bad reading of a passage or just cutting it. In short, and especially with new music, score and recording don’t always match. In fact, in my experience, they rarely match. My aesthetics, however, at least in terms of composition, do not play a role in making these changes.
Per Weber’s suggestion, I visited noahsweber.com/supplement using two different browsers, but could not access his materials. I presume he is waiting until publication of his critique before posting the information. Thus, I cannot comment on the particulars. I will add, however, that the note for the contrabass that he states is out of range is not out of range. As a professional contrabassist for many years, I can play it and so can most of the bassists I know.
I wish that Weber had read Computer Models of Musical Creativity rather than continually quoting from the popular press. In my book he would have found that I am very much involved in Emily Howell’s composing process. I am “her” audience so to speak. Without going into details, I “carrot and stick” her through the composing process. Works do not appear at the “push of a button.” Though my input does not ever give predictable results, it does nudge Emily’s output more towards what I prefer. All this occurs long before the score exits the printer. Since this process is presented in detail in a number of sources, I am a bit surprised at Weber’s lack of knowledge on this point. I also feel humans compose using trial and error. Beethoven’s sketchbooks represent a perfect example of this.
Weber comments on the effects my work may have on living composers citing an article by Milton Babbitt. Interestingly Babbitt, who died recently, did not title his “Who Cares if You Listen?” The editors did that in order to create controversy and sell more copies of their magazine. Without his permission I might add. The title is an oft quoted source for his supposed belief in music created solely by science. I suggest that anyone interested in this subject read the article before judging it on its title alone. Babbitt makes quite legitimate arguments for a different way of looking at and listening to music.I have heard the words “truly original music” many times over the years. They’re often used to suggest that humans, as opposed to anything non-human, have the ability to pull rabbits out of hats by developing completely new things, when in fact, we humans are, like everything else in the universe, combining and recombining already extant things to produce new and original things. I created a free program (again see Computer Models of Musical Creativity) that identifies sources of music touted to be “truly” original. For example, the so-called Tristan chord, supposedly originated by Wagner, existed long before him with roots in Beethoven, Liszt, and others and of whose music Wagner was fully aware. Spohr, for example, wrote an earlier opera that begins so similarly to Tristan that it’s difficult to believe that Wagner didn’t lift his music directly from that work.
Beneath a lot of what Weber says about Emily Howell rests this notion of humans versus machines. I find it confusing that when we use computers for bookkeeping, Internet shopping, email, and so forth, it’s called a tool. But when some people use computers for creative work, computers suddenly become beings in their own right, apparently operating as something far greater than tools. This is nonsense, of course. Computers and all the software running on them were and are created by humans. Computers simply provide the same speed and accuracy they do for bookkeeping, Internet shopping, email, and so forth. The only thing computers do is carry out instructions. The fact that I cannot predict what my program’s outcomes will be is not because my program has a mind of its own, but that its computations are so fast and complex that I can’t keep up with it. This is true of all computers everywhere. No computer anywhere operates independently of its creators and users.
I like the word “algorithm” because it existed long before computers and makes clearer the idea of computers as tools. Algorithms are recipes, sets of rules designed to reach some kind of desired result. Beethoven used algorithms limiting his harmonic vocabulary, his orchestration, his form, and so on. We all use algorithms. DNA is an algorithm. The only difference between Beethoven’s algorithms and a composer who uses computer algorithms is that Beethoven used software and computer composers use both software and hardware. I could have produced all the output of my computer programs were I to have the time, patience, and accuracy necessary to do so. I don’t. It would have taken me several lifetimes.
Weber claims—through another popular press quotation—that I am “audacious.” I think he would not find me so audacious if he’d read Computer Models of Musical Creativity. Emily Howell’s creative process is actually quite simple. She does not respond to “hoots and hollers,” but keyboard input. Weber asks, “Is there nothing to stop Emily Howell from churning out soulless, conformist tripe?” Again I suggest that recombination is a powerful process. “Churning out?” Well, I suppose one could call it that, though as I discuss in my books and articles, it takes a great deal of time and effort to produce the resultant music. “Soulless?” Well, I’m glad that someone knows what soul is. I surely don’t. Maybe Weber can enlighten me. For example, where exactly is soul located? Do only humans have souls? As for “conformist,” I don’t think that recombining groups of music necessarily conforms to anything. For example, sulfur and carbon are relatively simple elements. Potassium nitrate a compound. Mixed together in the proper proportion (10% sulphur, 15% carbon, and 75% potassium nitrate) forms gunpowder. Otherwise the ingredients remain inert. Mixing the ingredients just right produces the lethal “new” combination.
Finally, let me respond to a number of quotes from Weber’s critique. To begin, “She is nothing but a computer program.” This statement suggests that Weber imagines computer programs to be relatively worthless things, even though they pervade most of our lives in critically important ways. Weber states that I insinuate that “academia is reluctant to acknowledge my brilliance.” To the contrary, I feel that my many writings on the subject indicate the opposite, at least in regard to “my brilliance.” I have no idea what importance, if any, my work has or will have in the future. I am definitely passionate about it, but at the same time cautious in my claims. Weber argues that Emily has an “inability to ever truly create original works.” My previous comments regarding gunpowder relate to this. But as further testimony to the lack of validity of his statement, most scientists would argue that recombinance represents the core of originality. The pressures exerted during the Big Bang produced oxygen from the combination of hydrogen and helium. For that matter, every other element and compound in our universe results from the same processes. I previously mentioned DNA, but I’ll add here that each human being begins with genetic recombination during crossover. In short, recombination represents one, if not the sole source of originality we know of, and is the core of all my programs. I suppose I should be dismayed by Weber’s final comment that, in essence, I have spent the last thirty-one years of my life creating “little more than novelties . . . dwarfed by their potential negative consequences.” I hope he will not be too disappointed to learn that I am just now beginning a recording of Opus 4 and soon thereafter opera 5 and 6 for the next CD of Emily Howell’s music.
Again, I thank Weber for his critique of Emily Howell. Few people have taken the time to ponder the concerns he has so articulately stated. If they had, maybe more dialog would take place over what I think are important issues. Because of his work, this now may occur.
David Cope is Dickerson Emeriti Professor at the University of California at Santa Cruz where he teaches theory and composition. His over seventy published compositions have received thousands of performances throughout the U.S. and abroad, including those by the Vermont, Pittsburgh, Indianapolis, Cabrillo Festival, and Santa Cruz Symphony Orchestras, as well as numerous university orchestras and wind ensembles. His books Computers and Musical Style, Experiments in Musical Intelligence, The Algorithmic Composer, Virtual Music, and Computer Models of Musical Creativity, describe the computer program Experiments in Musical Intelligence, created by Cope in 1981, which functions by inheriting a composer’s style and then composing new music in that style.