Tag Archive | Economics

Experiments in Agent-based Sonic Composition

John Cage’s “Music of Changes,” which was composed using a random component from the iChing.
I perform and write music, normally acoustic, and usually for a single guitar, harmonica, and voice. I am traditional in my choice of instruments, they are basically “old” technology. On the other hand, I am also fascinated by the idea of robotics in music. The idea of artificial, autonomous music creators that work alongside human musicians. John Cage used the iChing to make choices about musical form in some of his compositions, including “Music of Changes” above, which has some of that flavor. It is music that is composed, not just performed, by a partially artificial means–by a non-human actor, the iChing.
In my work as an economist, I develop autonomous software programs that simulate economic actors in a process called agent-based modeling – the construction of independent pieces of software, which simulate real agents in the world, that interact and form patterns that transcend any single agent’s behavior. Recently I realized that agent-based modeling might be able to be applied to the construction of music: creating individual artificial decision makers which might together construct a piece of music that transcends what any one of them can do.
Think of a swarm of bees or a school of fish. Once biologists thought that schools of fish had a `leader fish,’ a single fish that would direct how the school would move. Biologists also once thought that the queen bee was the `leader’ of the hive, that it directed behavior of the bees in the hive. Both of these beliefs have been shown to be false. There is no leader in a school of fish. On the contrary, each fish responds to local information and then the co-ordination which arises on the school level emerges from this system of individual choices. The same with bees…the queen plays a part in the hive, like all the bees play parts, but there is no sense in which she directs the others. There is no bee that is in charge.
Here is a video of my colleague Hiroki Sayama’s `Swarm Chemistry’ in action. The specks you see on the screen are individual agents, dumb agents, who react to their environment, which is other local agents. There are no leaders here, there is only group behavior.

In this clip, you can see the swarms which emerge. The music is incidental in this clip; not a result of the swarm behavior.
I have begun an experiment in agent-based sonic composition with the idea of emergent behavior and agent-based modeling in mind. In this video I show my initial foray into this world:

The agents in this video are small triangles that seek a well, and eventually learn (sometimes more effectively, sometimes less effectively) where that well is. What I have done to add a sonic component is to assign each agent an instrument, and assign the agent’s proximity to the well to the pitch of the note they create.
“Random” sounds created by a computer are nothing new. And, frankly, I find them uninteresting. No depth, no humanity. But I think agent-based sonic composition might be something different. These agents are not simply random (although indeed, their behavior has something of a random component, or seemingly random component). They are goal-seeking, they are purposeful, and the sound they generate is a function of their effectiveness and path in pursuing that goal. I think this purposefulness can be heard in the sound the create. There certainly isn’t a melody, but there is a story being told, some kind of struggle being documented.

Swarms, too, are not simply random. Though swarms may be composed of elements have that have randomness in them, they are also structured. If Music is sound with structure, and complex systems is the study of emergent structure, there could be a genuinely interesting music that might emerge from a well-constructed agent-based approach to sonic composition.
I’m not convinced what I have is there yet. There are not interesting interactions between these agents, and there is not a structure to their sound that has depth – yet. Perhaps the next step is to tie the goals of the agents more explicitly to music making. Perhaps there can be melodic agent who moves on a predetermined path, and the other agents try to follow that agent, and hence the sound that comes out documents their struggle. Maybe the agents’ notes should be restricted to scales, so that it sounds less chromatic. Or, perhaps, as I suggest in the video, there can be some agents which control rhythm and others that control pitch.
To be clear: I wouldn’t just listen to this. I don’t know if I would call it “music” yet. But I think it may get there some day.

Andreas Duus Pape: is an economist and a musician.  As an economist, he studies microeconomic theory and game theory—that is, the analysis of strategy and the construction of models to understand social phenomena—and the theory of individual choice, including how to forecast the behavior of agents who construct models of social phenomena.  As a musician, he plays folk in the tradition of Dylan and Guthrie, blues in the tradition of Williamson and McTell, and country in the tradition of Nelson and Cash.  He plays acoustic guitar, harmonica, and voice: although the technology of his musical production is a hundred years old, his ideas are often quite modern, and he covers songs as old as early last century and as recent as this one.  Pape is also an assistant Professor in the department of Economics at Binghamton University, where he teaches microeconomic theory at the undergraduate and graduate level.  He is a faculty member of the Collective Dynamics of Complex Systems (CoCo) Research Group: http://coco.binghamton.edu and considers complex systems and agent-based modeling to be central to his research

Head Games?: The Strategic View of Liveness and Performance

When I tell people that I am an economist and a musician, they usually have one of two reactions. Either they tell me that I must be crazy, or conflicted—that the two things can’t possibly go together—or they immediately start talking about ticket prices, drops in CD sales, 360 deals. I however, refuse both stances. The connection that I see between what I study as an economist and how I perform as a musician is the element of strategy.

Andreas Pape performs at The Beef, Binghamton, NY, 10/16/10

Performance, in my view, is the willful construction of a series of events to create a particular mental state for the witnesses. This is the strategic view of performance. I am a game theorist, and game theory is the foundation of the strategic view. Game theory is based on the idea that games are a metaphor for human interaction generally. It is essentially the study of strategy: the chess player imagines different actions he can take, and imagines how his opponent will respond in each case, and uses those forecasts to make his original choice; that’s strategic thinking. In “Singing to my Imagined Listener,” I describe rehearsal as playing to an imagined audience member, judging her response, and adjusting accordingly. That is exactly the strategic view.

I got the opportunity to explore this synergy between live musical performance and economics in an intimate and visceral way a couple of weeks ago—February 9th, 2011, to be exact—when I was asked to speak to a small group of students at Binghamton University who study live performance in an English course called “Representation and Popular Music” taught by Jennifer Stoever-Ackerman. I thought I would reproduce some of that talk here, via video clips, in order to breathe elements of aurality and liveness into the words that follow, which meditate on strategic differences between liveness and recording in the game that we call music.

But first, a song.

In this clip, I play the guitar and harmonica and sing a song of mine called “Sittin’ on the Mailtrain.” I strategically start off with a song that has a jarring chord in each line: my intent is to make the audience feel a bit uncomfortable with this shared live experience, so that they are more inclined to look at it with fresh eyes. At minute 7:45, I point out that I am giving a live performance to those in the room, and recording to those in the future. That’s you. As I say at minute 8:07: “There’s an audience here, in the room, and there is an audience out there in the future, who are experiencing this, but clearly in a different way than you are experiencing it.”

And, I think that you are. Even now. However self-aware and live-esque, this recorded object cannot reproduce the physicalness and immediacy of performance.

Performance is standing in front of people, feeling nervous or confident, holding a guitar, forming words, reading faces, projecting to the back of the room or getting quiet. Performance is hitting taut strings pulled across a wooden box at specific times and with a certain speed, vibrating vocal chords in a certain way, holding ones hands out to make a point, or inhaling a wail out of a blues harp. It is a series of events that are a part of a human life, in the sense that life is a series of physical moments. Agency in that visceral present moment is the essential difference between live performance and recording. Like Kathleen Hanna (frontwoman of Bikini Kill, Julie Ruin, and Le Tigre) wrote in “On Not Playing Dead” in 1999, “[O]ne thing I do as a performer is to stay physically present on stage, and that means being in the now. (Oh my god, I sound like such a hippie.)”

Halfway through my talk, however, my computer interjects with a pre-programmed dialogue that complicates Hanna’s claims. Watch here:

“Excuse me, I’d like to make a point,” my computer says aloud, for all to hear. “I felt it was important to point out that this is not exactly a live event. This is a recording, in some way.”

“Sure,” I reply, according to script.

“You typed this in to simulate this conversation that you’re having right now.”

“Yes,” I reply, “it’s scripted. Did anyone not know this was scripted?” I look questioningly at the students assembled in front of me.

“But this [lecture],” my computer points out, “is basically a recording. It’s an encoding of a particular process. So [the] physical body and mind [of the performer] decodes this script into a process, just like the CD player decodes the CD into a process! So how is this performance not a recording?”

“A recording encodes a performance, and a performance decodes a recording,” I say.

There is a strong way that any “live” performance has a recorded aspect to it and vice versa. The decoding of a CD is a performance, akin to my live performance. My computer worked from a script in the computer science sense—a set of essentially English-language commands that it followed to reconstruct a set of sounds. That is not a traditional recording, in that it is not direct storage of sound waves in magnetic tape or record grooves. However, it is functionally a recording: the user presses a button and a predictable and pre-specified series of sounds emerge.

The Computer Performs its Script, The Beef, Binghamton, NY, October 16, 2010

If we agree that this computer script is a recording in this sense, then we are compelled to accept the next step—that if I, a human, am following a script, that I, too, am simply decoding a recording. That is, I had an idea about how “Sittin’ on the Mailtrain” would go and this idea was necessarily encoded in my mind; then I unpacked this encoding by arranging physical objects, namely my fingers and my voice, to create a song. The song followed an encoding in my head, like when you put a CD in a CD player. The encoding is unpacked, and ultimately results in the same thing: some vibrating object that vibrates the air which then vibrates the audience’s ears in a relentlessly physical way.

The future of performance lies in acknowledging the interrelationship of liveness and recording and further blurring the boundaries between them. The podcasts produced by the lo-fi movement are a key part of this new relationship. I am part of this artistic movement, which asserts the primacy of performance over recordings while also using recording technology to foster and promote liveness. Lo-fi’s hallmarks are: smaller numbers of performers in groups (often solo acts), an emphasis on live recordings complete with audience noise, low production quality (“Background hiss”), and a large number of recordings that often include many versions of the same song. The primacy of performance means the definitive versions of lo-fi songs are not located in recordings that live performances then try (and often fail) to recreate. Rather, the most recently performed version is the “master.” The performance you just watched of the song “Sittin’ on the Mailtrain” for example, was the most definitive version on February 9th. Today just may bring a new definitive version.

Lo-Fi Picture of Pape performing at The Beef, Binghamton, NY 10/16/10

In the aesthetics of the lo-fi movement, the life of the performer is treated not as a series of objects, but rather a series of events, which can be attended or subscribed to, like a podcast. No doubt, each episode is recorded, and the audience receives it as a recording. However, these recordings are meant to be listened to once or twice and let go; they are intended to be ephemeral. A podcast, when viewed as a process over time rather than a possession, is no doubt a performance; the audience can respond from one episode to the next via comments, email, Twitter, etc. and the performer can react. What you are reading and viewing here is simultaneously a lo-fi recording event and a lo-fi performance. This. This blogpost, you reading it, the videos you can watch and listen to, my comments on it here, your comments below that you can post, you sharing it on Facebook or Twitter. You can even follow the traces of this performance through my own Twitter feed.

Strategically, I think the podcast model is the next logical step in the Lo-fi aesthetic. Standup comedy (one of my favorite kinds of live performance) is making this transition as we speak. The old model for the young comedian was to develop “an act” that one (hopefully) toured with, perhaps releasing a comedy CD or landing a role on a sitcom. The new model is a couple of comics releasing a conversational podcast once a week, responding to their biggest fans, giving a raw, intimate, unpolished performance of improvisational humor and riffing, and convincing their fans to become members; a membership that occasionally awards the listener with additional content, but more often only a sense of satisfaction that one gets from supporting something one loves. See, for example, the podcast “empire” of Jesse Thorn at maximumfun.org which includes live comedy podcasts, or the political humor of wearecitizenradio.com, which is also member-supported. What’s interesting, here, is that a pure donation podcast model is enough for some comics to make a living. Ironically, using recording to give primacy to performance, serves the artist. Yesterday’s recording can be taken away from the artist, but tomorrow’s performance cannot.

As for my own future? My own “tomorrow’s performance”? On February 25, 2011, at the Eastern Economics Association Meetings, I will perform a similar event, called this time “Rhetoric, Choice Theory, and Performance.” I will perform music and discuss the strategic view. Economists are not used to thinking seriously about performance nor are they used to thinking seriously about sound. I intend to change that, one strategic moment at a time.

Additional footage in which I define strategy and game theory, and discuss what the strategic view of performance has to say about my references to Dylan and Guthrie:

Add to FacebookAdd to DiggAdd to Del.icio.usAdd to StumbleuponAdd to RedditAdd to BlinklistAdd to TwitterAdd to TechnoratiAdd to Yahoo BuzzAdd to Newsvine

Like This!

%d bloggers like this: