Music doesn't evolve

[This article has been republished with permission from Broad Street Review.]

At Imperial College in London, Bob McCallum spends his days researching mosquitoes. At home, he puts composers under his microscope. And has concluded, according to BBC News, that we composers are unnecessary.

I don’t know how many mosquitoes have died in his experiments, but he’s killed off my species by way of a computer program he wrote called “DarwinTunes.”

That’s what he’s saying, anyway. Music evolves, he says, and DarwinTunes proves it.

Here’s how it works. The software randomly generates two noise-loops from randomly selected sounds. These combine and generate four loops, then eight, 16 and, pretty soon, 100.

I thought the doubling sequence goes from 2 to 4, then 8, 16, 32, 64 and then 128, not 100. But I’m a composer, not a mathematician.

Anyway, DarwinTunes, or McCallum, mixes in other mutations along the way, and whatever the number, there’s a whole lotta shakin’ goin’ on before he hits the “Off” button.

‘Pretty horrible’

Since, as McCallum himself concedes, the examples are “pretty horrible,” and since DarwinTunes doesn’t accomplish anything more than what it’s already done, McCallum calls in volunteers. Perhaps he filters out composers or filters in mathematicians, but whoever’s there listens and chooses what they like most (or hate least). Then McCallum throws the selected loops back into the program.

DarwinTunes goes back to work, and McCallum waits. Something interesting, he says, starts to happen “after about 3,000 generations… There starts to be a kick drum or a bass drum.”

And that’s it. I read the article several times and, yes, that’s it.

Tweaking the results

Let’s review. Bob McCallum creates a computer program. It uses instruments and pitches that McCallum has implanted in its code, in a mathematical process McCallum has written. It runs the results through signal and amplification generators McCallum has already connected to it.

Then McCallum stops the process, tweaks the result and brings in other folks to tweak it further. McCallum runs it again.

This process continues for months, until McCallum is attracted to something, oh, perhaps, drum-like. Then he halts the process.

Well, he’s demonstrated something, all right— but it’s not evolution. To me it looks an awful lot like intelligent design, or maybe The Bob of the Gaps or McCallum ex machina.

A million monkeys on a million typewriters could never, ever type “To be, or not to be, that is the question” because, first of all, even humans rarely put the paper in correctly, and second of all, even if a monkey happened to get it right, it would never know it.

Who tells the monkey to stop? Who tells DarwinTunes to stop? Who changes the ribbons?

Who decides?

Every step in the DarwinTunes process is a step that a composer already takes, incalculably faster and more efficiently. All of McCallum’s coding, generating, processing and volunteering are just fig leaves covering up this truth: In music, somebody must make a decision.

The smallest decision is the original idea. Ideas are a dime a dozen. The bigger decision is what to do with the idea, and the biggest decision is when to stop— none of which DarwinTunes can do.

Going McCallum one better, composers even do a better job of getting rid of composers. We’ve been doing it ever since we invented the repeat sign and “Vamp Till Fade.”

McCallum may not have heard, but John Cage eliminated the composer about 60 years ago. In his pieces he sets wheels in motion— be they musicians, microphones, radios, silence— and leaves them to play out for the audience, if there be any (and if there isn’t, that’s just fine with Cage). McCallum could’ve put on a Cage concert and saved himself loads of volunteer-hours.

Would it be Mozart?

Nevertheless, Armand Leroi— professor of Evolutionary Developmental Biology, also at Imperial College— is jazzed.

“I’ve no doubt that if we ran this experiment for longer, using bigger, faster computers, and millions of people rather than thousands, and for years, instead of months, we could evolve fantastic music,” says Leroi. “Would it be Mozart? No, I don’t think so. It would have no composer behind it; it wouldn’t be the act of any individual musical genius. It would just be the people’s music in its purest form.”

Did he say Mozart? You couldn’t evolve the “In-A-Gadda-Da-Vida” drum solo (bup-bup-bup-bup, bup-bup-bup, budilleeyup-bup-bup-bup, bup-bup-bup) this way. Please, professor, never again speak of Mozart.

There is no People’s Music. I have no idea what “music in its purest form” means— and neither, I suspect, does Dr. Leroi.

All music— even folk music— is created by a person, never by People (unless it’s a small group of people playing together, and somebody else is buying). These persons make decisions. If the music’s good, somebody else will repeat it.

Is music improving?

Music changes, of course. But it doesn’t evolve.

Music isn’t better now than the music of centuries ago, however much it has changed (and it hasn’t changed much). There is as much music in Machaut as in Mahler.

Today’s music isn’t even more complicated. Try singing Ockeghem without having your life flash before your eyes. Follow, if you can, a Senegalese griot plucking a harp made out of a gourd, or beat two sticks to a tune from the mountains of Transylvania without getting lost. Do that, and suddenly “modern” music doesn’t look so difficult.

On behalf of composers everywhere, let me say this to Dr. McCallum: When you evolve a mosquito into a fruit fly… no, better yet, when you write a program that evolves a paper on evolutionary developmental biology, that will attract our attention. Until then, we’re safe.

Dr. McCallum’s premise was all wrong, anyway. Music never comes from noise. It comes from a human soul, singing.

4 thoughts on “Music doesn't evolve

  1. Nate

    Something to bear in mind is that biological evolution doesn’t make organisms “better” either. It drives them to a state in which they are better able to survive and reproduce within their environments. In this more accurately Darwinian sense, music absolutely evolves.

    McCallum does sound awesomely terrible….

    Reply
    1. Kile Smith Post author

      Hi Nate! “Survival of the fittest” always seemed tautological to me, as we define the fit as those who survive, and we define surviving as being fit. The problem with that in biology and music is that we look around us, and figure that what we see now must be the best (or better than what came before), because after all, it’s here now, and other things aren’t. Leaving biology aside, I reject that in music, and the more I learn about music, the more I reject it. Music changes, but for every “advance” in one direction (harmonic complexity, e.g.), we “lose” something else (can’t dance to it, e.g.!).

      I certainly agree that composers learn from past composers, and music changes. But it doesn’t get better (even in a survivability context), and in any case change comes from individual decision-making, not from any “natural” processes within the music. Thanks for your note!

  2. Darwin Tunes (@darwintunes)

    There was only one tweak in DarwinTunes – the recombination mechanism was changed at around 3000 generations from a random approach to a more biologically-inspired approach (swapping like with like). There has never been a stop-listen-tweak cycle – that’s not how we scientists roll. DarwinTunes is now at over 7000 generations and still evolving – so your point about choosing where to stop doesn’t make sense to me.

    Of course you can liken DarwinTunes to intelligent design if it has only one listener/rater – I’d say that was more like breeding (as in dogs, horses, cereal crops etc), but we’re splitting hairs. With a larger internet audience all providing independent feedback/ratings it’s much more like natural selection – no one person can have a “grand plan” and expect it to get followed through to completion.

    Also, pitches (and tempo and time signature) are indeed hard-wired in the software but “instruments” in the sense of trumpet/piano/marimba or even synth patch X, Y and Z are definitely not. The additive sine wave synthesis is completely under evolutionary control.

    We’ve collectively made music without a composer, but as you say, composers are much more efficient at making music and I don’t think that’s going to change very soon.

    Reply
    1. Kile Smith Post author

      Thanks for the clarification, which reveals the mechanism of DarwinTunes to be somewhat different from what the BBC News article made it out to be. But it’s still design, is it not, for thousands of people to do the work one composer does? (Knowing how I sometimes stumble over musical expression myself, I call the design “intelligent” wryly.)

      But–even for the brief loops–calling it music-making by audience, rather than composing by composer, doesn’t change its nature. The Yellow River Piano Concerto was composed under Mao by committee. There were (I think) five composers working on it, and while I’m sure they figured out some division of labor, it was still composing. An improvising band does the same thing, composing under pre-set conditions.

      It is the making of choices that is art. DarwinTunes offers up ideas (somewhat automated along a biological model), but can never eliminate the choosing. DarwinTunes hasn’t “made music without a composer.” A committee–a large, growing committee operating under pre-set conditions–takes the tiniest bits of the composing process, and makes choices. And if choice is necessary, then not only isn’t the composer going anywhere anytime soon, the composer is necessary.

Leave a Reply