The organic world – animals, plants, viruses – is the product of Darwinian evolution by natural selection. Natural selection expresses the idea that organisms (more accurately their genes) vary and that variability has consequences. Some variants are bad and go extinct; others are good and do exceptionally well. This process, repeated for two billion years, has given us the splendours of life on earth.
It has also given us the splendours of human culture. This may seem like a bold claim, but it is self-evidently true. People copy cultural artifacts – words, songs, images, ideas – all the time from other people. Copying is imperfect: there is “mutation”. Some cultural mutants do better than others: most die but some are immensely successful; they catch on; they become hits. This process, repeated for fifty thousand years, has given us all that we make, say and do; it is the process of “cultural evolution”.
However, the underlying mechanisms are poorly understood. For example, how important is human creative input compared to audience selection? Is progress smooth and continuous or step-like? We set up DarwinTunes as a test-bed for the evolution of music, the oldest and most widespread form of culture; and, thanks to your participation, we’re starting to get answers.
However, I am doubtful as for the validity of the conclusions drawn from this study, or at least from what I heard on radio this morning.
Having read the introduction, you must be thinking that I only despise a project that is not mine. Not at all! I actually find it very interesting. I am just unsure if the conclusions drawn from the study are really founded.
But let us start talking about the experiment before going further. Here, I am going to sum up and will gladly accept corrections if I dare say anything mistaken or incomplete.
The whole study is based on a selection and combination algorithm: a first series of samples is generated. They sound like chimes, for instance. Those samples are submitted for evaluation to the cohort for the study. Each sample must get note, from “Unbearable” to “I love it.”
The best noted samples are kept and randomly mixed together to create a new series of samples. Those samples are in turn noted, the best ones are kept and mixed together and submitted for evaluation, …
This algorithm is eventually a perfect example of a genetic algorithm: at each generation, only the best fit elements are reproduced to the next generation, though mutated.
After a couple of generations (well, actually a little more than “a couple”), you can see that the mean notes given to the samples stabilize to a fixed value.
The conclusions explained this morning on the radio (when I first heard of this experiment) is that composers and artists write and play their music according to audience’s reception. The music evolved the latest centuries because the audience became larger. I think this point of view is awfully reductive.
They also explained that much of the generated samples contained several chords which can be found in popular songs.
Now I skimmed through the paper the experiment authors wrote (and skimmed only, which is why I would not mind some correction from someone more familiar with the subject), the conclusion is slightly different. Yes, audience’s reception do have an impact on music, but that we know: mercantile producers for instance accept music only when they think they can sell it (unfortunately); people share the music they like, not the one making their ears bleed, …
Music also evolved because throughout time, people could not always write it down (by lack of a nomenclature, for instance) and transmitted it by ear only. It thus introduced variations (mutations). Those variations pleased each artist, since they searched for the best way to fill the gaps in what they were playing, and a track could then evolve from an artist to another.
The main idea remains the same in all cases though: music is what it is now because of Darwinist selection.
I agree with what the authors said — to some extent. This should not be thought of as a generality, however. There is more than selection to creativity.
Is the algorithm properly used?
This is the main point that makes me uneasy. Let’s imagine a child who do not yet know the world. They get out of the house after a rain and discovers the road is wet. Later, playing with the hose, they spray water on the road. Look! It is wet again!
Is it safe then to assume that it means a wet road always means that someone directed a hose towards it? Of course not. Yet, this is what we do here!
We use a genetic algorithm, designed to mimic natural evolution (meaning Darwinist selection), and see that the final result look (actually sounds) like what we know. Therefore, what we know must come from natural evolution…
Really? Am I the only one disturbed here? Taking back the hose example, we just had one possible cause. Only one, not event confirmed, just possible.
Genetic algorithms are powerful tools (I wish I was given the occasion to design one myself). Still, one must be careful of the conclusions drawn from its result.
Yet, I was also frustrated because of the explanation I heard, which focused on composers.
Is creativity selection?
This is eventually part of the question asked here. Do composers write to please their audience? Of course, some do. Some of those are more mercantile than artists.
Artists are those I am interested in. Do they submit their work, see how the audience react and then take back bits and pieces of what people liked to create a “new” piece?
I dare say there is more to creativity than that. Of course, at first, music may have been a mere imitation of pleasant sounds, which combined… Actually, yes, I think DarwinTunes perfectly well describe the beginnings of music.
But not now. Some people try to experiment with new things, things they have not heard of and yet crave for. Yes, music is born through selection: through selection of what composers love. And yes, it is doomed to be short-lived if producers do not find it to their own taste. So selection does have a role in it. But the selection is evolving too, as highlighted the authors: sharing has evolved; music can be freely distributed.
Is current music the result of the audience’s taste?
The radio speaker seemed to me to imply that classical music for instance may not have been what it was, had it been exposed to a larger audience. Can we really expect the tastes of another era to be the same as ours?
A simple example is the beauty of women. Some centuries ago, they had to be a bit plump, have large hips fit to bear children, … While in the past decades, model companies almost seemed to praise anorexia.
Selection in each era may lead to different result. The size of the selecting population is not the only criterium to take into account, the elements of this population are important too and their feeling towards the selected are key elements.
Though enraged by the explanation I heard on radio, skimming through the authors‘ paper made me a little more confident about their deductions. I must try to read it more thoroughly when I have some time to spare, and make sure I agree with them while what I heard included some shortcuts.
One key point though: the authors focused on music, not on composers. Using an algorithm as a composer and deduce about composers’ behaviour would be unwise.