Re: algorithmic music

Subject: Re: algorithmic music
From: Michael Gogins (
Date: Sat Oct 02 2004 - 09:28:37 EDT

Thanks for your response.

As you probably know, this Faustian quest to rationalize everything ran
square into Goedel's theorem, which rationally proves that you can't
rationalize everything or, to be quite precise and technical about it,
proves that there exist true theorems that can't be proved.

Then as you note, chaos theory showed how strictly deterministic dynamical
systems can produce literally unpredictable sequences.

These two facts are related. Gregory Chaitin has some very interesting
theorems about the nature of "mathematical randomness" which illuminate

For algorithmic composition, this suggests that the quest to compose by
rationalization is neither quixotic nor irrational, but can take advantage
of mathematical chaos to guarantee the production of surprising results.

A monist would, indeed, assume that "intuition" and "creativity" are not
supernatural but, rather, must be grounded in these phenomena of chaos and
computational irreducibility.

It seems clear to me that "reason" and "logic" are not diametrically opposed
to "intuition" and "creativity" but that both are different experential
facets of one process of thinking.

For algorithmic composers, I believe the critical problem has moved beyond
production and is now evaluation, and this seems to be what you are saying
also. We can generate anything, but almost all of it is noise or, at best,
wallpaper music. We need some means of selection that throws out the noise
and keeps the best music, and in a manner that reflects our artistic

However, I don't agree that AI has anything to offer in the direct sense (I
am sure it does in other senses). For, I don't think reason can rationalize
itself -- precisely because of incompleteness. But perhaps this is what you
are saying, also, in "I want musical AI to be a reality because I have no
idea what will come of it."

What I am pursing is methods that speed up and focus the selective
process -- parametric composition, evolutionary composition, and so on. I
am also interested in building music-theoretical constraints into
algorithmic composing systems, not to impose a specific style, but to avoid
what commonly are felt to be 'bad moves' in most styles.

----- Original Message -----
From: "Eliot Handelman" <>
To: <>
Sent: Saturday, October 02, 2004 3:14 AM
Subject: Re: algorithmic music

> wrote:
> >Algorithmic composition flowers from deep roots in the long-standing,
> >Faustian dream of Western civilization to rationalize nature and
> >everything.
> >
> Ok, but there's an equally long-standing engagement with chaos in western
> culture going back to the greeks and continually present. for xenakis
> chaos undoubtedly
> has the mysterious role of preceding and so enabling creation --
> conisder gendyn here. As far
> as I know, very few people attempt algorithmic composing as a way of
> rationalizing
> composing -- it's almost always the opposite, because one hopes for
> unexpected
> results. The problem of rationaliztion is whether you can create pieces
> on the order of
> Beethoven or Strauss, and I don't mean by recombining sampled music.
> There's probably
> only a few theorists who try this -- consider "melisma," a program by
> temperley. the results
> are wooden.
> My feeling is that rationaliztion could create chaos, which is why I'm
> for it. It's clear that
> computer music will progress only through a ton of analysis, of sound,
> psychoacoustics, perception,
> of music itself, and ultimately the mind. I want musical AI to be a
> reality because I have no
> idea what will come of it.
> -- eliot

This archive was generated by hypermail 2b27 : Sat Dec 22 2007 - 01:46:04 EST