Eric Lyon - Ricerca sull'intelligenza artificiale specializzata al MIT

Eric Lyon
Bregman Electronic Music Studio
Department of Music
Dartmouth College
Hanover, NH 03755 USA
eric.lyon@dartmouth.edu

Dartmouth Symposium
on the Future
of Computer
Music Software:
A Panel Discussion

[Editor’s note: This article is an edited transcript
of a panel discussion, moderated by Eric Lyon, at
Dartmouth College in October 2001. See ‘‘About
This Issue’’ on page 1 for more information.]

Eric Lyon: I think it’s fair to say that without the
work of today’s speakers, computer music as we
know it would not exist. Now that we have them
all together, we have an unprecedented opportunity
both to focus on their individual achievements
and to assess the bigger picture of where computer
music software is today and where it might be
heading.

At the end of a 1989 comprehensive survey of
computer music languages and systems, Gareth
Loy wrote the following: ‘‘It is remarkable how few
of the languages and systems documented above
are generally available.’’ Among the reasons postu-
lated for software’s demise were the obsolescence
of the host computer system, competition from
other languages, and discontinuation of support by
the programs’ creators. This symposium takes
Gareth’s observations as a point of departure. Ma
we turn the question around and ask not why so
much software has disappeared, but why a few ex-
ceptions have been so persistent and have devel-
oped a loyal and even fanatical following among
computer musicians.

In the world of recorded music, which comprises

most of what we listen to these days, the term
‘‘computer music’’ is redundant. The prevalence of
the use of computers in today’s music demands an-
other distinction; at its outset computer music
meant experimental music, carried out in laborato-
ries and universities. This experimental work con-
tinues here and at many other institutions, Ma
most of today’s computer music is created in the

Computer Music Journal, 26:4, pag. 13–30, Inverno 2002
(cid:1) 2002 Istituto di Tecnologia del Massachussetts.

ﬁeld of entertainment, whether ﬁlm music or the
various technology-drenched genres of rap, techno,
rock, and pop.

The distinction between experimental music and
what might be termed ‘‘normative music’’—that is,
music based on accepted stylistic norms—is mir-
rored in our software. On the normative side of
software are utility programs such as mixers, se-
quencers, and reverberators. On the experimental
side are the programs that we discuss today. Questo
software is open, extensible, and invariably used in
ways unanticipated by its creators. While such soft-
ware does not command a market on the scale of
normative utility programs, it is arguably much
more inﬂuential in the long run, as it facilitates the
creation of the music which today exists only in
our collective imagination. And the experiments of
today will lead inevitably to the norms of tomor-
row. This debt is even occasionally acknowledged,
such as when the Beatles put Stockhausen on the
cover of their album Sgt. Pepper’s Lonely Hearts
Club Band, or more recently when Radiohead sam-
pled (and credited) Paul Lansky on their album Kid
A which went platinum.

My hope is that this will be a little less like a
panel discussion and more like a town meeting,
given the people in the audience. I am going to
start by asking one or two questions, then open it
up to questions from the audience.

I would like to start by discussing various aspects

of different kinds of software, different speciﬁc
pieces that we know and love. Perhaps this is more
or less about survival paradigms, which is to say
that speciﬁc pieces of software and ideas have been
ﬁltered through—have made it through—the 40-
odd years of computer music. In particular, we see
the ‘‘Music N’’ paradigm represented by the Music
N environments and of course Csound, and we
see the graphic layout and connection of signal of

Lyon

D
o
w
N
o
UN
D
e
D

F
R
o
M
H

T
T

:
/
/

D
io
R
e
C
T
.

io
T
.

e
D
tu
/
C
o
M

j
/

UN
R
T
io
C
e
–
P
D

F
/

2
6
4
1
3
1
8
5
3
6
1
1
0
1
4
8
9
2
6
0
2
3
2
0
9
9
1
3
4
7
P
D

B
sì
G
tu
e
S
T

o
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

Figura 1. Max Mathews
and Jon Appleton.

nent can be worked on separately so that you don’t
have to keep ‘‘glomming’’ on things to one environ-
ment to make them do what you want.
David Zicarelli: I don’t know if this is related to
what you are asking, but I am currently working on
a project to make Csound work inside MSP. It
won’t be ideal, because of the way each program is
written, but at least you have a beginning, and it
certainly should be possible to do that with other
systems. I see two directions for Max: one is that,
because the API is relatively simple and well-
documented, it should be possible to incorporate
other software into it as an object, such as Csound
or even the SuperCollider language, and the other
thing is that Max/MSP is eventually going to be a
library that can be used for the purpose of custom-
izing other software. The ﬁrst example of this al-
ready exists, called Pluggo, which is a runtime
environment for VST plug-ins that can work inside
a sequencer so you develop software in Max and
then put it in a particular location on your com-
puter and open it as a plug-in to process audio in an
audio sequencing environment. The same kind of
thing could be true of a Web browser or video ef-
fects program, movie editor, or something like that.
So, I am not sure that’s cannibalism, but at least it
is a way that things can work together.

Unfortunately, as any programmer knows, the in-
terface between the two worlds is always where all
the action is, and it’s usually where all the bugs
are, so the impulse of the programmer to do this
because it seems easy is usually rewarded many
times over with incredibly difﬁcult implementa-
tion problems.
Lyon: Barry, I would be curious to hear your reac-
zione, since in a sense, without even attempting to
fare così, Csound by absorbing the music and the
world around it has necessarily taken in so many
other things.
Barry Vercoe: Yes, well I guess the secret is to keep
remaking yourself. And it is not necessarily true
that you have to, in an industry analogy, gobble up
all the dot-coms in order to survive; you can actu-
ally do your own thinking and grow along particu-
lar lines that are probably consistent. For

Max/MSP, and ﬁnally we see the pattern-based and
syntactical sort of object-oriented idea of Super-
Collider.

To start our discussion, I would like to ask how
you feel about the idea of ‘‘software cannibalism’’
and how that may suggest a direction for the fu-
ture. In other words, now that we have software
models that are very clear and palpable, it also
seems that we are opening them up and trying to
grab ideas from other sources and absorb them into
our languages.
Max Mathews: The more the better, because that’s
progress as well. I think the model for that is the
mathematician, which I believe Miller [Puckette]
È, and mathematicians do not go back to axioms to
prove things if possible. They ﬁnd somebody else’s
theorem and just continue from there. Ovviamente
that imposes rather strong requirements on the
theorems. They really have to be true under all
sorts of conditions to not introduce errors into
other theorems that depend on them. So if canni-
balism is going to be successful, then our algo-
rithms have to be very robust with regard to the
environments to which we are going to subject
them.
James McCartney: Another idea that is the oppo-
site of cannibalism and absorption is deconstruc-
zione, like the CARL system; and the way I am
doing SuperCollider now is to divide things up into
their simplest components. That way, each compo-

Computer Music Journal

D
o
w
N
o
UN
D
e
D

F
R
o
M
H

T
T

:
/
/

D
io
R
e
C
T
.

io
T
.

e
D
tu
/
C
o
M

j
/

UN
R
T
io
C
e
–
P
D

F
/

2
6
4
1
3
1
8
5
3
6
1
1
0
1
4
8
9
2
6
0
2
3
2
0
9
9
1
3
4
7
P
D

B
sì
G
tu
e
S
T

o
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

instance, we might say that SAOL is an outgrowth
of NetSound, when in point of fact, there are a lot
of deﬁciencies in the SAOL structure that I wish
weren’t there. We really did know better than that,
but there were reasons why I needed to let a stu-
dent go ahead and do something on his own. So it’s
ended up that the SAOL structure is much less efﬁ-
cient than Csound, certainly than XTCsound. IO
think that if I had it to do over again, I would sim-
ply take the XTCsound card away from hand-held
devices and put it into MPEG. It would have been
twice as fast, and what I see happening right now is
that the sound community is going back and forth
trying to get to the bottom of things that have al-
ready been solved many times over. When you
adopt a particular line and think through all these
things carefully, then you don’t necessarily have to
get involved in cannibalism.
Lyon: Gareth?
Gareth Loy: There is something that I don’t like
about the notion of cannibalism, which is that it
excludes the possibility of growth—except, I sup-
pose, it nourishes the organism that consumes the
other. I would prefer to couch it in terms of genetic
evolution. It seems to me what’s important is what
goes on in our minds, and our software is an epi-
phenomenon really, so in a sense, we are all like a
big genetic organism thinking about these things.
There are certain strains of elements of our musical
beings that seem right because they help us solve
the problems that we need to solve, ad esempio, for in-
stance, the unit generator. There are many other
variations that come along, from other domains
perhaps, of learning that we add to the mix to solve
particular aesthetic issues that we have. So it
doesn’t have this quality of digesting what went be-
fore us so much as being a part of an organism that
we elaborate and evolve depending on the environ-
ment in which we ﬁnd ourselves.
Vercoe: I might add one more thing. It’s nice when
one is meandering around and it’s nice at times to
close the circle, complete a circle, close the loop, O
something, just what we heard a few minutes ago.
It is interesting because it’s my belief that Max
grew out of the experiments that Miller and I were

running in Paris. I am not sure if that’s true, Ma
that’s where the genesis seemed to be.
Miller Puckette: I have yet to sort it out. It borrows
from that, but also from the synthesizer idea and
also from RTSKED and who knows—it’s too com-
plicated to ﬁgure it out.
Vercoe: Yes, but the fact that David might now
consider ‘‘burying’’ Csound inside of Max things is
nicely self-symmetric, I suppose. So it may be self-
cannibalism or something; there is a nice com-
pleteness about things like that happening within
the community. I think these things are driven by
the user-created process, by what composers want,
what composers need. The most expansive and cre-
ative innovations in Csound or in Music11 really
came from the workshops that we were running at
MIT for many years. That’s when the need for new
things would arise, and that’s when the new things
would happen. That becomes a wealth of experi-
ence that is folded into the language that’s cer-
tainly not just me, it’s just dozens and dozens of
other people’s ideas of what they would want me
to do, and you don’t necessarily want to try to re-
work all of that from scratch.
Lyon: You have all developed systems over long pe-
riods of time, and they’ve all proved themselves. IO
would like to ask if you have any fantasy of what it
might be like to throw away everything that you
have done and just come at the question from a
completely different angle. If you were to do that—
if suddenly all the code that you had written acci-
dentally was erased from every hard drive—do you
have any concept of what might be a completely
different way to approach the problems that we
deal with?
McCartney: Well, I think most of us have already
rewritten our software too many times.
Loy: Personally, I think I would start playing my
guitar more . . .
Puckette: I don’t think it is a problem of throwing
your code away, and indeed any good programmer
throws away code. Any good composer throws
away paper music as well. And what you don’t
throw away is your memory. You can’t. E, In
fact if I threw out Pd and started over again at this

Lyon

D
o
w
N
o
UN
D
e
D

F
R
o
M
H

T
T

:
/
/

D
io
R
e
C
T
.

io
T
.

e
D
tu
/
C
o
M

j
/

UN
R
T
io
C
e
–
P
D

F
/

2
6
4
1
3
1
8
5
3
6
1
1
0
1
4
8
9
2
6
0
2
3
2
0
9
9
1
3
4
7
P
D

B
sì
G
tu
e
S
T

o
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

Figura 2. James McCart-
ney, Max Mathews, E
Gareth Loy.

D
o
w
N
o
UN
D
e
D

F
R
o
M
H

T
T

:
/
/

D
io
R
e
C
T
.

io
T
.

e
D
tu
/
C
o
M

j
/

UN
R
T
io
C
e
–
P
D

F
/

2
6
4
1
3
1
8
5
3
6
1
1
0
1
4
8
9
2
6
0
2
3
2
0
9
9
1
3
4
7
P
D

B
sì
G
tu
e
S
T

o
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

point, I would like to think that I would probably
end up with something rather close to Pd again,
closer to Pd than Pd was even to Max. Di conseguenza,
in some sense I am incapable now of going back in
that same particular ﬁeld and coming up with
something truly revolutionary, in the sense of be-
ing revolutionarily different from Max. I would
simply have to change ﬁelds, which would be fun
but not something that I want to do right now.
Zicarelli: I have this idea for a program that I have
never implemented, which I have called BVM, for
Bad Version of Max. The idea is to exploit this ten-
sion that I talked about earlier where Max is—as
Miller said earlier—the white rectangle on a piece
of paper. I wanted to write a program that was al-
ways doing something and then you used its inter-
face to modify its default behavior and something
else until you got what you wanted. It would actu-
ally start out by doing something that you didn’t
like, rather than Max, which starts out by not do-
ing anything.
Mathews: I certainly agree with Miller that throw-
ing away the code wouldn’t clear my memory, E,
although I might bitch about having to rewrite it, IO
also have the feeling that at least my code over
time gets dirty with various patches and things and
probably bugs that I haven’t found, and that it’s a

good idea to throw away the code and start com-
pletely over and rewrite something that is cleaner.
And if I have the guts to do it myself, I will proba-
bly even do it myself before the cataclysm in the
world that destroys the whole code. I suppose
that’s not really the basic answer to what your
question is, and a program like Csound is probably
a very effective program because it keeps absorbing
objects, and the number of objects gets bigger and
bigger. Ma, having extra objects around probably
doesn’t interfere with the subset of objects that you
are using, at least not very much. So having a prop-
erly modularized program seems like one of the
very important things . . . but I wish Microsoft
would throw away Windows and start over.
Lyon: The last question that I am going to ask be-
fore opening it up is actually a meta-question. IO
would like invite each of you to ask each other
questions about what you would like to see in the
future.
Loy: I would like to ask how we balance the tech-
nical requirements of formal structures in the com-
puter. Because computers operate only on formal
systems, how do we balance that with the informal
aspects of music composition? Suppose you start
with a little motive do re mi do. Clearly, you have
the idea of sequence going on here but there is not

Computer Music Journal

much harmony there yet, and if I reinforce the idea
of sequence by going re mi fa re, now you have a
sense of expectation. I’ve got your attention, and I
believe that music is about the manipulation of
peoples’ attention. So, if I went mi fa sol mi, IO
would lose your attention because you are now
able to predict where I am going. Invece, say I go
mi fa sol sol do. What happens here is at the two
and a half times repetition through this formal
structure—it’s not even an integer multiple of a
formal structure—I break it off with cadence. Questo
brings the vertical harmonic structure to the fore-
ground, whereas over here, in the ﬁrst example, IO
was focused on horizontal sequencing as the princi-
ple thing that I wanted you to digest. I wanted you
to get that I was doing a sequence only until I
could catch you off guard, just before you would
start to tune me out, by shifting the focus to an-
other representational system that would allow me
to pull your attention back. And I believe that this
is endemic to all music, that even if you look at
the most formally crisp music of Bach or Mozart,
still what’s going on is a series of cataclysmic
breaks in formal representation of the underlying
compositional idea. And how can we get there from
where we are?

For instance, the systems that I developed and
which I saw demonstrated this afternoon are very
good at bringing up patches that implement some
idea—that implement some formal system, maybe
even with random input, so that it appears that
there is more information in the formal system
than there actually is. Ma, how can we allow these
things to shift levels musically in a meaningful
way so that we can bring this kind of surprise to
the process?
Mathews: Do you want your surprises at the level
of the composing process and possibly algorithms
for composing or helping composing, or do you
want your surprises at the timbral level, or do you
want them everywhere?
Loy: I just thought of this as an example to facili-
tate the question, and really I would expect the an-
swer to occur at all levels. I could imagine also in
an improvisational setting a performer might want

to use a system to accompany a structure that they
were creating and maybe suggest ideas to them
through an accompaniment. It could arise in a
number of different ways. How can we get a handle
on the shifting focus that often happens in music
in order to manipulate attention?
Audience Member: I don’t think you should wait
for the machine to give the answer to that ques-
zione. It should be the composer who makes those
decisions.
Loy: What if I were attempting to model the com-
positional process the way Lejaren Hiller did? His
experiments with the Illiac Suite were not so much
compositions as experiments, in the sense that he
was trying to model the compositional process ex-
ternally. And that is essentially the kind of ques-
tion I am getting at here. I come up with examples
like this all the time. The question is, how can I
get the computer to represent them so that I could
study them as a model for the process?
Audience Member: It seems to me as a composer
Quello, as we understand these models better (a` la
David Cope’s musical style generators and the
beautifully detailed examples in James’s pro-
grams)—even as we understand Bach or Mozart—
we keep changing the rules. So this problem
interests me from the standpoint of a teacher and
from a historical standpoint. But as a composer I
am very grateful that when I solve a problem like
Questo, I am forced to create a new one for myself so
that my music evolves or our music evolves and
i cambiamenti. Band-in-a-Box, Reason, Acid—these are
great new techno tools that do a lot of stuff—even
Aphex Twin’s use of SuperCollider—and generate a
lot of interesting new contemporary pop. They are
being absorbed by a popular audience and under-
stood as a kind of vocabulary. Ma, I think the great
thing about our music and our musical output is
that when we understand it from a formal stand-
point, that we break the mold, and we continue to
do that.
Audience Member: I think the artistic and aes-
thetic failure of David Cope’s experiments in artiﬁ-
cial intelligence simply underscore the question
that Gareth Loy has posed: the issue still remains.

Lyon

D
o
w
N
o
UN
D
e
D

F
R
o
M
H

T
T

:
/
/

D
io
R
e
C
T
.

io
T
.

e
D
tu
/
C
o
M

j
/

UN
R
T
io
C
e
–
P
D

F
/

2
6
4
1
3
1
8
5
3
6
1
1
0
1
4
8
9
2
6
0
2
3
2
0
9
9
1
3
4
7
P
D

B
sì
G
tu
e
S
T

o
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

More useful to me would be, because we are sitting
in the presence of a distinguished panel of people
who have all created very valuable tools for music,
for us to learn which speciﬁc musical compositions
that used your software do you feel most perfectly
implemented the ideas contained in that software.
It is not an aesthetic question; I am not interested
in whether everybody thinks they are beautiful,
Ma, more what you felt best used the techniques
that you built into the software. And I also ask that
you excuse yourselves from naming your own
works, Gareth and Barry.
Vercoe: Good question. I can answer that pretty
clearly. For me, the two examples of good use of
Music11 and Csound are In Winter Shine by Jim
Dashow—he wrote that in Music11—and the other
is called Sphaera by William Albright, who is now
deceased. Both of these composers worked at the
studio at MIT, and I think what they did was to
beneﬁt from being around the studio where there
were lots of ideas ﬂoating around with the stu-
dents, interacting with the students. Bill Albright
had lots of interaction with Miller, for instance.
The creative process of the composers builds from
the environment in which they ﬁnd themselves.
That’s part of the process I was referring to before
when I say software needs to grow from something
E, I think, that all the software that I have done
has grown from either my musical ideas or other
people’s musical ideas and needs. So there’s really a
symbiosis that happens between composers and
software developers. When that is working right,
you will get both the best compositions and the
best software. And I think that is why I would pick
those two pieces as examples of good use of my
stuff.
Loy: For the CARL system, I think my favorite
compositions were the ones that the guy sitting at
the end of the table did. Eric Lyon and Chris Pen-
rose were two students who did really excellent
work with the CARL system. It seemed like they
really got it and were able to use it in ways that no-
body else seemed quite able to match.
Lyon: Thank you very much—what a compliment!
Mathews: Jon [Appleton], would you remind me of

the name of the famous composer that I met at
your house—Einstein on the Beach—minimalist?
Philip Glass? Yes. Now, I myself ﬁnd Mozart de-
lightfully surprising at almost all times. I ﬁnd the
Beatles somewhat surprising and rather pleasant
looking back at them, 40 years later. (I didn’t like
them so much at the beginning.) I ﬁnd Glass’s mu-
sic very unsurprising, and I don’t think that he in-
tends it to be surprising, or at least if it is
surprising, the surprises come very slowly.
Audience Member: Well, there could be a problem
with too big a surprise, right? So, it has to be just
the right kind of surprise, not too big and not too
little. Has anybody done any music analysis of the
‘‘index of surprise’’ in music? Analysts are doing
everything else; they must have done that, pure.
Loy: David Huron, perhaps? He has put a lot of
thought into that, as a matter of fact. He has a the-
ory that the element of surprise is a combination of
linear sequential and leap-step types of patterns,
and he has fairly convincing proof that the thing
that results is the fractal behavior of music over
varying expanses of time resolution.
Audience Member: For every analysis, there can be
synthesis; has he turned that around and tried to
synthesize it back, and is it indeed surprising to
people? Does it keep your attention?
Loy: There is also the work of Voss and Clark, who
in their early work on fractal music did three ex-
amples: one musical example where they chose the
pitches based on a white-noise distribution, an-
other on a Brownian distribution, and the third on
a fractal distribution. People tended to prefer the
fractal one because, like you were saying, it man-
ages to balance the element of surprise against con-
tinuity.
Audience Member: I think David Rosenboom did
something with surprise back in the 1970s or
1980S. He was trying to make a sequencer that
maximized surprise.
Puckette: It’s interesting that surprise is not the
same thing as information. So the sequence of
notes that would carry the most surprise in
information-theoretic terms would be a memory-
less random process. And that is not actually what

Computer Music Journal

D
o
w
N
o
UN
D
e
D

F
R
o
M
H

T
T

:
/
/

D
io
R
e
C
T
.

io
T
.

e
D
tu
/
C
o
M

j
/

UN
R
T
io
C
e
–
P
D

F
/

2
6
4
1
3
1
8
5
3
6
1
1
0
1
4
8
9
2
6
0
2
3
2
0
9
9
1
3
4
7
P
D

B
sì
G
tu
e
S
T

o
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

would be the most surprising, because after you’d
heard ten or twenty notes of it, you would know
exactly what the rest of it would sound like. There
is a funny paradox there that people like Xenakis
had a way of getting around, but I don’t know how
to even describe how they did it.
Lyon: I think another point with respect to surprise
is that there are really two different kinds that we
are familiar with in music. One, Ovviamente, is what
we consider a surprising event in a piece of music,
but the second time we hear it we know it is com-
ing, so it’s not really a surprise anymore. And the
other thing would be to make some kind of musi-
cal machine that would always be generating some-
thing new, which we would be able to interpret as
interesting enough to keep our attention. It seems
to me that the ﬁrst case imposes a more stringent
aesthetic requirement that the element of surprise,
even after it is gone, still has some kind of impact
on us, as being a deviation from a norm that we
have absorbed. A good example, for those of you
who know the last movement of Beethoven’s
Eighth Symphony where just before you have the
big tutti, you get that unexpected sharp
doesn’t make any sense at all until much later in
the movement. Every time you hear it, you know it
is coming, but it hits you, because it’s such an un-
usual event.
Loy: Surprise is relative. I remember that when I
took a course in four part Bach-style counterpoint,
as I got my ear more attuned to that style, I began
to discover the surprising things that Bach put into
these chorales which at the ﬁrst sounded just ba-
nal.
Audience Member: I am a cognitive psychologist,
and it seems to me that musical surprise occurs in
the most satisfying way when we acquire some sort
of schema based on the tonality or the sounds of
our culture, whatever that culture is. And this
schema leads us to expect certain rhythmic pat-
terns and certain tonal and melodic patterns, E
they have to be violated in very particular ways. If
you hear something that violates it too much all at
once as you’ve been saying, it is not satisfying. If
the violations come in just the right amount and at

Quello

ˆ5

Figura 3. David Zicarelli.

just the right time, we ﬁnd that very pleasing. E
with respect to the point that you made about
maintaining surprise upon repeated listenings,
when we watch Charlie Chaplin and we see him
walking across the room and there is a banana peel
on the ﬂoor, even if you have seen that ten times
before, it’s still funny to watch him slip on the ba-
nana peel. There is something about this schema
that you are not supposed to fall down if you’re a
distinguished person or you’re not supposed to
twist this note in this way if you are a composer.
That violation, even though you know it is coming,
is very satisfying, and I think that’s what the great
composers dance on the edge of: somehow ﬁnding
a way to tweak your expectation so that it’s just fa-
miliar enough and just different enough.
Puckette: It is also important that it not just be any
sorpresa, but that it be a surprise that reveals some-
thing. When you hear a great moment in music, Esso
is because there’s a feeling of your eyes opening.
That perhaps is why random sequencing, although
surprising, is not musical in that it doesn’t reveal
as it surprises. Now what that means I can’t get
anywhere near to saying. What this has to do with
computer music software is quite unclear, except
perhaps that the software designer better stay as far
away from it as possible!

Lyon

D
o
w
N
o
UN
D
e
D

F
R
o
M
H

T
T

:
/
/

D
io
R
e
C
T
.

io
T
.

e
D
tu
/
C
o
M

j
/

UN
R
T
io
C
e
–
P
D

F
/

2
6
4
1
3
1
8
5
3
6
1
1
0
1
4
8
9
2
6
0
2
3
2
0
9
9
1
3
4
7
P
D

B
sì
G
tu
e
S
T

o
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

Loy: Let me come back to the question then, be-
cause I’m afraid the question was not well posed. IO
ﬁnd myself, when writing composition languages
and synthesis systems, in the predicament of hav-
ing to stand the system on its ear to introduce the
kinds of metamorphoses that would be musically
signiﬁcant. I am wondering if the rest of you have
encountered these kinds of dilemmas and what you
have done in your software designs to take account
of these kinds of things. Is that a fair question?
Zicarelli: It is a fair question. My response would
be that I try to avoid those problems—to stay away
from those problems as much as possible, and this
isn’t exactly related to what you said. One of the
things that I have always liked about Max is that it
has no musical concepts in it at all. That has also
been one of its more controversial aspects. There
was a discussion—I think it was in the ICMA jour-
nal; ‘‘The Mins of Max’’—and the criticism was
that this program has nothing to do with music,
and there is no way to represent musical structures
in it. I think that introducing concepts that explic-
itly had to do with musical structures into the fun-
damental core of the environment would be a
disaster. That’s not to say that those things
couldn’t be added on by people who are motivated
to do it, but it is personally something that I want
to stay as far away from as possible.
Mathews: Gareth, I think that you need to add an
adjective to this word ‘‘surprise’’ that you have
brought up, and the word that I might throw in
would be ‘‘enduring.’’ Now, music is to me a differ-
ent animal than most anything else, like a book. IO
can play a Mozart sonata a thousand times until I
really know exactly what is coming next, but I am
still pleasantly surprised by it. I can enjoy playing
it for the thousandth time far more than when I
played it for the ﬁrst time because, more or less,
my ﬁngers go automatically. Tuttavia, I can’t re-
ally read a novel a thousand times; I can read it
two or three times, but that’s it. E così, this sur-
prise function must be built into a different cortical
level than memory function, let’s say.
Audience Member: The critique of Max that David
mentions is a crucial point that really applies to all

these languages. Fundamentally, they are not com-
position languages: they are instrument design lan-
guages. They’re languages that give you a way to
design different kinds of instruments and play
them in various different ways. They do not have
anything to do with music at the fundamental
level, just in the same way that putting together
some strings and bits of wood and various things
has nothing intrinsically to do with music. If you
are making a violin, that is not the music, the mu-
sic is what you do with it afterwards. And so these
kinds of discussions really go on a layer above what
all of these languages are actually doing.
Vercoe: Let me just give you an example of an in-
novation that happened in Music11 when I was
writing Synapse. In Music360, which came from
Music4B, there was a thing called envlp—an enve-
lope in which you could describe how long it was
and what kind of decay there was at the end and
how long the decay would last. I found, when I was
writing and working closely with the violist Mar-
cus Thompson, that he had a way of articulating a
phrase that my system didn’t have. I had conceived
of envelopes the way a pianist would conceive an
envelope: you play a note and it has a certain rate
of decay and it may be pitch-dependent or what-
ever. A pianist would articulate by releasing the
notes early so you get the articulation of a phrase
by early releases. That was built into the structure
of Music360 and the early part of Music11. Che cosa
my violist was doing was articulating little patterns
with something else; he has a way of being able—
as near as I could pinpoint it, over a certain phrase
that he was wanting to articulate—to drop each
note down to, let’s say, 20 percent of the attack
amplitude of the note, no matter how long the note
era. He had a consistency of coherence about this
that is just very different from the way a pianist
would approach the same thing. I found that what I
had to do was to invent another kind of envelope to
be able to match what he was doing with the com-
puter part. And that gave rise to a thing called
envelopex, which has an exponential decay that
you could specify as part of the input argument.

Now, that was a discovery I made working with
a musician playing an instrument who had control

Computer Music Journal

D
o
w
N
o
UN
D
e
D

F
R
o
M
H

T
T

:
/
/

D
io
R
e
C
T
.

io
T
.

e
D
tu
/
C
o
M

j
/

UN
R
T
io
C
e
–
P
D

F
/

2
6
4
1
3
1
8
5
3
6
1
1
0
1
4
8
9
2
6
0
2
3
2
0
9
9
1
3
4
7
P
D

B
sì
G
tu
e
S
T

o
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

over things like that. This is the kind of evolution
of ideas that comes from the act of creation, perfor-
mance, rehearsals, and trying to bridge the gap be-
tween performance traditions which have so many
centuries of development—just as instruments, IO
suppose, have developed through a process of ex-
perimentation and natural selection. And until
computer music can go through that same process
of natural selection and evolution, then tape and
instruments are going to be quite some distance
apart. So, I have had a love for combining these
things, not because I just didn’t like computer mu-
sic by itself, but because it was a challenge to actu-
ally try and match that which has been going on in
the instrumental performance traditions all this
time. [Applause.]
Audience Member: I have a two-part question. IL
ﬁrst part of the question is directed to Professor
Mathews, and the second part is directed at the
panel in general. Professor Mathews, you have been
described in the past as the ‘‘grandfather of com-
puter music,’’ and I am curious how you view your
inﬂuence upon the landscape of music today, and I
ask the panel in general what you think you might
be doing if he hadn’t paved the way.
Mathews: Well, the answer to the panel part of the
question is that if I hadn’t done something ﬁrst,
somebody else would have done it very shortly
thereafter. I think my only claim to this recogni-
tion is simply that I happened to be in the right
place at the right time, and maybe, that I also had
the motivation of enjoying the violin but never be-
ing able to play it very well—so I wanted to make
an instrument that was easier to play. I was ﬁrst
called the father of computer music, then the
grandfather [laughter], but I gave some talks in
Svizzera, and a publicity man from a depart-
ment store ﬁlled the halls by advertising me as the
‘‘great-grandfather of techno’’ [more laughter]. Now
I was happy to do that—it ﬁlled the hall with 18-
year-olds, and I think they enjoyed it somewhat—
but I’ve never done a techno piece in my life!
Lyon: That’s actually not true. You made Bicycle
Built for Two, which is a kind of techno piece. It’s
the ﬁrst techno piece.

Mathews: Is it? Now why do you call it a techno
piece?
Lyon: It is a rhythmically quantized setting of a
pop tune. And it has that robotic synthesized voice
singing the melody.
Mathews: Well thank you, Eric!
Lyon: That also inspired a scene from the movie
2001, as I understand.
Mathews: They grabbed that vocal piece Daisy Bell
of which I did the least important part. The inter-
esting part was done by John Kelly and Carol Loch-
baum, and that’s really an interesting technical
thing because Kelly and Lochbaum used a physical
model of the vocal tract—a tube model—and what
the computer did was to manipulate the shape of
the tube appropriately to the sounds they wanted
to make, and they put in the consonants also. If
you listen to that, it’s really of remarkable quality
compared with things that came long after. . . .
That’s why the computer in 2001 sang that as a
swan song.
Lyon: Any other reactions to the question?
Zicarelli: Well, I wanted to hear what everyone was
going to do if Max didn’t exist. But I guess no one
wants to answer that.
Loy: I think I would have ended up as a failed com-
poser of motets.
Audience Member: When computer music started,
the emphasis was on a score with notes, and the
notes had start times, duration, and pitches, Quale
was very traditional. At that time, the electronic
music community was doing things that were non-
traditional. So, when this came out, a lot of com-
posers—not all—said, ‘‘This is too traditional, pure
note-oriented; we don’t want to do the things that
are note-oriented, we want to do the things that are
more organic.’’ I kept telling them, ‘‘Well, you can
make it organic; a note can be a whole piece.’’ But
they couldn’t get that idea. They kept thinking that
it must be a sequence of notes and this score is re-
ally imposing a structure on our ideas. It was simi-
lar to the notion that we shouldn’t play a key-
board—a lot of people were saying that we
shouldn’t play a keyboard at that point, and they
were ﬁddling with tape, which was more abstract

Lyon

D
o
w
N
o
UN
D
e
D

F
R
o
M
H

T
T

:
/
/

D
io
R
e
C
T
.

io
T
.

e
D
tu
/
C
o
M

j
/

UN
R
T
io
C
e
–
P
D

F
/

2
6
4
1
3
1
8
5
3
6
1
1
0
1
4
8
9
2
6
0
2
3
2
0
9
9
1
3
4
7
P
D

B
sì
G
tu
e
S
T

o
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

Figura 4. Gareth Loy,
Miller Puckette, David
Zicarelli, Max Mathews,
and Barry Vercoe.

D
o
w
N
o
UN
D
e
D

F
R
o
M
H

T
T

:
/
/

D
io
R
e
C
T
.

io
T
.

e
D
tu
/
C
o
M

j
/

UN
R
T
io
C
e
–
P
D

F
/

2
6
4
1
3
1
8
5
3
6
1
1
0
1
4
8
9
2
6
0
2
3
2
0
9
9
1
3
4
7
P
D

B
sì
G
tu
e
S
T

o
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

and gets us out of the habits of traditional music.
The Music N programs all have that score, so they
have notes, and is that a hindrance? I see these new
programs like SuperCollider and so forth, and they
don’t have notes. Wasn’t Cmix one of the ﬁrst pro-
grams that got away from the note concept? Well,
that’s the question, Poi: what is a note? When I
look at what James [McCartney] has up there, Esso
doesn’t look like a note. And when I look at a
graphic ﬂow diagram, I don’t see notes.
McCartney: A note could just be setting some con-
troller that is not necessarily creating a discrete
event.
Audience Member: You’re getting away from syn-
chronizing the start of the note with other parame-
ters that are changing and so forth. That makes it
more free-ﬂowing.
Puckette: There is an important psychological ele-
ment to it, which is that if you do have individual
cards and there is a stack of them, there is a sort of
linear sequencing. And also of course, the synthe-
sizer people—Buchla and Moog—were putting key-
boards on their synthesizers at about the same
time, and if you listen to the history of computer
music, you hear this similar regression right around
1968 or so—right around when Switched-on Bach

was coming out. It didn’t have so much to do with
computer music as it did with ‘‘note happiness,’’
which lasted at least a couple of decades. Now, In
the clubs, they are starting to get away from that.
Mathews: I agree that making a score with notes is
a real constraint on the kinds of music the system
is able to produce. E, in my Music V anyhow, Esso
was a very hard constraint, because I guess there
was a slight facility for describing things that af-
fected a sequence of notes as a function of time.
But it was a very difﬁcult facility to use, and in tra-
ditional instruments the sound of a given note re-
ally depends on how the preceding note at least
ended and what the next note is—and that really
was not possible to put into Music V. Now, my
recollection, Barry, is that you improved that
greatly in Csound and had a lot of ability to put in
large-scale time functions that would affect a
whole phrase or a section of the piece. And you
probably had abilities to, let’s say, play a note and
maintain the ﬁnal conditions of the state variables
in the synthesis and use those as the initial condi-
tions for the next note.
Vercoe: Yes, the basic principle was that unit gen-
erators could retain the state information or clear it
out from event to event.

Computer Music Journal

Audience Member: In the spirit of the symposium,
which is to assemble some of the leading music
technologists of the digital era, I wanted to ask a
historical question of the panel. In all of human
history, what do you think are the technological
advances that have had the greatest impact or in-
ﬂuence on music making? I am particularly inter-
ested to see how the answers fall out in terms of
very old things like the ability to drill holes in
pieces of wood and make ﬂutes versus newer
things, like magnetic tape or digital computers.
Vercoe: That of course depends on which culture
you live in. Looking at the Western tradition, con-
out going back too far, I would say it was around
1580. The viol had been known as an instrument
primarily to double the voices. What happened
around 1580 is that instrument makers made two
changes as part of their experiment in natural se-
lection. One change was to rout the frets, and the
other one was to substitute a very ornate—I guess
it’s the beginning of Baroque thinking—resonating
box. What we can see now from the work of, Dire,
Steve McAdams is that when you move a funda-
mental around so that the rich harmonic structure
about that fundamental is being moved in and out
of very sharp resonance peaks (as occurs in a violin
resonance box), then you have rapid changes of the
relative strength of the component partials. For
some reason that we have now begun to under-
stand in some way (but the people at the time in
1580 only knew by experiment), that kind of sound
is a very good one to separate from other instru-
mental sounds. That instrument can stand out
from the ensemble in a way that the most of the
viol family was not able.

You suddenly have in the combination of those
two changes the development of a new instrument
which they called the violin. And since it was an
instrument that then could stand out like a sore
thumb, if necessary, that began to raise new musi-
cal traditions: you get the beginning of the concerto
and so forth. Ma, along with that came the devel-
opment of instrumental music in its own right—a
change of musical style—the beginnings of all
kinds of new musical forms, and suddenly from

1600 onwards, you get the instrumental period of
music as opposed to what you would classify as the
vocal period. And I would say that that was a big
innovation in the history of Western music. Com-
posers could still write choral music, but they
could write for a different kind of ensemble that
was an interesting one to sustain in a concert.
Now, you had instrumental music by itself or cho-
ral music by itself or the combination.

A revolution didn’t occur again until the twenti-

eth century, when in addition to those two forces
you now had the addition of an electronic medium.
It was many decades before instrumental music be-
came mature, and I think that we can see the same
thing with the development of the electronic me-
dium. So we’re in the midst of another revolution
in music that I think is equally important. And just
as we got new musical forms in the sixteenth cen-
tury, we naturally have to expect big changes in
the way in which electronic music is going to in-
ﬂuence the course of musical evolution in our
time. We haven’t seen the last of it yet; we are still
in the middle of it all. But I would expect that this
would certainly qualify as being something equiva-
lent to the impact that the development of instru-
mental techniques—instrument-making
techniques—had in the late sixteenth century.
Zicarelli: Well, I think I understand computers and
digital electronics fairly well, so I can’t think of
them as a profound invention. What I am im-
pressed with is analog electronics and transducers.
I took Barry’s workshop in 1982, and this guy came
in that worked for a company that made D/A con-
verters. He basically said, ‘‘You know, people say
there is a 16-bit D/A, but they’re lying; there isn’t
any such thing.’’ Now we seem to have these 24-
bit, 96 kHz D/A converters, and I don’t hear any-
one saying that they’re an illusion. And I see the
great emphasis in the emotions that people place
on better D/A conversions, better speakers, better
microphones . . . It seems as though people care
a lot more about that than someone’s ﬁlter algo-
rithm.
Puckette: I head someone say about a year ago that
there might have been music before there was such

Lyon

D
o
w
N
o
UN
D
e
D

F
R
o
M
H

T
T

:
/
/

D
io
R
e
C
T
.

io
T
.

e
D
tu
/
C
o
M

j
/

UN
R
T
io
C
e
–
P
D

F
/

2
6
4
1
3
1
8
5
3
6
1
1
0
1
4
8
9
2
6
0
2
3
2
0
9
9
1
3
4
7
P
D

B
sì
G
tu
e
S
T

o
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

a thing as words and grammar. And if that is true—
and I like to think that it is true—probably the in-
vention of language itself would have had a more
profound effect on musical thought and practice
than anything else that’s happened since. Fire cer-
tainly had very little effect on the way people make
music! [Laughter.] So, that might be the answer; if
not that, then I would say recording.
Mathews: Mugnaio, did you read a little article in the
New York Times Science page a couple of weeks
ago where they have some evidence that it’s a sin-
gle gene that facilitates the development of lan-
guage? Now, the single gene doesn’t do it all, but it
turns on a bunch of other genes that are involved
in it. E, if you don’t have this one gene, Poi
they don’t get turned on—that was my interpreta-
tion of it. So, if that gene mutated after the musical
gene or whatever turns it on, then music could
have come ﬁrst. There are plenty of things like
birds and wolves, which as far as I know don’t have
language but have music—maybe even dolphins
and whales.
Loy: I can’t go too far back, before the beginnings of
language or the sounds of animals, but I can go
back to the tenth century. My favorite, as I am a
composer, is the work of Guido d’Arezzo, who did
two important things. First of all, he invented the
solmization syllables as a device for remembering
plainchant melodies. People could do it on their
ﬁngers, and that would help them remember what
note came next. He is also the great-grandfather of
the compositional algorithm. There is a clear ex-
ample of how belief in what you’re doing informs
how you go about doing it. His primary goal was to
set the musical text of the liturgy; that is to say
that the text was the jewel, and the music was the
setting for it. He wanted the words to be the main
focus and the music to be its setting, so he sug-
gested the following method to aspiring composers.
Find the ﬁrst vowel in the ﬁrst word of the Latin
phrase that you are going to set. There are 5 possi-
ble vowels, and in those days there were 15 possi-
ble notes that you could sing (a compass of two
octaves). And so you have three choices: you can
place this vowel in one of three places within these

15 notes. And then you go onto the next vowel,
and that gives you one of three possible notes.

Ovviamente, you can do the math and ﬁgure out
that there are still a very large number of composi-
tions that can result from this procedure. I don’t
believe I’ve ever heard of a compositional algo-
rithm that precedes this, so that would make him
the originator of this concept.
McCartney: I think the one thing that’s affected
computer music most during its existence is not
any invention but just the fact that Moore’s Law
has held, where a computer will continue to double
capacity every 18 months. So that’s really enabled
computer music to go from requiring a US$ 200-an- hour computer to laptop performances. I hope that will continue. Mathews: That certainly is my technical answer. I think that the computer programs that have been developed are pretty interesting as musical instru- menti, making sounds and timbres. They are pretty uninteresting—at least to me—as compositional al- gorithms. E, maybe they are uninteresting be- cause we haven’t found or written the right algorithms; that’s a more difﬁcult problem than just making timbres. But maybe they are uninter- esting because we really personally want to write the music and we don’t want to give that job to any old computer—or for that matter give it to any other person. Audience Member: This question gets away from the actual programs themselves and goes towards the problem of distribution. I know Miller Puckette was talking about making computers accessible to people who have limited resources, and I know that the MIT Media Lab is looking at expanding into other economic environments. Do you have any ideas about ways to maximize access to these pro- grams so that the cost can come down? Puckette: In my own work, the answer is going to be different for each piece of software. The barrier is one of culture—of being able to really use the programs, more than just being able to get your hands on them. I don’t have a good answer for that yet, and I think that is a really crucial question for me right now. How on earth are we going to make 24 Computer Music Journal l D o w n o a d e d f r o m h t t p : / / d i r e c t . m i t . e d u / c o m j / l a r t i c e – p d f / / / / / 2 6 4 1 3 1 8 5 3 6 1 1 0 1 4 8 9 2 6 0 2 3 2 0 9 9 1 3 4 7 p d . f b y g u e s t t o n 0 8 S e p e m b e r 2 0 2 3 it so that people can just download the program that really does just give you a blank sheet of paper to work on and that allows you to put your own will on it? I don’t know the answer to that; I cer- tainly think that’s one of the main issues right now that we face. Mathews: Moore’s Law is moving through the world in a very good direction in terms of making things more available. Certainly at the beginning, the entry price was several million dollars, and no artist could afford that, but a few sneaked in and worked at universities or companies. The PDP-11 and Music11 brought the entry-level costs down to about US$ 100,000 to make useful music. Charles
Dodge had that facility at Brooklyn College for a
long time at about that cost, which he was able to
raise. And then the chips and the DX7 came along
and brought the entry costs down to US$ 2,000, and then these popular musicians jumped in, and the number of people using these machines in- creased by much more than the factor of 50:1 that the price went down. There was an increase of maybe 5,000 O 50,000:1 in terms of usage. I guess the entry cost hasn’t come down to much less than a couple of thousand dollars, but the value of a cou- ple thousand dollars has certainly decreased a lot in the last couple of decades. And the power of what you are getting for these dollars has increased re- markably. Lyon: What seems to me quite interesting is that Moore’s Law does not pertain to software. There is some relationship between increased hardware power and software development, but it is nonlin- ear. As an example, I recall a very ﬁne piece of soft- ware called ‘‘denoise’’ that was written as part of CARL by Mark Dolson. This was essentially an adaptive noise-reduction system based on spectral analysis. The price of that program, as with every- thing else in CARL, was zero. I recall being very surprised and somewhat amused in the early 1990s when I ﬁrst discovered things called plug-ins, and these were essentially symbiotes that connect to larger programs—commercial programs such as ProTools—and they provide a particular DSP ser- vice for a price. And I was somewhat nonplussed to discover that there was a plug-in—I forget what it was called—that essentially did what denoise did, and its price was about US$ 1,000. That’s an exam-
ple of a non-Moore’s Law progression for software.
Three of the major pieces of software we’ve dis-
cussed here are free, and the others certainly cost
less than a good ﬂute.
Mathews: I think the major cost of a lot of software
is not the dollars that you have to spend for it, Ma
the number of hours or weeks that you have to
spend learning how to use it. And that’s a tough
thing to reduce without imposing restrictions on
the person that you don’t want to impose.
Zicarelli: It’s difﬁcult to learn how to use software
by yourself in many cases, pure. And all of the soft-
ware that we are talking about here was developed
at institutions where the author was there to help.
In various ways, there have been people who have
made efforts to translate the technology out of that
setting into places where you didn’t have to know
the author personally to use the program. And this
is the major difﬁculty involved in releasing or com-
mercializing software. Beyond releasing it, Poi
there is the question of training and how to distrib-
ute expertise in using the software, and that’s cur-
rently not free. But there are lots of kind-hearted
people who seem to be sharing their expertise in
various ways such as the Internet and user groups.
The problem is that if you take any one random
person in any one random place in the world and
look at their access to the technology, it’s probably
not very good. But if you happen to be in the right
place at the right time, in many cases someone can
do something for you.
Audience Member: I feel a little compelled to com-
pare access and costs in computer music to other
musical pursuits or other disciplines. I’d also like
to compare educational access to buying a ﬂute or
any other sort of traditional instrument. You can
pick up a used computer or a free computer; Esso
might not be the latest, but it’s probably only a few
years old and still very useful. Paint for painting
and stone for sculpture aren’t cheap, and neither
come with user manuals and tutorials the way that
Max and Pd and Csound do.

Lyon

D
o
w
N
o
UN
D
e
D

F
R
o
M
H

T
T

:
/
/

D
io
R
e
C
T
.

io
T
.

e
D
tu
/
C
o
M

j
/

UN
R
T
io
C
e
–
P
D

F
/

2
6
4
1
3
1
8
5
3
6
1
1
0
1
4
8
9
2
6
0
2
3
2
0
9
9
1
3
4
7
P
D

B
sì
G
tu
e
S
T

o
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

Figura 5. Miller Puckette.

someone to learn, that the eleventh fact would
somehow relate to the previous ones in some co-
herent fashion, so that you really could bootstrap
yourself into a wider horizon. In definitiva, I think,
computer software is an oral tradition. We can get
somewhere without having someone to answer our
informal questions about it. But there are so many
questions at so many levels that I’ve gotten that I
no longer believe it’s possible for me to thread a
single linguistic thread through the middle of what
everybody wants to know.
Audience Member: Earlier, you were talking about
what composers had done with your software, E
we glanced over that question. I think two of you
answered that question, and then it disappeared.
McCartney: I think that, at least for my software,
the people who seem to have learned it the fastest
and gotten the deepest into it are the people doing
the live techno laptop stuff. They seem to be the
most caffeinated groups, so they’re actually the
people who have written the most code.
Puckette: I think your question leads to a different
issue. We’ve talked quite a bit about software, Ma
what we haven’t mentioned is that—almost regard-
less of what software we are using—I don’t think
that we know on a musical level what to do with
computers. We don’t yet know what the modality
is or the modalities are that people will ﬁnd to in-
corporate computers in their music. I think that
perhaps there is some software that hasn’t yet been
invented that will allow that, but I don’t feel that
modo. I feel that in fact there’s a real question about
where to go now and how to use this tool, the pro-
grammable instrument. People will do interesting
things in spite of the limitations of the current sys-
tems, certainly.
Audience Member: You’ve all authored some really
extraordinary software. Would you in retrospect
have done anything differently to protect your own
intellectual property? I want to tie that into how
the MPEG-4 standard might return intellectual
property to composers and authors in a meaningful
modo.
Vercoe: I don’t know about returning to composers
and authors—that’s really between those compos-

Vercoe: As you look at the results of synthesis pro-
grams over the years and the beneﬁt of a hundred
years hence, I wonder which pieces will have sur-
vived. We are talking about software survival here,
and let’s also focus on which pieces of music will
survive. I don’t know how to guess at that, but it
may well be that the enduring pieces will come
from the systems that were the hardest to access.
I’m not talking so much about the cost of them—
just the hardest to get at, where you had to really
sweat through what you had to say to say it in the
way that you wanted to say it. I wouldn’t say that
making things easier and easier is going to give rise
to lots and lots of new, long-surviving music. For
certain, there will be lots of activity, but there’s a
fear I have that so much of this will be noise. E
what the history of music will eventually sort out
is which of these things were the giant steps that
will just keep coming back. And I think it’s wise
for us to think about that—just what engenders the
very clear statements from composers, which is af-
ter all why we’ve ever done this. We’re not just
writing software to keep machines busy.
Loy: There was a bit of philosophy that we had in
the design of the CARL system to this end. Since it
was designed in a university environment, it was
designed to create a community of users. We al-
ways tried to keep in mind that if we had, Dire, ten
facts about a system that we were going to require

Computer Music Journal

D
o
w
N
o
UN
D
e
D

F
R
o
M
H

T
T

:
/
/

D
io
R
e
C
T
.

io
T
.

e
D
tu
/
C
o
M

j
/

UN
R
T
io
C
e
–
P
D

F
/

2
6
4
1
3
1
8
5
3
6
1
1
0
1
4
8
9
2
6
0
2
3
2
0
9
9
1
3
4
7
P
D

B
sì
G
tu
e
S
T

o
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

ers and the music industry. I can tell you regarding
the software itself that, unlike most of the rest of
MPEG-4—which was contributed by industries and
various forces—the core music synthesis part of it,
the SAOL stuff, I put immediately into the public
domain. This has always been my policy from Mu-
sic 360: just to get it out there. Ovviamente, questo è
the sort of thing one can do from an ivory tower.
Corporate research groups cannot afford to do
things like that. So the other parts of the MPEG-4
standard do in fact have licensing attached. But I
think in general, wherever it is possible, we should
get something out there and just spread the word.
McCartney: I think that if a lot of things in this
ﬁeld had been patented that are pretty basic, IL
ﬁeld would have been stopped cold from advancing.
So I don’t want to patent anything that I come up
con, just because I don’t want to stop anybody
else cold. I really don’t like the idea of protecting
patents in this area, but I know it is done a lot.
Zicarelli: I am similarly also opposed to software
patents, and I think that they probably should
never have existed. And as far as why I’ve fallen
into being a commercial person rather than being
an academic person, you know I am a bad teacher.
Every time I’ve applied for an academic job, I never
got anywhere. I am not sure I would even be good
at the cocktail parties. My life just took a certain
sentiero, and I am not sure that I ever really examined
it from ﬁrst principles of the ideals of intellectual
property issues. But what I try to do is give every-
thing that people request. If there is some piece of
source code that someone wants, I usually give it
to them, but I don’t give away the ability to violate
the copyright on producing the commercial thing
that allows me to do the work I’m doing. As long
as there is one way that I can trick people into giv-
ing me money to continue to do the work that I
like, I’ll probably do that. I try to give away as
much—other than that one little trick—as I possi-
bly can, just because it’s great to see people doing
things with that information. If someone thinks
the information is valuable, I’m happy to share it as
much as other people in this ﬁeld. And everyone
sitting at this table has shared information that al-
lows me to do the work that I do.

Audience Member: I hope James gives everyone the
same opportunity to learn from his source code as
he learned from Barry’s.
McCartney: Well, actually if you download version
3 of SuperCollider, all the primitives are there.
There is actually a lot you can read; I just don’t
give you the engine in the middle. [Editor’s note:
Supercollider was made free as of June 2002, and it
was announced that the source code would soon be
available.]
Loy: The general issue with what to do about pro-
tection depends of course on what it is you are try-
ing to accomplish. You can either keep something
a trade secret, attempt to patent it and capitalize
on a license agreement, or simply make a wall of
patents around its core technology to guarantee a
revenue stream. Or if you get your revenue stream
from an academic institution, as Miller was saying
earlier, it is a really great opportunity to do what
you think is the right thing. That was also pretty
much the inspiration for the CARL system. Noi
took our salaries from the grants that we got and
from what facilities the university gave us, E
what we produced was then put into the public do-
main, because we were compensated for it. But if
you’re not compensated by an academic institu-
zione, you must have some way to guarantee your
revenue stream. And it’s really hard, as David was
saying, to support users in the ﬁeld. I have worked
with some companies in Silicon Valley that have
the same problem where they have a technology
that is difﬁcult to penetrate. Their product fails be-
cause they cannot get enough technical support
personnel, and so the business fails. This is a seri-
ous problem. You have to have enough return on
your investment to be able hire the technical sup-
port required for your product to ﬂourish. These are
tough problems. It’s not easy to know what the
best way to go forward is.
Vercoe: The solution is to have someone else write
a book on your stuff.
Audience Member: I would like to ask the panel
what they think of the open-source paradigm—how
that’s going to inﬂuence future development of
computer music software.

Lyon

D
o
w
N
o
UN
D
e
D

F
R
o
M
H

T
T

:
/
/

D
io
R
e
C
T
.

io
T
.

e
D
tu
/
C
o
M

j
/

UN
R
T
io
C
e
–
P
D

F
/

2
6
4
1
3
1
8
5
3
6
1
1
0
1
4
8
9
2
6
0
2
3
2
0
9
9
1
3
4
7
P
D

B
sì
G
tu
e
S
T

o
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

Puckette: I think it is certainly true that the open-
source movement is a very powerful source toward
higher-quality software, and I would turn the ques-
tion around and ask if open-source music wouldn’t
also be a very good thing. If composers starting
making music available in such a way that you
could absolutely use it for anything, it would be
amazing how fast the musical culture would then
develop. The problem is that we’re stuck on this
outdated notion that the only way a composer or
instrumentalist can survive is by forcing people to
pay them for airtime. That’s a revenue stream
modello, but I don’t think it’s a very good one. E
the thing that made it quit being so good is that
you quit distributing physical hardware with it. In
other words, I think a copyright makes sense when
you are talking about paper or a compact disc. IO
don’t think it’s a good thing to impose on things
that are distributed by the network. And I think
that the Internet is a good way of distributing mu-
sic right now. I would rather see copyrighters not
even get mixed up in the issue. I would rather see
it tied to physical media. D'altra parte, com-
posers can’t really survive anyway; the only people
that are really milking us are the studios—but also
the content holders like Sony. So that’s the main
development that I’m hoping to see in the future,
and I know that right now the GATT organization
is working as hard as they can to prevent us from
having this and also from having open-source soft-
ware. It’s not clear whether the open-source move-
ment (or the open-music movement) will really
succeed against the commercial interests that are
already against it, but I certainly hope so. I think it
is everyone’s duty to do whatever they feel they
can do in keeping all manner of intellectual prop-
erty restrictions out of the stream of innovation—
even if it’s in their income stream.
Audience Member: I would like to ask the panel
how computer music has changed in the last 30
years. We know that user interfaces and programs
and synthesis techniques particularly have a large
inﬂuence on the type of music that composers pro-
duce.
Vercoe: There’s a very big difference; there is a
whole lot more audio processing of live and re-

corded sounds. That’s just part of the evolution of
this phenomenon of computer music, so certainly
things keep on growing. I can’t predict what is go-
ing to happen in the next ten years, but it will be
something signiﬁcant, I am sure. But there will be
the classics that will survive the decades, and all
we can do is keep writing whatever we write and
hope that one of our pieces will become a classic.
Loy: At some point, quantitative changes in proces-
sor speed and disk sizes effect a qualitative change.
For instance, Max went from being a MIDI control
language to a digital synthesis engine, but it
couldn’t do that with the processors that it was
originally designed for, because there wasn’t
enough horsepower there. So you get this qualita-
tive shift then from a controller of synthesizers to a
general-purpose synthesis engine under the hood.
Puckette: In other words, we’ve only got to undo
the MIDI revolution!
Audience Member: Do you think the MIDI revolu-
tion was a step forward or backward?
Puckette: I think it was a huge step sideways.
Audience Member: As software developers, you’ve
probably all experienced this: when you put out a
new feature, the user community generates ﬁve
more requests for other new features. It’s been my
experience that the user demand for new things
and the ideas that ﬂow your way is pretty strong. Is
it a fact that you can’t keep up with what the users
would like to see, and are you frustrated by that?
What do you think is inhibiting you from making
more rapid progress with your software projects?
Zicarelli: If you implement a feature in x amount
of time, then you get 5x. So as x decreases, you just
get more feature requests, so you could never catch
su, even if you were more productive.
Audience Member: Some of the features are in high
demand; let’s just focus on those. I would think in
your case that Max on Windows has been an issue.
I think there has been a group of people who would
like to have seen that happen a few years ago. Cer-
tainly, there are always these individual quirky re-
quests that come your way that are not worth
pursuing. I am thinking more of large-scale re-
quests, where more than half the user community
just wants it.

Computer Music Journal

D
o
w
N
o
UN
D
e
D

F
R
o
M
H

T
T

:
/
/

D
io
R
e
C
T
.

io
T
.

e
D
tu
/
C
o
M

j
/

UN
R
T
io
C
e
–
P
D

F
/

2
6
4
1
3
1
8
5
3
6
1
1
0
1
4
8
9
2
6
0
2
3
2
0
9
9
1
3
4
7
P
D

B
sì
G
tu
e
S
T

o
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

Figura 6. Eric Lyon and
Gareth Loy.

action. There is no way to predict how these things
are going to be used.
Puckette: I am partial to Fat Boy Slim. I don’t think
he uses my work at all, but whatever he’s got that’s
playing right now is pretty cool.
Lyon: James, you have a lot of people that you’ve
contacted directly who are using your software.
McCartney: Well, I don’t care about what some-
one’s particular style is so much. Sometimes peo-
ple play me these things they did, and sometimes I
like them, and sometimes I don’t; but it doesn’t re-
ally matter, as long as they were able to use it, it’s
good. Sometimes people send me things in which
they have used my program in some way that I
think is really ridiculous. But they sometimes pose
the most interesting problems.
Vercoe: I’ve not usually paid much attention to
how my stuff ever got used in the popular ﬁeld, nor
have I paid much attention to the popular music
styles, so I’m often caught off-guard when some-
thing does happen. But I will answer two questions
at once: the one that was just posed, and also about
the music of Miller Puckette. I can tell you about
one of the really interesting, exciting things that I
experienced when Miller was around at MIT. Lui
wrote a little piece of music called Cat’s Ass. Quello
really was quite neat, and I thought it was a new
use of the technology coming from someone who
never claimed to be a composer, and yet it showed
the kind of musical insights that he has. And that’s
why he’s been such a contributor to the ﬁeld.
Thank you, Mugnaio. You didn’t expect that one, did
you?
Puckette: No. I don’t distribute anything that I
have ever done. But for the record, it’s Cat, spazio,
Sass.
Lyon: Gordon Mumma once told me that, in his
opinion, every creative act was essentially a form
of subversion. And of course, as you know, Gordon
is one of the most deeply technologically involved
composers of the last century. When he said that,
Mumma was working with a very cheap commer-
cial FM tone generator and was digging around the
operating system, discovering all kinds of bugs,
things just not working the way they were supposed

Lyon

McCartney: It’s triage!
Zicarelli: I was going to say that that was the num-
ber one frustration, but I don’t want to—there must
be something else that’s frustrating, pure. You have
a program that works on one computer, and the
stupidest person in the world can say, ‘‘Why don’t
you make the same program work on this other
computer?’’ The level of thought behind the sug-
gestion is inversely proportional to the effort the
suggestion requires to implement it. So that’s
what’s frustrating to me—this really stupid idea to
just make it work on another computer. There is
absolutely no creative work in that—it’s the most
horrible work in the world, and yet it’s the thing
that people want the most. The creative, difﬁcult,
interesting problems that might even be quick to
do—no one cares about as much.
Audience Member: I’d like to know what the panel
thinks of today’s technology-based popular music
like rap and techno, E, in a broader sense, how
you feel about the fact that your innovations have
been absorbed into popular music and popular cul-
ture.
Mathews: Well, I myself don’t really enjoy popular
music, but I’m delighted that anything from our
technology is being used by popular musicians. In
fatto, that’s the biggest ﬁeld of users. And I’m de-
lighted if they have an audience that goes for it.
Loy: This is an example of a dynamic system in ac-
zione . . . a chaotic, non-linear dynamic system in

D
o
w
N
o
UN
D
e
D

F
R
o
M
H

T
T

:
/
/

D
io
R
e
C
T
.

io
T
.

e
D
tu
/
C
o
M

j
/

UN
R
T
io
C
e
–
P
D

F
/

2
6
4
1
3
1
8
5
3
6
1
1
0
1
4
8
9
2
6
0
2
3
2
0
9
9
1
3
4
7
P
D

B
sì
G
tu
e
S
T

o
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

to work, and then making that a part of the piece.
It seems to me that one of the real questions is, ‘‘Is
it possible to subvert this software, and is it desir-
able?’’ For example, one model that we see is the
model of a blank piece of paper; how attractive or
even possible is it to subvert a blank piece of pa-
per? D'altra parte, the Music N model, al-
most by its nature, seems to be a challenge thrown
in the face of composers. Namely, here we have
this new way of making music, and yet we are sup-
posed to write a score for orchestra. That’s ridicu-
lous! Now you can take that model and subvert it
by doing things like having a million notes, so that
all of a sudden you have something that an orches-
tra wouldn’t do. You have instruments behaving so

a note is not what we usually think of as a note.
There are all these possibilities for subversion in
that ﬁrst paradigm, so I think I ﬁnd myself acciden-
tally asking the last question. How do you feel
about the idea of subversion of your own software?
McCartney: I think my users are secretly trying to
subvert it in every way possible. I guess the whole
glitch community will take all of your unit genera-
tors and ﬁnd out where you didn’t write them cor-
rectly and exploit that.
Puckette: Yes, that oscillator of yours is going to
fail as soon as the frequency goes above the sam-
pling rate!
Lyon: I would like to thank you all for a remark-
able and enlightening day!