colonel@gloria.UUCP (Col. G. L. Sicherman) (05/06/85)
[Do not remove this tag under penalty of law] > "To think -- and I have heard this suggested -- that we might soon be > able to command a preprogrammed mass-produced mail-order twenty-dollar > desk-model "music box" to bring forth from its sterile circuitry pieces > which Chopin or Bach might have written had they lived longer is a > grotesque and shameful misestimation of the depth of the human spirit. > A ``program'' which could produce music as they did would have to wander > around the world on its own, fighting its way through the maze of life > and feeling every moment of it. It would have to understand the joy and > loneliness of a chilly night wind, the longing for a cherished hand, the &c. &c. &c. Hofstadter is merely taking a stand. "Sterile circuitry," "depth of the human spirit," and so on, are empty rhetoric. Dr. Geoffrey Jefferson wrote something similar in 1949: "Not until a machine can write a sonnet or compose a concerto because of thoughts and emotions felt, and not by the chance fall of symbols, could we agree that machine equals brain ... " [_British Med. J.,_ 25 June 1949; quoted with disapproval in Turing's "Computing Machinery and Intelligence."] While Jefferson's article suffers from a few misconceptions about computers, it is philosophically superior to Turing's. Unfortunately, philosophers who write about computer-generated music seldom under- stand the nature of music. -- Col. G. L. Sicherman ...{rocksvax|decvax}!sunybcs!colonel
ee163acp@sdcc13.UUCP (DARIN JOHNSON) (05/08/85)
In article <711@gloria.UUCP>, colonel@gloria.UUCP (Col. G. L. Sicherman) writes: > [Do not remove this tag under penalty of law] > > > "To think -- and I have heard this suggested -- that we might soon be > > able to command a preprogrammed mass-produced mail-order twenty-dollar > > desk-model "music box" to bring forth from its sterile circuitry pieces > > which Chopin or Bach might have written had they lived longer is a > > grotesque and shameful misestimation of the depth of the human spirit. > > A ``program'' which could produce music as they did would have to wander > > around the world on its own, fighting its way through the maze of life > > and feeling every moment of it. It would have to understand the joy and > > loneliness of a chilly night wind, the longing for a cherished hand, the > &c. &c. &c. > > Hofstadter is merely taking a stand. "Sterile circuitry," "depth of the > human spirit," and so on, are empty rhetoric. Dr. Geoffrey Jefferson > wrote something similar in 1949: > > "Not until a machine can write a sonnet or compose a > concerto because of thoughts and emotions felt, and not > by the chance fall of symbols, could we agree that > machine equals brain ... " [_British Med. J.,_ 25 June 1949; > quoted with disapproval in > Turing's "Computing Machinery > and Intelligence."] > > While Jefferson's article suffers from a few misconceptions about > computers, it is philosophically superior to Turing's. Unfortunately, > philosophers who write about computer-generated music seldom under- > stand the nature of music. > -- > Col. G. L. Sicherman > ...{rocksvax|decvax}!sunybcs!colonel Ok, as a Computer Engineering major, with a minor in Music, I can safely assume that some type of (perhaps enjoyable) music may one day be created by computer. However, I seriously doubt that any great piece of music would emerge. It would be a simple academic matter to analyze any piece of music thus generated (in the program notes: .. bar 35 demonstrates the unlikely occurence of particular element leading the composer to choose this particular passage... ) One would no longer need to argue and debate as to what is really happening in the process of music making, just generate the same set of input and random numbers, and you can see exactly what happened and why. To me this is what is refered to as sterile. Admittedly, Beethoven tended to rewrite pieces over and over until they were "finished", perhaps ending up with the same as he would if he used another 'set of input'. If any great piece of music were to come out of computers, I would hope the program was highly interactive, taking out the great percentage of the drudgery accompaning this creative process. In any event, I would predict that perhaps a future assignment in some university music class would be to tell the difference between a 'human' piece and a computer generated one. I would hope the entire class could get this one right. Darin Johnson
srt@ucla-cs.UUCP (05/10/85)
In article <234@sdcc13.UUCP> ee163acp@sdcc13.UUCP (DARIN JOHNSON) writes: > >Ok, as a Computer Engineering major, with a minor in Music, I can safely >assume that some type of (perhaps enjoyable) music may one day be created >by computer. However, I seriously doubt that any great piece of music >would emerge... > <various discussion of music by computer> > ...In any event, I would predict that perhaps a >future assignment in some university music class would be to tell the >difference between a 'human' piece and a computer generated one. I >would hope the entire class could get this one right. > Not to get involved in a long discussion of AI and philosophy, but... Your assumption is that the computer would generate music by the machine equivalent of a throw of dice. I think most people would agree that this method is unlikely to produce great music. However, suppose we built a machine with all the memories and music knowledge of Beethoven, and then let this machine loose to create music by the same techniques of inspiration and experimentation that human composers use. It might be able to create a great piece of music AND tell you why: ``This passage evokes in me the image of my late wife as she stood in the afternoon sun in my study'' and ``Ah, now this passage I stole from my earlier work...'' and so on. This seems like a more likely method to produce, via machine, truly creative works of art. In short, if we could make a machine that duplicated the creative mechanics of a human at an appropriate level, it might be able to create music indistinguishable from human-composed music. Then, since it does just what a human does, can you still say it is not being creative? Well, it's an open question. Secondly, you make the assumption that art isn't Art unless it is created with purpose and intention. That is a philosophical viewpoint, and you should at least be aware that others may not share it with you. They may consider a great piece of music to be art even if it was created completely randomly (or "mechanically" whatever that means). Scott R. Turner ARPA: srt@LOCUS.UCLA.EDU UUCP: ...!{cepu,ihnp4,trwspp,ucbvax}!ucla-cs!srt SPUDNET: ...eye%srt@russet.spud
ee163acp@sdcc13.UUCP (DARIN JOHNSON) (05/10/85)
In article <5327@ucla-cs.ARPA>, srt@ucla-cs.UUCP writes: > In article <234@sdcc13.UUCP> ee163acp@sdcc13.UUCP (DARIN JOHNSON) writes: > > > >Ok, as a Computer Engineering major, with a minor in Music, I can safely > >assume that some type of (perhaps enjoyable) music may one day be created > >by computer. However, I seriously doubt that any great piece of music > >would emerge... > > <various discussion of music by computer> > > ...In any event, I would predict that perhaps a > >future assignment in some university music class would be to tell the > >difference between a 'human' piece and a computer generated one. I > >would hope the entire class could get this one right. > > > > Not to get involved in a long discussion of AI and philosophy, but... > > Your assumption is that the computer would generate music by the machine > equivalent of a throw of dice. I think most people would agree that this > method is unlikely to produce great music. However, suppose we built a > machine with all the memories and music knowledge of Beethoven, and then > let this machine loose to create music by the same techniques of inspiration > and experimentation that human composers use. It might be able to create > a great piece of music AND tell you why: I didn't mean to imply that the program would be totally random, but there would have to be a tiny bit somewhere, else the input would completely determine the output. (Unless there is a new concept around) Darin Johnson
hestenes@sdcsla.UUCP (Eric Hestenes) (05/10/85)
> In article <234@sdcc13.UUCP> ee163acp@sdcc13.UUCP (DARIN JOHNSON) writes: > > > >Ok, as a Computer Engineering major, with a minor in Music, I can safely > >assume that some type of (perhaps enjoyable) music may one day be created > >by computer. However, I seriously doubt that any great piece of music > >would emerge... > > <various discussion of music by computer> > > > > Not to get involved in a long discussion of AI and philosophy, but... > > Your assumption is that the computer would generate music by the machine > equivalent of a throw of dice. I think most people would agree that this > > Scott R. Turner I think you underestimate human ability to create. One could likewise say that we never would have realistic computer graphics made by a computer. One could say that the best computer graphics could only be created by the best human graphics artist. However, the best,most realistic graphics made today is made with the computer, not exclusively by humans. Fractal geometry provides mathematical principles that describe real-world scenery far better than the hands of most artists. And it draws scenery with such precision that it makes human efforts look "like the throw of the dice". What is to keep us from discovering priciples of music that would allow us to create the best in auditory mosaics with a computer? NOTHING. My bet is the computer that makes the best music will be the computer that incorporates the most understanding of human cognition. It is quite likely that there are a small set of unifying principles of cognition that govern differences betweeen good music and bad music, just as we can now mathematically describe scenery in a recursive way that beautifully matches the ability of the comuter.And perhaps there are other mathematical properties of the real world, like fractal geometry, that are relevant to creating music, and that are waiting to be discovered. The point is that the problem of creating great music on a computer isn't a function of the technology. Rather we are waiting for an innovator to tell us that, indeed, the world is not flat, that we can fly to the moon, that we can invent the bomb. It is easy the be a skeptic. It is hard to replace an older idea with one that better fits the problem. Eric Hestenes ARPA: sdcsla!hestenes@nprdc or hestenes@nprdc other: ucbvax!sdcsvax!sdcsla!hestenes
dave@rocksvax.UUCP (05/13/85)
/***** rocksvax:net.ai / ucla-cs!srt / 1:36 am May 12, 1985*/ >method is unlikely to produce great music. However, suppose we built a >machine with all the memories and music knowledge of Beethoven, and then >let this machine loose to create music by the same techniques of inspiration >and experimentation that human composers use. It might be able to create >a great piece of music AND tell you why: ``This passage evokes in me the I believe the band Boston used to use techiques like this to produce their music. The analyzed all the hit songs and figured out what people liked in them and generated songs that constistantly became instant hits. The music, although not great, always sounded OK when first heard but quickly became boring. These folks I heard were MIT people and may have used these AI techniques to help them. So ucla-cs!srt's idea of seeding a program to generate music based on past music is not to wild of idea, we probably have heard the results of such techiques already!! Dave arpa: Sewhuk.HENR@Xerox.ARPA uucp: {allegra,rochester,amd,sunybcs}!rocksvax!dave
srt@ucla-cs.UUCP (05/13/85)
In article <239@sdcc13.UUCP> ee163acp@sdcc13.UUCP (DARIN JOHNSON) writes: > > <a discussion of computer music, randomness> > >I didn't mean to imply that the program would be totally random, but >there would have to be a tiny bit somewhere, else the input would >completely determine the output. (Unless there is a new concept around) The same can be said for human beings, right? Do we have a "random" element, or does our input completely determine our output? The point is that if we could build a computer program that could experience as wide a range of existence as a human, then its output would be no more a trivial function of its input than would be a human's. I think one problem you may have is that you believe you could look at the code for a music generating program and understand what output would be produced. This simply isn't true anymore (see, for instance, Lenat's thesis (AM) or Dyer's (BORIS), both programs did things that surprised their authors), and certainly wouldn't be true for a program that could produce genius level work. It would be as complex and opaque as a human mind. Scott R. Turner ARPA: (now) srt@UCLA-LOCUS.ARPA (soon) srt@LOCUS.UCLA.EDU UUCP: ...!{cepu,ihnp4,trwspp,ucbvax}!ucla-cs!srt SPUDNET: ...eye%srt@russet.spud
mmt@dciem.UUCP (Martin Taylor) (05/14/85)
>graphics artist. However, the best,most realistic graphics made today >is made with the computer, not exclusively by humans. Fractal geometry >provides mathematical principles that describe real-world scenery far >better than the hands of most artists. And it draws scenery with such >precision that it makes human efforts look "like the throw of the dice". > What is to keep us from discovering priciples of music that would >allow us to create the best in auditory mosaics with a computer? NOTHING. >My bet is the computer that makes the best music will be the computer >that incorporates the most understanding of human cognition. It is quite Although I agree with the conclusion, I can't agree that fractal geometry provides more realistic descriptions of scenery than human artists. Real scenery is not fractal except in unusual circumstances. The effects of erosion or biological growth patterns are usually scale-dependent, and directed. For example, folded strata dominate most mountain structures, and the relative hardness and friability of the strata determine the visual appearance of a mountain. Alps don't look like Rockies, for example. One side of a mountain may be almost a mirror-smooth slab, whereas the other shows a dipping saddle of cliffs. Fractal geometry provides aesthetically pleasing pictures that sort-of look right, especially when compared with previous computer-based graphic methods. But it's a long way from there to realism. Music doesn't require realism, and I do think that the right computer-based approach may well provide music of deep aesthetic satisfaction (i.e. "great"). (Of course, visual art doesn't require realism, either). -- Martin Taylor {allegra,linus,ihnp4,floyd,ubc-vision}!utzoo!dciem!mmt {uw-beaver,qucis,watmath}!utcsri!dciem!mmt
dpb@philabs.UUCP (Paul Benjamin) (05/14/85)
>Fractal geometry provides mathematical principles that describe >real-world scenery far better than the hands of most artists. And >it draws scenery with such precision that it makes human efforts >look "like the throw of the dice". Statements like this illustrate the absurdity of this whole discussion. Fractal geometry describes scenery "better" than most artists? Precision makes human efforts look random? If I were to take an Instamatic outside and shoot a picture, I can easily obtain more precision than any of the great master painters of history. Does this make me better than they are? If I write a program that generates music without ever violating any of the rules of, say, the contrapuntal rules of Palestrina's era, will the music therefore be "better" than his? Maybe so, if you are looking for "mathematical principles" that formulate "good" music or art. But as long as the evaluation of art is subjective, (How can you like that music? It sounds like the wail of a dying baboon...) there will never be one fixed best way to make art, nor will there be a theory of art. The idea that we will discover a theory of music which will allow us to produce truly great (not derivative) works is ridiculous. If you think about it, you will realize that the great paintings, music, literature, and sculpture were produced by artists who used their personal knowledge of themselves, and of the world, to express their emotions, thoughts, and desires through an artistic medium. If an artist is just deficient in understanding the human condition, it doesn't matter how learned he is in the techniques of his art, he won't produce anything great (see Salieri vs. Mozart). The bottom line is that when we refer to art, we mean human art, i.e., art which appeals to people on a personal level. For a person to create great human art, he must have a great human understanding. There is no reason to expect this not to hold for any other type of system, e.g., computers.
hestenes@sdcsla.UUCP (Eric Hestenes) (05/15/85)
> I said: > >graphics artist. However, the best,most realistic graphics made today > >is made with the computer, not exclusively by humans. Fractal geometry > >provides mathematical principles that describe real-world scenery far > >better than the hands of most artists. And it draws scenery with such > >precision that it makes human efforts look "like the throw of the dice". Martin Taylor said: > Although I agree with the conclusion, I can't agree that fractal geometry > provides more realistic descriptions of scenery than human artists. Real > scenery is not fractal except in unusual circumstances. The effects of Whether or not fractals are precisely accurate, the point relevant to AI is that an understanding of the world ( versus artistry ) facilitates creating very realistic pictures ( a form of art ), without needing to dive into a lot of labor involving imitating human artistic endeavors. > erosion or biological growth patterns are usually scale-dependent, and > directed. For example, folded strata dominate most mountain structures, > and the relative hardness and friability of the strata determine the > visual appearance of a mountain. Alps don't look like Rockies, for example. > One side of a mountain may be almost a mirror-smooth slab, whereas the > other shows a dipping saddle of cliffs. Fractal geometry provides > aesthetically pleasing pictures that sort-of look right, especially > when compared with previous computer-based graphic methods. But it's > a long way from there to realism. Wrong. There are important areas where one cannot describe the scenery at all without the aid of fractal geometry. Mountain landscapes are just one example. Another good example of the impact of fractals is in predicting the shape of coastlines bordering major oceans ( pacific, atlantic, indian ) and the shape of islands. Without fractal geometry we have no coherent theory for predicting the shape of coastlines, islands, etc. With it, we can generate by computer near exact predictons of the shape of major continents. > > Music doesn't require realism, and I do think that the right computer-based > approach may well provide music of deep aesthetic satisfaction (i.e. "great"). > (Of course, visual art doesn't require realism, either). > -- > > Martin Taylor > {allegra,linus,ihnp4,floyd,ubc-vision}!utzoo!dciem!mmt > {uw-beaver,qucis,watmath}!utcsri!dciem!mmt Who says music doesn't require realism? Perhaps someone who also does not think fractal geometry is required to describe nature. One may ASSUME that realism isn't important, but this assumption doesn't provide evidence against the argument the realism MIGHT BE important. Read Mandelbenoit, The Fractal Geometry of Nature, for the real scoop. eric ------- hestenes@nprdc ucbvax!sdcsvax!sdcsla!hestenes
mmt@dciem.UUCP (Martin Taylor) (05/20/85)
>> Although I agree with the conclusion, I can't agree that fractal geometry >> provides more realistic descriptions of scenery than human artists. Real >> scenery is not fractal except in unusual circumstances. The effects of . . . >> other shows a dipping saddle of cliffs. Fractal geometry provides >> aesthetically pleasing pictures that sort-of look right, especially >> when compared with previous computer-based graphic methods. But it's >> a long way from there to realism. > >Wrong. There are important areas where one cannot describe the scenery at all >without the aid of fractal geometry. Mountain landscapes are just one example. >Another good example of the impact of fractals is in predicting the shape of >coastlines bordering major oceans ( pacific, atlantic, indian ) and the shape >of islands. Without fractal geometry we have no coherent theory for predicting >the shape of coastlines, islands, etc. With it, we can generate by computer >near exact predictons of the shape of major continents. > > >> >> Music doesn't require realism, and I do think that the right computer-based >> approach may well provide music of deep aesthetic satisfaction (i.e. "great"). >> -- >> >> {uw-beaver,qucis,watmath}!utcsri!dciem!mmt > >Who says music doesn't require realism? Perhaps someone who also >does not think fractal geometry is required to describe nature. >One may ASSUME that realism isn't important, but this assumption doesn't >provide evidence against the argument the realism MIGHT BE important. > > > >Read Mandelbenoit, The Fractal Geometry of Nature, for the real scoop. > >eric >------- If you are going to appeal to authority, at least try to get the name of the authority correct -- Benoit Mandelbrot. And yes, I have read his books, which is the reason I posted my original commnet. Fractal geometry provides pictures of landforms that look nice to the untrained eye, but confuse the geographer, in much the same way that a chessboard with randomly placed pieces looks nice to a non-chess-player, but confuses the chess master. It's better than a checkerboard with lumps of things on it, but it isn't a realistic *chess* board. Likewise, fractally derived landforms are better than previous computer-based efforts, but they shouldn't be confused with realism. Fractal geometry is a great step forward in understanding the geometry of nature, but don't be beguiled into believing it to be the end of the journey. (Anyway, what could "realism" mean in music? Using the call of the cuckoo in music representing "Forest Murmurs"?) -- Martin Taylor {allegra,linus,ihnp4,floyd,ubc-vision}!utzoo!dciem!mmt {uw-beaver,qucis,watmath}!utcsri!dciem!mmt
dgary@ecsvax.UUCP (D Gary Grady) (05/20/85)
All this discussion of the accuracy of fractals in modelling landscapes
reminds me of something I saw sometime ago in a graphics journal that
quoted a Lucasfilm chap saying that fractals were "obsolete" in computer
graphics (or something like that). They had supposedly come up with
something "better" called "graphtals." Anybody familiar with this?
--
D Gary Grady
Duke U Comp Center, Durham, NC 27706
(919) 684-3695
USENET: {seismo,decvax,ihnp4,akgua,etc.}!mcnc!ecsvax!dgarybrian@digi-g.UUCP (Merlyn Leroy) (05/21/85)
>I didn't mean to imply that the [music generating] program would be totally >random, but there would have to be a tiny bit somewhere, else the input would >completely determine the output. (Unless there is a new concept around) > >Darin Johnson Just being deterministic doesn't mean the output would be predictable. Humans might be deterministic (see net.philosophy). Since music quality is a matter of opinion, I won't try to predict if a computer can produce "good" music. Suffice to say it will be better than Michael Jackson. Merlyn Leroy
dbkay@dartvax.UUCP (David B. Kay) (05/23/85)
> Since music quality > is a matter of opinion, I won't try to predict if a computer can produce > "good" music. Suffice to say it will be better than Michael Jackson. > > Merlyn Leroy I don't know. I wrote a program once (a quick and short little beast) that wrote fugues. Well, things in the form of fugues, I suppose I should qualify, as it didn't really work with a theme. It worked by generating melodies using a random number generator coupled with given probabilities for the existance of certain intervals. Once it had its four melodies, it checked to see if they were compatable by (a) generating all of the possible pairs of notes from the four, (b) looking up each in a table that would tell the proabability that such a harmony would be accepted -- set high for consonant stuff, and lower for discordant things, and, if a melody line failed, it went back to the beginning and recalculated it. The program came to a stop when it the major chord of the key...C, in the program's case, for simplicity. I'm hardly one to make the judgement, but I really liked the peice that was generated. It's definitely better than anything I've written, and I think it really has some merit. Members of the music department here were taken aback that it had been written by a computer after they heard it; they had been told that it was a student composition, and they remarked that it was 'acceptable,' or 'passing minimalism.' I was really impressed. Kind of a small first step in a Turing test... Anyhow, the point of all of this rambling is that a computer already can compose decent music, (certainly more complex than MJ) and that my approach, which combines determaninsm with randomness, seems to bear fruit. DB Kay HB 1660 CSNet: dbkay@dartmouth Dartmouth College Hanover, NH 03755 "Look on my works, oh ye mighty, and despair. I know I do..."
bob@cadovax.UUCP (Bob "Kat" Kaplan) (05/30/85)
This discussion does not belong in net.books. Please restrict it to net.ai. -- Bob Kaplan "To be completely safe is to be completely dead."