The John Cageblog

First, some useless knowledge: John Cage is the 116th most-recorded composer according to Arkivmusic’s count of recordings.  That puts him just above Alessandro Scarlatti and ready to surpass John Phillip Sousa at any moment!  This entertains me far too much.  I won’t be satisfied until he moves up 35 places and unseats CPE Bach! :-D

Some of you may wonder about that “CageMap” link to your left.  My original purpose in designing it was to plot out the relationships between different Cage works on an axis where x refers to the degree to which performance of a work is indeterminate and y refers to the degree to which chance operations was used in the composition process.  I wanted to expand it by color coding based on instrumentation, but I was unable to find an easy way to make those colors work.

So this evening I hit upon another idea.  I dug up my old program from a pattern recognition class that does k-means clustering on n-dimensional data fed into it.  I thought it would be fun to see how an automatic algorithm groups Cage’s output.  As a test, I placed the first 50 works (alphabetically from my index) into a spreadsheet, and rated all 50 on 8 attributes, referring to the indeterminacy and chance operations used (as mentioned) and use of instrument categories: voice, keyboard, strings, winds, percussion and electronics.  The numbers are somewhat arbitrary (for example, solo piano works had a 2 for keyboard and 0 for all else, but prepared piano has a 2 for piano and 2 for percussion).  It is moderately ad-hoc, but I thought the results were sufficiently promising to mention here.  

After some testing with variable numbers of clusters, I found that 8 groups produces an interesting result, in that the groupings in my mind seem fairly intuitive.  I have given each group a category name and listed the works included in each of them below:  I’ve put asterisks by the names of groupings that I think will be subsumed into larger groups once all the works are added to the clustering program.  

Vocal Music
Alphabet, Apartment House 1776, Aria, ear for EAR

Traditional (more or less) Piano Music
Ad Lib, Cheap Imitation for piano, Chess Pieces, Crete, Dad, Dream, Three Easy Pieces,

Wind Ensemble Music (???)*
Eight, Composition for Three Voices

Percussion Music
Amores, And the Earth Shall Bear Again, A Chant With Claps, The City Wears a Slouch Hat, First Construction, Second Construction, Third Construction, Credo in US, Three Dances, Four Dances, Daughters of the Lonesome Isle, Double Music, Bacchanale, A Book of Music

Orchestral Music
Atlas Eclipticalis, Cheap Imitation for orchestra, Concerto for Piano and Orchestra, Concert for Prepared Piano and Orchestra, Dance/4 Orchestras, Sixteen Dances

Indeterminate Instrumental Music*
The Beatles 1962-1970, Composed Improvisations, ASLSP, Cheap Imitation for violin, Chorals

Indeterminate Performances*
Alla ricarco della silenzio perduto, Branches, But What About the Noise..., Child of Tree

Indeterminate Music with Electronics
Bird Cage, Cartridge Music, A Dip in the Lake (all parts), Electronic Music for Piano

Now, I think if I were to do this for all 246 works in my listing, I will have something akin to the Cage Map (but it may not be quite as pretty) and a thematic catalog to boot.  Cage would probably hate me for it though ;-)  Still, for me I find it easy to think about Cage’s early work (he did serial or serial-like music and worked with percussion ensembles, then invented the prepared piano, then moved into chance compositions and “idea pieces” that became progressively more indeterminate by the late 60s).  But I have a much poorer grasp of Cage’s output after 1970.  Why is this?  Well, the popular discussions in encyclopedias and such ignore post-1970 Cage.  Even his biography, The Roaring Silence, seems to degenerate into a tiring list of Cage’s itinerary during the late 70s, 80s and 90s.  Also, a fair amount of his 80s output is poorly documented and/or unrecorded.  I feel like the threads are there to trace but I am just not seeing them.

I’ll work on this a little more tomorrow evening I hope.  Fortunately for me, this sort of music clustering is directly related to my thesis (except in the thesis I am generating the characteristics from audio files and using a better algorithm) so it’s not quite as useless as my glassy-eyed readers might be thinking ;-)