While the act of musical composition has historically been a chore to define (the classical image of the white-maned recluse hunched over his desk has always been dubious at best), the invention and proliferation of sound recording technology in the 20th century, and its continued development in the 21st, has done much to raise deeper questions about the relationship between composition, improvisation, and performance.
What do we mean when we call something a composition? Certainly, few would be so bold as to assert that this term carries with it implications regarding style, depth, or complexity, but I think that it’s fair to say that this term often conjures images of musical notation, dusty scores, or some other form of written documentation. While it will be obvious to most contemporary musicians that compositions do not necessarily need to be notated in order to be organized and constructed with purpose, I think the idea of documentation plays an interesting role in this particular contemplation.
The idea behind notation is that it removes musical ideas from the linear arrow of time (well, sort of) and freezes them in some semi-permanent state so that they become repeatable and can be spread more easily (and intact) through geographical space. Sound recording, then, possesses the same core functionality as notation, but raises even more questions. What happens when a performance (especially one by the original composer), with all its nuances, microscopic (or grand) moments of improvisation, and incidental sonic happenings, is documented on a recording? All these things that were seemingly incidental and fleeting then become permanent fixtures in the musical work and possess the same sense of permanence as any written notation. Is it then the reducible and previously-constructed elements of the performance – harmonic progressions, melody, rhythm, etc. – that function as the composition and the nuance, happenstance, and improvisation that function as the performance?
In live music, the line seems slightly easier to draw (though I’m sure there are many who would argue against this), but when a work is recorded and becomes a musical document, it requires different considerations. What happens when a great improvisation is recorded and preserved? While many of the musical ideas may have been inspired in and carried out in the moment, upon being recorded they become as permanent as any notated score. (Consider how many Miles Davis, John Coltrane, or Jimi Hendrix solos have essentially become etudes for young aspiring musicians.) Moving forward, does it then function, culturally, as a composition regardless of the original intent? Does the original intent negate the consequences of the sound recording? What does it mean if the musician was aware that the improvisation was being recorded? Of course, I do not intend to say that improvisations are not, in some form or another, organized and purposeful. Rather, I think this raises interesting questions about the artistic consequences of musical documentation and whether a musical work can be defined clearly as a composition, performance, or improvisation based solely on its intended function at the moment of conception.
Hearing Is Not Listening 
These are great – and complicated – questions. I appreciate the work that you do here in drawing notation and recording together as two types of musical document. There’s a lot of cultural inertia for us to overcome in order to pull these terms out of the hierarchical relationship that they have long shared. The solitary composer’s scribblings certainly seem more intentional, important, and *serious* than recordings. That distinction continues to break down.
An important wrinkle to add here: we know that scores are inherently incomplete. It’s impossible to notate all aspects of sound, and it’s impossible for a score to generate an “ideal” realization. By the same token, recordings are never neutral. Something is always lost in the process of translating vibration to electricity and back (with a long signal chain in between), meaning that recordings are inherently biased and also incomplete. Sometimes this biasing is intentional, and sometimes the “capture” simply doesn’t provide a complete analogue of the experience. In this way, notation and recording are even more alike…