random thoughts

Thoughts for the day:
Assuming that neural recognition follows an associative model (key->value pairs), does human memory loss due to non-use more closely resemble magnetic drift on a hard disk platter, or discharge of a capacitive cell such as DRAM?


When listening to an improvisational piece and trying to transcribe the melody or changes/modes, I find myself applying the same techniques I use when analyzing binary-only software. First I try to get an overall feel for what is going on (the Zen aspect). Then I start to dig into the minutia (in the code: watching variables, in the music: determining directions of transitions, listening for intervals). Sometimes I will hit a part that defies intuition in both cases, and yet after spending enough time with it, even if all of the details of the passage or code segment have not been worked out to my liking, I am able to form a theory of what is happening. Usually that theory is good enough to garner what feels like a deeper understanding of the piece, even if I'm not sure it is what the author intended when he conceived the idea that ended up as that section.

In some ways the problems are similar; you are taking a work which has already been transformed from its editable “source code” to the form that serves the most useful end. A difference is that source code to binary form can be thought of as a lossy conversion, because all the comments, structure, and conventions of the source code has been utterly lost; while in music, notes to performance is usually an embellishing or improvising process, where the musician actually adds to what is written on the page.

But perhaps a program binary can be thought of as a performance by the compiler – taking the “notes” the programmer gave it, it uses knowledge about the desired result that only it possesses in order to provide a suitable result. In the compiler's case, this is knowledge about the machine language and pipeline architecture of the target to produce the highest performing binary. In the musician's case, this is knowledge about the styles and rhythms that will provoke the widest range of emotions in the audience.

This has interesting implications for copyright law, since a musician can copyright a performance separately from the composition. Should a compiler author be able to assert copyright over a binary produced/”performed” by his compiler? The same mechanisms seem to be at work in both cases.

Or is the compiler simply the instrument, and the person doing the compiling the performer? The compiler has intimate knowledge of how to generate the end result, just as an instrument has intrinsic sound generation properties. But the person doing the compiling also has intimate knowledge of the compiler's characteristics and capabilities, and how to make best use of the compiler for a given piece of code; just like a musician has intimate knowledge about the capabilities of his instrument and knows how to use it for creating music. What if an automated process invokes the compile? What about machine composition? Can it be copyrighted as a performance?

It would almost seem that any non-lossy transformation could be categorized as a performance.

Leave a Reply