I value the way that computer music lends itself to being transmitted through time as more than just audio recordings. The programs that make up a computer-based work are more durable. More like text, like scores.
An audio recording goes out into the world and people listen to it until people stop caring about it enough that it degrades or gets lost. It's rare that a recording by itself is compelling enough that people spend their time transcribing it or or otherwise doing their own versions of it.
A score, though, has this kind of engagement as its premise. I think it offers richer possibilities for connecting and generating meaning, especially in a musical tradition that's so based on structural aspects of sound and timbre.
[EDIT] To clarify... this may seem like a weird criterion for valuing computer-based work, because 99% of it goes out into the world just like anything else, as audio recordings. But, computers offer the possibility of getting pieces out in some more semantic way, and I want to promote that idea.
I more or less got this concern started in my head when helping put on a concert of a Luigi Nono work, turning the score into Max/MSP patches.
Magnus Bugge just released a new EP on Bandcamp—his Aalto patches are available as part of the download. Hard to imagine doing this with analog modulars!