[eu-gene] Psst...Wanna Buy an Algorithm? (Working Title)

[alex]

On Sat, 2005-10-15 at 11:23 +0100, nick collins wrote:
> Excuse the academic style in the below, I can't help myself these 
> days...

It's a pleasure as always :)

> One of the things that originally appealed to me about algorithmic 
> music was the capacity to provide refreshing stimulus, within a given 
> style. This chimes well with the reasons performers improvise to keep 
> renewing their idiom (Bailey 1980). But also, subtle, controlled 
> variation is useful to avoid the fixed repetitions that otherwise lose 
> their vitality and promote switched off listening states. 

But in order to build a recording, the composer has to listen to the
same thing over and over again.  It would seem to me that in this manner
a good composer will naturally find repetitive structures that can stand
repeat listens.  It's often said that a good recording 'grows' on you,
at first it sounds dull, but warms with repeated listens.

Also, there is a lot of music.  There isn't the need to have
compositions that dynamically play variations of themselves, when we
could listen to something new, once our ears tire of a particular
musical structure.

> (Interestingly, performers often dislike recordings because they are 
> forced to settle on one eternal immutable performance, and the 
> associated stress can lead to extensive dishonest multiple takes, 
> post-production editing etc- its all very career oriented in a music 
> industry set up around the fixed product).

Sure, but I think performers are happy with live recordings when things
go well, and I don't think it's dishonest to build a recording in the
way you describe, a studio recording is completely different from a live
recording, I think people understand that.

> I have also argued before though that coarse 'kind of' relations are 
> enough for some listeners to judge music, and thus that it is 
> incredibly hard to build algorithmic works that really do shift their 
> idiom sufficiently between runs to keep surprising you.

Yes, it's still the same composition that you're listening to.

> But I think I'd 
> still say there is a lot of healthy mileage in algorithmic music (even 
> without any external interaction) that I prefer to recorded works, they 
> are like 'live recordings', if only that term wasn't already 
> appropriated...

But isn't algorithmic music without any external interaction equivalent
to a recording?  Since each time you run it, it produces exactly the
same music?

alex