At 11:28 18/03/97 -0800, you wrote:

>No one seems to be mentioning that MIDI is exactly what finalized the
>devision between the "Academic Curiosities" and the pop world in the
>FIRST PLACE. Lest we forget...
>

Well, a lot of that division is self-inflicted as far as the academic
community is concerned. 

Csound is supposed to be flexible. But consider how easy it is to use a
MIDI sequencer to put a score together instead of typing it in line by line.

You get:

control over individual events
block moves of events
block transposition of events
block modifications of events
more continuous control channels than most people will ever need 
(and these can either be edited graphically, or typed in by hand if you
prefer)
fine control of tempo changes
'logical' editing functions, which let you change specific classes of
events in specific ways

and so on.

All you have to do is to define instruments that respond to these different
streams of information. Definitions of tonality and/or pitch are entirely
down to you. If you treat the MIDI events in the abstract, the same way you
would treat p-fields, then you get ease of use *and* flexibility.

Yes, there are limitations, but they are relatively trivial compared to the
time saved. And in practice there won't be that many situations where
you'll find they are a problem. Some for sure - but most, or even many? No.

Now, of course you can create software to write a score file, but how much
flexibility does that really give you in comparison? All you're really
doing is exploring different parameter positioning algorithms. Whether you
use neural nets, stochastic techniques, genetic algorthms, chaotic systems,
or whatever, this does not, on the whole, seem to make for interesting
music. (Although as an intellectual exercise, akin to solving chess
problems, it may have its attractions.) 

The irony is that for all of Csound's supposed flexibility, its limitations
- the lack of dynamic wavetables, a-rate conditionals, a-rate loops,
instrument 'chunking' and nesting, the fiendishly obsessive score structure
- make it a hugely clumsy and slow way to make interesting sounds. 

People attack GM's cheesiness, but the sounds that most people seem to get
out of Csound are cheesier still. (There are one or two exceptions to this,
but not many.) In theory you can do anything you like with it. In practice,
unless you have limitless amounts of time to experiment, or own an SGI, or
are only after some squeaky sound effects, it's a rather blunt and
unresponsive implement. Good for teaching, maybe. But not so good for
genuinely original music. (If not, how come so much of the music made with
it sounds so similar?)

And GM? Well, yes, it would be silly to include GM ugens when it's so easy
to create a GM orc set and pass that round the net. It would be much more
interesting to concentrate on Csound's shortcomings and bring it kicking
and screaming out of the constraining mainframe-minded 60s mindset it
inhabits now into something closer to the present. 

Really, adding new ugens is only papering over the cracks. (Although I have
to admit that Richard Karpen's new pvoc ugens are a *lot* of fun.) IMO a
look at how to overcome the limitations in the control structures and the
table system would be a very fruitful path to follow at this point.

R.