Apologies for the very late reply on this. Things move pretty quickly here (as well as in the modular universe), so consider this the nonlinear narrative portion of the thread. This is a thread I’ve thought a lot about and thanks to all who contributed.
@papernoise, you brought up a number of really great points, and there are a few I’m using as a jumping off point.
In terms of playable vs an attended self-playing instrument, I think it comes down to control. How many levels of control does the designer of the instrument want? For me personally, certain modules inability to control everything consistently with precise repeatability (according to spell check, that’s a word) is part of the reason why I love it.
As a drummer and percussionist, I’m always thinking of my instrument as a collection of sounds, so yes, a collection of multiple voices can still be called an instrument. From a classical percussion ensemble to an improvising noise band, I’ll consider the role I anticipate playing (rhythmic, melodic, soloist, embellishment, noises, etc) and bring a collection of elements that suit those needs the best. Sometimes things line up perfect and other times I need to turn my floor tom into a melodic underpinning for a sparse ambient drone. Different voices from my instrument.
Modular is mostly an open system and even a collection of things can represent an identity (or maybe an instrument with a split personality). By design, modules from various makers/designers are capable of being used together. It’s really what the musician or performer brings to it (ie what is patched where) that helps describe the identity of the instrument.
Even if you have a system made up of a single manufacturer, you are interacting with a vast array of ideas from many different people.
Modular brings to the forefront the element of powerful control over sound. Your view of control (How much? By whom? In what contexts?) influences the modules you include, which ones you prefer, and the way they are incorporated.