Hi David
If we lived in a ubiquitous OSC world you'd have a point, but we do not. Not only do we not have general OSC use, but many of the existing OSC implementations really just ship MIDI over it, with a few higher resolution control signals attached. This is simply not sufficient if one wants to create an instrument that can compete with the Violin, Guitar or Saxophone for emotional expression. And sending this data effectively over existing MIDI is simply not possible.
It is important to remember that OSC is just a network transport definition really - there is not yet a broadly accepted schema for sending highly expressive notes. I can tell you from our internal struggles with this (we use a system not dissimilar to OSC for our own inter-agent transport, hence my comments previously about a migration to OSC from this before we open it out), it is not trivial to get right. MIDI makes some powerful simplifying assumptions about what constitutes a musical note. These assumptions make writing software for it a whole lot easier, but those same assumptions also cripple it in many important ways. And we have now seen compelling evidence that applications written around MIDI are really going to struggle to migrate to a richer event model, one that can accommodate true expression. The absence of a single OSC DAW illustrates this point - even a 'MIDI over OSC' one. They will manage it, but it's going to take a while.
If you are happy with the level of expression offered by MIDI than all of this is of course irrelevant. Personally I am not, and the main reason that EigenD does all the things it does is because it does these things with very high data rates, with a smarter idea of what a note is. This turns out to be quite hard, and as far as I am aware nothing else quite does what it does as yet, with the possible exception of MAX/MSP, which is now also doing OSC.
You say that you can't see why we can't just send the raw data from the controller to existing applications. This is quite simple. There is not one application than can use it effectively. If we send the full data steam from our keys to any of the existing DAW's out there they just fall over, even with small numbers of keys down an at the reduced resolution MIDI offers. Our MIDI gateway throws a LOT of data away. Things are a little better when we host the AU/VST but we have to throw data away even then. It is possible to play expressively using MIDI, but not as expressively as one should be able to by a long way.
I have lost count of the times over the last few years where I have wished for a world in which all software instruments and DAW's were able to do this. It would have saved us a huge amount of work. But sadly, it is not the case. But we can hope - I think that OSC does potentially represent a new beginning. It does need some things that it doesn't have right now, like idiomatic bi-directionality, decent discovery to enable zero conf abilities, a better canonical note model than MIDI, but all these things potentially represent backwards compatible idioms within the current protocol (not incompatible extensions, which is what I think you were worried about), which is very promising.
Sorry, long ramble. I must have answered this question a hundred times now - many people have asked it and as yet I have yet to hear a compelling reason why we're on the wrong track with where we're headed.
John
UK
