Dear Stéphane,
I understand that you don't wish always to refine a piece but go on to the next project. But I think you may be missing an opportunity. Unless one is going to be rigorous like Pierre Barbaud and regard the computed output as the final statement, I believe the result of an algorithmic process does need a kind of critical 'massaging'. It's not just the notation but reacting to the reference recording / output. Even if you never imagine your work will reach live performance, or you prefer (as some composers do) performance from a computer driving a sampled instrument so every nuance can be under the composers control, l find that taking the piece into notation (and this could be a graphic display) and away from the tyranny of realtime playback allows you to critically reflect on what might be improved. The code is never going to construct the perfect output because it's difficult to connect the process with those areas of personal experience and intuition that inform composition. You and I both have prior experience through listening and study of the two-piano medium - Mozart's D major sonata, Messiaen's Vision de L'Amen, Bartok's Sonata, Stockhausen's Mantra, Debussy Blanc en Noir, Stravinsky's Concerto, Ligeti 3 Pieces. Although we don't think of these pieces necessarily when we start composing they are part of the history of this medium and I'm sure we make unconscious creative decisions based on this prior knowledge - even, for example, if we think we will in a particular piece avoid what Stravinsky does (overdoes) with scales. I'm very aware that the output of a compiled program so often needs adjusting in regard to register placement of pitches. Many such adjustments lie beyond sensible programming . . . and this is why having yards of OMN as part of a program is not always helpful . . . and you, to your credit do avoid that - so the argument of the composition can be 'read' and enjoyed. I really congratulate you on this. It's my goal too. The code is a stage towards a final result and can rarely 'be' the final result. I think one learns so much - for future projects - from being critical and using one's ear and 'craft' making adjustments.
Nigel