Obviously, Brian and I see many things (though not all) rather differently. This is partly because he has much more detailed knowledge of DAWS than I do and knows how to get the best out of them but it may also simply be that we are looking for rather different things from our music. First of all is the actual process of composing where DaddyO really hits the nail on the head. My ability to compose immediately took a considerable leap forward the minute I could use notation software to do it because I could see visually how the individual lines flowed in and out so it became much easier to write contrapuntally in a semi-competent manner. It’s possible to see immediately in an orchestral setting how various instrument combinations can sound together, assuming a decent VST.
On mixing being 90% of the battle, I see no evidence for this with the best modern libraries. With Sibelius, I did use Cubase’s mixer for modest adjustments to sound but as Dorico has the main essentials built in, this is barely necessary any more. How to isolate things or best place in the field has far more to do with good orchestration and a decent supplied acoustic – here the VSL Synchron stage player is close to outstanding in my view. As for shaping sounds, this is mainly down to the recorded samples. Of course if you’re using Halion or other basic libraries which do little shaping then it’s pretty difficult but why make things hard for yourself?
Pitch and rubato are two other things which have quite correctly been mentioned. Advanced libraries have random pitch imperfections built in so why would we want to waste time creating them with a DAW unless there are very specific requirements? Rubato has been supported in notation software packages for quite a while. Ok, the quality is open to debate but will continue to improve, particularly when the tools can be applied to selected passages. As two live performances are never alike, then why should computer ones be? Where tempo changes are explicitly required, they can of course be marked and exactly calibrated in the score.
On this March Trio thing, we’re talking about a) dynamics and b) patch change if we really want to completely change an attack. Dorico’s Note Length feature could easily be programmed to use a particular switch for notes of a certain length. Assuming you really want to use a specific cresc/dim combination for different notes (which to me sounds rather crude) then you can make it in one and then duplicate it in the automation lane. However, you can’t do all the parts at once and Note Length programming cannot yet be used for selected sections only. So yes, of course there are many things which can still be more flexible and faster with a DAW even within my sort of composing environment-- I doubt anyone disputes that!
I would agree that the majority of scores created by composers don’t sound that great. In general this is a mixture of lack of expertise in the composer of how to use the libraries or simply using ones which can’t by their nature sound decent in the first place. However, we sometimes forget that only first class musicians can really make music come to life as well – others can be struggling to play the piece at all.