There are two main ways I work, both of which have their upsides and downsides. The first way is relatively quick, uninterrupted, and takes place over a single session. If work on the music began around 10am, I might be done by 2pm. Not done done, but having decided upon most of the piece’s main sounds and structures. The upsides of this way of working are:
it’s a continuous flow state,
it’s a way to commit to decisions now, not put them off,
because of the flow and quick decision-making, one idea often quickly leads to another,
it’s exciting!
The downside of this way of working is that it feels frantic because there’s a self-imposed urgency to
finish something right now
find the “right” sound right now
understand where the music is going right now, which often leads to
following conventions of musical form (e.g. intro theme, development, surprise ending, etc.) that aren’t necessary.
The second way I work is slow, interrupted, and takes place over multiple sessions. This way of working isn’t as exciting as the first way, but it’s more objective in the sense that I have space to consider details that I ignored while working quickly. For example, maybe I want to work in a series of volume automations to parts in a piece. This requires a global view of the the music in its entirety and some patience to carefully draw in the volume changes just so. On a recent project, this task took me about a week. Maybe I could have done it in a day, but frankly an hour or so was all I could muster before I wanted to move on to something else. Such editing tasks have to be done and I’ll do them, but I limit my exposure because I’m biased towards the exciting.
My ways of working quickly-in-a-single-session and slowly-over-time roughly correspond to what the psychologist Daniel Kahneman describes as System 1 and System 2 types of thinking in his book Thinking, Fast and Slow. System 1 is the intuitive type that jumps to conclusions based on limited evidence, while System 2 is the deliberate type that proceeds by cautious reasoning. Kahneman offers dozens of case studies to illustrate the shortcomings of System 1 thinking, from exaggerating the coherence of what we hear, focusing and cognitive illusions, to the limitations of the “insider’s view” and the sunk-cost fallacy. We have, says Kahneman, summing up our proneness to error, an “almost unlimited ability to ignore our ignorance” (201).
With System 1 and System 2 thinking in mind, I alternate between working quickly and working slowly. Working quickly is my preferred production tempo, because it feels intuitive, it’s uninterrupted, and its results often surprise me. But I always revisit this work the next day, the next month, or the next year. If the music felt so exciting then, let’s reassess it now from the perspective of System 2 thinking. Take your time with the music, finesse its details, and make it better.
brettworks 
As you describe these two approaches, I think of the way Miles Davis choose to work for about half of his career. The recordings we as listeners know of his begin as spontaneous playing, and he seemed to work more at choosing a variety of musicians than on instructing them directly/explicitly on what he wanted them to play beforehand. Then he largely entrusted others to chop, edit, and assemble the resulting recordings into pieces or suites that would become tracks on the albums.
In a sense you are playing both roles. As an in the box player + bandleader you are selecting the sounds and playing with them, and then the later role is to assemble the structures that the pieces imply or later imagination constructs.
Wishing you a productive new year!
That’s a great point about Miles choosing his personnel “palette” and then letting them–or engineer Ted Macero–do their thing. And yes, the producer selects, plays, and orchestrates/arranges sounds to conjure some kind of performance or enchantment that is either realistic (acoustic) or else fantastical enough so that the listener doesn’t really care how it was constructed. In an ideal world, the music would come together like Miles’ band doing it in one take, but that’s probably rare. It reminds me of some things I touch on in my book: the idea of the producer approximating the dynamics of a band without conventions (musical role swapping: bass player becomes the drummer, etc.), and also the idea of the producer negotiating the uncanny valley so that the music doesn’t become a grotesque approximation of a live ensemble. And this–being a player/bandleader/structure assembler–requires time, I think, and also trust that process will reveal routes forward. Thanks for your comments!