I know from a few "behind-the-scenes" documentations that most of the composers use loads of reverb while composing.
I think you can compose without reverb. I often add reverb in the mastering stage, which gives the piece a lot more realism.
If you know how to compose, you don't need reverb to write. No amount of ambience is going to make you a better composer/orchestrator; what it will do is help you mix better, but that comes later anyway.
I use the same reverbs when composing at the sequencer as I want for the final product. I don't have the time in my schedule to make what I'm "composing" any less quality than what the final product should be. Often what started out as "just noodling" became the final cue. Of course some reverb tweaking may happen by the end, but essentially it is what I started with. I also have a template setup with this for my orchestral sequencing. Before I had a template I still used a hall reverb on the master bus - at minimum - while composing.
IMO the MIDI orchestra is acting as a synergistic group, not as individual entities. Without interaction with the room the sum will not be the same. I find that I will perform a line differently with reverb than without. While using reverb I feel I get better note transistions. I could see where soloing an instrument without reverb to edit may help in some cases, but there is also a point of diminishing returns concerning editing. I can become too focused on editing minutiae (3rd oboe in a tutti) that will just be masked anyway. I could also see having to potentially remix more after adding reverb. Adding reverb is going to change how an instrument sits in the mix.
For me, I like to perform the line, fix any "fat finger" mistakes, and move on. A good reverb is the MIDI orchestra's friend. It should work with the instruments, not just cover up weak spots.
While composing: I composer with a pencil and paper. If the lights go out I'm fine. Except when working with drums/loops/bass etc...
While squencing: Right now I'm using EW products, which come with reverb out of the box. It would be a waste of time and rather foolish of me to take out the natural tails (reverb) from EW (btw Gold), in order to work dry! But I don't care much, as I have the score next to me, and know pretty much what I'm expecting to hear...
A while back I was using different samples on a studio, and then I was working completly dry, exactly for the reason not to be fooled by the reverb. Later I added 2-3 different "places" of reverb in differnet sections of the orchestra (violins, a bit less, woods a bit more, and a bit more for the perc and brass... anything solo would take even less to bring it in the front). That kind of thing...
Of course all the above apply since I don't even have keyboard in my studio, and thus I'm importing midi files from the scores I work in Finale, into Cubase...
Otherwise while playing I would expect one to have reverb on, in order to take the final feeling... :-/
I tend to always sequence with a temporary reverb, something close to what I will use at the end, but less power hungry. I do this because with samples, (since in your mockup you're not writing for actual instruments but many combinations of samples -- perhaps many combined which will create the illusion of a a single instrument line) the way the samples react with reverb has everything to do with which samples you use to create that illusion.
I work almost exclusively with MIDI, and orchestral emulations in particular. I never have reverb going while I'm working on a piece. I want to hear how the various lines are really interacting, what the specific note choices are doing for the work, and reverb isn't in that equation for me.
Reverb to me is a production tool for when I'm mixing and working towards a final 2 track. At that point I've shifted my focus to making the composition into a performance, with venue emulation, which includes the distance of each instrument from the listener, and that of course involves the use of reverb.
I know singers often want to hear some reverb in the cans when they do vocal dubs--it's freeing to them to feel like they're in a performance space, rather than a cramped studio space. People using reverb while composing and putting together MIDI tracks are probably wanting that same psychological boost, and as such, there's nothing wrong with it. It just hasn't really ever occured to me to listen "wet" as I'm writing.