(To tune of "I read the news today...")
Actually, I didn't sell them. I put them up on eBay, and I'm betting I'll have to relist at least once, and at the end of it I'll still own one or two of them.
I had a pretty good rack going at one point. All Roland rack-mount synths, topped by a Roland sampler keyboard. Plus one Korg piano module (the venerable P3). It was a bit of a maze of cables, what with MIDI daisy-chains and all the audio connections to my Mackie mixer and a couple extra insert reverbs. Even more mess when you add the Octopad and pedals.
It took a bit of time in OMS and other applications, too, getting all the patch names entered and the routing organized so everything showed up correctly. And adjusting the internal reverb and chorus setting during a song remained a bit of a pain, since that was all System Exclusive Message stuff.
But, basically, once the rack was set up, I had most of the instruments I needed right at my fingertips.
This is the advantage, still, of hardware synths; that you can audition patches in a fraction of the time. No waiting for stuff to stream off the hard disk into RAM.
And there was a subtler artistic advantage as well. Back in the rack days, megabytes of patch memory was considered big. This meant, generally, fewer samples spread further. And shorter samples as well, looped instead of left to play out the full development of the tone. So realism was less. But playability was higher.
So compare this to a real instrumentalist. They can put lots of character into their tone and inflection. This is what makes real instruments, well, real. But at the same time a skilled player can modulate her tone to blend with the instruments around her (or to seat correctly in the band/orchestral mix).
Whereas a synth doesn't have this skill. It plays the same kind of note all the time (unless you, the keyboardist/programmer/arranger, alter it). Which can be done, certainly, but takes more time. And often more CPU cycles. So the simpler, pared-down sounds actually made you more productive. The music was a bit less realistic but you could write it much faster -- as fast, really, as you could write for humans.
Oddly enough, my old rack was actually superior in polyphony as well. A couple of reasons. First is that my present computers aren't particularly fast. Another is that patch size -- and the necessary CPU cycles for data handling -- has gone up by magnitudes. A minor point is also that sequencing on MIDI hardware promotes patch-changes. Software synth is almost always done with new instances for each new instrument, even if it will only play a single note.
I did push the polyphony on my rack, of course. One always does. I was doing orchestral stuff on it and I was doing it classically; instead of playing block chords into a "String Orchestra" patch, I'd play basically monophonic lines (and some double-stops) for each part; First Violins in two or more divisi, Second Violins, Violas, 'Cellos, Double Basses. So if a section had two simultaneous notes, I'd have two different instances of a patch that sounded like less than the full section, and each would play monophonically.
And I was using fingered tremolo as well!
(One nice thing about patch switching is if I had some of the strings switch to pizz, I'd actually change patches on that MIDI channel and that dedicated sequencer track. Which added to the realism; you couldn't just have a couple of bursts of pizz in the middle of an arco section, but instead, you had to think about what the real players were doing. Same for, say, the trumpets picking up their mutes.)
A composer friend of mine settled on his own hardware rack (organized around a D50) several years back. He's modified it a little over the years but the huge production advantage to him has been that he knows the sounds and can go to them without a lot of time wasted auditioning patches (or worse yet, editing them!)
My directions were always different. I would hear the sound first in my head, and do what it took to approximate it with the tools I had. I did go back to a small collection of favorite patches over and over, but I would also do things like edit a D-110 handclap sample to make it higher pitched. Or route a flute to one of the special outputs so I could run it through a hardware reverb unit.
Going into software synthesis allowed me to move towards writing for smaller ensembles with much more control over the precise tone. When I do software synth now, most of the outputs are routed through various effects processors as well as contouring EQ and the like.
But it is much slower. And I miss those days of just being able to dial up my usual set of strings (I even saved this as a blank sequence) and start writing.
With the rack dismantled, many of the pieces I wrote in the past only exist now as audio files. It is too much work to try to adapt what I could do with the patches in the rack to what I have available on my (somewhat thinner and very much more eclectic) virtual rack. I have a few of the old ones in a folder on my DropBox right now.
But, really, I took up arranging music largely to get used to mixing, and now I have plenty of opportunities to mix actual live musicians. So I don't do anywhere near as much music, and most of that is for specific designs.