Warning: I've had a lot of thoughts on this matter. I've tried to edit it as to be intelligible, but it's late and I've had a long day. Please forgive any excessive rambling. ;P
I too have been sorely disappointed by many drum-synths/sequencers. I think Reason is probably the most well-organized drum programming platform ever. Unfortunately, everything that Reason can produce ultimately sounds like it was made in Reason which to my ears is not spectacular.
Those are my opinions and are purely a matter of taste, but it illustrates the major problem of most drum synths or tools in general. Because of the current cut-throat audio market most devices are usually fixed in some way, or are by definition a big bucket of parts which ends up excluding newbs or those who have to make music on strict deadlines, or otherwise want to be able to throw a range of drum tones together very quickly.
The Scope platform is not immune from this, but because it is in itself semi-modular, and because of the way the Modular platform sits on top of the parent OS, and because the DevKit is about to be more available to the common (wo)man it allows for a slightly more open and egalitarian option. One that I think would (if properly advertised to the outside world) could help the Scope Platform as a whole, and be in itself far more competitive against other similar products.
My thoughts on what I think would be the best drum synth ever would be to not make it one end all be all device, but to focus initially on more devices in the vein of KickMe, but instead of making them .dev's, make them individual modules. That way, at the very least, people can pick them up cheap as little $10-#30 (or whatever) additions to the Module library, or buy the grand bundles.
But it also would allow a drum PLATFORM to develop.
To continue evolving the idea, it could lead to the development of semi-modular shells (or kits if you will). So one could choose from different workflow designs, or perhaps even within each kit/shell design presetable options allowing for variations in appearance or signal flow.
In truth, it's not much more than extending the original Red Dwarf idea, but here it would be more specifically geared towards rhythmic construction and manipulation instead of the typical virtual-analog, and being able to slap your drum/sample tones and effects into a predictable signal chain so you can be making beats in under a minute or two.
Another way of thinking about it would be to imagine if our old friend Angus at fxpansion had built DR-008, not as it's own platform where he would have to woo other developers to develop modules for his creation, but instead built DR-008 to piggyback on top of Reaktor, opening up development both the heavier coding of VST programming, but also the quick and dirty visual programming of Reaktor, for which there is already an immense user base.
So, if we imagine that this theoretical playground got some footing, someone could potentially pick a workflow shell, load their drumsynths (or modular sample players, or something yet unimagined) into the various banks/channels, toss in a few Red-Dwarf and/or standard inserts, allowing people to use their ModII/III/Flexor filters, delays, timestrechers, whatever, over a single drum, the entire kit, whatever blows one's skirt up. And voila! You've got a custom kit which is usable.
This concept leaves room for any number of new sequencer/matrix/groove-sculpting designs to be implemented, while leaving current ones valid and useable where appropriate. Swap in whatever is most suitable for what you're doing.
The most important idea is that it stay Modular at every level, so that it can extend beyond drums, but new, flexible, ways of building sounds of any kind. No one is left unhappy because their prize feature isn't in the otherwise perfect device, and no developers have to feel like it's some gargantuan project which would require an unfathomable dedication of time and energy.
JB, Stephen Hummle, the Flexor folks, or anyone I've left unmentioned could make a drum synth module or 2, or a new algorythm for allowing a 2nd cymbal hit to alter the ring of the 1st hit (so it doesn't play like two separate hits or cut the first one off) or something along those lines. Maybe, one could load midi-controllable groove modules or arpeggiators in place of traditional matrix sequencers. Or one could run a programmed beat in a matrix through the groove editor to produce realtime variations.
Others could design new shells and interesting signal paths, or even just cool skins. Different shells would naturally excel at creating rhythms in different ways.
Hell if developers were able (or would even want) to disassemble drum synths they've already made into modular-friendly components, that might reinvigorate (or resurrect) their sales for a lot of hard work already done.
If there's one thing that this thread has demonstrated, it's that we've all got a lot of devices that do certain things well, and other things poorly, or at least short of our imagination or expectations.
In my opinion, it would make more sense to avoid the singular device(s) and instead have the drum concept mirror the over all Modular concept and be an open sub-platform, as it could grow at it's own pace, brach off in any number of directions, and would be immediately useful within the current Modular paradigms while the greater schemas and goal posts are still being developed.

Kind of like a signature line of (semi)Modular, specifically for drums.
Did that make sense? Does it sound cool? Did I just make a big deal of something stupid and obvious? I don't care right now. I'm going to bed. I'll read yalls tomorrow once I've slept.
My $.87.
Sam
<font size=-1>[ This Message was edited by: dehuszar on 2004-06-19 01:02 ]</font>
<font size=-1>[ This Message was edited by: dehuszar on 2004-06-19 01:19 ]</font>