trtzbass
And I think we are saying the same thing: if you keep the legacy code, the software does not evolve
Although there are definite advantages to a "build-things-from-the-ground up" approach, software like SONAR is modular. Something like ARA integration can work with existing code (I'm frankly surprised that more DAWs haven't done it). Also I think the speed improvements that you've been seeing lately come from replacing sections of old code. Fundamental changes do take a while, though...Ripple Editing is a case in point. I suspect the more radical a change (e.g., scalable graphics), the more time it would take.
But really, I know nothing about code other than at one point, I was able to bounce pixels around on a Commodore-64's screen

. So I get to make requests while being blissfully ignorant of the consequences. When I wrote the manual for Ableton Live 2 I asked to make the Solo button be recordable as part of a performance. I figured that would be easy. Instead, I was told it was just about impossible because it was always intended to be a diagnostic tool in the studio, not a live performance feature.
When I've asked Noel about the feasibility of certain features, sometimes something I assume is super-complex he does in a day, but then he tells me something I assume is simple would take months of development. I'm glad my responsibility is restricted to
using (and occasionally abusing) software

.