DB3 is slow going, because I get bored along the way and start doing novel, untested things
My current plan is to have something (either an early alpha for DB3 or at least a physics playground) by summer 2015. I'm approximately on track for that, unless I find a new interesting problem to grab my attention
. I'm at the beginning of the process of gluing all the bits of a proper physics simulation together (graphics, collision detection, collision response, etc.), but there's a few open questions I don't have answers for just yet.
The first drop of DB3 proper won't really scale well to lots of bots (it's possible you'll need fewer than 50 bots to get realtime-ish speeds), and may or may not have fluid physics, but should have bots running around using DNA, a simple combat model, and multibot physics of some sort (the ability to create articulated structures from multiple bots). On top of that simple base I'll start adding things and streamlining the performance, and development should be at least more public if not more rapid.
