As I wrote on my blog, I think that Cocoon has to change fundamentally (focus on RESTful services, layered architecture and reuse in every Java environment) in order to survive in the medium to long term. Staying with Cocoon 2.x will mostly please already existing users but won't be very attractive for others.

I wasn't very eager to go for Cocoon 3 at this point of time myself. Cocoon 3 has been developed completely in the spare time of volunteers and releasing it will put some pressure on it that I would have liked to avoid it. But keeping it in the whiteboard without a release isn't a solution as well.

Cocoon 3 is a chance without a guarantee of success. But it's better than just waiting for some kind of miracle and doing nothing while Cocoon and its community are fading away.