The huge problem with thinking about evolution, emergent processes and behaviours is that we are usually dealing with the end result and trying to reverse engineer the development based on what we know of the result

However, evolving strategies and emergent processes don't work that way. What you end up with is conditioned by a previous condition and constrained by all the failed options to which the current situation might give rise. In trying many of them and failing, the contestants take themselves out of the game but sometimes the "successful" players abandon some of the scaffolding that got them where they are. The latest research into what birds use their wings for, turns a lot of thinking upside down, where apparently a lot of it belongs.