I certainly agree with your point about the goal and the issue in some companies. However, having seen Microsoft in action from the inside, any collaborations

I certainly agree with your point about the goal and the issue in some companies. However, having seen Microsoft in action from the inside, any collaborations - most of which are _required_ at gunpoint from top-down - result in a horde of program managers shuttling themselves from building to building "owning" the X->Y relationship, one each per group-pair and arrow direction. It's a win-win for them, as those are a quicker path to promotion than trying to build up internal scope or working with the external market (godforbid <grin>). Frankly, it's really, really expensive and painful for ~2000 developers on three different teams to each be adding a slew of features with one of those teams building on top of and coordinating the other two. Even with huge amounts of communication, tracking, and the best of intents, until the products at the bottom of the chain settle down, you can't make any progress at the top. And when the top is finally making progress, you actually can put developers in front of APIs (in our case) and then you find out that the stuff at the bottom is totally unusable, but it's too late, because they finally stopped churning and want to ship soon. I don't claim there's an easy answer, but throwing more people or tools at the coordination problem doesn't (I've seen it tried).