In my experience, OSS projects that require copyright assignment often fail to attract a significant amount of developers and community buy-in. This tends to leave them in a weak position compared to non-CLA projects, and virtually ensures that only a single commercial entity will dedicate significant resources to it. Eventually this single entity goes away or changes focus, and the project is either liberated from the CLA (best case) or left to rot.

I'm sure Mir has its technical merits, and they have great people working on it, but I'd place my chips with Wayland long term due to the above.

Thanks for a thoughtful post, by the way -- and all the other hard word you contribute on a daily basis. It's very much appreciated.