Xmas, I have no disagreements with nearly everything you said, but that's not the world in which we currently live. As if to add insult to injury, it seems unlikely that we're going back to that world in the foreseeable future. As for voicing your disagreement? The really sad part of this is that the overwhelming majority of folks who use these services have already "voted" for you, and Cloud is doing it.
My only single disagreement with your reply was here: "someone else (quite possible, IMO)" doing it -- I can't see how a 3rd party would be able to compete here. The amassed knowledge of billions of voices being translated and parsed every day in a cloud infrastructure will always result in a better response. A local voice recognition system will never be able to leverage that experience, and yet it will also never have the benefit of being hooked into the OS and the ecosystem of apps in the same way that the Big Three can do with their native services and API hooks.
Sure, you can buy Dragon's Naturally Speaking and build macros to have it do simple things via emulation of clicks on a GUI, but it can't touch the integration that Cortana has (and so will Siri) at the lowest levels of the apps that have been programmed to integrate with the common APIs, and with the OS itself that has the personal assistants woven in to the cogs, nuts and bolts of what makes your PC work.
And since that third party will need to monetize this application somehow, they''ll be trying to sell it in competition with the native featureset of an operating system. Sure, people buy DVD playing software too, but that problem has been solved for a decade now. Speech recognition is hard to get right (especially contextual clues) and even harder to apply to a system that is meant for hand-driven input. Unless your 3rd party reverse-engineers all the Cortana API hooks in Windows, I really can't see where they'd fit in.
