So much of Natal is unknown - I think that's the real problem.
I want to see how the voice/facial/gesture recognition works in and out of games. In some games, the D-pad is used to select different weapons, so voice recognition could be used for that. Natal can "see" you, so it could provide instant feedback, in the same way that Burnout Paradise does if you have a Live Vision Camera attached. Will I be able to launch games and apps with speech? Will I be able to invite people to games by saying "Invite mattclarkie" or whatever? How well will it work if I'm holding something in my hand(s)? If I'm playing a driving game, for example, could I hold the Wii Wheel from Mario Kart, so at least I feel like I'm turning a wheel? If that works, then even holding a frisbee as a steering wheel would work, lol. There is a TON of scope there - things that I think a lot of people haven't even considered - but it's whether or not they implement that stuff and how well it works that matters.