They've made two statements that don't mesh or don't actually mean anything together. Here are the statements.
1. Kinect will offer 1:1 motion.
2. Kinect will learn your movements and surroundings.
Here is why they either don't mesh or don't mean anything.
1:1 motion implies that when I make a movement the game will make the IDENTICAL movement EXACTLY as I made it. If Kinect is to learn anything about my movements or surroundings it is either doing so pointlessly or it is doing it with the intent to 'make things easier for me'. If this is the case, then it can not possibly be 1:1. It is the PROPERTY of 1:1 motion that makes this absolutely impossible.
If you're confused, you should be, because this is what they have said without actually saying it.
Either Kinect isn't 1:1 and only functions as an interpretation of your movement (WAGGLE WAGGLE WAGGLE) OR it is 1:1, it learns about your movements and surroundings, and doesn't do anything with it at all. Those are the only two possibilities.
Reasoning:
1. The very property of 1:1 motion is that it is impossible for the controller (in this case, the camera) to interpret your movement as anything other than it is.
2. Something that involves imprecise movement (not 1:1) uses gestures or 'stances' to interpret how the character should behave in the game. Obviously it is possible for something to 'learn' that my crouch is different than someone elses crouch and then be like 'ok, so this is his crouch' and then have the character perform said gesture based action.
2a. By this logic, I could eventually teach the Kinect to crouch by raising my hands in to the air.
So, it's either 1:1, or it isn't. They've made contradictory statements on the nature of the device and I'm curious to see which it is. Before I was under the belief that it was 1:1. Now, I'm not so sure.