Milt,
I for one would think twice before trying to argue aerodynamics or dynamics of flight with mr. Farley here. He's one of those people who have an annoying habit of spoling any and all discussions by actually knowing the facts.
An axis is not a physical thing. You can define an axis arbitrarily and have it far, far away from the CoG. This, however, is making things hard on yourself if you're interested in flight dynamics. Having all axes used be the rotational axes, through the CoG, makes things a lot easier*!
The definition of e g the yaw axis then will simply be the axis an aircraft will rotate around when you apply a yaw moment to the airframe.
That way, you can divide the forces acting upon an aircraft in flight into torque couples (moments) and forces. The former will rotate the aircraft about the CoG (i e the axes) and the latter will accelerate the CoG.
If the axes don't go through the CoG, you will have to consider acceleration of the CoG every time the attitude is changing... and you'll need to deal with torque about your origo to keep the attitude constant when the CoG is accelerating.
And no, with a CoG-fixed origo the axes don't move with alpha, beta or anything else, nor do they need to move relative to the airframe to achieve this stunning simplicity - provided the CoG doesn't move.
Mass does come into play even when you are only changing the attitude of the aircraft. Mass is what causes rotational inertia.
An interesting note on the actual axis to roll around is that some jet fighter FCS's change the roll axis when you go to guns mode. Rather than rolling around the CoG, the FCS makes the aircraft roll around the barrel of the gun. Neat trick, IMO.
Cheers,
Fred
*) When working with the dynamics, that is. The origin of the coordinate system used for component placement and so on in construction will probably very rarely be anywhere near the CoG - but that coordinate system is not the one we're talking about here, is it?