More precisely, Dean advocates putting the arrow pass at dimensional center. Technically, from the of the arrow at least, the limbs are of equal lenght. That the bending portion of the lower limb is shorter (part of it being occupied by handle) means, particularly on a shorter bow, it needs to be a little wider and carefully tillered to avoid overworking it. I try to visualize the apex of the tiller curve at the arrow pass and it all seems to work out.
In fact, on a bow where arrow pass is arbitrarily positioned 2" above dimensional center, the upper limb is "shorter" in terms of the geometry at the string grooves, even if the bending portion is "the same." That's why, I think, Dean and others assert such bows may tend to fatigue the upper limb more quickly. Even though they have "equal" bending lenght, the positive tiller required to offset the asym of the geometry means they have to bend farther, causing more set. Since we can't abide arrow nock downward travel on the power stroke, for want of interference, "negative" tiller is not an option.
In practice I make 'em all different ways and honestly any difference in performance I'd have a hard time quantifying. But I do prefer arrow pass closer to center.