Let's suppose you had two 35mm cameras loaded with 100-speed film.  One has a grainy "high-speed" film from the 1930s.  The other is loaded with T-Max 100.  Both cameras have a 50mm lens attached, and both cameras have similar weight.  Both are shot at identical shutter speeds and apertures.  It would be incorrect to conclude that one camera would exhibit more shake than the other, right?  Then why do people continue to believe that higher resolution digital sensors have more motion blur?

There is a persistent internet theory about the D2x - that it somehow harder to handhold than DX sensor cameras with lower pixel counts.  The theory is that because the D2x packs more pixels into a smaller space, there is more "magnification," and therefore more shake.  This makes no sense, because motion blur is disconnected from what we can call "grain size."

1.  Camera shake is always present at some level.  There is no single point recorded in a frame because your imaging surface always records the subject in motion.  You are always getting arcs, circles, or lines, depending on the direction and speed of the apparent subject motion.

Apparent subject motion can originate in the subject or the camera (or both), but the effect is always the same.  You can increase the shutter speed, and this will help, but there is always some apparent subject motion, no matter how steady a tripod or how "fixed" the subject.  The key is controlling it.

The 1/shutter speed rule does not eliminate movement.  It simply pins it down to an acceptable level.  The rule was calculated from assumptions about an acceptable circle (or arc) of confusion in output.  That, in turn, is in turn driven by an assumed relationship between output size and viewing distance.  If you are more demanding than what the 1/shutter speed rule anticipates, you need to increase the shutter speed.  This principle is identical in operation to computing depth of field.

2.  Apparent subject motion is amplified with a reduction in viewing angle.  Mount a 100mm lens on a 35mm camera, and it exhibits x level of apparent subject motion.  Put the same lens on a camera with a 24x16 sensor, and because the angle of view is reduced, you get 1.5x the amount of apparent subject motion, or the equivalent of a 150mm lens on the 35mm camera.  The factor is relative and related to how much it is necessary to enlarge the sensor's physical size to reach the same output.

3.  "Motion blur" is typically (and incorrectly) evaluated at 1:1 pixels.  When you look at images that way, the ultimate magnification depends on the pixel count of the sensor.  This is the conceptual equivalent of basing a paper enlargement on grain size.  Using the 1:1 method, the more pixels on a sensor of given physical size, the more magnification occurs.  The correct way to compare for motion blur is to adjust the 12mp shot down to around 66%, so that magnification on screen ends up proportional to the sensor's physical size, not inversely proportional to pixel size.  Hmm... where was that blur again?

So you can see that keeping the lens constant, switching from a 36x24 imaging surface to a 24x16 one will cause more blur, but there is simply no basis to conclude that increasing the resolution between two 24x16 sensors would.  So where does the D2x shake come from?  I would look at fatigue and other human factors.