Glabella I agree, the observed movement between exposures appears to be much larger than you'd get from what most people call dithering. In normal dithering, the shift is literally only a few pixels one way or another , and the goal is to ensure that your sensor noise doesn't stack directly on top of itself (which would reinforce it). By ensuring you align on the real target details, the noise never lines up so it is reduced, while the target details are reinforced.
It is possible that they're purposely making much larger jumps, not only to remove noise around center, but to ensure that there's more overlap as the sky rotates, because this is an alt/az scope. Otherwise you'd almost immediately start seeing the edges creep in around all the corners, but we don't generally see that unless stacking for several hours. Also, something I've seen in Mosaic mode is that on the first pass, there seems to be more noise along all the edges relative to the centers, which gets cleaned up with subsequent overlapping passes, so these larger jumps in normal non-mosaic staking mode may just be part of that edge cleanup process. Since it "moves between frames" they chose to call it dithering.
