Wait, no difference? The focus is clearly sharper in the second (5.0) photo. Look at the words on the box in particular, but it's also apparent in the train. It looks like Lollipop improves the focus lock at the least.
I agree. I took a small portion of the train and clipped out the same bit and pasted them side-by-side and magnified the image 4x.
The difference between 4.4.4 on the left and 5.0 on the right is astonishing.
Small sample, granted, but you can take most sections of the photo and do the same.
EDIT: Looking at the pictures overall, I was struggling to see if maybe the focus point was just wrong, but I can't find much in the 4.4.4 picture that is in better focus anywhere than the 5.0 shot. So either the 4.4.4 just happened to pick a bad focus spot as a one-time issue, or 5.0 has markedly increased focus lock.
The only "on par" difference I could think of is if the car was moving more slowly on one, and the camera attempted to focus on the car rather than the train box. That is somewhat supported by the better focus on the word USED on the cardboard immediately adjacent to the car, which is closer and would be a different focal point (top: 4.4.4, bottom: 5.0)
So it is possible that either the focus algorithm on 5.0 takes more of the image into account, or that the 4.4.4 shot merely got "unlucky" in having the car become part of the focus algorithm while the 5.0 shot didn't see the car due to timing.
Having said all that, the blur on both cars is similar enough that they appear to be moving at about the same speed, and they are both close enough to the center of the actual shot to have the same weight focus-wise, I'd think.