The blue object is a very simple rectangle in two point perspective.
Next I zoom in on it: 
When I measure the sides, I see is that the back side is shorter than the front side. It is foreshortened. The right side is closer to viewer than the left side. I would think the left side should be shorter because it is farther from the viewer. But it doesn't work. The left side is longer. So the farther side is not foreshortened. Isn't this a contradiction?
Where is my logic wrong?
I don’t see a contradiction. Closer line segments are only larger under the same angle of viewing. In your example, the difference between the viewing angles has a larger effect than the distance. This is roughly because you are closer to the right vanishing point than the left one.
To better see this, let’s take the example to the extreme: Take any non-quadratic rectangle and place it such that your line of sight goes almost along one of the shorter edges. You will see something like this:
Now you go further such that your line of sight is exactly along that edge. Now every point of that edge is closer to your eye than any point of the opposing edge, yet the length of the projection of that edge in your vision is zero, while the opposite edge has a finite (and thus infinitely larger) length.
I just think the difference is very small and your drawing isn't precise enough for you to measure it.
Which application did you draw this in and how do you measure?
It looks like the left vanishing point is a tiny bit lower than the right one. I can see the right one is a few pixels above the horizon.
If I take the closeup and stretch it vertically it's easier to see the details.
Notice how the blue shape is skewed compared to where the red lines actually intersect?
All these little inaccuracies add up.
