I believe length contraction always makes more sense when integrated with reminders of relativity of Simultaneity. Let's say the engines are at the back end of each rocket. For the viewer "A" in the initial frame, they begin moving and continue accelerating simultaneously, and clocks next to the engines are seen in synch. However right away, "A" will measure that for each rocket, a clock by the engine is running faster than the clock by the head of that same rocket. I have never seen this pointed out in a discussion of this paradox, and I think that this may be one reason people so often are mistaken about this situation. The idea that "both spaceships accelerate and keep their clocks synchronized" to "A" distracts the reader and puts one in the mind of a situation without SR. The reminder that within each rocket, the clocks appear out of synch to "A" might snap the view back to SR. We might attempt to apply this same logic to the rope, but right away we run into a problem, since the back end of the rope is connected to the front of the back rocket, and the front of the rope is connected to the back of the front rocket. If the back end of the rope was accelerating in synch with the backs of the rockets, the rope back clock could not be connected to the back rocket front clock, since they are out of synch. And the similar applies to the front of the rope being attached to the front ship back clock. We are asking of a suggested nonstressed rope that it fulfill contradictory requirements.