"it would initially seem that the container ship’s movements were so completely unpredictable that the Fitzgerald’s bridge team would have been unable to know what was going to happen next."
Well, that's part of the problem with that 'first' version of events --it requires some exotic motivation. The alternative version that Stephen Davies pointed to makes much more sense.
Let's assume, as the Japanese Coast Guard has now officially suggested, that the timing has been reported wrong. The collision occurred about an hour earlier than originally believed. The initial change in course, a sharp turn to starboard by the container ship, ACX Crystal, was the collision itself, followed by a return to original course presumably managed by the vessel's autopilot. A while later, the ACX Crystal slows, comes about, and returns to the approximate location of the collision. They spend a while there and later eventually head northeast into the bay where they wait until morning. That all fits nicely, and it does not demand any unpredictable, erratic course changes by ACX Crystal. Naturally, aboard the container ship, there would have been no one standing watch in the vessel's bow in the middle of the night, and with a typical pile of cargo on deck, the bridge would never have seen the Fitzgerald. I'm sure they have cameras, but if no one was looking at the right time... Assuming a single person on watch on the bridge had fallen asleep (or was watching a movie, studying for a celestial navigation exam, doing a crossword puzzle, or otherwise not paying attention), it's not hard to imagine that the collision could have been initially misunderstood as some machinery problem: "what the devil was that groaning noise, and why did we suddenly veer to starboard and slow down? Did we lose a propeller blade?". The Fitzgerald would have been left in their wake to port aft, and if no one looked in that direction immediately, and if they were watching AIS traffic instead of radar (USN vessels normally, though not always, do not broadcast AIS data), then the severely damaged destroyer might have been unobserved. Only after an inspection would they have realized that there was serious damage to the port bow, and only then would they begin to consider the awful possibility that they had collided with, perhaps even sunk, another vessel.
Seeing how fast ACX Crystal was travelling, looking more at the damage to the Fitzgerald, and considering that this is a hardened vessel designed for battle, it really is just lucky that the Fitzgerald wasn't sunk out-right or cut in half. I am reminded of a collision I read about years ago where the Australian aircraft carrier HMAS Melbourne ran over and cut in half the destroyer USS Evans in the Vietnam era, leaving half of the destroyer eerily afloat, decapitated and dead in the water: Melbourne-Evans collision on Wikipedia. Though USN brass have announced that USS Fitzgerald will be repaired and put to sea again, I would not be surprised if that proves to be a poor bargain and instead the vessel is de-commissioned.
A celestial detail: several media accounts have described a "moonless night". Using a collision time of 1630 UT on 16 June, I find the Moon about 22° high in the southeast from the accident location and 59% illuminated. Am I doing something wrong, or are those moonless reports wrong?