@A0283: I see your points. Just a thought: once things fell apart, the movement of the water and "how long a given bit floats" may offer sufficient explanation for how the dispersal came about. Or not. You are more likely right in your estimates, as you've examined it in more detail.
At first thought. There might be two 'theoretical reasons' that could explain the 2,000 m. First floating, and second forward speed. But both start with hitting the water and losing the THS and the bottom of the tail with the APU.
On the first, floating. So you have an open tube ... which if it fills, will fill quickly ... if not, the plane might be weighed nose down and keep the open end up for a while (until the swell flows in) and might float around a bit ... at the 1-2 knots published current that distance would take roughly 30 minutes to 60 minutes.
On the second. The forward speed would combine losing the tail followed by skimming over the surface like a flat stone. At say 60-120 knots that would be roughly 2,000 m in roughly 0.6-1.2 min. The impression is that the plane went up and down quite vertical (not far from LKP) and pretty fast. So, that seems to be less probable.... In spite of the fact that the lower aft section may still have been attached (as mentioned, the aft top panel stayed connected to the tail/rudder section).
What it does not explain is the apparent stripping of most interior components..... Which gives a bit more probability on the side of a higher speed and above break up. Also, skimming has lower probability, when reviewing earlier widebody waterlandings.
Theory and practice can be quite different of course. But it gives some idea I guess. And no more than 'some' idea ... at this stage and with the publicly available information.