Investigators flew the same exact flight path in another 777 and it should have been apparent if primary radar misreported altitude +/- such a large margin at the distances observed. If they observed primary radar lose no accuracy at that distance, why judge the 45k as inaccurate?

You are assuming that the altitude reading would always be off by (close to) the same amount.

Can you expound on that? Does it have to do with atmospheric curving of the radar beam?

> The use of primary radar requires a great deal of signal power, because objects further from the antenna will reflect or send back a weak signal. At longer distances from the antenna, radar becomes unreliable as a way to determine aircraft position with only reflected signals.

http://www.wisegeek.com/what-is-a-primary-radar.htm

Radar derives altitude from measurements of distance and elevation angle. The greater the distance, the less sensitive elevation angle is to vertical displacement (simple trigonometry), and hence signal:error decreases.

I think it has to do with electromagnetic ducting or other clutter effects:

http://www.radartutorial.eu/07.waves/wa17.en.html

If the last effect on the page were in effect, you might get a return, know it's relative bearing well, it's range a little less well, and it's altitude not very well at all.