Why?
Well, check out this wiremap shot, of a map that's absolutely flat (all heightmap values are equal, period), with no shadowmap on...

Interesting, how many polygons we're using.
If I put POV down to the lowest value, I assume it turns into a pair of triangles, but this is only POV 32, which is barely acceptable. Can somebody explain to me again why we aren't storing a series of LODs for large sectors, and then using the LODs for sectors depending on distance from camera --> some user variable? I know that this is supposed to be hot shit, but well... er... that's something like a thousand triangles. For a square.
Here's what it looks like, with ViewRadius 2:

Note the effect on my raw FPS. Let me just turn it off. I have a better idea, although it may take some custom tools for us to do it right.