Google continues to use the polar coordinates style Sphere mesh for Google earth.
The superior solution would to use a geodesic sphere where each polygon is an equilateral triangle. As you zoom in the triangles would tessellate into smaller and smaller ones, like how it is already done with the LOD, but it will always be equilateral triangles.
If the polar coordinates system was canned and they switched over to the geodesic sphere mesh,
1. There would be no shearing tearing and z-fighting near the poles.
2. They could afford more polys because they could be more easily referenced from to create new mesh data quickly.
3. it would increase texture mapping accuracy and reduce stretching.
Those are my thoughts, if anyone else has any more to add, I'd like to hear it.