Thanks for the positive comments guys!

As far as a breakdown of the back-end technology, we basically just have plugged into our existing Map server architecture, but are now using Google Earth as the user interface instead of a web browser.

I wrote a paper for the 2005 ESRI User Conference about our existing internet mapping solution, it should provide some insight into what we are doing, you can download it here:

http://gis.esri.com/library/userconf/proc05/papers/pap1297.pdf

The NetworkLink and GroundOverlay tools are very powerful ways to quickly get real GIS maps inside of Google Earth. GE presents an excellent interface for navigation and rendering (3D is really just a bonus.) I think by far GE's most powerful feature is it's speedy and intuitive of navigation. There is simply no comparison to any existing mapping tools.

Though, there are some areas that we would love to see improved for KML. If I could suggest some features they would be:

- Pass range/tilt angle along with the BBOX variable in the NetworkLink call

- Allow NetworkLinks to be called simultanously, not one at a time

- Don't flush "old" NetworkLink KML until new one is ready (actually I think it might do this for features already but not for Ground Overlays)

- Ability to load <description></description> via a URL for points (this could be HUGE for allowing integration into new and existing GIS apps)

Our solution is now very reliant on "traditional" GIS tools to do the grunt work and produce useful information and maps dynamically, nothing has changed with that. What is different is we now have a great platform with which to display these maps and information. Google Earth does that very, very well.

Thanks!
-Phillip