The key to Visual Fusion is the visual part. And, we're always looking for ways to make your applications speedier and more responsive. So, what's the point of sending all the geographic data for the intricate coastline of a stretch of the Ohio River if you are zoomed out so far that one screen pixel is bigger than a bunch of it's twists and turns? There is no point (unless somehow you get paid for unnecessarily enormous network traffic), they just take longer to fetch from the server and bog things down in the client.
In Visual Fusion 5.0 there is a snazzy new server-side generalization method that is...aware of your zoom level. So, it knows which twists and turns in your geographic data are sub-pixel in scale and, therefore, superfluous. And out they go! I should mention that this feature is optional, and if you need out-of-scale precision, then you can have it...for whatever reason.
What's the point of precision greater than the device showing you the data can handle? I ask you!!