It's not a totally crazy idea, but you have to balance the cost of
transmitting data about your scene against the cost of rendering that
data. In a scene that's complex or highly dynamic (i.e. changes a lot
from frame to frame), sending all that information will almost
certainly be the bottleneck.
Render farms at Pixar, etc. get on the positive side of this equation
by having frames that are very expensive to render, and what I assume
is the fastest networking money can buy. Analogously, Folding@Home
has chunks of data that are small but complex to process. All told, I
think the class of problems for which this approach works is fairly
small.
Zach