The effort to convert change the way we broadcast television is problematic enough. But we need to be careful about citing physics as a justification for policies that are motivated by social and political concerns. Even if it's not really about physics we should still attempt to get the facts right.
One problem is that Shannon’s Law says nothing about the economics of “spectrum” – it doesn’t even define the physics of “spectrum”. It’s just a mathematical model! We can argue about the physics of wireless signaling but the more important issue is that bits are bits are bits. It doesn’t matter whether they are wireless bits or wired bits or blue bits or pink bits or …
We need to heed the lesson of the Internet – applications should be independent of the accidental properties of the transport. Instead we find out what works and what doesn’t and adapt. If we have to depend on the particular characteristics of each hop we’ll continue to have a brittle system that depends on every hop working perfectly even if we don’t know what path we’re using! Instead the Internet has succeeded because we can communicate as long as we find any path that works.
Rather than spending all our efforts on intermediate issues we should focus on moving towards a bit-commons whether within the constraints of spectrum allocation, inefficient as it may be, or using other signaling techniques that do not rely on regulation. We can then treat access points as routers between segments that might or might not use wires or glass to facilitate signaling.
If you do want to understand the physics you should read these comments by David Reed. It’s also useful to recognize that as much as we focus on how to detect available “whitespace” we can’t take into account legacy receivers that are confused by new signals. These are the “bubble babies” of policies that seek to protect the past against the innovation and frustrate progress.