Broadcast has bandwidth. Computers have brains. Why are these two competing? Broadcasting a signal that means something to computers, and having computers then marshal that for users is way more profitable.
Easy place to start - start hiring a small TV station for an hour in the morning, somewhere in amongst the informercials, and broadcast a signal that people with TV Tuners can decode. Using some variation of colour 3D barcodes, say. Have the TV Tuner decode software use this hour to pump content into squid or some other form of web cache.
Monetisation strategy: give people the ability to vote on what gets broadcast based on their cache logs.
Bittorrent does the same thing more easily, but neither works for live broadcasts. The technical solution to the HD bandwidth issue is probably ISP level multicast, but they'd still only be able to deliver the most popular content, making them effectively "a la carte" cable companies.
Perhaps a frame based variation on the bittorrent protocol is the way to go, where you're constantly torrenting sequential blocks of content and viewing with a slight delay. I think people would find that acceptable for live events that aren't popular in their current location, but broadcast is still superior for mass appeal content.
Where there is a price difference, there is an arbitrage opportunity. Sure it's a bunch of work, but it only has to be done once. With something like software radio it can even be invisible to the end user. It becomes a usb dongle that feeds them high def youtube et al.
Easy place to start - start hiring a small TV station for an hour in the morning, somewhere in amongst the informercials, and broadcast a signal that people with TV Tuners can decode. Using some variation of colour 3D barcodes, say. Have the TV Tuner decode software use this hour to pump content into squid or some other form of web cache.
Monetisation strategy: give people the ability to vote on what gets broadcast based on their cache logs.