The only relevant measure of broadband adoption for my proposed system would be how many households A: *do not* have broadband and B: *do* have TV beyond basic over the air broadcast. The reason being that anyone who is still watching over-the-air broadcast isn't paying for TV and isn't a part of the mainstream TV ecosystem in general. I'd bet the TV watching hours of those without cable/satellite are much lower, so really they don't need to figure significantly into TV company's consideration of how to evolve their business IMO.
Anyway I think it's safe to say the results for people without broadband and with cable TV are much lower than even the 40% low-end number for dial-up users (which I think is out of date anyway - last stat I read is 66% "broadband" penetration in the US). Now whether "broadband" as they define it would support decent on-demand TV is another question, but given the quality of broadcast non-HD TV, you'd be surprised how little bandwidth might do, for a single channel at a time. But that's less important really. Remember that I'm postulating that they use existing cable bandwidth (anyone who has cable TV) as one of the main delivery channels, at least to cable-served areas. Satellite would have a harder time doing this since their bandwidth model is different, but if they got rid of the massive one-to-many streaming of TV they have now, they could probably still up their per-customer satellite internet speed to levels high enough to stream a single 1080P show per user. That would be enough.
The trick is getting it all into people's homes and onto their other devices (e.g. cell phone, laptop, etc.) in an easy-to-use and relatable way, while still providing appropriate controls. It could be done very simply with existing tech if you could magically cut through a lot of red tape. The idea of a massive catalog of well organized and instantly watchable media is not unprecedented - Netflix is basically already doing this. Just imagine it extended to include new shows, as they debut. Realistically it would unfortunately involve a huge amount of contract negotiations with different providers and whatnot. Perhaps Google has a legitimate start with Google TV, but remember that for now that's little more than a content aggregator from existing sources, and a front-end for existing TV. When a Google TV type system can actually give you control of anything you want to watch, on-demand, that will be the real revolution.
Most TV is massively pirated now anyway. Torrent sites aren't going to be suddenly so much better just because we have 100% on-demand TV. If anything they could be further limited, the TV companies could charge per-stream for example (a stream being a single channel being watched, so if you want to watch 4 channels you pay more) and more easily encrypted. The way it is now someone with AT&T u-verse can have their DVR recording 4 shows at once, 24/7, they're already decrypted to their DVR device, and then they can be simultaneously ripping those recorded streams on their computer and uploading them in parallel. If anything the system I'm proposing would give TV providers *more* control. And I'm ok with that precisely because it also gives *me* more control. It's win-win, as I said. They can lock it down with all the DRM they want to prevent me uploading it en-masse, but put in *reasonable* ability to A: view my TV on multiple devices (cell phone, computer, TV) *and* perhaps even to send/loan a show to a neighbor/friend one at a time, and you've sold me. I'll pay $100/mo for that service happily. Know how much I pay for TV now? $0. In fact I have never in my life paid for TV, and probably never will as long as it's as crappy as it is. And most people I know who do have cable either get it free, steal it, or have only very basic cable. It's just not worth it otherwise.
As for problems with the financial model and how TV shows would get made, I would ask how they get made now. Cable/satellite service costs money, *plus* there are ads, *plus* they have extra cost on-demand stuff. They have existing funding models for making TV shows. None of that needs to go away. I'm not really suggesting fundamental changes in how things are charged for. You can still charge $50+/mo for "cable", and have ads, so TV shows are funded in the same ways they are now. But if you're doing it in a totally on-demand system, you get A: more relevant ads B: your ads are harder to get rid of with DVR-type technology or to otherwise skip, mute is the only hinderance, and thus your ads are more effective, and C: much more accurate stats about viewership and therefore better decision making capability for which shows to make and which to cancel. And if it's the aggregate service providers, like Comcast, who are collecting the viewership data, they can sell that to the content creators just like Neilson data is sold, to better inform content creation decisions.;
I can see that somehow I'm not explaining what's in my head very well as none of what has been postulated as issues really comes in the way of what I'm proposing. The biggest changes necessary for this would be in the back-end business deals, the odd ways in which cable companies get "bundles" of channels, and schlock shows get made. That's the stuff that's going to be hard for the indsutry to let go of, and some companies will inevitably suffer as a result. I certainly have no illusions that everyone who is in the game now would stay on it. But I do think the primary content delivery companies would thrive, and that's why I'm a little saddened that Comcast, AT&T, and the like are not being a little bit more forward-thinking with their offerings. It seems to me that they can afford to do so, they have the upper hand. They should force the content creation industry to work by more sensible rules and move to an entirely on-demand model. The world will end up a better place if they did.