The first one works if you do it properly. One of the lines of business just operates the physical infrastructure. It not only doesn't provide video service, it doesn't even provide internet service -- the end user is never their customer, their only customers are third party ISPs, who all get the same terms.
The physical infrastructure is what makes everything on top of it possible, and at the same time it is one of the most expensive pieces to build, upgrade, and maintain. There needs to be adequate returns on investment in the infrastructure, and that's hard to do in the unbundled model.
If you want high-quality infrastructure, you need to let the infrastructure builder capture a sufficient portion of the value created by that infrastructure. Consider trains. Trains create value for people using the train to get around, but also value for the real estate by the train station. If you only let the train company recover from the rider, you won't incentives that reflect the true value of investing in the train.
Look at Japan, where train service is exceptional. The train company gets to charge both sides. They charge fares to riders, but also are major land owners around train stations, and charge rent to the businesses who benefit from having train service near them. If you had your model in Japan, where train companies just own the tracks and get a modest fee for allowing people to use them, you would not see that kind of high-quality train service.
Why is it hard? You charge a price that reflects the cost of building the infrastructure. 200Mbps service can have a higher price than 50Mbps service, providing an incentive to build the capacity necessary to offer it. Which is still true even if the price for the faster service falls to approach the price for the slower service as upgrade cost is paid off. And then there is another upgrade and 1000Mbps service becomes the more expensive one compared to the now more affordable 200Mbps etc.
> Consider trains. Trains create value for people using the train to get around, but also value for the real estate by the train station. If you only let the train company recover from the rider, you won't incentives that reflect the true value of investing in the train.
You're essentially arguing against there being efficient Coasian bargaining. But the arguments against it are usually related to transaction costs, which don't seem to help you here. If Comcast can't charge anything to Netflix and as a result Netflix service is slightly less expensive to the user and internet service is slightly more expensive to the user, the user is not engaging in any new transactions and the net to the user is approximately zero. Moreover, then there is no transaction happening between Comcast and Netflix, which reduces the overall number of transactions that have to occur.
The situation right now is so ridiculous that some states have laws outright barring municipal ISPs. If a breakup into regional private monopolies is the stepping stone to reducing ISP influence on government enough to get a more sane legal framework, I'll take it.
In Canada, there's a bunch of smaller cable resellers. For instance, Shaw does cable/phone/tv and also has some wifi infrastructure deals with various commercial businesses, and provides "Shaw Go Wifi" access to internet subscribers.
In my local area, we also have LightSpeed which resells access to Shaw for a cheaper rate, but you need to buy your own modem, and there's no wifi access around town. Trade-offs, and a slightly more healthy marketplace.
An antitrust breakup of Comcast would be one or both of:
1. splitting off different lines of business, such as the content production parts from the physical cable service part, or
2. splitting them into different companies geographically.
Neither of these would do much to increase your ISP options.