[AnthonyMouse]

> If by "natural monopoly" you are implying a monopoly arising without government intervention, I'm pretty sure you are completely wrong on that point.

The reason telecommunications service is a natural monopoly is that the fixed costs are extraordinarily high and the variable costs are trivial. In consequence it is not profitable to build a competing network once one already exists, because all you accomplish is to lay out an enormous capital expenditure in order to enter into an aggressive price war with the incumbent, since making $10/month from a customer that costs you less than that to service once the network is already built is still more profitable than losing that customer to the competitor and making $0. So the result will always be the same: Either there will be a monopoly (or equivalently a cartel), or there will be a price war until every competing provider either goes out of business or joins the cartel and it reverts to a lack of competition. And prospective market entrants understand this dynamic ahead of time and choose not to participate. The only reason we even have the duopoly that exists today instead of a straight monopoly is that the two networks were originally built to provide different non-competing types of service, so now they effectively act as a cartel with the knowledge that no one else is likely to enter the market.

It is certainly true that government intervention has been involved in the creation of that infrastructure -- it's plausible to say that it wouldn't have been cost effective without it. But that only means that in the alternative "free market" for telecommunications service, the infrastructure cost would have been even higher because the provider would have had no subsidies and would have to buy all the land without eminent domain and pay off holdouts etc.

[jessaustin]

Don't assume that the network must be entirely owned by one party. That is the situation right now in the cell and broadcast markets, but it is caused by regulation rather than anything "natural". It isn't the 1940s anymore; wideband radio technology exists that would allow the sharing of all spectrum rather than the single-use requirements we have. After that is adopted the first-mile connection is to a WISP, which is connected by fiber to the rest of the internet.

[AnthonyMouse]

Wireless is not a panacea even with spread spectrum. You still have a finite amount of radio spectrum you have to share with everybody within range. The amount of bandwidth you can serve to each customer with a fiber-terminating central office every square mile is limited in practice by the cost of installing fiber. The amount of bandwidth you can serve to each customer with a WISP every square mile is limited in practice by the laws of physics -- the only way to get more is to build a higher density of wireless base stations, and in higher population density areas the cost of that is on the same order as installing fiber to the customer premises. Moreover, the cost effectiveness of using wireless for the last mile goes down as the demand for bandwidth per customer increases over time, because you have to increase the density of wireless base stations and run new cable to each of them in order to expand capacity rather than only replacing the terminating equipment at both ends of an existing cable.

Wireless is a great solution to providing high bandwidth services in rural areas where the cost of laying new cable is prohibitive. It's not a solution for cities.

[jessaustin]

It's true that there is a rural/urban breakover point; I'd contend that is at far greater densities than most people think, especially if we made use of the entire spectrum rather than the drips and drabs meted out so far by the FCC. The portion of the population getting shafted right now, all the "rural" (and most suburban, and urban in Texas-style "sprawling" cities) folks, is definitely a majority. "Laws of physics" arguments aren't convincing at this point, since engineers have had their hands tied by FCC regs. Let a big market for this type of radio tech exist for a decade, and then we'll see what can be done.

[AnthonyMouse]

You can't use the entire spectrum without addressing the sunk costs problem. All the existing users of the spectrum would have to throw out their equipment and replace it with spread spectrum equipment. Nobody wants to do that, especially when the benefit of them doing it goes primarily to third parties. You want to talk about phasing it out in favor of spread spectrum, fine, but that's going to take years or decades. What do we do in the meantime?

The next piece of trouble is, if you really want high power unlicensed spectrum, what do you do about contention? What happens when scientists start setting up transmitters to continuously send 50Tbps of scientific data from remote monitoring stations to the regional university's data center? Or people set up a distributed mesh network of Tor nodes? Or someone decides to operate a wireless "cable TV" franchise and starts transmitting 50,000 streams of 4K HD?

Don't get me wrong, I think you have the right idea -- spread spectrum is the way to go for wireless and there is room for a lot of innovation there if the FCC would get out of the way. Especially in the way of short and medium range mesh networks. It just isn't a replacement for fiber to the premises.