Foreword: I don't pretend to be 100% precise with the following arguments; I am very well aware that there are many details and exceptions. I just hope to be "best effort" right -sort of-.
This is a plot of nominal commercial access throughput for different technologies (fixed and mobile (let me call them wireless)) that have been introduced in Italy from 110bps to 20Mbps. It confirms Nielsen Law of 50% to 100% increase per annum. (n.b.: fixed just DSL, just copper, it would look different with FTTB, FTTC, FFTH)
Taking advantage of the copper already installed, three families of technologies have been introduced: plain modems, ISDN and DSL.
On the wireless side the access throughput trend was diverging compared to the fixed tend (i.e. the fixed had higher trhoughput) until the introducion of WCDMA and HSxPA. (if you take away the last points of wireless and replot the trendline, you see it's diverging)
Though Moore's law is active, on the fixed network we cannot indefinitely benefit from it as there are physical limitations of the copper, unless we shorten it, that is progressively substituting with fibre, but there are good economic arguments that rather that shortening it to the curb, we should shorten it to nil, building FTTH which effectively will move the problem essentially in the domain of Moore's law (first and photonics later).
Problem is that considering an approach similar to FTTC with VDSL, you can get comparable nominal access throughputs with wireless, for much lower capex and opex. So, why restructuring the fixed network with FTTC/VDSL when you can go wireless ? This is a question that increasingly surfaces in operations of Telcos (and Telco gear suppliers).
In order for the fixed network to be clearly better than wireless, it needs to become FTTB/FTTH, stepping away from DSL. But is this going to be enough to attract "subscribers" compared to the capex-cheaper wireless option ? (an option which, BTW, will allow you for mobility ? (precisation: albeit with a inferior effective throughput)). This is clearly unproven.
Think of it: Wireless access connects people, Fixed access connects places.
This is the reason why they're pushing HDTV. Though you carry your computer with you, you don't carry the TV set with you.
It's not an easy dilemma for the operator: go for fixed or go for wireless ? it's a bet that requires a lot of money and needs a couple of years to prove which is the winning side. Of course, the best thing would be to have both, like we have (almost) today with DSL and HSxPA (and there is in some countries), but that costs a lot and its sustainability is questioned as more and more "subscribers" leave their fixed access for the more practical wireless option.
But fixed network costs are essentially fixed and attribution to subscribers is a matter of cost allocation; when one user cancels the subscription to the fixed network (saving money), the operator looses remuneration but keeps the costs (unless all users on the same exchange cancel).
It's a classical Tragedy of the Commons, draining available resources: for the user the marginal benefit of "unplugging" is 1 and the marginal cost is 1/N where N is the affected population. So this will happen.
There is one more consequence which is user contendability. While fixed access is essentially a local monopoly, wireless access is not. So the fixed operator's arm counts all the users losses; but churning customers may choose other wireless access providers, so the chance for limiting damages from moving customers from the same operator's fixed business unit to the wireless one, is limited.
Furthermore, the effecive availability of switching possibility amongst different wireless operators thanks to number and credit portability induces competition and price reduction.
This is the reason why operators are trying to find reasons to keep connected, not only people but also places.
Once it was clear: voice was mobile and Internet was fixed. Now this is quickly becoming not true.
But for BTS (wireless radio access points) to proliferate as needed, wireless operators need fibre to feed them, though not until the customer premises but somewhat shorter.
The access portion of the fixed network may well die, like a kind of Telco-ALS (Amiotrophic Lateral Sclerosis).
The opportunity for traffic management, in the operator's view, will not disappear until there is unlimited capacity (which is far from happening in the wireless access), but all the society-related issues are there (access control, wiretapping, innovation, competition, etc.).
So we need the fixed network but the present structure remunerability is challenged. Like roads.
In conclusion, I believe that if we want Network Neutrality (which I believe being almost essential for society progress) being a basic requirement for networks, we ought to accept fixed network being a market failure natural monopoly and hence some kind of public policy must intervene.
p.s. I'm not arguing that this is STRICTLY so in ALL countries, and in ALL places of a country; there are some differences from country to country with many factors interfering from orography to population distribution to age of cities, to ... I'm just trying to add a dimension to the discussion.