End-to-end QoS for some applications and/or data streams is achievable only by tweaking BGP (the Internet’s network-to-network routing protocol) or by private contractual agreements between networks. The core is already all fiber, as is most of the middle, but it’s not the wiring that makes QoS happen or not happen, it’s the traffic load.

The media currently in play for the last mile are wireless spectrum, coaxial cable, twisted pair copper, fiber optics and power lines. The media in play for middle mile and backbone are the same minus twisted pair and coax. The only one of these media that offers a consistent QoS, high speeds, low maintenance and is, most importantly, future-proof within a reasonable time-span is fiber.

If we set a policy goal that we want broadband internet for all and we don’t want to constantly be reshape the infrastructure to meet new demand and new bandwidth eating killer apps, we need to look to fiber for all. Just as rural electrification allowed for a competitive marketplace for refrigerators and the other home appliances we enjoy reasonably cheaply today, so would a uniform fiber infrastructure shift the market competition to the application market which has low sunk costs and barriers to entry and away from the infrastructure.

We have an infrastructure market failure. Let’s roll cheap fiber out to every home and let the tax revenue and positive externalities from the resulting application market cover the cost.

All “innovations,” loosely defined a popular new applications, alter the mix of traffic within and across the private networks that make up the Internet, and it always takes a lot of engineering and a great deal of investment to digest them. The publicly-owned networks don’t contribute to this engineering ecosystem, preferring to ride on the backs of the capitalists who do the hard work.

Similarly, every ISP network connected to the others through the Internet is engaged in a never-ending struggle to increase capacity, and the concomitant struggle to find the capital it takes to do so (and I don’t need to tell you that, of course, as you’re in the middle of it.)

It becomes more clear each day that the real problems with American broadband are not so much at the first and last mile as they are at the backhaul and peering points, so all this net neutrality emphasis on “end-to-end principles” is nothing more than a distraction. The “net-to-net principle” is the important one.

The idea of the Internet protocol was to allow traffic to be carried between data networks that were NOT all owned and managed by a single entity, and had DIFFERENT rules of operation — drafted not by some central authority or by the government but by the owners of the networks. This decentralization of control and of authority may be anathema to those whose business is centralized regulation (or lawyers, such as Ms. Crawford, who can be assured of lifetime employment dealing with regulatory bodies), but it is part and parcel of the way the Internet works.

The only situation in which government should intervene in the Internet is in the case of anticompetitive tactics by providers — which are concentrated not on the level of local ISPs such as Comcast but in the markets for backbone bandwidth and wholesale transport. In short, Ms. Crawford and the “network neutrality” lobbyists are barking very much up the wrong tree.

By the way, in your message, there is one thing that you neglected to mention about “P4P.” Because it attempts to lighten the load on providers by causing file transfers to happen within their own networks, it raises two concerns. First of all, it does nothing to ease problems in situations where there are real constraints on bandwidth — e.g. for cellular data. For cellular carriers, the most scarce and precious resource is spectrum — and even if data flows only within the carrier’s network, it’s still consumed as excessively by P4P as by any other form of P2P.

Secondly, P4P discriminates against small, competitive, and independent providers. After all, the chances are far smaller that a file is available on the same provider’s network when the provider itself has fewer subscribers! The result of this, if you work out the statistics, is that requiring providers to allow P4P imposes a burden that is roughly inversely proportional to the SQUARE of the number of subscribers. In other words, if Comcast and “Mom and Pop Internet” (a competitive ISP which is 1/100th the size of Comcast) both allow P4P traffic on their networks, the backbone traffic cost to “Mom and Pop” network is 10,000 times as great. So, we may well see larger ISPs embracing P4P as an anticompetitive tool — a good reason NOT to require providers to carry it.