IPUrbia

The Net Neutrality debate has been stumbling along for several years now, and still there is no real consensus as to what, if anything, is needed as a course of action.

The inherent obstacle, from both a technical and business level, to an open access approach to broadband engineering is that the fundamental economics of broadband are based on over subscription. It is very common (and usually feasible) to have hundreds or thousands of broadband customers (each with, say, 3-5Mb/s of service) going through a backhaul of 50-150Mb/s. This model assumes that each customer will only use their connection some of the time and, assuming that not all of the ISPs customers use the service at the same time, is generally a good model. This is roughly analogous to many business model that operates on a subscription basis. Every year, my local theme park (Six Flags over Georgia) sells season passes to hundreds of thousands of people with the understanding that only a small percentage show up on a given day. Gym club memberships, NetFlix, and telephony/cell phone providers all leverage this principle to maximize profit.

In the end, these services operate on a principle of engineering a solution that roughly takes into account both the expected behavior of the customer and what can actually be profitable. Both are important for a business to be both sustainable and desirable. Some services tend to err on one side or the other–the Department of Transportation for Atlanta, for example, seems to engineer its road system for what it can afford, as evidenced by the obvious usage in excess of capacity.

As far as Internet access goes, the conflict arises because of the evolution of the expected behavior of the customer with, necessarily, a corresponding shift in the economic realities of providing service. The oversubscription model based on web surfing (a human driven, intermittent activity) is heavily challenged by “peer 2 peer” video transfers (an automated, 24×7 constant load on the network). It is (very) roughly analogous to replacing your normal telephone usage with an autodialer or, in terms of a road system, replacing your morning/evening commute in a Honda with having a robot drive a semi around the city 24×7. Simple handwaving and long-winded discussion of “principles” doesn’t change the reality that the networks simply can’t handle an increased load without being redesigned and upgraded.

I’m going to detour for a moment and bring in my main point of discussion: the Internet as a platform. In the computer world, a platform generally means a set of hardware / software that enables other software to run. This is often used in terms of “PCs + Windows”. The main necessity here is consistency–if other people are going to be using a given “platform” as a basis for building not only their offerings, but their livelihoods, then it is necessary that it there is enough consistency that they can write their software on Tuesday and run it on Wednesday without the “platform” having morphed into something else entirely. No building houses on sand, here–the specifications need to be defined, communicated, and can only change at well defined intervals that give everyone involved an opportunity to update.

Over the last 5-10 years, the insurgent platform has been “The Internet”, or, probably more accurately, “The Web”. I think it is safe to say that by and large, the “Web as a platform” has arrived: we have the basis (”PC + Internet access + web browser”) that, thanks to (relatively) open and consistent standards, enables millions of people to build business models around delivering applications and content over “the Web”. Other people have tried various approaches to take advantage of the Internet as a platform, largely unsuccessfully. Vonage is a prime example–an Internet-based telephone service provider that has tried to leverage the Internet as a platform to deliver telephone service. What it ran up against is the inherent problem of using the Internet in this fashion–the Internet may be many things, but it is not consistent, especially in all the areas that actually matter to voice-based applications. It is not just a matter of “open standards”–the implementation of these standards has to be open and consistent, as well. Some ISPs are better than others, some backbone providers are better than others, and, in the end, you get a service that, regardless of whatever engineering efforts Vonage could (and has) thrown at the problem, will only work, at best, most of the time under most circumstances. Vuze, an Internet-based video content distributor, is the latest iteration of a company building a model on the “Internet” as a platform and running into the very problem that we’ve been describing–most providers will allow their service to run over the network, most of the time.

Given the real work technical and economic realities, “allow” is probably not the best description of the situation. In the end, the problem comes from the inescapable fact that the “service providers” built their networks to enable the “web” as a platform, not necessarily the “Internet” as a platform. Whereas most things that run in a web browser and generally require a large degree of human interaction work well, everything that falls outside of that box is a house built on sand, and is generally plagued by reliability and consistency problems. We are, unfortunately, many years away from the Internet as a platform–we’ll probably have the “web as a platform”, and “p2p as a platform” and “voice over IP as a platform” and so forth before we can reliably say “any Internet-based application will reliably work over the Internet”.

What does this mean for Network Neutrality? It means that disclosure is a very inadequate approach to this problem, that simply requiring Internet providers to be more open about disclosing their limitations, in the end, produces a very fractured Internet where most applications work only some of the time on some provider’s networks. This is huge step backwards and heads in the wrong direction of the ultimate goal: “all of the time on all provider’s networks”. For the Internet as a platform to be the reality, this does require consistency, not just within the individual service providers, but between service providers as well. This does require the FCC and other “oversight” organizations to mandate this openness and consistency, for regulatory entities to step up and, well, regulate. Given that the current FCC administration under Kevin Martin has apparently mis-understood their mandate to regulate as meaning “deregulation”, I’m only hopeful that a new administration and the increased pressure from Congress will change things…next year.

However, this doesn’t need to be all at once. For a myriad of reasons, we’re just simply not at the technical point of the Internet being able to support every possible type of application we throw at it. It is a bit absurd to require my ISP to support “24×7x365 HD television service”, given both the technical and economic absurdities of that, to date. But, we are at the point of being able to set in stone “the Web as a platform”; “voice over IP as a platform” shouldn’t be far behind; in the end, regulatory pressure is a necessity to force both consistency and progress out of an increasingly monopolistic (and monopolistic-minded) market. A clear mandate for today and a clear roadmap for tomorrow ensures that we get to the eventual point of the “Internet as a platform” with the ability to support any application and any service with a high degree of reliability.