Brian's Brain

The latest case study in my longstanding Network Neutrality scrutiny and coverage comes in two different implementations, albeit both from the same service provider. In late July, Comcast was verbally spanked (albeit not fined) by the FCC on a 3-2 vote for discriminatory throttling of Bittorrent traffic, although the legality of the FCC's action is debatable (and the lack of a corresponding fiscal penalty also begs a 'so what's the point' query). In response, Comcast has announced by-year-end plans to still dynamically throttle heavy network users, albeit on a protocol-agnostic basis, to 'DSL-equivalent' speeds for 10-20 minutes at a time.

More recently, Comcast last week made official the long-rumoured 250 GByte per/month bandwidth cap on its residential service, beginning October 1. At first glance, 250 GBytes seems like a whole lot of data, and as such I agree with Peter Glaskowsky that the cap won't affect the vast majority of today's Internet users. Unlike Glaskowsky, though, I also concur with Om Malik...Comcast's move effectively muzzles the next generation of Internet usage evolution (which power users like me are already tapping into), particularly as it relates to online video downloads and streaming, and specifically to high-def video.

Let's quantify my concern. Look back at my past trials of Apple TV, VUDU and the Xbox Marketplace, and you'll encounter a range of high-def Hollywood movie payloads reflective of varying play times, frame sizes, codecs, and encoded bitrate aggressiveness:

Average them out and you end up with a 4.25 GByte 'typical' file size. Now oversimplify the average consumer scenario, assuming that the Comcast broadband tether will only be used for video distribution, and you end up with the ability to download or stream 58 movies per month, or roughly 2 per day.

Again, that seems like a lot. But of course, consumers aren't using their Internet connections solely for high-def video downloads. An increasing number of homes have multiple LAN clients simultaneously tapping into that capped collective bandwith, too. I'd tell you exactly how many Internet-capable widgets are under my roof, but it's embarrassingly excessive...at least given current mainstream metrics, that is. But remember, today's power users are tomorrow's typical consumers. And anyway, as the per-movie file size assumption increases, the per-month movie allotment proportionally plummets. At 6.1 GBytes, for example, you're only able to download or stream 40 feature films per month before Comcast starts complaining...

...which leads to one of my objections to Comcast's cap. Fundamentally, it's going to encourage content providers to compress video in an increasingly aggressive manner in order to fly under service providers' radar, with increasingly egregious visible and audible side effects. As I wrote back in late June:

Economic factors, coupled with consumer ignorance, make the approach quite tempting to a variety of content providers and their delivery intermediary partners.

And what of my network neutrality angle on Comcast's actions? Consider, for example, that the vast majority of Comcast cable Internet subscribers are also Comcast cable television subscribers. Now consider that whereas cable Internet-based video content viewing is bandwidth-capped, cable television viewing is not. Voila; an unjust in-house service emphasis, and one that from a societal standpoint is even more disconcerting if Comcast were, say, to decide to favor one political party over another from a cable television coverage standpoint in the ramp-up to the upcoming November U.S. elections.

The service unfairness argument may conceptually also hold true for VoIP, for example, although given the substantial bitrate disparity between it and high-def video, the practical likelihood that VoIP will push a customer over the cap is remote. Still, will Comcast be more forgiving if one of its broadband customers exceeds the monthly bandwidth allocation due to Comcast VoIP use versus with, say, Vonage? And Comcast's stance is particularly baffling when you consider that Internet backbone capability growth continues to outpace users' incremental bandwidth demands.

Lest you think the scenario I paint is too futuristic to be of practical concern, I'll offer up some relevant datapoints from just the past few days:

• All of the major movie studios have now pulled out of South Korea from a DVD distribution standpoint. Pragmatically, this probably has at least something to do with rampant continued copyright violations by optical disc pirates. But it also has a great deal to do with the pervasiveness of high-speed broadband as an alternative content distribution scheme.
• Just yesterday, a spokesperson from Samsung (a particularly loyal historical supporter of Blu-ray) forecasted that the format might be dead in as little as five years in the face of the online alternative. Lessee, where (and when) have I heard that before?
• And further to the burgeoning momentum of online video, as I forecasted last week, Amazon just went public with its Video On Demand streaming service as a more platform-agnostic adjunct to Unbox.

As I've said innumerable times before, I understand well the technical and economic underpinnings to the desire to restrict per-user bandwidth consumption. See, for example, this excellent recent overview from Ars Technica if you need a tutorial from a backbone standpoint. However, I remain greatly concerned about the societal access-limits-on-diverse-information-perspectives impacts of such moves, as well as their negative impact on the free-market development of new and improved Internet services.

If you're curious to see if (and how) your ISP is mucking with your Internet connectivity, you might want to give the Electronic Frontier Foundation's 'Switzerland' utility a whirl (more from Slashdot). Don't forget to bring your favorite compiler along, though...all that's currently available is source code. With all due respect to the EFF and its honorable intentions, which I've steadfastly supported over the years, I daresay that the organization's aspirations in this particular regard will largely be for naught as long as it doesn't distribute O/S-tailored binaries.