Award Winning Blog

Friday, November 9, 2007

Response to Two Columns on Comcast’s Traffic Management Tactics

Two columnists have offered their perspective on the Comcast peer-to-peer traffic management issue. See Larry Seltzer, eWeek.com, Network Policies Should Be Open, Not Neutral (Nov. 6, 2007) available at: http://www.eweek.com/article2/0,1895,2213092,00.asp and George Ou, ZDNet Blog, A Rational Debate on Comcast Traffic Management (Nov. 6, 2007); available at: http://blogs.zdnet.com/Ou/?p=852,

I agree with much of what they wrote, particularly the view that an ISP has a duty to disclose to subscribers what traffic management tactics the ISP can elect to use. Apparently Comcast and others do not want to commit to disclosing what traffic management tactics it might use ostensibly to preclude the onset of actual congestion. No one has disputed that Comcast forged TCP reset packets even though it appears that Comcast could have handled the actually occurring traffic volume without having to degrade anyone’s traffic.

On the other hand I endorse traffic management tactics that respond to actual congestion. I probably part company with some network neutrality advocates by endorsing an ISP offering premium services at a higher rate to power users, provided the ISP does not deliberately degrade service to standard service subscribers.

Neither of the authors addressed whether the current language, or the likely replacement language in subscription agreements, constitutes full disclosure that is fair. From my perspective ISPs cannot have it both ways by marketing “All You Can Eat” unlimited service and “blazing fast” bit rates only to establish, but not disclose quotas and bit rate throttle scenarios. That comes across as a classic “bait and switch” maneuver. ISPs should not be able to insert binding, “take it leave it” terms and conditions that include reserving the option of using “traffic management, “traffic shaping” and “rate-limiting” without defining the terms. ISPs should have to specify what these terms mean specifically as relates to monthly throughput quotas and bit rate throttling, and when such service degradation kicks in.

As the Internet matures and diversifies ISPs should have the option of targeting different consumer segments. As a light to moderate user of Internet access, I do not want to subsidize heavy users, nor do I want them bogging down the network and adversely impacting my service. But I also do not want a trigger happy ISP ready to punish power users regardless of whether these users have made it impossible or even difficult for the ISP to provide an adequate level of service.

I appreciate that ISPs need to recoup their sizeable network investments that seem to grow as more subscribers access bandwidth intensive services. But forging TCP resend packets comes across as a sneaky way to delay having to upgrade networks, or to establish the need for surcharges or rate increases.

Tuesday, November 6, 2007

In Praise of Relatively Dumb Pipes

Comcast's furtive and undisclosed traffic manipulation reminds me of a curious, red herring asserted by some incumbent carriers and their sponsored researchers: that without complete freedom to vertically and horizontally integrate the carriers would lose synergies, efficiencies and be relegated to operating "dumb pipes." For example, see Adam Thierer, Are "Dumb Pipe" Mandates Smart Public Policy? Vertical Integration, Net Neutrality, and the Network Layers Model, 3 Journal on Telecommunications & High Technology Law 275 (2005)

Constructing and operating the pipes instead of creating the stuff that traverses them gets a bad rap. It may not be sexy, but it probably has less risk. But of course with less risk comes less reward, and suddenly no one in the telecommunications business is content with that. So incumbent carriers assert that convergence and competitive necessity requires them to add "value" to the pipes.

Put another way, they would assert that any limitation on a carrier's "right" to add value is an unconstitutional taking. Of course we used to have common carriers that operated as neutral conduits carrying the content produced by someone else, but apparently that is an anachronism now.

The dumb pipe argument comes across to me as disingenuous. Would anyone buy an argument from an electricity carrier that it should not have to provide a neutral conduit for the carriage of electricity? It would seem that everyone makes more money and has more fun using the electricity to make something more valuable than just carrying electrons.

So it appears with Comcast. Hellbent to cash in on convergence, or at least generate greater returns for its pipe investment, Comcast wants to operate a non-neutral network with all sorts of intelligent packet sniffers ready to prioritize or degrade traffic. And I thought consumers would beat a path to Comcast instead of Verizon, because Comcast offered faster and better service. Who would want that when they can have a smart pipeline whose genius owners stand ready to delay and drop packets according to some secret and real smart plan?