Divide and Conquer

I’ve argued that the Network Neutrality battle is really about a number of different issues, some of them having little to do with transparency, blocking, and unreasonable discrimination.   Taken together, they are a hairball of  corporate tussles and consumer grievances.   Sometimes, the best way to solve a complex problem is to break it into simpler problems.  This is one of those cases.

FCC Chairman Tom Wheeler has started to do that.  Last Friday, he announced an investigation into the set of issues surrounding interconnection, or “peering”.    To consumers,  recent periods of unacceptable video performance involving Netflix, their transit provider Cogent, Comcast and Verizon look like unreasonable discrimination.      They are actually caused by congestion at points of interconnection between Cogent and the broadband ISPs.  This in turn is caused by a commercial dispute over terms of peering arrangements.    This issue has overshadowed and confused the Network Neutrality (or, as the FCC calls it “Preserving the Open Internet”) debate.

The Internet is literally just that: a network of interconnected networks.  It is an abstraction, realized as individual networks,  the physical connectivity between them, the data exchanged among them using the Border Gateway Protocol (BGP) and some administrative legerdemain.    While it is a truism that “nobody owns the Internet”,  some entity owns each of the networks that comprise the Internet.

Interconnection is a dark corner of the Internet in need of sunlight.  To tell the truth, a lot of it was a complete mystery to me until fairly recently.  Brough Turner wrote up a great tutorial on the topic.   Please read it before continuing with this screed.

The current interconnection mess  is yet another example of the Internet’s growth pains.  Friendly handshake agreements that “I’ll take your traffic and you’ll take mine, and it will all even out”  functioned well in the environment in which they evolved.  This reciprocity worked  because the economic interests of the networks were symmetrical, with each benefiting equally from the interconnection, and each bearing similar costs.   To the extent that there was paid peering, overage charges or port congestion charges,  it wasn’t a serious problem that pricing and terms were individualized and confidential.   In a world where traffic flow has become highly asymmetrical, and the interests of networks which are content providers diverge from the interests of networks that are access providers, the old, informal system has become a point of contention.

To be clear, a press release and a news conference do not constitute a regulatory action by the FCC.  They must adhere to a process specified under the law and their own procedures.   So this is Step 0 of the process.  What it means is that some of the beleaguered staff (presumably most of the from the Wireline Competition Bureau) are conducting an investigation into the way interconnection works.  They have access to confidential contracts, such as the one between Netflix and Comcast, and get to hear all sides of the story.  The information they gather will inform the formal regulatory process if and when it goes forward.

This is a positive development.  If it plays out to its logical conclusion, everybody will have the rules in writing, and temptations to use interconnection as an anti-competitive weapon will be thwarted.   Stay tuned.

By the way, thanks to Jim Sackman for the shout-out.  Jim and I used to work for competing organizations, and either because of or despite that,  share a lot of perspective.  He tries to blog on Network Neutrality every Friday, and it’s well worth a read.



One thought on “Divide and Conquer”

  1. Dan, the FCC lacks a good, objective, convergence framework that can handle both qualitative and quantitative inputs to assess critical supply and demand trends driving costs and hence pricing.

    It starts with understanding how internet (data) peering grew out of the competitive voice networks and how the latter developed to begin with. Initially data peering was a pimple on the voice elephant’s behind. Now it is the elephant. And whereas most data was essentially 1 or 1+ way and store-and-forward, today the traffic is increasingly 2-way and real-time.

    But at the end of the day it comes down to cost. Because of 30 years of competition, the cost to transit the WAN for a voice minute is $0.0000004 (100,000 minutes for a penny). But in the MAN the cost is closer to $0.001 (10 calls per penny). That’s the single biggest issue with respect to both absolute and relative issues when it comes to peering.

Leave a Reply

Your email address will not be published. Required fields are marked *