Feb 26

Andy Oram

Andy Oram

Network neutrality: code words and conniving at yesterday's FCC hearing (Part 2 of 2)

Yesterday I summarized the public FCC hearing about bandwidth at the Harvard Law School, and referred readers to a more comprehensive background article. In this article I'll highlight some of the rhetoric at the meeting, which shows that network providers' traffic shaping is no more sophisticated or devious than the shaping of public perceptions by policy-makers and advocates.

Broadband: quick and dirty

Our information highway has hit a dead-end due to a couple of cheap tricks that incumbents took to meet public demand for more bandwidth in the late 1990s. At this time. when the need was to serve up graphics-rich web sites, the public made a bad deal, paying cable companies to convert a channel or two to Internet access and phone companies to adapt their copper wires to high-frequency digital transmissions (asymmetric in both cases). This meant nobody had to pay for fiber to the home, but the trade-offs caught up with us when streaming media became popular.

Comcast's much-criticized throttling of BitTorrent streams is a finger in the dike. Naturally, when put under the spotlight, they tried to minimize the extent of the traffic changes in their submission to the FCC:

More specifically, Comcast's network management practices (1) only affect the protocols that have a demonstrated history of generating excessive burdens on the network; (2) only manage those protocols during periods of heavy network traffic; (3) only manage uploads; (4) only manage uploads when the customer is not simultaneously downloading (i.e., when the customer's computer is most likely unattended) ("unidirectional sessions" or "unidirectional uploads"); and (5) only delay those protocols until such time as usage drops below an established threshold of simultaneous unidirectional sessions.

In his spoken testimony, Executive Vice President David L. Cohen also stressed that the restrictions on peer-to-peer protocols affected only a small minority of customers. This is certainly true (although it isn't smart to try to keep blocking secret when the small minority of customers affected are precisely those who have the technical sophistication to uncover the secret).

But peer-to-peer offers obvious advantages to sites trying to compete with YouTube without incurring its enormous bandwidth costs, so the small minority will eventually become the mainstream.

Truly on the edge

Things are even harder if you're a small rural provider, like the outspoken Brett Glass, who has been single-handedly broadcasting the view of mom-and-pops in the Internet service business when others speak from theory. Brett's opposition to network neutrality is expressed in numerous postings to forums as well as comments to the FCC.

Brett, who is also known as a FreeBSD expert and outspoken proponent of its approach to free software, has kept a wireless ISP called LARIAT going in Laramie, Wyoming against all odds for years. Nobody at the FCC hearing spoke for Brett: not the large incumbent providers, and not the network neutrality advocates (although they support an environment that would let many LARIATs flourish).

Brett couldn't handle increased traffic by buying more bandwidth. Where he's located, there is no more bandwidth. He already has to string a line 50 miles to pay inflated prices to an incumbent phone company (because the three local fiber providers are owned by a single company that denies him access). Although more backbone bandwidth is available, the price is prohibitive. Furthermore, the last mile for him is wireless, and the narrow range of spectrum available to the public imposes its own limits.

What does righteousness buy you?

Essentially, whether you're Brett Glass making $5 per customer each month in Laramie or Comcast providing a cable channel for asynchronous Internet traffic (probably at artificially low prices, subsidized by the high cost of cable TV), you are underprovisioning in the hope of making giving customers just the access they need. So I snorted at yesterday's hearing when FCC chair Kevin Martin posed a leading question to a peer-to-peer video provider:

The service doesn't let customers exceed any limit imposed by their providers, does it? It only lets customers efficiently use the bandwidth that they've already purchased.

This is like analyzing a bank failure after a run on the bank by asking:

The withdrawals didn't let customers take out more money than existed in their accounts, does it? They only let customers efficiently use the money that they've already invested.

Martin knew very well that current U.S. networks can't support sustained use of the full capacity, any more than banks can pay out all their customers' funds.

No, this is not the open network that network neutrality experts describe as the true, pure Internet. Perhaps they are right that the service should not be marketed as Internet service. But no change in terminology will help the situation until we invest in better networks. Righteousness and $29 a month can buy you 10 Megs, downstream.

A possible fix: easier interconnection

I'm lucky: I am in one of the areas where Verizon finally strung fiber (some 20 years after the telcos first promised to do it). I welcome this investment, but in the short run it will widen the gap between information haves and information have-nots, such as the people just an hour's drive West of me, represented at the FCC hearing by Massachusetts state rep Daniel E. Bosley. They're still living with dial-up there.

Will the incumbents' fiber plans eventually turn a luxury item into a commodity, as happened with automobiles and cell phones? High-speed networking is more likely to resemble electrification, which spread to remote areas either through government funding or community self-help measures.

I said earlier in this article that the public made a bad deal with cable modems and ADSL. Congress and the FCC made a bad deal too. Congress recognized in the 1996 Telecom act that competition was important in order to build out networks, but assumed that a few large companies would compete with each other.

Congress also recognized that nobody could build a whole parallel network from scratch, so they ordered the FCC to enforce interconnection rights. The idea was that small ISPs could get a start by stringing some of their own lines (and now setting up wireless stations), by leasing and reselling others, and by attaching their systems to the incumbents' networks.

The FCC came up with no less than 14 interconnection points that incumbents were forced to offer competitors, but happened to omit the one or two interconnection points that could really make a difference.

Perhaps the worse lapse of the FCC was to limit interconnections to phone companies. Nobody could string a fiber to the phone company's box without incurring enormous financial and legal costs. This rule is like saying that nobody can earn interest on a bank account unless they're also a bank.

In other countries where interconnection is easier, many small institutions run fiber, and bandwidth penetration tends to be higher than the U.S.

Markey lays the groundwork

I've chosen to finish my examination of yesterday's FCC hearing where it began: a speech by Representative Edward Markey, head of the House's Subcommittee on Telecommunications and the Internet. Markey is my representative, and I greatly admire his work in the telecom area as well as others. Still, I had to analyze his charming and stirring speech from a rhetorical perspective. In about fifteen minutes he managed to slip in quite a lot of code words--seemingly everyday terms that embody significant and long-held agendas. Looking at these words and their true meanings can help to summarize the competing viewpoints in today's Internet debates.

Internet providers are access providers, not service providers

This assertion denies to Comcast and Verizon the right to promote services and content by altering traffic on the underlying network. More deeply, it puts forward a doctrine that is popular among network neutrality advocates but by no means universally accepted in the Internet community: that the physical conduit (and minimal software protocols to convey data) should be administered by a different entity than the information flowing over the conduit.

At its most extreme, advocates of the doctrine would either have governments provide the physical network, or shake up the industry again like the famous 1982 Final Judgment that split AT&T apart. Network providers would have to divest themselves of either their physical network or their content and service divisions.

Competition is the solution, because there was no creativity without competition

Thus did Markey anticipate the many calls by panelists for a return to the original competitive premises of the 1986 Telecom Act. I have already explained how the Congress applauded competition more as an ideal than as a reality, and how widely we missed the mark in the implementation of the act. Still, history shows the power of competition. The question is whether we can recover from the exhausting battles of the past decade and succeed where we failed before.

Solve problems through high speed as opposed to rationing

This doctrine holds that network providers should pay for whatever bandwidth is needed to support their customers. Such was the conclusion reached by the research consortium Internet2 after decades of trying to implement Quality of Service through resource reservation and traffic shaping. The powerful lessons of this experiment were picked up by others in the Internet community as proof that we can solve our bandwidth needs only by adding more.

Of course, life is always easy in the absence of scarcity. But is it any more feasible to keep adding bandwidth to the Internet than it is, say, to keep adding lanes to roads or docking gates to airports?

Advocates say yes, approaching the solution from several standpoints. Bundles of fiber could offer bandwidth so far beyond current needs that we can assume they'll serve us throughout our lifetimes. Protocols and brand new technologies (such as optical switches) can theoretically match or exceed the magic of Moore's Law in computer processors. And if we just do have to empty out our pockets and pay for expensive bandwidth, it can create so many other efficiencies in the economy that it pays for itself.

And thus started a hearing that revealed more about the problems participants would like to have than the problems we actually face. I haven't bothered to report the traditional "government regulation doesn't work" doctrine that we heard briefly from Republican commissioner Robert M. McDowell as well as from Comcast's Cohen. The interesting fact is that this claim wasn't raised particularly often; it seemed as if everybody realized the status quo is unacceptable and that the FCC has to do something.

Certainly a lot is at stake. It was revealed the next day that Comcast tried to pack the audience. What bothers me more is that everybody tried to pack our heads with fine notions and not honor us with explanations of the hard trade-offs they entail.

tags: comcast, internet2, markey, verizon  | comments: 1   | Sphere It

Previous  |  Next

0 TrackBacks

TrackBack URL for this entry: http://blogs.oreilly.com/cgi-bin/mt/mt-t.cgi/6341

Comments: 1

  Michael Weinberg [02.27.08 03:15 PM]

The Markey Bill is an important first step in bringing elements of internet policy under the same non-discrimination policy that governed older telecommunications. We have an analysis of the "real" purpose of the bill on our policy blog;

Post A Comment:

 (please be patient, comments may take awhile to post)

Type the characters you see in the picture above.