Thanks!

An anti-Net Neutrality Op-Ed in today’s Seattle Times, “Hysteria Makes for Bad Law“, by information technology provider Avis Yates Rivers, is so riddled with omissions and sloppy logic, I do not know where to start taking it apart. Willful failure to explain what’s happening makes bad law too, Ms. Rivers.

Guest Post by David Isenberg

The article ridiculously recasts Comcast’s recent interference with peer-to-peer traffic, in which an Associated Press reporter caught Comcast forging “reset” packets to disrupt transmissions from such applications as Bittorrent and Lotus Notes, as “a very good reporter’s innocent mistake.”

There was no mistake. Comcast was blocking its customers traffic based on the kind of traffic it was. (Further, Comcast never admitted doing it despite the red-handed evidence, and never renounced it. As far as we know, they’re still doing it.)

The supposed mistake was that the AP reporter did not wait long enough. The Op-Ed says, “Had the reporter simply waited, the file would have arrived unadulterated once the bandwidth became available.” Aha! Arrival time does not matter. What do we need the Internet for? Why not send the file by pony express?

The sentence above has another red herring, “once the bandwidth became available.” This is complete nonsense. The bandwidth was available — indeed, Comcast was sending its forged bogus signals over said precious bandwidth.

The heart of Yates Rivers rhetoric is the conflation of connection and application. She writes, “There’s nothing objectionable about [application] management as long as providers do so in a content-agnostic manner.” But, “content-agnostic” means more than, e.g., don’t favor Mitt over Rudy. Blocking certain applications while favoring others is itself a violation of content agnosticism. Bittorrent is currently the best way to deliver Internet video content. Was Comcast blocking Bittorrent to keep Internet video competition with cable TV at bay? The medium is the message.

It was much simpler back a few years ago when the Internet had to be stupid, that is, when computation was so scarce that you couldn’t do deep packet inspection at wire speeds. That little hiatus was when the Internet matured and proved that a network that would just deliver the packets was a fertile innovation environment indeed.

Today, simple is still better. If our Internet connection were a simple Ethernet interface, it’d cost about $15 per household. But what we get from the telco (or cableco) is an interface that supports telephony, TV and Internet. It costs ten times more. And when you start snooping into the Internet packets themselves, network complexity (hence infrastructure cost) goes up even more.

Andrew Odlyzko, in work that is at least a decade old, showed that most Internet resources are way over-provisioned, and that congestion virtually always comes from a single overloaded node or facility. As far as we know, this is still true. Where there is a spot overload, it is cheaper to simply add capacity than it is to manage the packets and what’s in them.

Another fact: It is approximately as cheap to implement a Gigabit link as a 56 kbit link. Despite the 18,000-fold increase in speed, most of the cost is in the right of way, the wire (or fiber), the power supply, the network-facing interface, the customer-facing interface and the modulation/demodulation logic. How fast the algorithms run is a trivial cost in the overall system. Indeed we have the **technology** to never be bandwidth limited ever again. And the technology is affordable. We could put Gigabit fiber Internet access into every household in the USA for about three Iraq-budget-months.

So mostly Yates Rivers is wrong when she says that bandwidth is finite. Where it is finite, the blame lies at the feet of the telcos . . . well, not really, they wouldn’t be so stupid as to build such abundance that they have nothing to sell anymore. The blame lies with our limited vision — we have affordable, mature technology that would make bandwidth scarcity as obsolete as horsepower from horses.

In those rare cases where a node or facility is a bottleneck and can’t be up-sized, you still don’t have to do what Comcast did. Instead, it is much simpler to count packets and have explicit tiers of service.

– David S. Isenberg spent 12 years at AT&T Bell Labs until his 1997 essay,”The Rise of the Stupid Network,” was received with acclaim everywhere in the global telecommunications community with one exception — at AT&T itself! So Isenberg left AT&T in 1998 to found isen.com, LLC, an independent telecom analysis firm. Isen.com is producing F2C: Freedom to Connect March 31 & April 1, 2008, in Washington DC. Read this and other writings at his popular blog.

This entry was posted on Thursday, December 20th, 2007 at 11:53 am and is filed under Uncategorized. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.