Friday, April 18, 2008 - The Progress & Freedom Foundation Blog

The Economist's Confused Straw Man

Often a good source of technology news, The Economist has it all wrong on the "exaflood" and new traffic management regulations now under consideration at the FCC. Most likely, the august magazine is just being lazy when it relays this erroneous message:

Once again, alarmists are issuing dire warnings about the internet collapsing under the weight of its traffic. But that’s nothing new: they’ve been doing so since the 1990s.

Bob Metcalfe, who invented the Ethernet protocol for local area networks, once claimed that the internet was about to be overwhelmed by e-mail traffic. That was in 1996. A year later, Dr Metcalfe not only admitted the error of his doomsday prediction, but literally ate his own words—grinding his speech from a year before with liquid in a blender and quaffing the lot to cheers from his audience.

The latest panic started with a scare-mongering story in the Wall Street Journal last year, which concerned the rise of internet video and the inability of the network to handle it, especially at network edges where the internet enters the home. The author talked of the “coming exaflood”, referring to the exabytes (ie, billions of gigabytes) of HD video users would soon be downloading.

Others in the industry have continued to fan the flames, with cable companies like Comcast wafting the hardest.

It's true Bob Metcalfe did worry of an "Internet collapse" in the mid-90s. But I -- as the author of that supposedly "scare-mongering" Wall Street Journal article -- have never suggested any such thing.

I've said the exaflood -- or this third phase of the Internet dominated by video content, which can consume a thousand times more bandwidth than second-phase applications -- will require much more broadband capacity across the network -- last mile, edge, core, and data center. I've estimated that some $100 billion in new investment will be required in the U.S. over the next five years.

The Economist makes the same point:

Verizon is spending $18 billion pushing its FiOS fibre network out into neighbourhoods. Despite its woes, Sprint is pressing ahead with WiMAX, the faster, longer-range version of WiFi. And Comcast is making good progress with the latest cable protocol, DOCSIS 3.0, which should allow it to offer 100mbps within a year or two.

Comcast has also made its peace with BitTorrent. The two companies are to collaborate on addressing some of the issues caused by P2P. It has also teamed up with an outfit called Pando Networks which has a nifty traffic-management technology known as P4P. The technology helps P2P file-sharers find each other more selectively, thus boosting their download speeds.

So, one might ask, what's the problem? Right. There is none. If we allow service providers to build more capacity -- as Verizon, AT&T, Cox, Comcast, and others are doing -- and if we allow creative partnerships among service providers, content companies, P2P app vendors, CDNs, and data center companies, we will build a thriving, rich, new, interactive, photorealistic Internet.

But then The Economist turns around and writes:

there’s still enough bandwidth over the last mile for current traffic.

Enough for current traffic? Well, kind of. Do you ever wait longer than you'd like for web-pages or photos to download? Try streaming an HD movie over a 768 kilobit connection. Or uploading family movies over a 256k upstream link. Or doing HD videoconferencing on today's modular DSL and cable modem networks. The Comcast-BitTorrent dust-up only happened because there wasn't "enough bandwidth over the last mile." And that's why the service providers are furiously building more, as The Economist correctly notes:

And soon there will be a whole lot more [bandwidth]—at least for Verizon, Sprint and even Comcast.

So, again, one asks, what's the problem? Moreover, we don't merely want "enough bandwidth" to meet current demand. Like abundant computer cycles or digital storage space, new bandwidth will drive new applications and services once unknown or thought wasteful.

At this point, it's hard to know what The Economist thinks or wants. It says video requires huge new bandwidth but that there's no exaflood. It says the service providers are building lots of new capacity, which should transcend any "traffic management" concerns, but then it calls for new regulations to solve the problem. Yet again, what problem?

Anyway, here was my comment on The Economist website:

Dear Economist,

You've got it all backwards. The Wall Street Journal article you refer to, the one predicting a "coming exaflood," was about the huge consumer and business opportunities presented by new video and rich-media applications and the fast fiber-speed networks required to distribute them. There was nothing "scare-mongering" about the article. There were no "dire predictions." I should know. I wrote it. The "exaflood" is merely a term for the torrent of bits created by new video applications and delivered by new broadband networks. Your own story confirms the significant increase in bandwidth needed to deliver robustly even current generation video, let alone HD and Ultra HD video beyond. That was the whole point of my article and subsequent studies. We need lots more bandwidth to accommodate this "third phase" of the Internet, where video, which requires between a hundred and a thousand times more bandwidth that "second phase" applications, will dominate. If we allow people to build new optical and wireless networks and the storage, caching, and data center facilities to serve up all this new content, we should be just fine. No "Internet collapses" or crashes. If we get in the way with intrusive new regulations, then of course we won't build new capacity as fast as we otherwise might, and it's the innovations on the edge that will suffer. The Internet won't collapse, but -- just as thousands of dot-coms did the last time we over-regulated the Net -- David Isenberg's favored edge companies, which I also favor, and which are counting on new big bandwidth networks, will crash.


Bret Swanson

posted by Bret Swanson @ 7:26 PM | Exaflood