Saturday, February 10, 2007

Net Neutrality is Politically Correct

"People who demand neutrality in any situation are usually not neutral but in favor of the status quo." -- Max Eastman

Andy asked me in the comments of my last post about net neutrality--what is it really and who cares?

The Internet Protocol treats all packets equally, but there are various reasons why someone might want a "fast lane" for packets. Content publishers and VOIP carriers deliver media streams whose packets expire if they arrive late. E-commerce vendors lose money when impatient shoppers abandon carts due to router congestion caused by all those media streams. So demand grows for express services, and a number of technology vendors, ranging from startups to Cisco, increasingly offer ISPs technologies for delivering them.

Net neutrality arose from a concern that these new services would change the nature of the internet. Privatize the highways, people worry, and only the wealthy will drive. Rich incumbents will control the pipes, while innovative challengers must compete for the scraps of spare bandwidth. Proponents of net neutrality would therefore impose regulation on tier one ISP's so that they cannot deliver any services that discriminate among packets. (Readers of Kurt Vonnegut, Jr. will recognize this line of thinking from his story Harrison Bergeron in Welcome to the Monkeyhouse.)

The call for net neutrality is superficially appealing, in the same way that it's easy to oppose free trade in defense of your countrymen's jobs. But just like regulating imports, regulating ISPs with rules on net neutrality is short-sighted and, in the long term, terrible for both businesses and consumers. It's politically incorrect to say so, it's likely to get me flamed, and Google won't like me for saying it, but it's true.

Market forces will, as they always have, drive innovation on the internet. ISPs will find ways to accelerate and guarantee delivery with all sorts of interesting new services, and businesses who can deliver more value to their consumers through better internet performance can afford to pay for them. Should Fedex have been prohibited from competing against the Post Office, so that "big corproations" wouldn't have an advantage over the little guy? Of course not.

Proponents of net neutrality would counter that Fedex doesn't hurt the performance of old-fashioned mail, while express lanes for packets will necessarily slow down "free packets" pushed to the back of the line. On the contrary: allowing ISPs to profit from delivering express services for special classes of traffic will directly lead to the rapid development of additional internet capacity. There is no limit to the number of lanes one can build on the information highway, unless of course you regulate and cripple the only entities capable of building those lanes.

So far, this may sound like a reasonable technical issue to debate. But the campaign for net neutrality has transcended logic, manuevering instead to prevail upon Congress with an emotional appeal to the voters. "If we are silent, if we don't stand up for Internet Freedom," warns Hollywood star Alyssa Milano, "corporations will take away our right to choose!" As always, it's easy and popular to demonize corporations.

In his letter to the public (a great PR play, and a nice pander to regulators who look for reasons to work), Eric Schmidt wrote that net neutrality is needed to prevent broadband carriers from controlling what people say or do online. As I have blogged before, Eric is certainly a genius (I can pander, too), but this call to fear is wrong on so many levels, not to mention egregiously hypocritical. (Remember China?)

For one thing, accelerating a stream of packets, even at the mythical expense of some random packets, does not "control what people do online." Also, ISPs are not public utilities; they are businesses whose owners--including individual investors and pension funds--have no legal obligation to amuse Eric with whatever internet sites he craves. (Should AOL and the mobile environments of AT&T and Verizon be legally forced to provide access to outside content?) Having said both those things, the market will not reward ISPs who effectively block or even slow access to the full array of web sites--there is demand for express traffic and free traffic, so both sevices should and would exist.

Finally, it isn't simple to decide what kinds of acceleration services count as neutral. Is Akamai style acceleration neutral? What about Netli style wormholes? What about monitoring services? Okay it's easy to say these don't count. Now what about treating video streams differently on an unpaid basis, to enhance customer satisfaction and better control traffic flows? What about building out IMS networks with ENUM servers for quality VOIP? How about blocking access to sites that are pornographic, violent, hateful, illegal (casinos), or ridden with viruses? What about selling IP service that can withstand DOS attacks? What about simply selling higher quality service plans that give more bursty bandwidth when needed? Who will authoritatively model the complexities that emerge to determine whether these upgrades are good or bad for overall network performance--Alyssa Milano??

Engineers have done pretty well so far building the internet without regulatory oversight. If we now erect a glorious bureaucracy of regulators who painstakingly review every upgrade to a broadband carrier, the one thing I am sure of is that US carriers will immediately lose market share to their competitors. The state of the U.S. internet backbone itself will freeze both in capacity and technology as the rest of the planet leapfrogs our creaky, petrified infrastructure.

"The Constitution is not neutral. It was designed to take the government off the backs of people." -- William Orville Douglas

Blogged with Flock


  1. Anonymous8:37 AM

    The core issue is "how do we pay for additional internet capacity and performance"?

    TANSTAAFL - We can either shift more cost to the distributors of high value content or we can go to a usage based billing model of a base level of service with additional charges for more data usage, or potentially, lower latency.

    The model is broken now because a very small minority of large bandwidth users are subsidized by the masses who read email and use a few web sites. The telcos are not going to raise the rates on the light users to subsidize the heavy users, so we either end up with lower caps for heavy users or they pay for additional quantity of data usage.

    This is also the fundamental flaw with thinking that TV distribution over the internet is a major threat to established TV distribution. Both legal short-form (most YouTube video) and mainly illegal long-form (BitTorrent copies of TV shows) distribution really only works as long as the percentage of the population using them is low. If this becomes more prevalent, the internet doesn't have nearly the capacity required. If there isn't more revenue involved, the ISPs aren't going to build more network capacity.


  2. Anonymous5:34 PM

    With regards to VoIP, ENUM, check out A private voice grid used by AT&T and other companies to route VoIP traffic, bypassing Internet. According to their web site, voice peering fabric routed 139 billion minutes in 2006.

  3. Anonymous5:22 AM

    From a purely technological standpoint, the net is built as a best effort mode of transport and at this point TCP makes the public internet "good enough" for voice applications which can not be buffered and CAN NOT arrive out of sequence. The "good enough" works much better for data transport. As far as capacity goes, current fiber capacity is under utilized and with advances in technology (particularly WDM technology that allows recognition of new light wavelength in the same strands of fiber) can increase the capacity almost infinitely. The bigger challenge is the last mile limitation and that is currently being addressed by several phone, cable, satellite providers and WiFi based ISPs.

    This by the way is all fueled by the consumer hunger for rich media content and the 1996 de-regulation of the telecom industry. So I say let the internet be!


  4. Yet, we can imagine a scenario where a provider moves to monopolize the infrastructure and create an untenable situation. It's like software patents: they are subject to abuse but, on balance, give small guys a chance to thrive against the excesses of a Microsoft of Google. Ultimately, the best outcome will be an efficient compromise between bandwidth providers and major consumers (a la Google). Regulation is not inherently flawed; bad regulation is.

  5. Anonymous3:29 AM

    This by the way is all fueled by the consumer hunger for rich media content and the 1996 de-regulation of the telecom industry. So I say let the internet be!
    Bali Hands