Net Neutrality: Too Much Spin, Not Enough Facts

7 Important Tech Regulatory Issues In 2015

(Click image for larger view and slideshow.)

The latest spate of arguments around net neutrality reveals that the blind partisanship and BS bias are escalating. The comic strip site The Oatmeal saying that net neutrality is about "freedom" is just as mindless as the infamous tweet by Sen. Ted Cruz (R-TX) comparing net neutrality to Obamacare.

If we truly want broadband innovation, we need to start digging into the facts and opening ourselves to modern ways of doing things, instead of posturing and trying to squeeze today's Internet into a 1996 regulatory framework by classifying it as a Title II utility service. Let's review three facts:

Fact No. 1: Traffic management is a necessary evil.
Robert Atkinson, president of the Information Technology & Innovation Foundation, made some astute points in a column he wrote recently for InformationWeek, in which he argued that broadband shouldn't be treated like electricity or cable TV. "Some bits, like email bits, can handle high levels of latency," Atkinson wrote, "while others, like video telephony bits, can't." Exactly.

It's just about impossible to let folks be "free" with a non-free resource. Bandwidth costs capital investment money, as any network engineer will tell you. And basic economics tells us that demand for any desirable good or service approaches infinity as its price approaches zero.

[Worrying that technology will bring doom and destruction is nothing new. See 8 Doomsday Predictions From Yesterday And Today.]

If there's no traffic management, then the only way for an operator to keep up with Internet demand is to invest continuously in network equipment and physical plant. And at some point, those investments cease to make economic sense. Besides, being smart about traffic management, either as an enterprise network engineer or as a public network operator, creates a better experience for customers.

But the truth about traffic management is that it's hard and imperfect. And it will always upset someone, because the engineer or operator has to make judgment calls about which traffic streams need managing.

Traffic management isn't the same thing as anti-competitive behavior. Trying to regulate traffic management is like trying to pass a policy that regulates which types of conversations people may have in the workplace, and for how long. If we're trying to prevent (note that I didn't say "regulate") anti-competitive behavior, there are better, more modern ways to influence behavior. Let's start with data.

Fact No. 2: Data is only as accurate as interpretation.
Data influences behavior, but so does the way you present data. Various reports on Netflix download speeds have shown massive fluctuations, leading to an emotional outcry that the network operators must be intentionally throttling Netflix's traffic to force it into paying for speedier interconnects. But base values and long-term trends are where the facts are, and the findings there aren't nearly so upsetting or scandalous.

When your base values are only 1 and 2 Mbit/s, massive percentage changes are inevitable. The Washington Post's Wonkblog reports that Netflix-provided download speeds on Comcast dropped from around 2 Mbit/s to 1.5 Mbit/s between January and March 2013. That's (gasp) a 25% decrease.

But look at the graph below for Netflix-provided ISP download speeds for all of 2013 and 2014 year-to-date. I've highlighted some points on the graph via line thickness and dotted lines. (You can download the numbers for yourself here.)

The diamonds mark the times when Netflix started paying an ISP for enhanced transport. The triangles are when the provider is using Netflix's free Open Connect peering arrangement. Some context questions may pop into your mind:

• Netflix was designed during a time when DSL was prevalent. So why is the Comcast dip from 2 Mbit/s to 1.5 Mbit/s such a big deal?
• If the end of net neutrality means impending Armageddon, why is the 22-month general speed trend up, even before those nasty peering agreements were signed?
• What's up with Cox Cable? I couldn't find any information about a peering agreement, but the speed with which it's delivering Netflix traffic increased steadily during this 22-month period.
• From June to July this year, there was a dip in Netflix performance on Time Warner Cable, Verizon FiOS, and AT&T U-verse that was similar to the January 2014 dip on Comcast, followed by a rise after Netflix signed peering agreements with those ISPs. Do you interpret this rise as a sign of "anti-competitive behavior," or do you interpret it as evidence that network performance improves when service providers have resources to invest in their infrastructures?

I have another question for the net neutrality faithful: Why is it OK for Netflix to invest in its own Open Connect peering arrangement (that costs Netflix money but doesn't cost the ISPs) in order to increase performance, but it's not OK for the service providers to ask Netflix to make a different type of investment?

Data is useful and influences behavior. Netflix is right to publish these stats; they keep service providers in line. I guarantee you that it's trotting out this performance data to cable companies and telcos in negotiations. But the data scarcely warrants the type of hysteria that John Oliver, The Oatmeal, and others have generated.

Comcast CEO Brian Roberts was recently quoted as saying that his company is a supporter of net neutrality. I don't take his assertion literally. I take it as: "We know that lots of people are watching us." The world has changed; it's impossible for a giant telco or cable company to get away with

1 of 2