It is rare for me to be on the same side of an issue as AT&T and Verizon and on the opposite side of Sprint and T-Mobile, but I think the new Net Neutrality rules that the FCC adopted this week are a mistake that will hurt consumers and the telecom industry.
I won’t take the time to go point-by-point through the various elements of the new rules. Plenty of people smarter than me on regulatory topics have written about that elsewhere. The two aspects that really have me concerned are:
- the inability to prioritize paid traffic
- the inability to impair or degrade traffic based on content, applications, etc.
I believe that these restrictions will lead to networks that will perform much more poorly than they need to.
The Importance of Prioritization
Thirteen years ago, while I was chief strategist for TeleChoice, I wrote a whitepaper using some tools that we had developed to evaluate the cost to build a network to handle the traffic that would be generated by increasingly fast broadband access networks.
In the paper I say “ATM, Frame Relay, and now MPLS have enabled carriers to have their customers prioritize traffic, which in turn gives the carriers more options in sizing their networks, however, customers have failed to seriously confront properly categorizing their traffic. There has been no need to because there was no penalty for just saying ‘It’s all important.’”
With the new rules, the FCC ensures that this will continue to be the case.
Think about it. If you live in a city that suffers from heavy highway traffic, if you’re sitting in slow traffic and you see a few cars zipping along in the HOV lane, don’t you wish you were allowed into that lane? Of course you do. Hopefully it even gets you to consider making the change necessary to use that lane. Why do HOV lanes even exist? Because it was deemed a positive outcome for everyone if more people would carpool to reduce the overall traffic. Reducing overall traffic would have many benefits including reducing the amount of money needed to be spent to make the highway big enough to handle the traffic and at the same time improving the highway experience for all travelers.
Continuing the analogy, if you’re sitting in slow traffic and you see an ambulance with its lights flashing driving up the shoulder to get a patient to the hospital, do you consider it an unfair use of highway resources that you aren’t allowed to use yourself? Hopefully not. You recognize that this is a particular use case that requires different handling.
Finally, extending the analogy one more time, as you’re sitting in that traffic (on a free highway) and you look over and see traffic zipping along on the expensive toll road that parallels the free highway, do you consider whether you can afford to switch to the toll road? I bet you at least think about it.
Analogies always break down at some point, so let me transition into explaining the problem that the new rules impose on all of us. Networks, like highways, have to be built with enough capacity to provide an acceptable level of service during peak traffic. Data access networks, unlike highways, have traffic levels that are very dynamic with sudden spikes and troughs that last seconds or less. While all telecommunications networks have predictable busy hour patterns, just like highways, unlike highways, the network user experience can be dramatically impacted by a sudden influx of traffic. This requires network operators to build enough capacity to handle the peak seconds and peak minutes reasonably well rather than just the peak hour.
Different network applications respond differently to network congestion. An e-mail that arrives in 30 seconds instead of 20 seconds will rarely (if ever) be noticed. A web page that loads in 5 seconds instead of 4 seconds will be easily forgiven. Video streaming of recorded content can be buffered to handle reasonable variations in network performance. But if a voice or video packet during a live conversation is delayed a few seconds, it can dramatically impact the user experience.
Thirteen years ago, I argued that failing to provide the right incentives for prioritizing traffic to take into account these differences could require 40% more investment in network capacity than if prioritization were enabled. In an industry that spends tens of billions of dollars each year in capacity, that’s a lot of money.
Why The New Rules Hurt Consumers and the Industry
Is the industry going to continue to invest in capacity? Yes. But the amount of revenue they can get from that capacity will place natural limits on how much investment they will make. And, without prioritization, for any given level of network investment, the experience that the user enjoys will be dramatically less acceptable than it could be.
Let’s just quickly look at the two approaches to prioritization I called out above that the new rules block.
Paid prioritization is a business mechanism for ensuring that end applications have the right performance to create the value implied by the end service provider. This is the toll road analogy, but probably a better analogy is when a supplier chooses to ship via air, train, truck, or ship. If what I’m promising is fresh seafood, I’d better put it on an airplane. If what I’m promising is inexpensive canned goods with a shelf life of years, I will choose the least expensive shipping method. Paid prioritization enables some service providers (e.g. Netflix or Skype) to offer a level of service that customers value and are willing to pay for that requires better than mediocre network performance, and for the service provider to pay for that better network performance to ensure that their customers get what they expect. The service provider (e.g. Netflix or Skype) builds their business model balancing the revenue from their customers with the cost of offering the service. This approach provides additional revenue to the network operators enabling them to invest in more capacity that benefits all customers.
Impairing or degrading traffic based on content or application is a technical mechanism that enables the network to handle traffic differently based on the performance requirements of the content or application. An e-mail can be delayed a few seconds so that a voice or video call can be handled without delay. This allows the capacity in the network to provide an optimized experience for all users.
Obviously, these mechanisms provide opportunities for abuse by the network operators, but to forbid them outright, I believe, is damaging to the industry and to consumers, and a mistake.