If there is one consensus in the very polarized technology world, it is that more and more people and more and more devices will be connecting to the internet. Consequently, the amount of internet traffic that will be delivered over these networks will explode – estimates by Cisco suggest that total internet traffic will grow on average 41% every year resulting in an almost inconceivable 327 exabytes per year in 2012. To put that into context, that’s the equivalent of ~84 billion DVDs!
The growing size and importance of the internet has pushed regulators and activists to advocate for new rules and regulations to preserve the internet’s independence and neutrality from political and corporate interests. This movement has been called “Net Neutrality” and seeks to make it impossible for network owners to gain too much power over the content and information that a consumer can access.
Now, I am a firm believer in the aspirations of net neutrality – I am no fan of “walled gardens” and am even less a fan of Comcast/Verizon/AT&T throttling access to content they don’t approve. I am also well aware that major network owners, at least those in the US, are granted public licenses by the FCC to operate, and thus have a moral and legal obligation to provide a public and valuable service. But, aspirations and obligations without a realistic assessment of technological and economic realities are meaningless, and I suspect, based on many of the arguments I’ve heard on the internet, that many net neutrality proponents don’t have a good grasp of why their Eden-esque visions of net neutrality may be missing some of the bigger picture needed to make sure their implementation is more effective than naive.
The fundamental reality around the economics of network construction is, in short, that it is expensive as hell. There are enormous upfront costs that need to be recouped, and there are plenty of maintenance costs and challenges. This has two direct consequences:
- Network providers tend to be large
- The cost of an incremental byte of data transmitted is marginal (because the equipment is already there), but the cost of incremental capacity is extremely high (because the provider has to buy new equipment, contract new construction/digging, bring it online, etc.)
What does this mean for net neutrality? The cynical view would be that the network providers are primarily interested in minimizing new capacity investments/maintenance costs and in finding ways to “jack up” prices on data transferred, for example by only allowing content and devices where the content providers/device manufacturers pay the network owner a handsome reward. There is no doubt that network carriers can be guilty of this mindset, but the carrier’s current reaction to the smartphone revolution by opening up their formerly “closed gardens” and the failure of walled gardens like AOL to dominate the internet service provider space suggests that something else is at play. In my mind, what will dominate the business priorities of the network providers as time goes on is one question: how will network carriers profitably provide access for the exabytes of content that will be traveling through their networks?
The best example of what may come is in the mobile phone space where AT&T has been caught off guard with the enormous network demands from Apple’s iPhone. It seems the general consensus is that AT&T has been behind on its network infrastructure investments, but the fact of the matter is it will be increasingly difficult for network providers to keep pace with the investments in capacity necessary to maintain service quality for its users. This is especially complicated in the mobile phone space because of three things:
- the amount of wireless spectrum available for use is being consumed faster than it is being made available (although this could be alleviated by new regulatory policies)
- the quality of network service is dependent on more than just the wireless spectrum, but also an extremely expensive-to-upgrade, rapidly saturating “backhaul” network
- the spectral efficiency (how much data you can cram down a particular wireless pipe) is reaching the theoretical limit defined by Shannon’s Law (chart below)
I’ll admit I haven’t fully crunched the numbers (its been my experience that its pretty difficult for a casual observer to find good data on the costs of upgrading network capacity or on the ability of femtocells/network topology changes to help carriers cope with new data demands), but I suspect that network carriers will soon hit a real wall in terms of their ability to profitably expand their networks to meet new demands. If this is true, then there really are only four options open in the long run for profit-seeking network carriers:
- Dramatically increase cost of network service – Ending “unlimited access” plans and increasing the average monthly service costs and cost per byte transmitted could alleviate the profitability constraint in the short-run; but, depending on the economics involved, if the network service reached a high enough price, new users would be deterred from using the service, resulting in user attrition for the service provider and the loss of a potentially valuable network access for society
- Introduce tiered traffic and practice heavy network management – Allowing the carrier to preference certain traffic (e.g., to users/businesses or content providers/application developers who have paid more, or to preference emergency/government network traffic, etc.) and practice heavier handed network management could alleviate the network traffic constraint without dramatically changing the cost of network access
- Some combination of (1) and (2)
- Government bailout
None of these options (especially #4) are especially attractive, but it is important for regulators and activists to keep these potential futures in mind when considering their proposals. I fear that the strict form of net neutrality espoused by many would drive carriers to adopt course #1: something which could cut short the full potential of the internet to only wealthy individuals and businesses. This blogger’s humble opinion is that the best outcome would be to use regulation to pursue a variant of #2 for the following reasons:
- AOL 2 is unlikely to happen: To believe that network carriers absent strongly worded neutrality regulations will rapidly clamp down on access is to believe that the failure of AOL’s walled garden and Verizon allowing Skype and Android on its network are flukes. The value from the openness of the “full Internet” is not trivial and no longer something that major network providers can ignore. And, based on Verizon’s recent promotion of the Droid by harping on problems with the closedness of Apple’s iPhone App Store, is something carriers have learned to use for their own purposes.
- Regulations can be implemented just around content neutrality: Of all the concerns that I have about network neutrality, the one that matters the most to me is equivalent treatment of content. After all, the internet in the US would suffer a great deal if every ISP/carrier only provided content sanctioned by just a single political party. The remainder of the provisions (universal device access, universal platform access, ability of third party networks to piggyback off of the main network) are much less important, especially if content neutrality is mandated, and implementation of them may even weaken the ability of the industry to create high-quality integrated/tailored products (e.g., requiring AT&T to allow the iPhone to access Verizon’s network as well would have significantly delayed the iPhone’s release and probably would have worsened its battery-life and introduced additional problems). Furthermore, the cost of implementing content neutrality is likely minimal, whereas the cost of requiring networks to support every possible platform/device/third-party are likely to be significantly more prohibitive.
- Enabled tiering/network management allows network providers to manage networks for higher quality experience: Strict network neutrality could limit the ability of network providers to manage the traffic on their networks, especially in times of peak demand, likely reducing quality of service for all parties. It also limits the ability of network providers to prioritize traffic that should be prioritized (e.g., emergency/government communications, traffic which businesses have network providers manage for them, or even packets related to keeping tabs on the network service quality or devices on the network) and potentially do basic things like block spam/DDoS attacks or manage the quality of internet experience by traffic type (the demands for web page viewing are very different from the demands of internet video and internet telephony).
I won’t claim to be the policy expert that knows what the right solution or sets of costs and tradeoffs are, but hopefully this gives a sense of some of the nuances behind the network neutrality debate.