The Consumer Cost of Inefficient Increases in Capacity Necessary to Meet QoS Expectations


                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                        

Net neutrality[1] will, almost certainly, increase the cost of end-consumer[2] connections. The logic behind this is simple and requires only minimal understanding of economics; in short: requiring equal treatment of all data will mean that capacity increases will be the only way to ensure quality of service for important or time-sensitive data[3]; such increases in capacity will, necessarily, be inefficient[4]; increased inefficiency means increased operating costs[5]; and the consumer always pays.

Equal Treatment of All[6] Data

"Net neutrality" means many things to many people. The plethora of definitions serves almost exclusively to confuse stakeholders and to give cover to Government mis-/malfeasance. So, I'm going to add another definition[7]:

net neutrality — a requirement that a network operator treat all data transiting its network, regardless of type, source, destination, or any other characteristic, equally

I believe that this definition strikes the appropriate balance between recommendations that do not go far enough (e.g., those that cover only consumer connections) and those that go too far (e.g., those that seek to micromanage network operators and their businesses). Anything short of this definition would leave open room for circumvention[8], and anything more encompassing than this definition would lead to unacceptable welfare losses due unnecessary increases both in the regulatory burden and in inefficiencies. In short, this definition is optimal.

Capacity and QoS

It should go without saying that not all data transiting a network are equal. However, net neutrality seeks precisely equal treatment for such clearly unequal data. This presents problems for end users and for network operators. End users wants their important or time-sensitive data ("ITSD") (e.g., 911 calls) to be treatment as ITSD, and network operators seek to be able to provide this quality of service ("QoS") to end users/customers. When a network operator is required to treat all data equally, the only way to ensure QoS for ITSD is to increase network capacity. However, such increases in capacity must, necessarily, be greater than necessary.

Without going into the technical details, suffice to say that networks typically do not have sufficient capacity for all end users to utilize one hundred percent of their allocated capacity at any given time. In the real world, a good example of this is the decreased connection quality and download speeds experienced when many users simultaneously make large demands on a given network (e.g., multiple people in the same household streaming Netflix simultaneously). In order to increase/ensure QoS for ITSD, network operators must increase capacity such that spikes in demand will have less[9] impact on network performance. This is necessarily inefficient.

By prioritizing certain types of data over others, network operators can ensure that networks with limited bandwidth will still be able to meet the needs and the demands of end users. This could take the form, e.g., of a Web page loading a few seconds slower to ensure a VoIP call is not interrupted or degraded. We all benefit from this kind of prioritization and many of our home networks do this automatically[10]. Net neutrality removes this possibility[11]. Without allowances for QoS, network operators must continually increase capacity beyond what is truly necessary. This is inefficient and expensive.

Inefficiency

Think of a network as being a highway and the bandwidth on the network as being the number of lanes. Almost any highway[12] will have peak and off-peak usage periods. During off-peak times, all drivers can coast along happily at the speed limit[13]; as we all know, this is frequently not the case during peak times/rush hours.

In essence, net neutrality would be the equivalent of requiring cities and States to construct highways of a sufficient size to accommodate rush-hour-level traffic flowing at the speed limit. As a native of Los Angeles, I cannot even begin to imagine what that sort of requirement would do to, say, the 110, the 405, or the 5. Even without the absurd mental image of a fifty-lane highway, the problems here should be apparent.

For networks, the infrastructure problem is not quite as challenging as it is for highways[14]; however, the cost issues are still present. Further, network infrastructure must be upgraded/replaced more frequently than transit infrastructure (we were driving cars more than a century ago; we were not streaming on-demand media twenty years ago). Just as we would not want our Governments to waste taxpayer resources on impractically, inefficiently oversized highways, we, as consumers, do not want network operators building and maintaining massively unnecessary capacity. The reason for this is straightforward: Cost.

The Consumer Always Pays

A truism in economics, this simple fact is vitally important to remember when proposing regulatory regimes. Larger, more capacious network infrastructure is, naturally, more costly. Network operators will pass these additional costs on to consumers. It is, to some degree, a deep irony that the net effect of net neutrality regulation will be to increase consumer prices, the very thing[15] many net neutrality advocates say they are attempting to avoid.

Further, the burden of these increased costs will fall disproportionately on the poor. If the world of the fever dreams of the most ardent net neutrality advocates were to come true and a tiered system of Internet access were implemented by all ISPs[16], the poor would very likely content themselves with a lower level of access[17], and, further, such 'bottom'- or 'basic'-tier access would likely be both sufficient and relatively cheap (i.e., a tiered system would likely actually benefit the poor). Under net neutrality, the cost of Internet connectivity will almost certainly see a relative increase. For those who are not at the bottom of the economic ladder, such (likely modest) increases in price will be quite bearable. The same is very likely not true for the poor, who have more limited discretionary incomes[18]. While not determinative or decisive, the disparate impact upon the poor must be considered.

Will the consumer be better off with or without net neutrality regulation? As the provision of Internet access is a service, this question should be the primary concern of those seeking to craft and to implement a net neutrality regulatory regime. As for cost, it seems almost certain that costs will increase for consumers if net neutrality regulations are implemented.

The central cost consideration, then, is this: Are the likely benefits to consumers from the implementation of net neutrality regulations greater than the expected increases in costs? It is an open question, and the answers are not so simple as many would have us believe. We should not act rashly and, in our haste, risk harms greater than the ones we seek to avoid.


  1. While this article will give a definition of this term, infra, a full examination of the term and its uses is beyond the scope of this article. ↩︎

  2. Whether individuals or businesses. ↩︎

  3. e.g., VoIP calls (particularly 911 calls). ↩︎

  4. i.e., they will exceed what is actually necessary to meet the reasonable needs/demands of consumers. ↩︎

  5. Both in terms of capital investments and ongoing upgrades/maintenance. ↩︎

  6. While exceptions would, almost certainly, be a part of any finalized version of net neutrality, this article deals with such exceptions only in passing. A subsequent article will deal with exceptions in more depth. ↩︎

  7. While the irony is not lost on me, I fully believe this should be the definition of net neutrality. ↩︎

  8. e.g., data originating and terminating within the same network could be treated preferentially by the network operator. ↩︎

  9. Increasing capacity to the point where spikes in demand will have no impact is virtually impossible (i.e., it is cost prohibitive to even attempt to do so). ↩︎

  10. Prioritizing VoIP is actually one of the most common QoS measures. ↩︎

  11. While exceptions could be made, this article does, as noted, supra, dwell on those exceptions or analyze them in detail. Nevertheless, there is a brief treatment of them, infra. ↩︎

  12. Except, perhaps, those in the most rural parts of the country where these concerns would never manifest. ↩︎

  13. That's my story and I'm sticking to it. ↩︎

  14. After all, newer, faster networks seldom use more physical space, which is generally a requirement for expanded/improved highways. ↩︎

  15. At least one such thing, and, arguably, one of their two primary objectives. ↩︎

  16. n.b., this is a virtual impossibility in a functioning market economy. ↩︎

  17. n.b., this is not to say that the poor do not want access to higher-tier services, but merely a recognition of the fact that basic Web browsing, email, et cetera, would almost certainly be included in any "basic" tier of access. ↩︎

  18. Internet access, no matter what some on the fringe scream at each other and passersby on the Internet, is not a right. ↩︎