Network Management Facts and the Tragedy of the Commons

The debate over elements of the Net Neutrality (NN) policy platform promoted by a handful of advocacy groups is now focused on what techniques, if any, Internet service providers (ISPs) should be permitted to use for the purpose of managing congestion over their private networks.  This ConsumerGram provides a set of facts on which there seems to be no reasonable basis for disagreement, shows how these facts are important in resolving the network management debate,  reviews the main elements of the positions of the contending parties, and offers a consumer welfare perspective on the issues and their possible resolution. This ConsumerGram shows that, ignoring the facts on net management can lead to the wrong conclusions and policies that ultimately reduce economic benefits for the vast majority of online consumers.

 

Issues to Be Resolved

 

A coalition of advocates united by concerns about threats they perceive to NN recently petitioned the Federal Communications Commission (FCC) to find unlawful Comcast’s interference with selected peer-to-peer (P2P) Internet traffic.  The interference involved applications enabled by P2P protocol that permit users to exchange large files including high resolution movies and other bandwidth intensive content, but requires so much bandwidth that it can slow down all traffic on the network.  Comcast claims the right to manage its network and asks the FCC to declare that its practices are reasonable and fully consistent with the FCC’s Broadband Policy Statement.  Dozens of parties commented on the petition and what follows summarizes some of the main positions and their implications for consumers.     

 

Critical Facts Not in Dispute 

 

  NN advocates opposing ISP efforts to manage their networks do not always openly concede or emphasize them, but neither do they deny an assortment of facts about the Internet.  These facts are critical to any reasoned consideration of how to balance contending stakeholder interests and claims.  We are aware of little disagreement over the following critical presumptions about the current state of the Internet:

 

• FACT 1.  The Internet has finite capacity at any point in time.  The ability of the “network of networks” to move information is not unlimited.  Its capacity is exhaustible by the sum of demands placed on it.      

 

• FACT 2.  The Internet is a common user network.  It is used for a variety of applications and by numerous users and is a common user network in the usual sense of the term. 

 

• FACT 3.  There are network externalities.  The implications of ISP user choices are not limited to individual users.  Private choice may create social benefits or costs that affect other users.  That implies that usage rights are interdependent. 

     

• FACT 4.  Usage is not uniform across all users.  Some users consume more Internet capacity than others and thereby create greater demands on the limited network.  Reports of the extent of inequality in usage vary, but there seems to be general acceptance of estimates that five percent of users absorb of half or more of total capacity.  Moreover, the data suggest that the appetite of so-called “bandwidth hogs” is (in large part because of P2P applications) growing faster than usage by less avid users so that the maldistribution is likely to grow over time.[1] 

 

• FACT 5.  There is always a potential congestion problem.  Congestion results in circumstances (time or location) in which demand is so great that capacity is inadequate to provide “acceptable” service.  The result is delay, failure to deliver or other dilution of service quality.  Congestion may be caused by individual users, but it harms all users.    

 

• FACT 6.  All common user networks are subject to management and usage rules.  There are some exceptions, but we cannot think of any but trivial ones.  The FCC and others have likened the Internet – with its last mile, middle mile and backbone links – to a road network with on-off ramps, local streets and by ways, and Interstate highways.  Power grids, air transport systems, waterways and other networks are all managed to maximize efficiency and keep traffic moving.  Indeed, engineers will tell you that all networks are designed with management in mind as it is simply too costly to build networks large enough to handle peak loads.

 

 

The Opposition to Network Management

 

            It is not possible to summarize briefly the case against network management advanced by NN advocates.  There are several elements to their positions, which are not always clearly and succinctly stated.  The degree of opposition and the arguments vary among those who oppose network management and some of the viewpoints have changed over time.  In that context, we attempt here to address its most important elements. 

 

Claim 1: Don’t Worry!  There Is and Will Be Enough Capacity!  Opponents of network management by ISPs argue that there is no need to manage traffic because periods of congestion are relatively rare and can be addressed, over time, by continually investing more money for bigger and faster networks. 

 

The demand for and supply of Internet capacity are driven by different forces and there is uncertainty about the extent to which they will match in the future.  A variety of predictions about growth of demand have been made by independent sources and, not surprisingly, given the degree of uncertainty involved, there is significant variation.  The methods and data inputs used in forecasting models show varying degrees of rigor ranging from bad to reasonable, and after eliminating extremes, the results range from 50% to 100% CAGR. Given market dynamics on both the supply and demand sides, the fact is that nobody knows with any degree of confidence. 

 

However, some things are known.  We know that the cost of forecasting and capacity planning errors are great and also that some errors are more costly than others.  Thus, the costs of insufficient capacity are borne by all users and disproportionately by particularly heavy users.  The costs of excess capacity are borne in the short run by ISP shareholders and creditors, but in the longer run by all of us as a result of the financial distress and spillovers. 

 

            We also know the enormous costs associated with the inclinations during the Internet Boom of the late 1990s to assume that:  “If we build it they will come!”  The economic wreckage of firms who overbuilt backbone networks is still being repaired. 

 

            We know that financial markets are reluctant to underwrite the investment needed to complete local broadband distribution networks, which some estimate would require $100 billion or more in additional investment. Nor is that investment a guarantee against congestion.  Japanese officials, for example, say that even their networks, which are widely touted as the fastest and most robust in the world, are straining to handle current traffic loads.  

 

Claim 2: Users Are Entitled to Applications, Services, and Content of their Choice without Regard to the Impact on Other Users.   The notion of an unlimited entitlement to any good or service or action, without regard to collateral or downstream consequences, is unprecedented.  There are no absolute, unmitigated rights anywhere, including, quite notably, in the Bill of Rights.  To proclaim such a right for bandwidth hogs in the context of their usage of the Internet is to reach well beyond precedent and reason, and to beg the issue of what limits are needed when other “rights” and the “rights” of others are fully considered.  The essence of consumer welfare oriented economic policy is to balance individual rights with those of the community. 

 

Claim 3: Network Management is Anticompetitive.  NN advocates emphasize that ISPs, may make claims about managing congestion, when in fact they are merely protecting their core video distribution business by thwarting delivery of competing services.  A broader, but related, charge is that ISPs, the major ones in particular, will, in the guise of reasonable network management handicap downstream providers in markets into which they intend to diversify.  Consumer welfare could of course be harmed were either of those to be the case.  But, to date, proponents of this view have been unable to offer any empirical evidence of such behavior.  Without evidence, it would be premature and counter to consumers’ interest to ban network management practices that ensure efficient operation of networks for subscribers’ benefit, simply because of some theoretical harm.  

 

            It is notable in this context that the FTC examined vertical integration issues, found contradictory indicators of consumer welfare and concluded: “The balance between competing incentives [in the context of vertical integration] raises complex empirical questions and may call for substantial additional study of the market generally, of local markets, or of particular transactions.”[3]  There may be a basis for concern and certainly a need for careful monitoring, but to date the Scotch Verdict applies to the charge made by NN advocates:  “Case not proven!” 

 

Claim 4: Network Management Discourages Innovation.  Advocates maintain that innovation in applications will be discouraged by any form of blocking or traffic shaping.  However, their argument focuses on innovation by applications and content providers, but does not recognize potential salutary impacts on innovation of reasonable network management practices that effectively expand the capacity to move additional traffic.  Network congestion, because of limits on management, can itself limit innovation by limiting the usefulness of new applications that cannot work on congested networks.  Conversely, if reasonable management facilitates a smoother flow of traffic, innovative new applications become more valuable.  Again, some balancing is required along with more analysis of the impact on different kinds, locations and timing of innovation of various network management practices. 

 

Most of the discussion about innovation has been marked by opinion, conjecture and expressions of fear, with almost no analysis in the network management context of the difficult questions related to incentives and opportunity for innovation, measuring the value of different kinds and loci of innovation, the trade-offs between innovation in the network and at the edges, and, finally, the effect of different rules respecting network management.  The challenge is compounded by the fact that the literature on innovation and its sources is not at all sharp in its policy implications and particularly not so in this context.  Again, “Case not proven!”     

 

 

Case for Network Management 

 

The core of the case for network management is that it delivers the best possible Internet experience to the largest group of users by ensuring that legal applications work as effectively as possible and that consumers will be able to access the Web sites of their choice without unnecessary delays or interruptions.  In addition to addressing congestion, network engineers also use a number of management techniques to guard against viruses, spam, identity theft, online fraud and malicious attacks that can shut down web sites.  Conversely, the absence of management means:

 

• Total consumer welfare from the use of the Internet will be substantially reduced;
• The typical consumer will suffer from delays and disruptions because of congestion created by a very small number of high-use consumers; and
• Effective Internet capacity will be reduced, effectively restricting the national benefits that can be delivered to all Americans by expanding access to the Internet.   

 

Tragedy of the Commons. In the 17^th century New England settlers in different villages set aside community-owned pastures where livestock of all villagers could graze.  In short order, the pastures (called “commons”) were barren.  The failure of private incentive to ration and preserve common user goods is widely known as the “tragedy of the commons.”  Such “tragedies” are known to materialize in a large number of circumstances involving common resources (sometimes called club goods) – fisheries, highways, the environment, national parks, and many others, including the Internet. 

 

The facts cited in the beginning of this ConsumerGram indicate a substantial likelihood of emergence of circumstances resembling the “Tragedy of the Commons.”  Specifically, use of the Internet is, because of the popularity of some new applications with a relatively small number of users, increasingly skewed in favor of the few at the expense of the many.  And the record indicates that these users are imposing costs on other users in an increasing number of circumstances.  The problem has all the earmarks of an emerging “Tragedy of the Commons.” 

 

Common Rules for Redistributing and Increasing Total Welfare.  Policy makers have long struggled with the challenge of defining when a redistribution of economic welfare among citizens is warranted.  Most policy changes help some citizens, but hurt others.  Few are win-win for all.  Every change imposes costs on some, while benefitting others.  The pivotal question here is whether managing the network on behalf of all users should be restricted by regulation supported by advocates for a few.  Consumer welfare economics holds that a change is good if the benefits enjoyed by the “winners” are sufficient to compensate for the losses of a few:  in this case, whether the value accruing to 95% of users is sufficient to offset the convenience lost by the bandwidth hogs who constitute 5% of users, but absorb 50% of the capacity.  Put differently, in the context of the historic standard of Jeremy Bentham – the greatest good for the greatest number – we are obliged whether the policy creates the most economic welfare.  Both the economic standard and the Benthamite design are apt here.

 

            Unequal Incidence of Restrictive Regulations.  Advocates of regulations restricting reasonable network management do not differentiate among different technology platforms and different ISPs.  The implicit suggestion is that the problem to be addressed, as well as the costs and benefits of doing so, are uniform across ISPs.  The facts are quite different.  The architecture, cost structure and services profiles of wireline (DSL or PONs) networks are very different from cable networks, and wireless networks are dramatically different from both.  These differences implicate substantial variation in the impact of restricting the ability of ISPs to manage networks.  Two stand out.   

 

            First, depending on the technology employed, wireless networks carry different formats in a common channel.  The significance is that congestion caused by, say, video downloads will infect ordinary voice service quality.  This contrasts with wired networks in which plain old voice services are insulated by engineering design from congestion originated by broadband data distribution hogs. 

 

            Second, the costs of regulation are particularly burdensome on smaller ISPs many of which provide service over wireless platforms.  There several thousand small ISPs service rural, remote areas.  By one recent account before Congress, there are about 8,000 wireless providers of broadband services to rural areas.  Virtually all of them practice fairly aggressive forms of network management as a means of holding down costs, minimizing the need for users to suffer delay and generally to maximize the value of the network to all users in the aggregate.  Stripped of the ability to practice reasonable traffic management, many small entities would have to raise rates, reduce QoS or, in the case of extreme congestion by a handful of users on a small network, simply shut down.  The ultimate losers are the consumers who subscribe to these small ISPs.  At a minimum, they would be deprived of a choice and might have to pay higher prices.  In some instances, that might lose Internet access altogether. 

 

Concluding Comments 

          Much has been made by opponents of network management of the restrictions ISPs might place on “freedoms” of a relatively small number of users (bandwidth hogs according to some) and of the loss of consumer welfare associated with such restrictions.  However, reasonable network management practices reflect efforts by prudent managers to maximize the total value of the network when the interests of all users are taken into consideration.  Most, if indeed not all, economic policies are undertaken in the name of the common good and invariably impose costs on some, so that others might benefit. 

 

So it should be in the case of network management.