Definition of Net Neutrality

The Heart of the Matter

Overview of the Net Neutrality Game

Outcome of the Game: Per-User vs. Per-Usage Internet Fees

A recent court decision struck a blow against net neutrality.  One account of the decision appeared in the WSJ, “Court Backs Comcast Over FCC on 'Net Neutrality’” by Amy Schatz:

A U.S. appeals court ruled Tuesday that the Federal Communications Commission overstepped when it cited cable-giant Comcast Corp. for slowing some Internet traffic on its network, dealing a blow to big Web commerce companies and other proponents of "net neutrality"…

At stake is how far the FCC can go to dictate the way Internet providers like AT&T Inc. and Verizon Communications Inc. manage traffic on their multibillion-dollar networks…

Big Internet providers say ordinary Web users have no reason to fear restrictions on legal content. But companies like Google Inc. and Amazon.com Inc. that want to profit from offering more Web video and other high-bandwidth services are concerned that some day the big telecommunications companies will use their power to restrict certain kinds of Web content or charge more to deliver it at high speeds…

Consumer groups warned that the decision strips the FCC of authority to enforce net-neutrality rules and means that Comcast and other Internet providers can block Web sites anytime. They called for the FCC to take a tougher approach toward regulating Internet providers.

Definition of Net Neutrality

Cybertelecom.org provides the best description of the concept of neutrality, together with all the arguments, both pro and con, that I’ve been able to find.  It turns out that the issue is not as cut and dried as I had originally thought it was.  The following passage provides the essence of what net neutrality means; the passage was taken directly from the Cybertelecom.org website, though for the sake of readability (I was going to say “for brevity’s sake”, but it’s not actually brief, so I chose a different term), I’ve omitted the cited references:

"Network Neutrality" is the idea that Internet access providers should not discriminate with regard to what applications an individual can use, or the content an individual can upload, download, or interacted with over the network. Individuals acquiring services from Internet access providers should be able to use the applications and devices of their choice, and interact with the content of their choice anywhere on the Internet.

The concept of "Network Neutrality" is essentially traditional Common Carriage. Common carriers are carriers of goods, people, and information such as trains, planes, buses, and telephone companies. They cannot discriminate with regard to what they carry or where they carry it. Common carriage embodies the ideal that the efficient movement of goods and information is essential to our economy, our culture, and our nation, and therefore carriers must not discriminate or favor particular content or individuals.

Telecommunication carriers (those communications carriers that transport information back and forth) are one type of common carriers and have been classified as such for 100 years. This status was essentially inherited from telegraph companies. Forty years ago, the FCC initiated the Computer Inquiries which established the telephone network as an open platform over which computer networks could be constructed. The FCC also resolved the Carterfone proceeding, holding that individuals could attach devices (ie, faxes, modems) of their choice to the telephone network. These proceedings created an environment where any computer network could be constructed for any purpose and go anywhere.

Computer networks which are provisioned over telecommunications services, and in particular, Internet service providers, were classified as Information Services and did not have telecommunications regulations imposed upon them. Because these ISPs were interstate networks, they fell under the jurisdiction of the FCC and not state public utility commissions. ISPs hold the same role as telegraph and telephone carriers, carrying information central or critical to our society and nation.

This policy moved from the dial-up world to the broadband world. However, when the Internet moved from the dial-up world to the broadband world, it moved from something that was done over the common carrier network, to being the network. The question posed was whether, with this metamorphosis from some thing over the network, to being the network, the Internet would take on the common carrier status.

Advocates argued that cable modem service and DSL should be classified as telecommunications carriers so that the Computer Inquiries would apply. This is also known as the Open Access debate. Here the FCC policy of an open communications carrier shifted dramatically; the underlying communications network had always been a common carrier for 150 years. But when broadband Internet became the underlying network, the concept of common carriage was eliminated. The FCC concluded that these new communications networks were "information services" which did not need to be shared, did not fall under the Computer Inquiries, and did not fall under the non-discrimination provisions of title II of the Communications Act.

This is a move from an Internet access service classified as an information service provisioned over a telecommunications network classified as a common carrier - to a network where the whole thing top to bottom is an unregulated information service…

Opponents of Network Neutrality make a semantical attack on the term itself, arguing that it is a made up term for a made up problem; that there is no accepted definition for "network neutrality;" and that "it is just a slogan."

Proponents have put forth their definitions of "Network Neutrality;" these definitions are generally consistent. These different proponents generally put forward the following tenants of NN:

     • Individuals can use the applications of their choice and can access any other individual or resource on the Internet

     • Network service providers cannot discriminate with regard to the use of applications or access of end points on the network.

           ° Network service providers cannot block competitive applications or services in order to gain a competitive advantage

     • If network service providers offer a preferred service level, this preferred service level is offered and is available to others on the same terms and conditions.

     • Individuals at the edge of the network can innovate without seeking approval of the network service provider.

It is important, as this highlights, to distinguish between policy and implementation. Policy is why something is done; implementation is how something is done. If you review the literature of the proponents of Network Neutrality, there is rough consensus as to policy - network neutrality is about preventing a network with market power from discriminating - acknowledging the important role a network plays as a carrier of information. There may be differences in how different advocates envision implementation. The point here is - a difference of opinion concerning implementation is not the same as discord concerning the meaning of network neutrality.
 
With regard to "network neutrality" as a made up term, this is a good point. "Network neutrality" became a term of use in order to avoid uttering "common carriage." But "network neutrality" is consistent with, and fits well within, the jurisprudence of "common carriage." If Network Neutrality is characterized as a free floating arbitrary term, it is easy to define it with a strawman definition and shoot it down. If Network Neutrality is seen in its true context of common carriage, then this becomes a richer dialogue of history, jurisprudence, and policy objectives.

The Heart of the Matter

The Internet providers are largely private entities, and they own the infrastructure (i.e., they paid for it) that supports the content supplied by content providers.  Together, the infrastructure and the content it supports constitute the Internet.  Internet providers claim that since they own the infrastructure, they should be free to manage it as they see fit.  However, in the words of NSFNET, many content providers and users claim

Infrastructures, for purposes such as transportation and communication, have long been vital to national welfare. They knit together a country's economy by facilitating the movement of people, products, services, and ideas, and play important roles in national security.

Given the importance of the Internet infrastructure to the nation’s economy and welfare, content providers argue, Internet providers should not be free to manage the infrastructure with a free hand.  Rather, there should be regulations that generally require Internet providers to give unfettered and non-discriminatory access to all users.

From an economic standpoint, the issue at the heart of the matter is the fact that the Internet “pipes” through which content moves have a limited capacity.  To the extent aggregate flows of content over the Internet (more than) fill the pipes, then how should the capacity be allocated among users?

Currently, content providers who require greater access to the Internet pay higher fees to Internet providers than do content providers who require less access.  However, access fees paid by users are largely on a per-user basis, rather than on a per-usage basis.  This means users who consume large amounts of content pay the same access fees as users who consume little content.  This would be fine if there were plenty of capacity to spare.  However, this is not the case.  Capacity is scarce, and the pipes become congested, at which time all Internet traffic slows down.

Given the scarcity of capacity, an economically fair way of allocating capacity across users would be to switch to a fee based access system in which users are charged based on the amount of capacity they use.  A usage-based access system would enable Internet providers to recover the costs associated with providing and maintaining Internet infrastructure, and at the same time it would conform with the tenants of net neutrality listed in the bullet points above.  In particular, users would have free access to any content they choose, but they would pay higher fees to access more infrastructure-intensive content.   

So now the question becomes: How would the current situation change if user access to the Internet were switched from the current per-user system to a usage-based system?

Overview of the Net Neutrality Game

As the basis of the net neutrality situation lays a perfect hardware-software game that I will refer to as the Net Neutrality Game. The players in the Net Neutrality Game are:

1.  Internet Providers, such as Comcast, Verizon, and AT&T, have invested in constructing the hardware infrastructure that serves as the backbone of the Internet.  Internet providers provide Internet connections services to users in exchange for a fee.  (Content providers are also users, but for simplification of analysis purposes, I exclude them from being users in this analysis.)  Each period (e.g., month), Internet providers:

must choose how much bandwidth to provide (HW) and what price to charge each user for bandwidth (P),

given the number of users (N) and the amount of content provided by content providers (SW).

2.  Content Providers, such as Google, Amazon, and YouTube, provide content to users in exchange for users’ viewing of advertisements.  (Content providers also provide content to users in exchange for a fee, but for simplification of analysis purposes, I exclude content-for-fee transactions from this analysis.) Each period, content providers:

must choose the amount of software to provide users (SW),

given the amount of bandwidth provided by Internet providers (HW), the price Internet providers charge users for connection services (P), and the number of users (N).

3. Users purchase Internet connection services from Internet providers, which enables them to interact with other users and to consume content over the Internet.  (Users also provide content for other users, but for simplification of analysis purposes, I exclude user-provided content from this analysis.) Each period Users:

must choose whether or not to purchase Internet consumption services from Internet providers,

given the amount of bandwidth available (HW), the price of Internet connection services (P), the amount of content provided by content providers (SW), and the number of other users (N).

This situation forms a game because each of the (sets of) players is independent from the others, each faces his own set of incentives, and each takes actions to optimize his own outcome (profit or well-being).  Yet, the outcome achieved by each player depends on the actions taken by the other players.  That is, while the players are independent entities, free to take whatever actions they choose, they are each dependent upon the others players and the actions these other players take for their outcomes.  The questions thus become: What actions will each player be led to take, and what will be the outcomes of the game?

So we have four variables that affect the incentives and actions of each set of players:

HW: The total amount of bandwidth available; 

HW/N:  The total amount of bandwidth available per user;

P: The price of Internet connection services

SW: The total amount of content available

N: The total number of users

 Let’s take a closer look at the incentives each set of players faces.

Users

Demand for Internet Connection Services, N:

N is a function of (N, SW, HW/N, P)

                                           (+) (+)       (+)     (–)

Internet connection services enable users to interact with other users and to consume content.  Users will be happier, and thus demand more Internet connections services when the net benefits from doing do – the benefits from interaction with other users plus the benefits from consumption of content less the price of Internet connection services – are higher.  Users can interact more with other users when there are more other users with Internet connection services.  Users can consume more content when there is more content available and when there is more bandwidth available per user.  Demand for connection services will thus be higher when (1) more users are connected to the Internet (N is larger), (2) more content is available to consumers (SW is larger), (3) bandwidth per user is greater (HW/N is larger), and (4) the price of Internet connection services is lower (P is lower).

The number of Internet users, N, affects user demand in two ways.  First when there are more users, then the user interaction experience is richer.  At that same time, however, when there are more users, the Internet pipes become more congested, which slows the flow of information.  It follows that an increase in the number of users will have a net positive effect on increasing the value to users when the amount of bandwidth available (HW) is increasing faster than the number of users (N).

So, having access to the Internet enables users to interact with other users and to consume content. Assume that users who use the Internet more intensely –- that is, interact more with other users and/or consume more content -– are the users who value Internet access the most.  Now line up all users based on their intensity of Internet use (which is the same as their willingness to pay for Internet access) for a fixed amount of bandwidth per user and content.  For a fixed supply of bandwidth and content, the user demand function for Internet connection services looks like green line N(HW₀,SW₀) in the following graph:

Under this original demand curve, for supplies of HW₀ and SW₀, N(HW₀, SW₀), if Internet Providers set the access fee at P₀, then N₀ users will sign up. For high amounts of bandwidth, HW₁ > HW₀, and content, SW₁ > SW₀, the user demand curve for Internet access will shift out, from N(HW₀, SW₀) to N(HW₁, SW₁).  With the new higher levels of bandwidth and content, the original users, N₀, will be willing to pay a higher price, P₁ for Internet access.  Alternatively, if the price of Internet access is held constant at the original level, P₀, then more users will sign up for Internet access, N₁, when the amount of bandwidth and content increase to HW₁ and SW₁.

Let’s now compare the number of users who sign up for Internet services when the price of connection services is fixed for all users, that is, on a per-user basis, relative to the case in which higher intensity users – that is, users who interact with more other users and/or consumer more content – pay a higher price for connection services than lower intensity users.  This comparison is illustrated in the following graph:

When all users pay a single, per-user price for Internet access P₀, then N₀ users sign up for access.  These N₀ users include all of the highest intensity users, plus the portion of low intensity users that have higher intensity use.

Now say that higher intensity users pay a higher price, P_H, for Internet connection services then do lower intensity users, who pay P_L, where P_H > P_L. When lower intensity users pay a lower price than they did in the previous scenario, P_L instead of P₀, then more lower intensity users can afford to sign up for access.  That is, instead of having just the higher intensity of the low intensity users access the Internet, N₀ - N1_H, as is the case under a single user fee system, more moderate intensity users in the lower intensity group, N_1L - N₀ can now afford access.

Internet Providers

Choose per-user price and, P, and bandwidth, HW, to maximize Internet Providers’ Profits

= Revenues from Sales of Internet Connection Services to Users – Cost of Providing Internet Infrastructure

= P x N – Cost(HW)

= P x N(N, SW, HW/N, P) – Cost(HW)

  (+)       (+)   (+)     (+)     (–)                 (–)

The per-user price of Internet connection services, P, affects Internet Providers’ profits in two ways.  First, higher per-user prices bring in more revenues, P x N, for a given number of users, N, which means to maximize profits Internet providers want to set their prices as high as possible.  However, higher prices also decrease user demand for Internet connections services, N, which means Internet providers want to set lower prices to draw in as many users as possible.  Internet services providers must therefore choose the price, P, that will balance out its impact on increasing revenue per user (higher P), with its impact on decreasing number of users (lower N).

The bandwidth Internet providers make available to users, HW, also affects users in two ways.  First, when there is more bandwidth users are able to interact with more other users and to consume more content during the period.  So by making more bandwidth available (increasing HW), Internet Providers will increase the number of users, N, thereby increasing revenues, P x N.  At the same time, however, there is a cost to providing bandwidth, so greater amounts of bandwidth, (increasing HW), will increase Internet Providers costs, Cost(HW), which will, in turn, decrease profits. Internet services providers must therefore choose the bandwidth, HW, that will balance out its impact on increasing number of users (higher N), with its impact on increasing costs (increasing Cost(HW)).

Note also that whatever the price and bandwidth Internet providers choose, their profits will be greater when content providers make more content available to users (SW is higher).  So Internet providers want content providers to provide as much content as possible.

Content Providers

Choose amount of content, SW, to maximize Content Providers’ Profits

= Ad Revenue per User x Number of Users – Cost of Providing Software

= R(N) x N – Cost(SW)

= R(N) x N(N, SW, HW/N, P) – Cost(SW)

       (+)        (+)   (+)     (+)      (–)               (–)

The profit model associated with providing content on the web to attract users and generate ad revenues is interesting.

Start with the fact that content is a public good, in the sense that once content providers post their content, content providers generate very little subsequent costs, no matter how many users consume their content.  That is, for normal (non-public) goods, such as an apple, if one user consumes the good (eats the apple), then other users are nor able to consume the good.  So one user’s gain is another’s loss.  With content, it’s different.  No matter how many users consume web content, an infinite number of other users are still able to consume that same content.  There’s no depletion.  What this means is that content providers pay a one-time cost to generate and post a quantity of content, then after that any number of users can consume that same quantity of content without the content provider having to pay any more costs.  In other words, the cost of generating content is (generally) independent of the number of users who consume that content.

The next interesting fact is that the ad revenues per user generated by content providers are larger when more users consume the content providers’ content.  In other words, when a 2^nd user consumes a content provider’s content, then the revenues generated from the first user’s consumption of the content goes up.  Graphically, this is how ad revenues work:

Starting with N users, ad revenue per user is R(N); so total ad revenues are R(N) x (N), equal to the shaded area labeled A in the diagram.  When more users, Δ, are attracted to the site, content providers can either raise the price each advertiser pays, due to the fact that there’s now a larger audience for the ads, or the content provider can sell more ads.  In either case, what happens is that when there are more users, N+ Δ instead of N, ad revenues per user increase, to R(N+Δ) from R(N), in addition to the added revenues generated from the added users, Δ.  What we have, then is that with more users, N+Δ,

•  The ad revenues generated from the initial users, N, increases from R(N) x N (= shaded area A) to R(N+Δ) x N (= shaded area A + B), plus

•  The new users, Δ, generate ad revenues of R(N+Δ) x Δ (= shaded area C).

What we are left is that for content providers, having more users increases revenues disproportionately without increasing costs.  More users is a win-win for content providers.  This means that content providers want as many users to have as much web access as possible; that is, content providers want Internet providers to provide as much bandwidth as possible while charging very low access fees.

In short, cheap, abundant bandwidth enables content, which attracts users. 

Outcome of the Game: Per-User vs. Per-Usage Internet Fees

A table summarizing the differences for each of the players under each of the two scenarios appears at the end of this section.

 Relative to the scenario in which all users pay a single Internet access fee—that is., a per-user fee--  a move to a per-usage based fee would lead the fee paid by low intensity users to decrease, while the fee paid by high intensity users would increase.  With lower prices for low intensity usage, new users would sign up for Internet connection services. The new users would be very low intensity users for whom the fixed price under the per-user based system was too high for them to benefit from paying to access the Internet, given their low usage.  At the other end of the spectrum, the number of high intensity users would not decrease, but their intensity of usage would fall.

Content providers generate revenues from ad placements.  The fees paid by advertisers are based on the number of times each ad is viewed, where fees are larger when ads are viewed more often.  Content providers’ revenues are therefore sensitive both (1) to the number of Internet users each period, and (2) to the intensity of each user’s Internet use.

With a move from the current per-user fee scenario to one based on usage, content providers would generate more revenues from the new low intensity users, but they would lose revenues from the decrease in intensity of use for high intensity users.  Content providers’ losses from the move will be larger when high intensity users are more sensitive to the price of connection services.

How would Internet providers fare under a move from per-user to per-usage based fees?  Under the current per-user fee scenario, bandwidth use by high intensity users is larger than it would be under a per-usage based fee scenario.  In contrast, low intensity users use less bandwidth under the current scenario.  A shift to the new scenario would thus entail greater bandwidth usage by low intensity users and less bandwidth usage by high intensity users. The greater is the difference in usage between high intensity users and low intensity users under a per-usage based fee system, the larger will be the “free up” in bandwidth under the per-usage based fee system.  With the free up in bandwidth, Internet providers will either experience lower operating costs if they keep the bandwidth levels the same, or users will experience faster connection services if Internet providers choose to invest the saving in increasing line capacity.

Let’s take a closer look at the user dynamics under the two systems.  Referring back to the second User Demand fro Internet Access graph, but adding a few labels, we have:

Consumer surplus is the benefits that consumers retain when they end up paying less for something than they would have been willing to pay.  For example, suppose you get a craving for chocolate and would be willing to pay $1.50 for a Hershey Bar, but the store charges only 75¢.   In this case, you get the benefit of eating the Hershey Bar, plus you get an extra 75¢ in surplus from having to pay only 75¢ when you would have been willing to pay $1.50.  Suppose further that it cost the store 50¢ to provide the Hershey Bar.  Then the distribution of the $1.50 value ends up as follows: the cost of the transaction is 50¢; 25¢ goes to the storekeeper as profit; and 75¢ remains with you as surplus.  The total net surplus, then, is the total value generated by the transaction, less the cost.  The total value is your surplus (75¢) plus the storekeeper’s profit (25¢), or $1.00, while the cost is 50¢, so the total net surplus generated by the transaction is 50¢.  In other words, the total net surplus is the net benefit to everyone as a whole (society) associated with your consuming the Hershey Bar.

Let’s now take a look at the total net surplus to Internet providers and users associated with the provision of Internet services under a per-user based fee system relative to that under a per-usage based system.

Under a per-user based fee system in which Internet providers provide HW in bandwidth at a cost of C(HW) and set a per-user fee of P₀, then N₀ users sign up, N_1H of which are high intensity users and N₀ – N_1H of which are the higher intensity usage group of the low intensity users.  Under this scenario, the social surplus is allocated as follows:

• Internet providers get D + E + G + H in revenues at a cost of C(HW);
• High intensity users get a surplus of A + B
• Low intensity users get a surplus of C
• For a total social surplus of A + B + C + D + E + G + H – C(HW)

Under a per-usage based fee system in which Internet providers provide HW in bandwidth at a cost of C(HW) and set a per-user fee of P_L for low intensity users and a price of P_L for high intensity users, then a total of N_1L users sign up, of which N_1L - N_1H are low intensity users and N_1H of which are high intensity.  Under this scenario, the social surplus is allocated as follows:

• Internet providers get B + D+ G + H + I in revenues a cost of C(HW);
• High intensity users get a surplus of A
• Low intensity users get a surplus of C + E + F
• For a total social surplus of A + B + C + D + E + F + G + H – C(HW)

Comparing the outcomes under the two scenarios, we see that in the per-usage fee case relative to the per-user fee case:

• Internet providers gain B from high intensity users but lose E to low intensity users
• High intensity users lose B to Internet providers
• Low intensity users gain E and F
• For a net gain to society of F

Now let’s consider the situation of content providers.  The consumer surplus generated by users associated with the purchase of Internet services provides a good proxy for content providers’ surplus.  That is, the greater is the degree of “under-pricing” of Internet services, relative to the price users would be willing to pay, then the higher will be their intensity of use, which is essentially the consumption of content, which translates into higher ad revenues generated by content providers.

So then with a lower price under a per-user fee than per-usage fee scenario, (1) high intensity users consume much more content under the per-user fee scenario, and (2) the higher intensity of the low intensity users consume more content, but (3) the lower intensity of the low intensity users choosing not to sign up for Internet services, consume less content.  The greater consumption of content by both (1) the high intensity users and (2) the higher intensity of the lower intensity users surely outweigh (3) the lesser consumption of content by the lower intensity of the low intensity users.  This means that under the per-user based fee system consumption of content is larger (probably much larger), and thus revenues generated from sales of advertisements are also higher for content providers.

An interesting implication from this is that in the net neutrality debate, users and content providers generally associate non-discriminatory Internet pricing, that is, a user-based fee system rather than a usage-based system, with greater access by all.  In fact, the analysis presented suggests just the opposite, that under a user-based system, lower intensity users end up subsidizing access by higher intensity users.  Since prices are “too high” for low intensity users, the lowest intensity usage portion of users end up getting priced out of the market.  In other words, in effect, fewer users have access to the Internet under a per-user based system than they would under a per-usage based system.

Go to Part 2