Should Free Press Get a Free Pass? Or Do Demands for Rules Demand Reasons for Rules?

An unstated, but inescapable, assumption underlying the Free Press letter seems to be that "Broadband Internet Services" should be a separate product market, like water, electricity, or natural gas.  Unlike, these other high-fixed cost network services, though, Free Press seems to be convinced that pricing for this service should be on a "per-household" basis, rather than in proportion to the user's contribution to peak demand, which imposes costs on both the provider and other users.  Using this analogy, every residential consumer of "electricity" would pay the same price to have "electric" service in their home.  For example, even a consumer like Richard Nixon, who "was known to keep the air conditioning cranked up high in the Oval Office so that he could build fires in the summer"--would pay the same as the consumer setting the temperature at a more reasonable level, and limiting their usage of discretionary power.  After all, "metered" service is just another way of pricing on a per-unit basis. 

The difference, though, between the electric power example, and the broadband situation is that the power "hog" can impose costs on the provider (who may have to temporarily activate a more expensive, natural gas-fired "peaking unit" to generate additional electricity).  The peak usage bandwidth hog, though, doesn't just impose costs on the broadband provider (who, in the case of cable, may have to temporarily divide a node), but the heavy user also imposes costs on other users of the shared resource--who may be using applications that tolerate less "latency" (packet delivery delay) or "jitter" (variability of packet delay--"micro-congestion").  These are applications that require higher bandwidth quality, as well as a minimal level of capacity.  Examples include VoIP, online gaming, and IPTV.   It is especially vexing that an organization, like Free Press, that actively promotes diversity of speech in the media, would seem indifferent, or even hostile, to alternative speech distribution technologies (like IPTV).

To support their assertion that marginal cost pricing is the correct public policy yardstick, and what that marginal cost is, Free Press relies on a well-intentioned, but poorly conceived, "back of the envelope" analysis from a New York Times blog. Thus, Free Press explains that, because the marginal cost--the incremental cost of transmitting information to and from one additional household to the Internet (once a network that passes all homes has been built and subscribership is near 100%)--of providing "broadband service" is relatively small (they estimate $1 to $2 per home), prices of broadband are arbitrarily high under a metered pricing approach.

On the other hand, if you don't think paying for units of usage is crazy, the TWC plan doesn't look so crazy.  According to TWC's Chief Operating Officer, under TWC's tiered pricing plan, 30% of TWC's broadband Internet customers use less than 1 gigabit of bandwidth per month. These customers will save some money, but will pay $15 per gigabit per month.  On the other hand, the heavy users, whose plan is priced at $75/100 gigabits per month will only pay 75 cents per gigabit.  The heaviest users can pay even less per gigabit per month, at $150 for unlimited capacity.  That doesn't seem too much different than most widgets:  buy one, pay more per unit; buy in bulk, get a discount per unit.

Free Press, though, does not limit its claims of unfair pricing to prices divorced from their erroneous estimates of what they believe the ISP's "marginal" costs should be.  Free Press also (briefly) attempts to dismiss the network providers' burdens of paying for a large fixed cost investment; an investment which is--unfortunately, and much to Free Press's consternation--capacity constrained.  Nonetheless, they wrongly trivialize this cost by resting their assertion that capacity-based pricing is anticompetitive because the anticipated "exaflood" of bandwidth demand has yet to materialize.  This is a misleading assertion. 

Publicly available traffic studies are scarce, but the numbers published by the Japanese Ministry of Internal Communications ("MIC") are considered extremely relevant because the total capacity available per subscriber in Japan is so much higher than in the US, and the Japanese broadband Internet market is more mature than in the US, yet the issue of peak traffic management remains as relevant in Japan as it is here.  While total traffic demand, for both uploading and downloading, has been steadily, and rapidly, increasing in Japan  (see figure 3), total capacity is not the only relevant measure of capacity constraint in a network.

Rather, in networks characterized by "peak" demand capacity constraints, it is the growth in peak demand that is the relevant concern, and this is the behavior sought to be influenced by price signals.  This is the motivating theory behind President Obama's focus on building a "smart grid" for America's energy demands.  The primary goal is not to expand the grid, simply to be able to accommodate limitless demand, but rather, to develop a grid that meets efficient demand for energy, by among other benefits, motivating and including the consumer to make efficient choices regarding demand and technology.  Again, using the latest aggregate traffic data from Japan, we can see that for the 3 years, 2005-2007, daily peak period traffic increased from approximately 1 1/3 times more than average daily traffic to approximately 150% of average daily traffic. (see slide 4 or p. 7 of 9 in the pdf).

That percentage difference, combined with total traffic growth, is enormous.  At the end of 2007, peak demand was approximately 300 Gbps versus average daily demand of 200 Gbps.  In other words, 100 Gbps of valuable network capacity was being used for only a few hours a day.  More efficient demand distribution, by "motivating and including" consumers in demand shaping--through price signals--could benefit all consumers, by using a smaller network more efficiently.

Finally, it's important to note that the pricing trials being conducted by some ISPs are just that--experiments.  As Free Press notes on the first page of their April 22nd letter, some ISPs are continuing to offer flat-rated service for all residential consumers.  If the experiments being complained of are truly anti-consumer, the ISP undertakes the pricing trial at its own peril.  If consumers do not respond well to the tiered pricing trials, the ISPs offering the trials will either modify their practices or lose customers.