Neutrality and non-neutrality, from the past to the future

                        Andrew Odlyzko


Most current discussions about telecom policy appear to lack proper
perspective, from both economic and historical points of view.  There
are extensive discussions of different approaches to net neutrality,
with many Europeans observers dismissing it as just an American issue.
On the other hand, some Americans wonder why Europeans have to make a
big issue out of international mobile roaming fees.  Simultaneously,
in seemingly totally different areas, such as pharmaceuticals, there
are hot controversies about wildly varying prices for the same drugs
from the same manufacturers in different countries.  What is generally
missing is a perspective that puts all these developments into a single
coherent view.  Yet such a perspective is not hard to present.

History, going back centuries, is full of similar pricing controversies,
in many cases amazingly similar to current ones.  This long history,
fortified by economic explanations for the existence of such
controversies, guarantees that they will persist into the indefinite
future, as no definitive solution exists.  They will therefore
provide generous meal tickets for generations of lobbyists, lawyers,
and economists.

The fundamental issue is the degree to which service or goods providers
are allowed to discriminate among their customers.  Basic economic
theory tells us, and empirical evidence reaffirms, that more control
allows providers to increase revenues and profits.  Both theory and
evidence also show that under some conditions, this is to the benefit
of customers.  The real question is about the appropriate balance, and
this question has been debated, explicitly or implicitly, over centuries.
The basic arguments have not varied much, the providers always pleading
high costs or need to expand, customers resisting on the grounds of
ruinous charges or unfairness.

Probably the only novel element in recent times has been the argument,
based on property rights, that providers should have complete freedom
in pricing.  This does seem to be new.  Traditionally, law has imposed
special non-discrimination and duty to serve obligations on those employed
in certain industries.  The understanding was that untrammeled control for
providers in such industries, in particular transportation, inn-keeping,
and communication, was similar to a lack not just of property rights
but of law itself.  It would expose society to arbitrary actions that
repressed economic activity, either by making enterprises unprofitable,
or introducing prohibitive degrees of uncertainty.

Aside from basic non-discrimination principles of law, there were often
statutory requirements that extended these principles and made them
explicit.  Yet many of these requirements compromised with the economic
need to allow some forms of discrimination.  Thus, for example, the 1833
charter of the London and Birmingham Railway in England envisaged a degree
of structural separation, with the company providing only the rails,
and other carriers being able to bring their own wagons and locomotives
onto those rails to provide actual carriage service for customers.
This charter (3 William IV, c.36) provided that the railway "shall not
partially raise or lower the Rates or Tolls payable under this Act,
but all such Rates and Tolls shall be so fixed as that the same shall be
taken from all Persons alike, under the same or similar Circumstances."
Yet the maximal tolls set by this Act (the railway could charge less, but
not more) varied by a factor of three between goods such as cotton clothes
and compost.   Also, the extent to which providers could discriminate,
or could engage in vertical integration, varied historically, depending
on economic circumstances.  Thus, for example, English canals were in
most cases barred from being carriers.  This prohibition was lifted in
the 1840s, when canals were impacted negatively by railway competition.
Moreover, even before this change, the tolls the canals collected depended
on the cargo (as with railways such as the London and Birmingham one),
and so were totally uncorrelated to costs, and thus required (as in the
case of railways) the equivalent of the modern "deep packet inspection"
to determine what was being carried.

These examples illustrate the general historical trend for society to
balance the competing incentives, depending on the circumstances of each
case.  In general, the greater the costs, the more power to discriminate
is given to the providers, although that is often controlled by fairness
rules.  (For example, private colleges in the United States practice
an extremely discriminatory policy, with affluent parents paying the
full tuition, and those earning less benefiting from discounts, called
scholarships.  However, this policy is carried out through policies
that those colleges assure the public are based on rules and "fair.")
Hence precedent and economic logic suggest that neutrality rules should be
based on detailed evidence of costs, something that is seldom seen today.

The process of finding a compromise between the incentives to discriminate
and to have equal treatment was seldom free of politics, and the outcomes
most likely were seldom exactly optimal.  This is nicely illustrated
by the following citation from a report of a UK 1867 Royal Commission,
writing about the Stockton and Darlington Railway, which opened in 1825,
and was a key step in the development of this industry:

      It is worthy of remark that in order to check the use of this
   line for conveying coals for shipping, and to confine it to inland
   traffic, parties interested in rival ports contrived to insert a
   clause limiting the charge for the haulage of all coal to Stockton
   for shipping, to one halfpenny per ton per mile, whereas the rate of
   fourpence per ton per mile was allowed for all coals transported for
   land sale.

     It was supposed by all parties that it would be impossible to carry
   coals at such a low rate without loss; but this rate not only turned
   out profitable, but formed ultimately the vital element in the success
   of the railway.

This quote illustrates not only the political interference in policy
making that was and surely will continue to be ubiquitous, but also the
difficulty in regulating rapidly developing technologies.  Tolls that were
"supposed by all parties" to be ruinous turned out to be profitable, as
incremental changes in the newly developed steam railway lowered costs.
Hence there is no reasonable hope of coming up with a fixed set of
rules that will work in general.  This means that regulations will
have to be flexible.  Yet they should not be arbitrary or too heavily
affected by politics, in order not to inhibit economic activity, another
difficult compromise.

That regulation will be required by society there seems little doubt.
Claims that there haven't been any serious problems concerning net
neutrality in telecom and so there is no need to regulate are easy
to dismiss.  Not only have there been problems recently, but history
shows that providers, if allowed, will discriminate, just as the
economic incentives suggest.  Very often they will do it even when
there is competition, much less a monopoly.  (There are many examples
of competition leading to increased discrimination.)  They will even
do it when there is regulation, by looking for ways around the rules.
Thus, for example, English canals that were barred from the business of
serving as carriers, sometimes managed to exert extra control by gaining
control of warehouses.  In some cases regulation simply could not deal
with the complexity of the industry at the desired level.  The early
English railways were obliged by their statutes to allow carriers to
use the rails for movement of their own wagons.  However, in practice
those railways frustrated such requirements by not providing water for
the carriers' locomotives or sidings for loading and unloading cargo.
The result was that Parliament conceded to railways monopoly carriage
rights.

Regulation in the future is likely to be even more difficult.  While the
world is not moving at the proverbial "Internet time," many changes
are taking place faster than before.  In addition, some of the most
interesting changes that could lead to a reshaping of our economy are
taking place in software, in the architecture of the information systems,
such as cloud computing and social networks.  In the traditional physical
world, of physical infrastructures, regulators had some tools at their
disposal that are less applicable now.  Service providers, even when
they clothed themselves in the mantle of private property rights,
frequently had to rely on the coercive power of the state to acquire
land or rights of way, and the state could condition the grant of such
rights on proper behavior.  In the online world, that is not available.
And it is very tricky, for example, to decide whether a change in a web
search algorithm is driven by the need to control activities that pollute
search results, or by the desire to enhance the prominence of a service
controlled by that search engine.  Thus we are bound to be involved in
very messy proceedings, and the only safe prediction is that lobbyists,
lawyers, and economists will do well.



Endnote:  The subject of this note is dealt with in more detail in two
papers by the author that appeared in Review of Network Economics:

1.  Network neutrality, search neutrality, and the never-ending conflict
between efficiency and fairness in markets,

     http://www.bepress.com/rne/vol8/iss1/4/

2.  The evolution of price discrimination in transportation and its
implications for the Internet,

     http://www.bepress.com/rne/vol3/iss3/4/