The Economics of Pornography

Text-only Preview

The Economics of Pornography

Kirk Doran

I. Introduction

The effect of pornography on societal wellbeing depends on two areas well-suited to economic
analysis: one, how it is distributed and consumed; and two, the externalities associated with these
processes. In this paper, I attempt to make an initial contribution to the study of these areas by
drawing on many underutilized data sources on a relatively new and significant branch of
pornography: internet pornography. First, I explain the main mechanisms of pornography
distribution on the internet. Next, I demonstrate that although the adult industry’s own estimate
of $2.5 billion in internet pornography revenues per year is often considered an exaggeration, it
corresponds well with independent data sources and is in fact a significant underestimate of total
pornography consumption: between 80% and 90% of internet pornography consumers seem to
consume primarily free internet pornography. My analysis suggests that there is not yet any
good statistical evidence of positive or negative externalities associated with the consumption of
pornography, but that several promising techniques exist for improving this research. However,
the relatively large market for help with overcoming pornography addiction strongly suggests
that there are large personal costs to pornography consumption, which opens up the possibility of
welfare-improving government regulation. I conclude this study by considering the possibility
of regulation of pornography in light of the information above. I suggest that eliminating
copyright will only succeed in limiting pornography consumption under certain conditions, but
that regulation of Internet Service Providers could quickly reduce internet pornography
consumption, especially the vast majority of such consumption that occurs for free.

II. Distribution and Consumption of Internet Pornography

How is internet pornography distributed?

Internet pornography does not differ substantially from other pornography in the manner of
production, but rather in the manner of distribution. Today, with high speed internet connections
quite common across the United States, internet pornography consists of (sometimes quite large)
digital files containing videos or photographs, distributed to consumers through at least 40,634
websites around the world (Tancer, 2008).1 These videos and photographs are sometimes

1 The National Academy of Sciences study cites 400,000 pornography websites worldwide as of January 2001,
information that was obtained from an interview with Bill Johnson, the director of Marketing of Flying Crododile,
Inc, operator of (Thornburgh and Lin, 2002). This number is older than the number in Bill
Tancer’s book, and furthermore, SexTracker does not seem to be a data analysis company, and therefore it may not
have access to data from actual ISPs, but rather may only be using data from the adult websites that register with
SexTracker. This would imply that the estimate by Johnson would probably be an extrapolation based on their own
subscribers. Therefore, the data from actual ISPs, analyzed by Bill Tancer in (Tancer, 2008), is likely to result in a
more accurate list of the total number of adult websites on the internet. Nevertheless, it must be considered a lower
bound, since it is certainly possible that the ISPs that subscribe to Hitwise have customers who do not make use of

identical to those that can be purchased through pornographic print media or pornographic video
vendors. In other cases, these videos and photographs are essentially only distributed through
the internet.

There are several ways to distribute this content on the internet. Pornographic photographs or
videos are stored on servers, often in the United States. Pornographic websites (e.g., contain links to this content. When a consumer searches for
pornography on the internet, he or she first finds the address of a likely website (often through a
search engine such as, and then attempts to view and/or download the
material linked to on that site. Viewing of photographs is typically flexible: they can either be
downloaded and stored on the consumer’s computer, or viewed directly from the website without
downloading. Some pornographic videos are stored on websites in a way that allows for them to
be downloaded and stored on the consumer’s own computer. But other videos are presented in a
“streaming” format that allows the video to be viewed on the consumer’s screen, while
preventing him or her from downloading the file containing the video. Finally, some streaming
videos claim to be live video feeds from a “webcam” that is currently filming pornographic

The websites that distribute this pornography may roughly be divided into three categories. First,
there are pay sites that allow users to consume pornographic content for a fee, often paid by
credit card over the internet. Second, there are free sites which allow consumers to view either
samples of pornographic material or the full photographs and videos without paying. Third,
there are pornographic videos, etc, uploaded to the various versions of Youtube that cater to
pornography consumers.

The pay sites usually charge anywhere from $10 to $100 per month for access (Caslon, 2008).
Some of these sites are connected with massive webs of gateway sites. A user who stumbles
upon one of these gateway sites will be lead by a series of links to the main site, where they will
have the opportunity to pay for pornographic content (Edelman, 2003). The free sites may be
primarily for-profit ventures, in which case they will typically be supported by advertising, or
simply personal home pages that contain (sometimes illegally copied) pornographic content.
The pornographic versions of Youtube are for-profit ventures, and the pornographic content on
them is sometimes produced by amateurs. Thus, there are several revenue models for internet
pornography: the end-user model (in which pay sites charge individual customers for personal
access to pornography), the advertiser model (in which websites distributing pornography
support themselves by selling advertising space, sometimes to other pornography sites), and the
model of various free pornography sites that may not be profitably supported by either end-user
subscriptions or advertising.

How much web pornography is being consumed for payment?

Before one can analyze the likely effects of pornography distribution and consumption on
society, it is important to get an estimate of the total amount of revenues involved in this
industry. The adult industry publication AVN reports that total internet pornography revenues

the gamut of adult websites, and furthermore it is possible that Hitwise does not make use of the full adult
consumption data in the information gleaned from its ISP subscribers.

for 2005 were about $2.5 billion (Johnston, 2007). Although many mainstream news sources
take AVN’s numbers at face value, the AVN revenues are frequently cited as gross
overestimates. In fact, some sources claim that the pornography industry has enough of an
incentive to exaggerate revenues that their numbers simply cannot be trusted (Ackman, 2001).
In light of this, it will be necessary to evaluate the reliability of the industry revenue reports
using independent data sources. If I can obtain reliable numbers on the size of this industry, then
I will already have made progress in understanding it’s effects – if I cannot obtain reliable
numbers, then there is little hope for assessing its impacts.

First, using information from the U.S. Census bureau on 2005 e-commerce revenues by industry
sector, I can see that the AVN report of $2.5 billion in internet pornography revenues amounts to
18% of the total e-commerce revenues earned from publishing, arts, entertainment, and
recreation services, 2.5% of the total e-commerce revenues earned from all business-to-consumer
“service” industries combined, and 1.3% of the total e-commerce revenues earned from all
business-to-consumer industries combined (U.S. Census, 2007), (U.S. Census, 2008). These
percentages do not seem too large.

Furthermore, I can use a trick to jointly determine the reliability of both the AVN revenue
numbers and the reliability of the Pew Internet and American Life (PIAL) Project’s internet
pornography consumption numbers. According to the PIAL 2005 May tracking poll, 66% of
Americans aged 18 or over used the internet, and 11.25% of these accessed internet pornography
(for a total of about 16.7 million internet pornography consumers). Using May 2004 tracking
poll data, I see that 20% of internet pornography consumers admit to paying for any online
content – combining this percentage with the May 2005 internet pornography consumption, I
obtain about 3.3 million paying internet pornography consumers in 2005. These numbers could
potentially be an underestimate, because of the fear of revealing oneself as an internet
pornography user during a telephone survey (Tancer, 2008). But I can divide the revenue
estimate by the pornography consumer estimate to evaluate whether either of them seems to be
biased in the directions we fear. If the numerator is really a gross overestimate (i.e., an order of
magnitude too high) and the denominator is really a gross underestimate (i.e., and order of
magnitude too low), then the resulting fraction should be at least two orders of magnitude larger
than reasonable.

So I divide the $2.5 billion by the total number of paying pornography customers from the Pew
data (about 3.3 million), to get $737 per paying customer per year. This works out to each
paying customer paying $61 per month. Caslon analytics reports that pay sites charge $10 to
$100 per month (and in 2007, charged $30 per month). So $61 per month per paying
customer is fairly reasonable. Thus, the $2.5 billion in annual internet pornography revenue
seems reasonable, as does the 3.3 million internet pornography paying customers. Since this was
calculated using the Pew pornography consumption data this also indirectly suggests that Pew’s
11% to 15% rate of total internet pornography consumption is also not considerably too low.2

2 Some examples might make this logic more clear. Suppose that the industry revenue numbers were correct, but
that the Pew data on paying customers was an underestimate by a factor of 10 (this could occur if twice as many
consumers actually view pornography as admitted it in the surveys, and if all of these consumers paid for
pornography consumption, rather than only 20%. This would result in $6.14 of pornography consumption per
month. This is outside the range of monthly rates reported by Caslon analytics, although it could be obtained if

How much web pornography is being consumed for free?

The numbers above reveal a potentially important fact about internet pornography consumption.
Of the 14% of internet users who admitted to consuming internet pornography during the May
2004 Pew survey, only 20% of these said that they pay for online content. This implies that the
majority of them were consuming pornography for free. This implication is backed up by the
February 2007 Pew survey. In this survey, 4.5% of internet users admitted to watching adult
videos from the internet. But only 10% of these pornographic video consumers said that they
ever pay to access online video. Finally, Bill Tancer reports that “hits” on internet pornography
websites are broadly distributed, more so than in internet retail (Tancer 2008).3 But Caslon
Analytics says that the industry is a winner takes all model in which a few firms bring in most of
the revenue. These two facts can be reconciled if a large number of the hits on pornography sites
are users downloading free material.4 In short, the Pew data and other evidence suggest that a
large number of internet pornography consumers – probably the vast majority – are usually
consuming free internet pornography. This free material consists of samples of the pay material,
illegally copied versions of the pay material, and amateur material.

Who is consuming internet pornography?

I argued above that the total number of pornography consumers and the total dollars of
pornography revenues could both be estimated fairly reliably by combining independent data
sources, in spite of the potential for aggregate measurement error. In order to further analyze the
demographic characteristics of pornography consumers, I would have to evaluate whether
measurement error in this pornography usage data is distributed unequally across demographic
groupings. For instance, are men more willing to admit the truth about their pornography
consumption than women? It is impossible to do justice to this question in this paper, but I can
obtain some leverage by once again comparing results across independent sources of data.

There are three main sources of data that I draw upon in this section. One is the General Social
Survey (GSS), administered since 1972 across the United States. The second is the Pew Internet
and American Life (PIAL) project telephone surveys, which have occurred several times a year
since 2001. Third, I draw upon data from Internet Service Providers collated by Bill Tancer of
Hitwise and discussed in (Tancer, 2008). Claims about pornography usage that are common
within each of these sources (over time) and across each of these sources are less likely to be
artifacts of chance or survey methodology.

people use the cheapest sites and spend half their months not subscribed. But if the revenue numbers were also
incorrect by a factor of ten (ten times too large), then this would imply 61 cents spent per month per paying
customer. This seems much too small. Thus, it seems likely that it is not the case that both the revenue data and the
paid pornography consumption data are severely biased in the directions sometimes feared in the literature on
pornography data.
3 “One of the things that stands out about the distribution of porn sites is just how fractionalized the spread is. The
top five hundred sites account for only 56 percent of all visits to the adult category. (In contrast, the top five
hundred retail sites account for 76 percent of all retail visits)” (Tancer, 2008).
4 It is also possible to reconcile these facts by Ben Edelman’s observation that there exist some central sites that
register innumerable gateway sites that link to the central site (Edelman, 2003).

Table 1 collects pornography usage data from the May 2004 PIAL survey and the GSS. It is
clear that in both surveys, men more frequently claim to be pornography users than women.
Furthermore the young are more likely than the old, and the married and widowed are less likely
than people of other marital status. Each of these patterns is largely confirmed by pornography
data from four other PIAL surveys (dating from February 2001 through May 2005).
Furthermore, the data from Internet Service Providers reported in (Tancer, 2008), which may be
less likely to have differential measurement error across demographic groups, confirms that men
are more likely to use internet pornography than women. The GSS also contains data about
religion, and Table 1 shows that admitted pornography consumption in the GSS varies
considerably across religious groups.

All three data sources are consistent with a decline over the last several years in the proportion of
people consuming pornography. In Figure 1a, I show the proportion of respondents who claim
to have seen an x-rated movie during the past year, measured since 1973 by the GSS. It is clear
that there has been a small decrease between 2004 and 2006. In Figure 2, I show the proportion
of respondents who report using the internet to visit adult websites, measured since Feb 2001 by
the PIAL. The PIAL data likewise show a small decrease from May 2004 to May 2005. Finally,
in Figure 3, I replicate the figure in (Tancer, 2008) showing the market share of visits to adult
websites from September 2005 to September 2007. It is clear that the adult website market share
has declined from 17 percent of all visits to websites in September 2005 to 11 percent by
September 2007. This decline in market share is consistent with constant consumption of
pornography and an increase in the consumption of other services, but it is also consistent with a
continued decline in pornography consumption.

Nevertheless, the GSS data in Figure 1a demonstrates that over the long term, the consumption
of pornography has increased since a low point in the late 1970s up until the latest year of 2006.
This is even more obvious in Figure 1b, where I restrict the GSS data to men aged 29 or under,
where the most recent pornography usage figure in 2006 is 54%, almost twice what it was during
the low point in 1978.

Summary facts on pornography distribution and consumption:

There are at least 40,634 websites that distribute pornographic material on the internet. About
11% of all internet visits are to one of these sites. About 14% of the online population in
America visits these sites (17 million Americans), spending on average 6.5 minutes per visit.5
About 80% to 90% of these people only access free pornographic material. This free material
consists of samples of the pay material, illegally copied versions of the pay material, and amateur
material. The remaining 3 million Americans who pay for internet pornography pay about $60
per month. This generates $2.5 billion in annual revenues for the internet pornography industry,
making internet pornography revenues a very small percentage of total e-commerce revenues.
While this number is frequently cited as an overestimate, it is reasonable in comparison with
independent data sources, and in fact it is a severe underestimate of the total amount of
pornography consumption, because of the prevalence of free material. Most of these revenues
are apparently earned by a small number of top firms (Caslon, 2008). Most of the consumption,
on the other hand, is spread widely across a number of websites, with the top 500 of these 40,000

5 The minutes per visit number is taken from (Tancer, 2008).

websites earning only 56% of the total traffic. Consumption of pornography is spread unevenly
according to gender, age, marital status, and religion, and has no clear relationship with
education or income. In spite of a recent decrease in pornography consumption, as of 2006 the
data is consistent with a considerably higher proportion of the population consuming
pornography than at the recent low point in the late 1970s, especially among young men.

III. The Effects of Pornography Consumption

Are there positive externalities from consuming pornography?

The most difficult requirement in studying the existence of externalities (positive or negative) to
the consumption of pornography is finding some random variation in access to, exposure to, or
temptation towards using pornography. Two recent economics papers, one a working paper, and
one a published paper, claim to have found such variation.

The first, by Todd Kendall, formerly an assistant professor at Clemson University, uses variation
in state internet usage over time to proxy for variation in the “price” of accessing pornography.
His first claim is that there exist unobservable reasons why residents in some states had an easier
time adopting the internet than those in other states; that these reasons are unrelated to the
outcomes that Kendall considers; and that, once a handful of other related variables are
controlled for, the remaining variation we see in the speed of internet adoption across states is
largely attributable to these reasons. His second claim is that this variation in the speed of
internet adoption across states led to variation in the rate of rapes across states. His final claim is
that it was internet pornography consumption that caused the link between internet usage and

Each of these claims is difficult to prove empirically, and the fact that Kendall has found some
evidence in support of them is impressive. Nevertheless, there remain important concerns about
his approach. The first problem is an old one, and therefore a boring one, but that does not make
it an irrelevant one: endogeneity. In an ideal experiment (ideal for identification, not ethics),
people would be randomly exposed to pornography, and their sexual and criminal behavior
would be subsequently observed. In Kendall’s natural experiment, random exposure to
pornography is replaced with differing rates of usage of the internet, which he believes should
create differing rates of access to pornography. Kendall is aware that the differing rates of
access to the internet are not necessarily unrelated to differing secular trends in sexual and
criminal behavior, including rape. On the contrary, changes in internet usage could be caused by
changes in other variables that are related to rape, such as preferences and unobserved
demographics. Normally, what a researcher would do in this circumstance is find an
“instrument” for internet access: a variable that would lead to random variation in internet usage
but that would not directly affect outcome variables such as rape. Kendall could not find such a
variable, so he relied on the older technique of simply including controls for whatever potentially
related variables he could find. Given the prospect for omitted variable bias here, this is a
serious issue.

The second issue is an odd one, and has been raised elsewhere (cite website): Kendall controls
simultaneously for changes in the internet usage within states over time, and changes in
computer usage within states over time. So the results are driven by states in which internet
usage has risen by more or less than would be predicted just by their increase in computer usage.
If it is unclear what makes some states have slower increases in internet usage than others do, it
is even more unclear what drives some states to experience slower increases in internet usage
relative to their speed of increase in computer usage than others do. What would make some
states have large increases in the number of people with computers but small increases in the
number who access the internet? Since we don’t know where this variation comes from, there is
no clear reason why we should assume that variation in this phenomenon is unrelated to
important omitted variables.

The second paper, by Winai Wongsurawat, is entitled “Pornography and Social Ills: Evidence
from the Early 1990s,” and was published in the Journal of Applied Economics in 2006.
Wongsurawat uses variation in availability of post office boxes as an instrument for variation in
the “cost” of subscriptions to Penthouse Magazaine. The theory is that, ceteris paribus, greater
ease of obtaining a private post office box will make it less likely that other people will notice
one’s Penthouse consumption, thereby lowering the overall psychological cost of subscribing. In
order for this instrument to be valid, it must be the case that it has a statistically significant effect
on Penthouse consumption (specifically, the effect of P.O. box availability on subscriptions
should be the main mechanism here), and it must also be the case that availability of post office
boxes is not related through another mechanism to the outcome of interest (in this case, rapes).

In the previous paper there was no instrument, but in this paper the instrument is not convincing.
Wongsurawat demonstrates that in fact P.O. Box availability does not have a strong effect on
Penthouse subscriptions, but rather has a borderline-strong effect on subscriptions plus individual
(e.g., newsstand) sales. This implies that the main effect of P.O. Box availability is on the
individual sales, which strongly suggests the existence of an unmeasured third variable affecting
both P.O. Boxes and individual sales. Furthermore, he admits that the F-statistic on the joint
significance of the instruments is not quite at the level sometimes used as a rule of thumb for
avoiding the weak instrument problem. Finally, Wongsurawat finds that P.O. box availability
has a statistically significant and apparently positive effect on Discover Magazine subscriptions.
Without running a two-stage least squares regression using Discover Magazine subscriptions as
an independent variable, he thus has no way of ruling out that Discover Magazine consumption
is what connects P.O. box availability to rapes, rather than Penthouse consumption.

While both (Kendall, 2007) and (Wongsurawat, 2006) point out inadequacies in much of the
statistical evidence in favor of negative externalities of consumption of pornography, their own
evidence for positive externalities of consumption of pornography is not convincing. In both
cases, their results are driven strongly by their proposed techniques for obtaining variation in the
“price” of pornography consumption, and in both cases this variation is unlikely to be
exogenous. Therefore, their results are not a substantial improvement on the older literature that
showed positive effects of pornography consumption on rape.

Nevertheless, there is more that can be done to use statistical evidence to measure the existence
of externalities associated with pornography consumption. The first step would be finding a

useful instrument for pornography availability or consumption: one that has a strong and
measurable effect on pornography usage, and is also unlikely to have an independent effect on
the outcomes of interest. One example of this would be variation in broadband internet
availability that comes from a known source. For instance, it seems likely that localities that
neighbor other areas that have already had broadband cables laid will be more likely to get
broadband access than other similar localities that happen not to have been near such areas.
Another example would be using the similarity in take-up rates of technology over time reported
in (Stevensen, 2008) as an instrument for internet access. In fact, this approach is being used in
ongoing work (Doran and Price, 2008).

What can economic consumption theory tell us about the consumption effects of

Economic consumption theory is based on the assumption that anything a person chooses to
consume while subject to some scarcity constraints will be (subject to those scarcity constraints)
the choice that makes that person best off. Under this assumption, it must be the case that those
who choose to consume pornography are improving their own well-being by doing so, and that
attempts to prevent them from consuming pornography by adding an additional constraint to
their decision-making (such as raising the price of pornography or making pornography
consumption a punishable offence) would necessarily reduce their well-being. As Samuel
Cameron explains in Economics Uncut, “But why, from an economist’s point of view should we
be spending time trying to define porn in the first place? If individuals are rational utility
maximizers, then why do they need to be barred from pornography?” (Cameron, 2006). The
answer, that Cameron himself later admits, is that pornography may be addictive, and that the
study of addiction in economics leaves open the possibility that it is optimal for the state to
attempt to reduce consumption.

In fact, there is a lot of money being made helping people overcome their “sex-addiction,” a
condition that increasingly refers to the use and abuse of internet pornography (Landau, 2008).
There is no money being made helping people with tooth brushing addiction, or kite flying
addiction. It is thus very unlikely that consuming pornography has no worse consequences for
the person doing the consuming than does brushing teeth or flying kites – otherwise no one
would pay psychologists, doctors, and authors to help them overcome what they perceive to be a
debilitating personal problem. (Landau, 2008) reports: “"We're seeing it with epidemic
proportions now, particularly with regards to cybersex," said Mark Schwartz, psychologist and
former director of the Masters and Johnson Institute in St. Louis, Missouri. "There isn't a week
that goes by where I don't get two calls" about sex addiction.” There are numerous self-help
books on breaking pornography addiction, in addition to numerous internet filters designed to
prevent temptation. All of this evidence should be sufficient to convince economists that using
pornography can result in great costs for the person consuming, and that these costs are high
enough that people will pay for such services as PureOnline’s $165 counseling session for
married men, or Covenent Eyes’ $55 per year accountability software.

In short, if standard economic consumption theory claims that a lower bound on the value of a
consumption good for a consumer is the price that that consumer is willing to pay for consuming
that good, then we are left with a contradiction regarding pornography. It appears that the same

person is willing to spend a positive amount of money to consume pornography and a positive
amount of money to be prevented from consuming pornography. The first expense would imply
that a constraint to prevent him from consuming pornography would have negative value, but the
second expense implies that such a constraint has positive value. Clearly, economists need to
make use of various models of addiction to properly understand the effects of pornography

Summary Facts on the Effects of Pornography Usage:

There is not yet convincing statistical evidence in favor of either positive or negative
externalities associated with consumption of pornography. However, there does appear to be a
strong incentive to prevent oneself from consuming pornography, and the monetary value of this
prevention is possibly much higher than the monetary value attached to the consumption of
pornography itself. This suggests that there may be large personal costs of consumption
associated with pornography, and opens up the possibility that it may be optimal for the state to
use regulation to limit the distribution and consumption of pornography.

IV. Regulation of Pornography Distribution and Consumption

Is it possible to reduce the negative effects of pornography usage through eliminating
copyright protection on pornographic material?

In Section II, I explained that between 80% and 90% of the people who consume internet
pornography generally consume free content. This content is presumably inferior in quality to
the pay content, but nevertheless it is the main source of internet pornography consumption in
America. Thus, if we want to know the effects of eliminating copyright on pornography
consumption, the most important question is: will the large number of consumers currently
accessing free content still be able to do so after elimination of copyright? I will attempt to
answer this question below.

This free content is a combination of legally posted content supported by advertising, legally
posted content serving as an advertisement (e.g., samples), illegally posted pay content, and
amateur content. Only the first two categories could possibly be negatively affected by
elimination of copyright: illegally posted material will just become legally posted material, and
amateur material will be posted for the same reasons it already has been. I have not found what
portion of total free pornography consumption comes from these two sources, but whatever level
it is at will be a lower bound for the amount of free pornography consumption that can persist
after elimination of copyright.

The revenues that pay for the bandwidth used up in distributing the first two categories of free
content will primarily come from end-users (since Caslon Analytics reports that much of the
advertising on pornography sites is from other pornography sites). So it remains to find out
whether the end-user revenue model can persist after elimination of copyright, whether some

other source of revenue can be found that would pay for the bandwidth, or whether these first
two categories of free content will disappear entirely.

A useful way of answering this question is to ask whether there already exists an industry that
distributes public domain works and supports itself through its own revenues. The answer is yes.
As (Beers, 2006) and (Sloan, 2007) attest, there is money to be made from reworking public
domain books and selling them, often over the internet. Furthermore, many major publishers sell
large numbers of public domain books. Jane Austen’s Pride and Prejudice is in the public
domain. A search of “Jane Austen Pride and Prejudice” in the “Books – Literature and Fiction –
Classics – British” section of produced 573 results. In addition to all of the older
editions no longer available except as used purchases, I counted at least 15 editions currently
available as new purchases, including several by major publishing houses. The bestselling
Dover thrift edition of three Austen works packaged together has a sales rank of 2,140 out of the
total number of books sold on, and according to (Rosenthal, 2008), this implies
yearly sales of about 3,300 copies. These works are not being printed and distributed out of
charity – it must be the case that the revenues are greater than the costs. Finally, the complete
text of Pride and Prejudice is available for free online at ( :// and at
( :// It is therefore clear that the fact that Pride and
Prejudice is in the public domain has not reduced its availability for consumption whatsoever. In
fact, it is currently quite cheap to consume Pride and Prejudice, and it would be much more
expensive if it was still under copyright (consider, for example, the works of J.K. Rowling).

How can this be? The answer is that publishing houses can make money on public domain
works by distinguishing their product from the other versions currently available. Many editions
of Pride and Prejudice contain special annotations and essays to make the text more readable and
interesting. Some contain illustrations, or are printed or digitally stored in an especially
attractive format. This is likely to be what would happen to the internet pornography industry
were copyright to be revoked. Some websites would package and organize the pornographic
material in a better way than others. They would make an extra effort to ensure that the digital
videos and photographs on their site were free of computer viruses. They would develop
interfaces that were easier and more pleasant to use. It is likely that their profits would go down.
But there would still be profits to be made from high quality distribution. The price would likely
decline, and the consumption of pornography would either remain constant or increase.

The same cannot be said for production. In the absence of copyright protection, the service that
is being profitably sold is efficient distribution, not quality production. Thus, there would be
little incentive to continue to produce works for profit. If the moral costs of producing
pornography are high, then eliminating copyright will thus reduce the total social costs of
pornography. However, the numbers of consumers of pornography range in the tens of millions.
Here is where the main social costs of pornography will be, and elimination of copyright will, in
the short term at least, either leave constant or increase that consumption, as the price of
consumption decreases. The one silver lining about copyright elimination will be that, as
decades pass, the lack of new production will make the older pornographic material seem dated.
Few people now consume the pornographic post cards of the late nineteenth century. Thus, after
the changes in tastes that occur over generations, it is possible that elimination of copyright will
succeed in reducing consumption of pornography.