The Rimm Study: Text

Context note: This article appeared on pages 1849 through 1915, and the appendicies on pages 1916 through 1934. Page numbers will be added as HTML anchors at a later date.

Finally, Reader Advisory: Due to the nature of the studied material, Minors and Those Who Are Going To Be Easily Offended are hereby warned of strong language and sexual content. Leave now if dealing with such things is going to be a problem for you.

Marketing Pornography on the Information Superhighway:

A Survey of 917,410 Images, Descriptions, Short Stories, and Animations Downloaded 8.5 Million Times by Consumers in Over 2000 Cities in Forty Countries, Provinces,and Territories

A. Pornography[1] on Computer Networks

In addition, the market for computer pornography is evolving rapidly. A decade ago, few people had access to the technology necessary to store, transmit, or receive pornographic images on computers. During the past few years, however, pornographers have begun to utilize computer networks and the unprecedented distribution channels they offer to penetrate markets throughout the world where public access to pornography has been historically restricted, including China, Saudi Arabia, Malaysia and Turkey.[3]

Computer pornographers are also moving from a market saturation policy to a market segmentation, or even individualized, marketing phase. Until now, most have saturated customers with tens of thousands of images, reasoning that their customers would inevitably find material that they liked. However, few customers have the patience or technical resources to perform the extensive database analysis necessary to quickly download only the images they prefer. Pornographers now have sufficient information to dramatically shrink the size of their portfolios, while at the same time increasing their subscriber revenues. A few have already begun to do so.

It is clear that pornography is being vigorously marketed in increasingly sophisticated ways and has now found a receptive audience in a wide variety of computer environments. According to industry experts,[4] and the pornographers themselves,[5] there are at least five factors, in addition to an increased focus on paraphilic content, which account for this recent explosion of pornography via computer networks. First, consumers enjoy considerable privacy on computer networks and can easily avoid the potential embarrassment of walking into an "adult" store to acquire pornography. Second, consumers have the ability to download only those images that they find most sexually arousing. Previously, a consumer had to purchase an entire magazine or video in order to gain access to a few desired depictions. Third, easy, discrete storage of pornographic images on a computer enables consumers to conceal them from family members, friends, and associates.[6] Fourth, the prevalence and fear of AIDS and other sexually transmitted diseases has helped pornographers to successfully market "modem sex" and autoeroticism as "safe" and viable alternatives to the dangers of "real" sex. Finally, new and highly advanced computer technologies are quickly being absorbed into the mainstream, permitting an ever-expanding audience to gain access to digitized pornography available on the "Information Superhighway."[7]

B. Scope of Current Study and Article[8]

For the Carnegie Mellon study, the research team downloaded all available pornographic images from five popular Usenet boards over a four month period. In addition, the team obtained descriptive listings from sixty-eight commercial "adult" BBS containing 450,620 pornographic images, animations, and text files that had been downloaded by consumers 6.4 million times;[10] six "adult" BBS with approximately 75,000 files for which only partial download information was available; and another twenty-seven "adult" BBS containing 391,790 files for which no consumer download information was available. Thus, a total of 917,410 descriptive listings were analyzed for content by the research team. Finally, approximately 10,000 actual images were randomly downloaded or obtained via adult BBS, the Usenet, or CD-ROM. These images were used to verify the accuracy of the written descriptions provided in the listings.

The research team's content survey of the images and descriptions for which all, only partial, or no download information was available suggests no substantive differences between the datasets. Accordingly, this article focuses on the 450,620 files for which complete download information was available. Of these, animations, text files, and images which were either ambiguously described or not described at all, were excluded. A total of 292,114 image descriptions remained and are discussed here. At least 36% of the images studied were identified as having been distributed by two or more "adult" BBS. These "duplicates"[11] enable researchers to compare how identical imagery is consumed on commercial BBS in different regions of the country.

With respect to users, the Carnegie Mellon research team was able to identify consumers of pedophilic and paraphilic pornography via computer in more than 2000 cities in all fifty states in the United States, most Canadian provinces, and forty countries, provinces, and territories around the world.

Part II of this article addresses three issues concerning pornography on the Usenet: (1) the percentage of all images available on the Usenet that are pornographic; (2) the popularity of pornographic boards in comparison to non-pornographic boards, at both a university studied and worldwide; and (3) the origins of pornographic imagery on the Usenet.

Part III, an analysis of commercial "adult" BBS, comprises the major portion of this article. It examines: (1) the availability and demand for hard-core, soft-core, paraphilic, pedophilic, and hebephilic imagery; (2) the concentration of market leaders; (3) market forces common to all "adult" BBS; (4) "adult" BBS demographics; and (5) the image portfolio and marketing strategies of the Amateur Action BBS as a case study.

Two important aspects of reliability and validity are carefully considered. First, how well do the written descriptions of the pornographic images correspond to the study's classifications? Second, how well do the written descriptions marketed by the pornographers correspond to the actual images? The linguistic parsing software developed at Carnegie Mellon University was found to offer a highly reliable estimate of the proportion of different types of pornographic images available on commercial "adult" BBS. Moreover, upon performing extensive validity testing, a panel of judges found a high correspondence between the pornographers" marketing descriptions and the activities depicted in the actual images.

Part IV presents a more informal discussion of the data including: the relationship between images and the words that describe them; the wide circulation of paraphilic imagery; the importance of descriptive lists; the sophistication of modern pornographers; privacy concerns; and a brief comparison between the Marquis de Sade and the market leader of computer pornography.

All BBS data was collected in May and June, 1994, unless otherwise noted. This article begins to address, but by no means exhausts, the breadth and depth of the data collected during the research.

C. Comparison to Other Studies

Numerous behavioral studies, conducted in field and laboratory settings, have previously attempted to determine various effects of pornography on its consumers.[13] However, the literature has largely neglected the study of actual consumption because little reliable data has been available. The accuracy of all prior studies has depended upon the honesty of replies people give when surveyed about their sexual tastes. In contrast, this study focuses entirely upon what people actually consume, not what they say they consume; it thus provides a more accurate measure of actual consumption.[14] This methodology is particularly important when analyzing such taboo imagery as incest, bestiality, coprophilia, urophilia, and torture.

The Carnegie Mellon study is also illuminating because its sample size is several orders of magnitude larger than previously published studies of either pornographic content or consumption.[15] Because the data is in many respects exhaustive, statistical techniques and assumptions that are commonly invoked to impute general consumer behavior are not necessary for this dataset. Thus, the research team considers the inferences drawn highly robust.>

The study results suggest a tremendous rift between the sexual activities in which Americans claim to engage, as reported most recently by the study Sex in America,[16] and the sexually explicit activities presented in images that many Americans consume.[17]

D. Implications of Study

Child Pornography:[19] On a practical level, the ease of copying and disseminating digitized child pornography presents unique law enforcement challenges because the seizure and eventual destruction of computerized child pornography may no longer appreciably reduce the amount of child pornographic imagery existing in the pedophile underground. Moreover, current bans on child pornography are justified in part as necessary to protect the actual child depicted in the photographs or videos. If technology advances, as it surely will, to allow the creation of pornographic images that do not depict actual children, this justification for prohibiting the dissemination of these images may no longer be compelling.[20]

Limiting Access by Minors: The widespread availability of pornography on computer networks may have a profound effect on those who wish to utilize the emerging National Information Infrastructure for non-pornographic purposes. For instance, primary and secondary schools are increasingly connecting their students to broader national and international computer networks, where pornography permeates the digital landscape. Given the current structure of the Internet, which often permits multiple routes around a blockade, there appears to be no simple practical solution to limiting access by minors to computer pornography.[21] Accordingly, policymakers, as well as parents and educators, may have to reconcile the unprecedented educational and cultural benefits offered by computer networks with their concerns about children's access to pornography.

Obscenity Standards: One of the most intriguing implications for criminal law is whether the computer transmission of obscene material is sufficiently similar to the transmission of obscenity through the mails or common carriers to apply the traditional analytical regimen to digitized pornography. In Cyberspace, this issue extends beyond the borders of the United States: the research team was able to identify consumers of paraphilic and pedophilic computer pornography from countries as diverse as China, Saudi Arabia, Turkey, South Africa, Chile, Malaysia, Hungary, Trinidad and Tobago, New Zealand, Hong Kong, Nigeria, and Japan. While the results discussed here may to some extent dispel the notion that hard-core pornography is consumed primarily in Western nations, the results also raise significant questions as to whether obscenity standards can be implemented on an international scale.

In the final analysis, if the application of digital imaging technology to pornography has any recurring theme, it is that traditional constitutional and law enforcement assumptions and conclusions must adapt to a new technology.

2. Other Implications

In addition, portions of the methodology utilized by the Carnegie Mellon study may also be applied to the study of products other than pornography that are marketed and distributed by computer. One could also use computer networks to study other activities sometimes considered threats to society, such as how to make a bomb or how to break into a "secure" computer system. Other issues, such as encryption technology, content flags, and authentication procedures, are discussed in this article.

E. Research Interests of the Carnegie Mellon Team

An unusual amount of data was freely available from commercial "adult" BBS, primarily as a consequence of the evolution of the online industry. Large commercial BBS such as America Online, CompuServe, and Prodigy do not carry hard-core pornographic imagery, either for legal or policy reasons. As a consequence, several thousand comparatively small "adult" BBS have sprung up across the country. For many entrepreneurs willing to risk violating local and federal obscenity statutes, these "adult" BBS have proven quite lucrative. In many instances, the research team was able to persuade the owners of these BBS to provide information about subscriber consumption habits.

The principal investigator, an electrical engineer with a background in broadband communications, began with an interest in digital image manipulation and the transmission of multimedia applications over computer networks in real-time. Other interests include examining how computer networks challenge researchers to develop new methodologies to monitor consumption habits, marketing techniques, and reliability and validity methods for the analysis of vast quantities of data. Indeed, in the coming decade, the modeling of consumer behavior on computer networks will likely prove fertile ground for market researchers, as businesses begin to reap the enormous potential of advertising and distributing goods and services online.[24]

An analysis of pornography in cyberspace provides a fascinating case study of many of the computer network-related legal and technological issues confronting businesses and policymakers today.

F. A Brief Sketch of the "Information Superhighway"[25]

Another interesting and important facet of Usenet is the administration of the various hierarchies. A hierarchy is determined by the leading portion of the newsgroup name. For instance, sci.* denotes the science hierarchy. A more pertinent example for the study's purposes is alt.*, or the "alternative" hierarchy. Most of the sexually explicit images on the Usenet are located in the alt.* hierarchy. All hierarchies, except the alt.* hierarchy, have a quasi-formal procedure for creating a new newsgroup. The alt.* hierarchy, by design, is less formal, and anyone, in essence, can create a newsgroup at any time. How widely a newsgroup is distributed or read is another matter entirely (e.g., it is not uncommon to see new newsgroups with the name alt.rosanne.flame.flame "created"). It is therefore extremely difficult to obtain informative numbers when analyzing the number or percentage of groups or traffic, except at a particular site.

2. Commercial Bulletin Board Systems (BBS)

Bulletin board systems differ from the Usenet in a number of ways. First, unlike the Usenet's diffuse configuration of newsgroups, BBS usually operate from a central locus. More centralized control of BBS files permits BBS operators to charge for their services, and this opportunity for profit has motivated a number of pornographers to operate "adult" BBS. This is another important contrast between the Usenet and BBS: while access to files on the Usenet is often free (particularly for academic users), a user must subscribe to a BBS in order to gain access to its files.

The reader should note that most of the commercial "adult" BBS discussed in this article did not offer their subscribers access to the Internet or Usenet at the time data was collected. However, in the past year, many BBS have begun to offer such access as part of their service, and some BBS can now also be accessed through the Internet via telnet. The leading "adult" BBS have recently formed a "BBS Direct" Network, which enables users from anywhere in the United States to pay only $30 per month for unlimited access to their BBS, pornographic image files, and the Internet.[27]

3. Downloading and Viewing Images

While widespread use of the Information Superhighway by the larger consumer market is still in its infancy, tools that make connecting to, navigating, and understanding complex international networks are developing rapidly. Many software programs which automatically cut, paste, and decode Usenet images are now available.

II. Usenet

There is one notable, though likely not significant, limitation to this method. Images are occasionally posted on newsgroups outside of the alt.binaries hierarchy, such as alt.sex.bestiality and alt.sex.fetish.watersports. Those newsgroups outside of the alt.binaries hierarchy known by the research team to contain pornography, such as the two mentioned, were included.

2. Popularity of Pornographic vs. Non-Pornographic Usenet Newsgroups

alt.binaries.pictures.erotica

alt.binaries.pictures.bestiality

alt.sex.fetish.watersports

alt.binaries.pictures.female

alt.binaries.pictures.tasteless

The research team classified images into two main categories: pornography which originated from "adult" BBS and pornography which did not. The images were said to originate from "adult" BBS if the name, logo, and telephone number of the BBS appeared next to or within the image.[34] Images with no BBS logo were classified in the second category. A third category was added to include individual portraits of women (and occasionally men), with varying degrees of nudity, but no sexual contact or lascivious exhibition. By most standards "PG" or "R" rated, this category was introduced in such a way as to enable those who do not consider such images "soft-core" pornography to exclude them from the final calculations. This study presents the results under both sets of assumptions.

B. Results

As discussed later in this article, multimedia applications that combine text, sound, and interactive graphics are currently being developed. These applications will no doubt increase the percentage of imagery, including pornography, found on the Usenet.[36]

It is informative, but difficult, to estimate the extent to which the Internet is being used to carry pornographic images. Unfortunately, no reliable data is available to answer this question. The Internet is a network of networks, and traffic between neighboring institutions may go over regional networks. Moreover, much of what a user sends from his or her computer never leaves the corporate or campus network from which it originates.

At one time, the NSFNet, supported by the National Science Foundation, served as a common backbone interconnecting all regional networks. That began to change in 1990 when the first commercial Internet backbone providers emerged, and on May 5, 1995, the NSFNet was fully decommissioned. As of the summer of 1994, when data was initially collected for this study, available statistics indicated by proportion the traffic that utilized the NSFNet backbone. Figure 1 makes this clear.

Figure 1 [Percentage of Internet Backbone Traffic, August, 1994.]

Within each of these categories, it is difficult to estimate the current percentage of traffic that may be pornographic.[37] The best data concerning network pornography consumption comes from the Usenet, which itself constitutes only 11.5% of Internet traffic. Of this 11.5%, approximately 3% by message count, but 22% by byte count (e.g., 2.5% of total Internet backbone traffic), is associated with Usenet newsgroups containing pornographic imagery.[38] However, as this study makes clear, studying pornography according to consumption, as opposed to availability, provides a much more revealing picture of the marketplace.

Most notably, the percentage of World Wide Web traffic passing through the Internet backbone has more than tripled, to 26.25% as of April, 1995. Accordingly, the research team undertook an initial survey of pornography on the World Wide Web (commonly referred to as the "Web"), which is included in Appendix C. It is essential to note that Usenet and the World Wide Web are merely different protocols.[39] Thus, studying the Usenet in combination with commercial "adult" BBS is highly informative of the broader picture and provides an excellent means of following the potential growth of pornography on computer networks.

2. Popularity of Pornographic vs. Non-pornographic Newsgroups

University Results. In an effort to present relatively current information, the results from August 1994 are analyzed here.[40] In broad terms, the research indicated that pornographic newsgroups are accessed more frequently during the school year than during summer recess. This suggests that, in comparison to teachers, faculty, and staff, a disproportionately large number of students access Usenet pornography.

Approximately 3600 Usenet newsgroups were available to the users studied. Of these, 104, or 2.88%, were identified as pornographic. Thirteen of the forty (32.5%) most frequently accessed newsgroups were identified as pornographic. The following is a list of the top forty newsgroups. (The newsgroups identified as having pornographic content are in bold):[41]

Table 1 [Top Forty Usenet Newsgroups at a University Studied]

The fact that alt.sex.stories is currently more popular than alt.sex.pictures.binaries.erotica has been often misinterpreted as an indication that stories are more popular than images. It is likely that users have been discouraged by the burdensome encoding and decoding tools necessary to transmit imagery over the e-mail network. As new tools become more refined and easier to use, binaries may overtake stories as the top pornographic choice.

Worldwide Results. The worldwide statistics suggest that Usenet hosts appear less willing to offer their readers access to pornographic newsgroups than other types of newsgroups. 81.2% of the sites offer access to non-pornographic newsgroups, whereas only 55.8% of the sites offered their readers access to the pornographic newsgroups. Of the forty most popular newsgroups worldwide, only one alt.binaries.pictures.erotica contained encoded pornographic images. Three others of the top forty contained sexually explicit stories and discussion. The top forty worldwide newsgroups are summarized in Table 2.

Table 2 [Top Forty Newsgroups in Order of Popularity, Worldwide]

The newsgroups are ranked in Table 2 by the estimated total number of readers worldwide. However, when the data is classified by percent of news readers who subscribe to the newsgroups, three of the five most popular newsgroups are pornographic.[42] Moreover, 20,644 of the 101,211 monthly Usenet posts in the top forty newsgroups, or 20.4%, are pornographic.

Comparisons: University and Worldwide Results. Six of the top forty newsgroups at the University are composed of encoded graphical images, all of which are pornographic. This is true for only one of the top forty Usenet groups worldwide. None of the top forty newsgroups at the University were geared toward new users, whereas three of the top ten Usenet newsgroups worldwide are designed to assist new users or answer general technical questions about the Usenet.

Policymakers might be equally interested in comparing the cost of pornographic and non-pornographic newsgroups. The primary cost factor in maintaining a newsgroup is the amount of storage space required. Image files require considerably more storage space per image posting than a text story or comment. The following chart lists the top forty Usenet newsgroups by volume:

Table 3 [Top Forty Newsgroups in Order of Traffic Volume, Worldwide]

In the top forty newsgroups, the percentage of total cost used for pornographic newsgroups is 34.2%. Of total space taken for image newsgroups, 76% of the space is used for pornographic images. Of total space taken for text newsgroups, 9.8% are pornographic.

3. Origins of Pornographic Imagery on the Usenet

Many of the ninety-two "adult" BBS identified on the Usenet were unabashedly advertising their products, posting "teaser" images on the Usenet as an essential part of their marketing strategy. However, the precise number could not be determined, because the research team could not distinguish between the posting of an image (or enthusiastic comments) from a subscriber and a posting by a BBS sysop.

Figure 2 [Relationship Between Pornography on the Usenet and Commercial BBS.]

Figure 2 illustrates the various paths of pornographic images along the Information Superhighway: The diagram can be elucidated in the following manner. An "adult" BBS sysop commissions photographers for original images; finds the images in magazines; purchases "adult" CD-ROMs; or pirates the images from other BBS. Where necessary, the sysop uses a color scanner to convert them to digital form, and adds his logo and telephone number using a software image manipulation package. After posting the images on his BBS, he writes a brief description of each image, which is included in his complete list of files. He may also upload them onto the Usenet or his customers may upload them onto the Usenet, with or without his knowledge and consent. The Usenet subscribers then download these images from the newsgroups indicated earlier and may then decide to call the BBS if they find the images appealing. When these new subscribers first log on, the menu asks where they heard of the BBS. This enables BBS to determine the extent to which the Usenet serves as an effective advertising outlet.

While such data was not available for this study, most BBS sysops with whom members of the research team "chatted" indicated no objection to their customers posting their images on the Usenet.[44] A BBS with an image portfolio of 10,000 can easily afford to give away a few of its images at no cost. Indeed, the Amateur Action BBS imprints "DISTRIBUTE FREELY!!" on many of its images and even encourages its subscribers to upload them onto other newsgroups.

The results from Part II demonstrate that pornography is widely accessed on the Usenet, and that commercial "adult" BBS are the Usenet's primary source of pornography. These commercial "adult" BBS, which comprise the major portion of the Carnegie Mellon study, will now be discussed.

III. Pornographic "Adult" Commercial BBS

In order to locate the commercial "adult" boards, the research team obtained listings of more than 5000 computer bulletin board services from Boardwatch magazine, BBS Monthly, the Usenet, and the bulletin boards themselves. The team eliminated all boards that did not advertise an "adult" selection. This narrowed the list to about 500 boards that focused primarily on marketing "adult" pornographic images, or general boards with large "adult" sections in addition to other general interest material. Two recently published texts identified another 500 "adult" BBS.45 The research team did not discover any other authoritative list of active "adult" BBS in the United States.

The research team contacted each of these approximately 1000 boards by modem or voice telephone. About half had gone out of business, leaving roughly 500 active boards for further study. Many of these were primarily "chat" boards, which customers use to communicate live with other customers. Still others had only one or two telephone lines and transmitted at 1200 or 2400 baud, suggesting they were smaller, more transient operations.[46]

Consequently, the research team decided that a BBS had to meet the following criteria to be included in this research project: (1) have been in business at least twenty-four months; (2) have at least four telephone lines for use by customers; (3) transmit at 14,400 baud or better; and (4) have at least 1000 "adult" images available for downloading. Because of the expenses involved in operating this type of large BBS, these criteria greatly decreased the likelihood that any BBS examined would have been a transient operation.[47] Exceptions were made in the following two cases: (1) if the BBS was located in a remote region of the United States where it was difficult to find another local source of computer pornography; and (2) if the BBS, based upon its new user menu, appeared to be aggressively expanding its customer base.[48] To the best of the research team's knowledge, the BBS included in this study comprise most of the medium- and large-sized "adult" BBS in the country that existed at the time of the research.

In order to collect descriptive lists of the pornographic images available on each BBS, as well as a representative sampling of the images themselves, the research team placed more than 300 hours in long distance telephone calls to the "adult" BBS selected by the team. Every BBS asked the members of the research team to provide a real name, address, business and home phone numbers, date of birth, password, and type of computer and modem. Most asked where the members of the research team had heard about their BBS, and approximately half of them required photocopies of a driver's license with proof of age before granting further access to their systems.[49] Still others asked for the user's mother's maiden name (purportedly in case the password was forgotten) and required users to read legal disclaimers related to pornographic files.[50] Members of the research team did not, as a rule, identify themselves as researchers.

2. Developing the Linguistic Classification Scheme for Analyzing Consumption of Pornographic Images

Figure 3. Structure of a Typical BBS Record Listing.

This listing structure provides considerable information about pornography consumption patterns. First, based on the number of downloads (5), it can be determined which images were the most popular. Second, the linguistic description, identified by numbers (4) and (7), permits the images to be classified into categories of certain sexual behaviors. In the above example, the linguistic description denotes fellatio. Third, based on the original posting date (3), researchers can trace over time the manner in which the BBS sysops increased or decreased their offerings of various categories of images, presumably in response to consumer demand. Fourth, while the name of the record (1) usually gave no indication of the content, it often did give an indication of the original source of the image. For instance, an image from the Amateur Action BBS always began with "AA-," followed by a number. An image from the now defunct Windy City Freedom Fortress BBS[52] always began with "WC," followed by a number. This was particularly helpful in determining how many of the images from larger BBS were "pirated"[53] and redistributed by smaller BBS in different regions of the country. Fifth, the date of the last download (6) indicates whether the older images are downloaded as frequently as the newer ones. And finally, the size of the file (2) made it possible to determine the maximum number of images a customer could download over time.[54]

It should be noted that the manner in which the record is presented to the customer often differs from the above sample record, which is presented in "double line" format. Many BBS either hide this information from their customers or do not provide it because of space or software limitations. For instance, the list of one file from Amateur Action BBS reads as follows:[55]

AA-2438 92K Cody is having an ORGASM! Fingers hole!

Classification Scheme. The initial image description classification scheme was adopted from the Dietz-Sears study.[56] However, the Dietz-Sears categories were often found to be either too general or too specific for the current study. For instance, Dietz-Sears divides "B&D/S&M" images into twenty different subcategories.[57] While this is useful in studying different types of "B&D/S&M" as an individual phenomenon, its effect is to diffuse the "B&D/S&M" imagery within the entire pornographic landscape. The Carnegie Mellon study therefore tabulates the twenty "B&D/S&M" subcategories as one category.[58] The criteria for adding additional categories to the Dietz-Sears model were that: (1) a particular word in the description described a sexual act which could not be included in the previously established categories; and (2) that word was repeated more than once in the list. The added categories include: incest, hair color, obese, pedo/hebephile,[59] "amazing," dogstyle, swing, muscular, sixty-nine, emotions, shower, outdoor, petting, panties, and whore.[60]

When completed, the computer "dictionary" included 3823 words spanning sixty-three basic categories. It is significant to note that pornographers rely not only on activities in identifying an image, but also on attributes (for example, hair color, muscularity) and emotions (for example, pleasure and pain) to market their images. For instance, in the following description, the pornographer intersects the sexual activity with the subject's hair color and body structure to market the image: "Brunette coed has tiny boobs! 3 inch wide cock in her ass." Similarly, pornographers use the intersection of certain sexual activities with sensations of pleasure or pain. "She swallows his whole cock! She is choking!" and "Cody is having an ORGASM! Fingers hole!" are two examples of this technique. Figure 4 makes this clear.

Figure 4. [Description Strategies Utilized to Market Pornography.]

Prioritization Scheme. Because pornographers described images with multiple words, prioritization schemes became necessary to avoid false categorizations. For instance, if the word "fucks" was assumed to imply the vaginal category, the computer dictionary would have classified such descriptions as "he fucks her ass" or "he fucks her mouth" or "she fucks herself with a dildo" as vaginal, when clearly they were not. The research team therefore established a priority scheme that checked for the word "fucks" appearing with the words "ass" or "face" or "dildo;" otherwise the image was classified as "vaginal." Without such a prioritization scheme, the word "fucks" would need to be excluded from the dictionary as too ambiguous. Under such exclusion, the percentage of vaginal sex images recorded would have been greatly and unrealistically diminished because there were relatively few other words describing vaginal sex. The only alternative would have been to require a human judge to review each description that contained the word "fucks" and classify it, which would have been inefficient given the large number of records to be examined. Problems that arose with other words and categories were handled in a similar manner.

Overlap. Overlap was permitted among categories that had the same precedence level. For instance, the following categories were identified in the same precedence level: orgy, asian, fisting, lesbian. A description such as "She fists her girlfriend's pussy," was thus classified as both a lesbian and a fisting image, while "Three Asians going at it in bed," was classified as both an asian and an orgy image. Without some overlap, the classification scheme would have proven less insightful.[61] However, the Dietz-Sears model permitted overlap in all cases, which could be misleading.[62] "She gets fucked by a horse" was classified by this study under bestiality, not vaginal intercourse, while "He sucks his boyfriend's dick" depicts homosexual activity, and was not classified by this study as a fellatio image.[63] A more difficult description to categorize is, "She licks her girlfriend's asshole." The image could be classified as oral, anal, or lesbian, but the computer classified it as lesbian, because lesbian is a more general term which may describe a variety of sexual acts, including oral, vaginal, and anal acts. Thus, general terms were given precedence over particular terms.

Uncategorized Images. An uncategorized category was created to identify image descriptions that the dictionary could not sort effectively. Uncategorized images accounted for the fewest number of downloads, probably because when pornographers failed to provide clear content descriptions, they did not effectively market their images. The absence of a description or the use of ambiguous words simply does not entice the customer to download the image. Therefore, there is no reason to believe the presence of these uncategorized images distorted this study's results. Nevertheless, the research team tried to decrease the number of uncategorized images to as low a percentage as practicable by repeatedly fine-tuning the dictionary and developing new categories until the number of uncategorized images fell below 16.1%, or 6.5% of images weighted by number of downloads. There were three reasons an image remained uncategorized. First, the description portion of the record was left blank. Blank descriptions accounted for 72% of what remained uncategorized. Second, the description contained deliberate or accidental misspellings, such as "She socks his cock," which the computer could not identify.[64] Third, descriptions such as "Blonde getting a cheeky dinner plate" were too ambiguous to assign.

Exceptions to Standard Methodology. Finally, an exceptions category was created to account for double meanings and idiosyncrasies in the English language. For instance "Bo Derek walks her dog on the beach," is clearly a portrait of a movie star, not a bestiality image. A manual operator identified exceptions among a random sampling of 40,000 descriptions, then assigned the descriptions to the proper category. The total percentage of exceptions was 1.5%.

3. Reliability and Validity of the Category Dictionary[65]

Reliability. The research team implemented a number of checking mechanisms in order to establish the accuracy of the classification scheme before finalizing it. First, four members of the research team checked a random sample of 10,000 descriptions and their classification by the computer dictionary. The researchers checked descriptions and classifications in groups of 2500 each, until the fourth group achieved what they considered a high degree of agreement between the computer dictionary and the researchers.

Next, two judges performed formal "reliability" tests. Reliability tests measured how accurately the computer classification scheme had sorted the linguistic image descriptions. The judges were trained in three practice runs of twenty descriptions each to clarify several points of procedure and definitions. They were to decide separately and without discussion whether the description and its classification matched (was accurate), did not match (was inaccurate or erroneous), or partially matched. They developed several rating rules. First, a match occurred when the major idea of the description corresponded to the definition of the classification. Second, a partial match was scored if there was at least one element described which was included in the definition. Sources of partial disagreement typically included descriptions such as the following: "He spreads her ass cheeks," which may or may not match anal sex; or, "Two hot blondes kiss each other and suck cock," which may partially match lesbian, but is also matched with orgy. Third, a non-match ocurred when the description had no correspondence to the classification definitions. For example, "She goes down on his love snake," which does not match bestiality; or, "Lady in black gets fucked doggy style," which does not match inter-racial.

The judges were given a stratified random list of 500 image descriptions, twenty from each of twenty-five categories. The judges strongly confirmed the reliability of the computer classification scheme for pedo/hebephile (92.6%) and most of the paraphilia classifications (83.6%). However, because of overlap, the reliability was not as great for classifications such as fellatio (73.8%), lesbian (78.2%), and portraits (75%).[66] The judges had difficulty agreeing, for instance, whether an image described as "She sucks his cock while fisting her girlfriend's pussy," was most appropriately classified as fellatio, orgy, lesbian, or fisting.

In the course of research, the team had the opportunity to discuss how pornography is marketed with many computer pornographers. The broader distinction they draw is between the "action" clearly depicted in hard-core materials (oral, vaginal, or anal penetration) and the display of nudity without clearly visible penetration or consummation of the sexual act in soft-core materials. Thus, the classification and prioritization scheme was revised to group the categories into four aggregate classes. The judges were then provided with a second stratified random list of 200 image descriptions, fifty from each of the following four classes.[67]

1. paraphilia, which included transvestite, transsexual, sadomasochism (b&d/s&m), fisting, urophilia, coprophilia, foreign objects, voyeurism,[68] bestiality, and incest.
2. pedo/hebephile, which included both nude portraits of young children in pre-pubescence, and hard-core sex acts involving young-looking boys and girls.
3. hard-core, (non-paraphilic), which included explicit sexual contact or penetration between two or more individuals, such as fellatio, vaginal intercourse, or anal penetration.
4. soft-core, which included nude and semi-nude portraits emphasizing large breasts, genitalia, or famous models, but with no penetration or erect penis visible.[69]

Figure 6. [Final Precedence Scheme Adopted for the Carnegie Mellon Study.]

The final precedence scheme reflects both consumer demand patterns and the differing legal definitions recognized for different materials. Because statutes and the courts have regulated much pedo/hebephilic imagery[72] according to a more easily recognized per se standard,[73] this study gives terms suggesting pedophilia and hebephilia the highest precedence in order to more easily identify them. In addition, during its preliminary data collection, the research team was somewhat surprised by the volume and degree of paraphilic material widely available and consumed on commercial "adult" BBS. Correspondingly, the precedence scheme identifies this material separately from hard-core materials, allowing the research team to identify the overall relevance of paraphilic material as defined by DSM-IV.[74]

The distinction in the final precedence scheme between hard-core and soft-core pornography merely acknowledges a dividing line drawn by many pornographers, statutes, and courts.[75]

Table 4 offers an example from each of the four classifications, and how the computer program classified them.

Table 4 [Precedence Scheme Example, Carnegie Mellon Parsing Software]

In addition to the four major classifications, the numbers for vaginal intercourse will be presented in this article, defined both "narrowly" and "broadly." When defined narrowly, no overlap was permitted of vaginal intercourse with other hard-core or paraphilic categories. When defined broadly, overlap was permitted with other hard-core and paraphilic categories, with the exception of bestiality. In both instances, vaginal intercourse has been counted as a subset of hard-core.

Validity. In addition to assessing the reliability of the classification scheme, the research team assessed the validity of the descriptions. This study defines validity as the extent to which the written descriptions correspond to the actual pornographic images. Reliability alone is sufficient if we are concerned only with categorizing how images are marketed; if we wish to make statements about the images themselves using linguistic categorization, we must also verify the validity of the descriptions. The use of verbal descriptions to classify images themselves would be invalid if pornographers are often mistaken in their verbal descriptions, or if they deliberately exaggerate or conceal the nature of their images.

There are at least two types of validity tests that could be performed on the Carnegie Mellon dataset. The more obvious validity test would be to have a panel of judges examine the actual images and determine the extent to which they match their classifications. However, such a test is limited to the categorization scheme adopted. As a rule, pornographers do not force images into categories; instead, they force images into words. Accordingly, a second validity test would be to have a panel of judges examine the actual images and determine the extent to which they match their descriptions, independent of the adopted classification schemes. Such a validity test is generic and enables policymakers to answer a broader question: to what extent do pornographers accurately describe their images? The research team decided to adopt the second approach; it thus preserves the world of pornographers and consumers, and makes no attempt to establish "intrinsic" or "absolute" classifications of images independent of the empirical data collected. However, because the validity was high, one may reasonably extrapolate that the images also correspond to the classifications discussed here to the extent that they were found reliable.

For the Usenet portion of this study, it was necessary to examine approximately 3500 actual images. For the commercial BBS portion, another 1500 actual images were examined. The research team found nearly 100% correspondence between the descriptions and the images for those images obtained from the Usenet, but these procedures would be appropriately defined as "informal."[76] Accordingly, a panel of three human judges invoked formal validity procedures. They were given a stratified random set of 400 images, 200 from the market leader, Amateur Action BBS, and 200 from the other thirty-four "adult" BBS, fifty of which were described by pornographers to depict vaginal intercourse. As with the reliability procedure, the judges were trained in three practice runs of twenty images each to clarify separately and without discussion whether the image and the description matched, did not match, or partially matched.

Of particular importance, from a legal perspective, is whether real men and women are depicted in sexual acts with actual animals (or other acts, which based on their descriptions, may be difficult for the reader to accept as having genuinely occurred). The judges took special care to be precise in their judgments by viewing the images on a twenty-inch color monitor and zooming in on certain areas when necessary. However, there is not always enough room in the description area to describe all aspects of the image; the pornographers then decide to emphasize those aspects of an image they believe the customer will find most arousing. For instance, upon examining the image described as, "BRUNETTE IS ON HER KNEES! SEXY REDHEAD FISTS HER ASSHOLE!" it was discovered that one woman was being fisted by another while fellating a man. In this case, the pornographer may have understood that the primary appeal of the image was rectal fisting, and that other possible classifications, such as fellatio, were only of secondary interest to the customer.

Accordingly, the judges developed several rating rules: First, a match occurred when each of the major ideas of the description corresponded to what was visible on the image. Some tolerance of wording was acceptable but nothing which contradicted the description was permitted. Second, a partial match was scored when at least one major element was described which did not appear in the image. Sources of partial disagreement typically included descriptions such as the following: "SHE FINISHES SHITTING! SHE HAS SHIT SPECS ON HER ASSHOLE!" (although a close-up of the rectum, one of the judges unable to perceive fecal matter); or "SHE CHOKES ON THICK DOG COCK! DOG SPERM ON HER SEXY LIPS!" (although a depiction of a woman performing fellatio on a dog, two of the judges unable to perceive the ejaculate); or "Brunette has big boobs! She pisses on her girlfriend's face!" (although a urophilic image, two of the judges concluded that the urine was aimed at the chest, not the face). Third, a non-match occurred when there was more than one major element which did not appear in the image. Fourth, the pornographers used code words which did not correspond to standard English. Accordingly, their usage pattern was accepted, because it would be known to those using the BBS. For example, the phrase "junior teen," when invoked by the market leader, Amateur Action BBS, described pre-pubescent children, rather than post-pubescent teenagers. Fifth, the judges also disregarded incestuous references, such as father, mother, daughter, and sister, as irrelevant because there was no way to establish the validity of these relationships. Finally, the judges considered a description accurate if it also contained some form of marketing braggadocio: "She is horny! Sucks thick horse cock like a vacuum cleaner!" (the judges agreeing that a woman was depicted performing fellatio on a horse but ignored the phrase "like a vacuum cleaner").

For the imagery from all thirty-five "adult" BBS, the judges unanimously agreed with the pornographers" descriptions on 84.5% of the images. They agreed partially with 12% of the images and disagreed unanimously with the remaining 3.5%. For the Amateur Action BBS imagery, the judges unanimously agreed with the pornographers" descriptions on 91.5% of the images. They were in partial agreement with the remaining 8.5%, of which 60% were unanimous and 40% suggested some disagreement among the judges. The judges suspected that two of the four hundred images examined may have been digitally manipulated. It is significant to note that none of the judges disagreed with any of the descriptions from Amateur Action BBS.

In sum, validity was very high, suggesting that pornographers, as a rule, take special care to describe their images with the high degree of accuracy they consider necessary to satisfy their customers.

B. Results

1. The "adult" BBS market is driven largely by the demand for paraphilic and pedo/hebephilic imagery. The availability[77] of, and demand for, vaginal sex imagery is relatively small.
2. A relatively small number of pornographic "adult" BBS dominate the market at the present time.[78] However, the widespread proliferation of CD-ROM technology now enables "start-up" BBS to accumulate large pornographic portfolios (50,000-plus images) at little cost.
3. Among the most successful "adult" BBS, a tremendous gap exists between the availability of, and consumption patterns for, the four classifications of imagery. In particular, availability of hard-core and soft-core images exceeds the demand for these types of images, while demand for paraphilic and pedo/hebephilic images exceeds their availability. The data suggests that computer pornographers have not yet optimized their image portfolios according to consumer demand patterns.
4. Regular mathematical relationships exist between the size of the BBS and their download activity for each of the four major classifications outlined in this study.
5. Changes in the portfolio of Amateur Action BBS, the market leader for pedo/hebephilic imagery and a leading supplier of the paraphilias, graphically illustrates the market forces common to all "adult" BBS.

Table 5 [Total Surveyed "Adult" BBS Files and Downloads by Classification]

Figure 7 [Total Surveyed "Adult" BBS Files and Downloads by Percentage.]

Some important observations can be drawn from the underlying data:

1. Pedo/hebephilic and paraphilic imagery accounts for 2,685,777 downloads, or 48.4%, of all downloads from commercial "adult" BBS.[79]
2. For hard-core and soft-core images, availability exceeds demand, while for paraphilic and pedo/hebephilic images, demand exceeds availability.[80]
3. The supply of vaginal sex imagery exceeds the demand. The total availability of vaginal sex imagery (a subset of the hard-core classification), defined narrowly,[81] was 6.9%, while the demand was even lower, at 4.6%. The availability of vaginal sex imagery, defined broadly[82] with overlap, was 10.1%, while the demand was again lower at 8.3%.

These five leading "adult" BBS were studied in further detail. Hard-core imagery was the most widely available, followed by soft-core, paraphilia and pedo/hebephilia. Figure 8 makes this pattern clear.

Figure 8 [Top Five "Adult" BBS by Availability Statistics.]

The exception to this general pattern is Amateur Action BBS, which offers an unusually high proportion of paraphilic and pedophilic imagery.

However, these availability figures should be compared to actual consumer consumption statistics for the four classifications of imagery, shown in Figure 9.

Figure 9 [Top Five "Adult" BBS by Consumption Statistics.]

Hard-core is in the greatest demand by consumers among top "adult" BBS, followed by paraphilia, pedo/hebephilia, and soft-core. It is significant to note that paraphilia, which the courts generally regulate according to community standards,[84] increased from third to second, and that pedo/hebephilic imagery, which is per se illegal in light of New York v. Ferber,[85] outmarketed the soft-core imagery despite (or perhaps because of) the difficulty in obtaining the former and widespread availability of the latter.

Even among the most successful "adult" BBS, a tremendous gap exists between the availability of and demand for each the four classifications of imagery. The data suggests that computer pornographers have not yet optimized their image portfolios according to consumer demand or capitalized on statistical techniques used by other industries to market their products.[86]

3. Market Forces Common to All Commercial "Adult" BBS

The research team then performed a statistical analysis on the data, which revealed that the largest "adult" BBS are not disproportionate to their competitors, but simply reflect the diversity in size that one expects among similar companies in a free market. The results suggest that the Amateur Action BBS, far from being an anomaly, is merely the most graphic instance of market forces common to all "adult" BBS.[87]

4. "Adult" BBS Demographics

As noted in the introduction, the Carnegie Mellon research team was able to identify consumers of pedophilic and paraphilic pornography via computer in more than 2000 cities in all fifty states in the United States, most Canadian provinces, and forty countries, provinces, and territories around the world. Appendix D contains a complete list of cities and countries where these consumers were identified. The demographics should not be understood as exhaustive; they merely indicate a floor of the extent to which pedophilic and paraphilic imagery pervades the digital landscape.

5. "Adult" BBS Market Leader: Amateur Action BBS--A Case Study

The intriguing questions the Thomas case raises are somewhat similar: can communities that desire to maintain certain standards still insulate themselves from pornography when it is distributed via computer? Will a fundamentally different test for obscenity be established in the future, given the significant changes in the means of pornography distribution, which is increasingly facilitated by computer networks?[93] Lastly, what will be the impact of such a test on communities which desire a different standard?[94]

In addition to these questions, it is significant to note that 116 Amateur Action BBS images were recently discovered on the Usenet. Two of these images reposted on the Usenet were among those recently found obscene by a Tennessee jury.[95]

The computer dictionary developed for this study was able to classify 99.4% of the 22,319 Amateur Action BBS images, which were downloaded 1.6 million times. The Amateur Action BBS images[96] were also analyzed according to a mean popularity index, which indicates the average number of times each file in each category was downloaded. This index is useful in determining whether a category is in low availability but high demand, or the reverse. The mean popularity indices show that the paraphilias were far more popular than any of the hard-core or soft-core classifications in other words, the demand for the paraphilias exceeded the availability. Finally, the research team developed numerous histograms to divide these categories in increments of ten percent, according to number of downloads. The histograms reveal that most of Thomas's files were in high demand, and suggest that forcing Thomas to remove a few "objectionable" files will have no measurable effect on his subscriber download habits or interest in his board.

Amateur Action BBS relies on three powerful marketing techniques to attract and maintain its large clientele, which ranged from 3500 to 10,700 individuals over a typical six month period. These techniques include a) portraying a "power imbalance" between the sexes, including a disproportionate representation of women in acts which may be considered degrading; b) deceitful marketing; and c) exploitation of children.[97] Examples of each technique are offered below.

The Power Imbalance: Fellatio-Choking. The power imbalance between the sexes is among the most debated and controversial subjects in Western culture. Imagery which may help define the acceptable boundaries of relationships between the sexes has also been the focus of heated discussion.[98] It is beyond the scope of this article to present an exhaustive analysis of sadomasochistic imagery discovered on computer networks. At present, Carnegie Mellon researchers are developing more than two dozen computer dictionaries that explore sub-classifications of B&D/S&M imagery.

One of these sub-classifications, which links fellatio with B&D/S&M imagery provides a salient example of how Thomas (and pornographers in general) may transform the same sexual act from "hard-core" pornography into "paraphilia" material, and thereby increase download activity.[99] Thomas markets 1113 images depicting some form of fellatio. These images were downloaded 38,611 times, for a mean popularity index of 34.7:1.[100] Clearly, fellatio is a relatively unpopular category on Thomas's board, accounting for only 5% of all images available and 2.4% of all downloads. When Thomas describes an image as "Horny sexy blonde sucks cock! She is rubbing her wet pussy!"[101] he generates an unusually low number of downloads.[102] Whenever he uses the word "choke" in his fellatio descriptions, however, he doubles his downloads. The mean popularity index for fellatio-choking imagery increases to 65:1. These results suggest at least some correlation between increased power imbalance between the sexes (man over woman) and increased consumption. The results also remind us that our classification of images is dependent upon how pornographers market them.[103]

The Power Imbalance: Degradation-Bestiality. Degradation of women is one of the most challenging issues for a researcher to define, because the term is emotionally and politically charged. However, the disproportion of imagery depicting women engaged in acts that would be considered degrading in most communities can be definitively established.[104] Because it is beyond the scope of this study to address such issues exhaustively a computer dictionary needs to be developed for each category only one instance will be examined in detail.[105] The research team chose to examine bestiality because demand for these types of images more than quadrupled in the six months following the obscenity convictions of Robert and Carleen Thomas.

Thomas offers 852 bestiality images, which were downloaded 122,057 times, making bestiality the second most popular image category on his BBS. The mean popularity index for bestiality was 143.2:1, the highest of all categories. The resolution (picture clarity) of these images ranges from very poor to exceptional, but the validity was near 100%.

Figure 10 [Amateur Action BBS Bestiality Portfolio. 852 Images, 122,057 Downloads; Chart Indicates % by Downloads.]

The pie chart in Figure 10 illustrates the variety of animals engaged in sexual acts with humans, the most popular of which was "BRUNETTE SLUT TAKES A HUGE HORSE COCK IN HER TIGHT PUSSY!"[106] The most popular image depicting a woman and a dog was "Super slut! Sucks dog cock and gets cum on her cute face!"[107] More importantly, 99.1% of the images depicted women engaged in sexual acts with animals, whereas only 0.9% depicted men engaged in sexual acts with animals. Even the most popular bestiality image involving a man had the following description: "He holds the big erect dog cock! Makes cute blonde suck it!"[108] Bestiality images involving only men and animals had the lowest mean popularity index. Thomas apparently realized that his customers did not want man-bestiality imagery after a series of such images that were posted in September 1993 failed to generate a significant number of downloads.[109] Not surprisingly, the least popular bestiality image involved a man: "Horny black guy fucks a young blonde! Big dog licks his ass!"[110] Animals other than dogs and horses depicted in sexual acts with women include monkeys, donkeys, snakes, goats, pigs, and cows.

Bestiality is a typical, but by no means extreme, example of Thomas's shrewd implementation of this marketing technique. Although nearly one-half of his entire portfolio depicts at least one man engaged in sexual acts with women, Thomas has not posted any man-bestiality images since September 1993. He continued to meet the intense demand for women-bestiality images up until the time of this study.

Deceitful Marketing. Verbal descriptions enable Thomas to market his portfolio as somewhat more taboo than may be the case.[111] Indeed, Thomas is one of the few sysops in the country to exploit fully the remarkable power of verbal descriptions to fulfill customer fantasies. The best illustration of this technique is provided by "incest" imagery. Thomas described 1234 images as sexual or highly suggestive acts among family members. These "incest" images were downloaded 167,534 times and accounted for 10.3% of all the downloads. The mean popularity index for "incest" images from Amateur Action BBS was 135.8:1. A second computer dictionary was developed for incest imagery, which was found to overlap considerably with other paraphilias. For instance, the most popular mother-daughter image involved bestiality: "She holds the dog cock! Inserts it in her daughter's ass!"[112] Thomas offers a triple-paraphilic combination with this popular image: "Mother watches daughter fuck herself with a screwdriver!"[113] The most popular father-daughter image was also the most popular sister-sister: "She is fisting her sister's cunt! Father fucks her asshole!"[114] The most popular mother-son image was also among the most popular father-daughter: "She sucks her son's cock! Father is fucking his daughter!"[115]

Because it is not clear whether the images depict sexual acts between genuine family members, Thomas services his clientele with a shrewd blend of exaggeration and deceit. Curiously, Thomas's descriptions of other categories achieved the highest reliability rating from the panel of judges, suggesting that his is among the most meticulously designed boards in the country.

Exploitation of Children.[116] Thomas has attempted to push the parameters and current interpretation of child pornography law. By offering more than 5000 images featuring the exhibition of genital areas of children,[117] being careful not to depict hard-core sex acts with pre-pubescent children, and terming many of the images "nudist" material, Thomas has attempted to skirt the attention of law enforcement. He repeatedly used marketing language in an attempt to convince his subscribers that his material depicting children was both sexually enticing and legal. At the same time, he boasts that his BBS is the "nastiest place on earth."[118]

The demand for soft-core pedophilic imagery exceeded availability by more than 25%, whereas availability exceeded demand for hard-core and most of the paraphilias.[119] Thomas is fully aware of this imbalance as evidenced by his use of a sophisticated third layer of category abstraction.[120] Most of the hard-core and paraphilic descriptions overlap with the following five sub-classifications: virgin, genital exhibition, no pubic hair, no breast development, and "nudist." For the sub-classification "genital exhibition," for instance, the computer dictionary identified 1361 lascivious descriptions of young girls which contained the following words: spread, spreads, spreading, close-up, kinky, kinkiest, candid, poses, posing, and open-leg. The most popular of these was "TENDER BRUNETTE JUNIOR TEEN! NO TITS! SHE IS SPREADING!"[121] Another such image, "INNOCENT BLONDE JUNIOR TEEN! NO TITS! NO PUSSY HAIR! GREAT!"[122] was widely distributed on the Usenet and via so-called drop or underground ftp sites.[123] The entire genital exhibition portfolio accounted for 216,672 downloads. For the subclassification "virgin," 2249 image descriptions contained the following words: sweet, tender, sprouting, budding, virgin, and innocent. These images accounted for 357,192 downloads. Thomas also relies heavily on overlapping descriptions within this third layer (e.g., no pubic hair and no breast development, nudist and sprouting, tender and spreading, etc.). Figure 11 shows the breakdown of how Thomas markets his pedophilic imagery to consumers.

Figure 11 [Amateur Action BBS Pedophile Portfolio. 5214 Images, 717,896 Downloads; Chart Indicates % by Downloads.]

Appendix B, which lists the most popular files for each of the Amateur Action BBS categories, indicates that of the 22,319 image descriptions analyzed, the following generated the most downloads: "Young amateur teen! No tits at all! She spreads her pussy!"[124] An analysis of Thomas's use of five of these terms young, amateur, teen, no tits, spreads revealed that this and similar combinations generated unusually intense download activity.

Clientele Distinctions Among Amateur Action BBS Subscribers. Thomas clearly understands that his BBS caters to a variety of clienteles. While he cleverly mixes incest with other paraphilias in any number of combinations, as noted above, he carefully avoids explicit overlap between pedophilia and the other paraphilias.[125] There is substantial evidence, based upon a contingency table analysis of category overlap, to suggest that Thomas caters to at least three separate classes of clientele: paraphilic, pedophilic, and hard-core. For instance, when posting a bestiality image, Thomas takes special care not to describe the female in pedophilic terms (for example, "teen," "young," "virgin," "budding," "no tits," "no pussy hair"). He is keenly aware that customers who seek bestiality (and more generally, paraphilic) images do not seek pedophilic images, and the converse.[126]

The number of subscribers to the Amateur Action BBS has varied significantly during the past year. In May 1994, when the first dataset was collected, Thomas had 3621 active subscribers. As of May 15, 1995, he had 10,687 subscribers[127] and had logged more than 1.16 million calls. 98.9% of the subscribers are male, and 1.1% are female. Many of these subscribers are system operators of other "adult" BBS, who download the images for redistribution to their own customers in various regions of the U.S.[128]

Although no definitive numbers are available, there is at least some evidence to suggest that Thomas's BBS is designed to be an advertising lure for his lucrative video business featuring nude children and paraphilic activities. The strong appeal of his BBS to tens of thousands of online consumers and Usenet aficionados around the world highlights the importance of computers as a medium of choice for the distribution and marketing of pedophilic and paraphilic imagery.[129]

IV. Conclusion

Because of this ability to play to consumer desires through the interplay of images and descriptions, the title of this study is "Marketing Pornography on the Information Superhighway." The description, "She hangs from the ceiling as weights are suspended from her pussy lips and nipples" may describe an inherently painful act; however, it is not marketed primarily as a painful act, but as one of restraint. There are many other B&D/S&M images that observers might categorize as painful, based upon their own impressions, but they are not marketed as such and thus not classified in this study as such. When the pornographers want to market an image as painful, when they want to emphasize its torturous aspects, they do so by using words such as pain, torture, abuse, torment, hurt, and suffer. When one actually views many of these images, moreover, there may be some ambiguity as to whether the model is actually in pain or merely acting.

B. Phenomenology of a Porngeist[130]

Ten years ago, college students could not type a few commands from their dormitory computers and receive instant access, via the Usenet, to thousands of bestiality images and stories. Until the recent explosion of World Wide Web sites, which enable users to view images with the click of a button, college students were among the first computer afficionados to have access to image decoders, powerful UNIX-based software and machines, and most advantageously, free and unlimited advice concerning the technical problems associated with downloading, cutting, and pasting these images. As mentioned in the introduction, the pervasiveness of digital pornography may have a profound effect on primary and secondary schools interested in offering their students access to computer networks. Yet, attempts at regulation may be ineffective; "the net interprets censorship as damage and routes around it" is a well known expression among Usenet enthusiasts. The current structure of the Usenet requires that individual sites choose between an "All things not expressly permitted are prohibited policy," or conversely, "All things not expressly prohibited are permitted."[134] A middle ground does not appear viable.

C. Digital Cultural Artifacts

The collapse in the past few years of several of the largest and most active boards in the country, such as the Windy City Freedom Fortress BBS (hebephilic, all paraphilias), The Atlantis BBS (bestiality), and Taste BBS (fisting), appears to have had no measurable effect on current market conditions. The images from Windy City and other defunct BBS are widely distributed via CD-ROM and the Usenet. Thus, it appears unlikely that the conviction of the Amateur Action BBS operator will do more than temporarily encourage other BBS operators to remove their paraphilic and pedophilic repertoires from public display.[138]

D. "Free" Pornography

Yet there is one notable difference between the manner in which pornographers capitalized upon emerging technologies to make a profit in the past, such as the printing press or the VCR, and the Information Superhighway of the 1990s. Pornographers are using the Usenet primarily as a vehicle to advertise their products to a large audience at no cost. Indeed, the concept of hard-core and paraphilic pornography which is widely distributed, yet "free" to the general public, is a novel one. Pornographers are keenly aware that their products remain "free" only to the extent that consumers opt not to subscribe to their BBS. A number of "adult" BBS have reported that their subscriptions "skyrocketed" after posting a few images on the Usenet.

In fact, the Usenet is ideally suited for market niching. A soft-core BBS may choose to advertise on alt.binaries.pictures, erotica.female, alt.binaries.pictures.erotica.blondes, or occasionally, alt.binaries.pictures.erotica. A hard-core BBS might advertise its images on alt.binaries.pictures.erotica. A B&D/S&M BBS may advertise on alt.sex.bondage or alt.binaries.pictures.erotica.bondage. A BBS specializing in any of the paraphilias may advertise on alt.sex.fetish.watersports, alt.sex.bestiality, alt.binaries.pictures.bestiality, or alt.binaries.pictures.tasteless. A gay BBS may advertise on alt.pictures.binaries.erotica.male. An Asian-focused BBS may advertise on alt.binaries.pictures.erotica.orientals.

Many network analysts expect, however, that the Usenet may collapse under its own weight. As more Americans utilize the Internet and Usenet, the proportion of new messages posted each day may be too large for individual sites to carry, or subscribers to read. The trend appears to be moving toward moderated newsgroups, which would exclude most of the alt hierarchy, including the sexually explicit newsgroups. However, mainstream pornography distributors, such as VCA Platinum Pictures, Vivid Video, the Odyssey group, and Leisure Time Entertainment, have begun to promote their products on the World Wide Web.[141] Subscribers are offered "tantalizing QuickTime movie previews of the hottest new releases" and can "browse through our adult video show rooms which feature everything from sizzling new XXX-rated titles, to exotic European releases, to sexy amateur home videos."[142] The companies boast that "shopping for adult videos and novelties has never been easier, more fun, more convenient or more private! And we have an electronic ordering system that makes shopping for adult products as simple, convenient and easy as clicking a button from the privacy of your own computer screen!"[143]

E. Privacy, Pornography and Computers

A great deal of information we consider to be highly personal, and of interest to ourselves and the town gossip our names, telephone numbers, marital status, educational accomplishments, job and credit histories, even medical, dental, and psychiatric records is now being sold on the open market to anyone who believes he or she might be able to use such information to turn a profit. These transactions usually take place without our knowledge or consent.[144]

The compilation of these statistics also raises significant privacy issues, which are not limited to pornography. Computer users normally maintain a logfile in their home directory (entitled .newsrc) that keeps track of the Usenet boards they read and their latest access of each board. At the university studied, these logfiles were publicly available to any authorized user of the campus network.[145] It is possible that any user could obtain another user's logfile and use this information to harass, embarrass, or blackmail unsuspecting users. For instance, if a female student subscribes to talk.abortion and alt.adoption.agency, a user could examine her logfile and infer that this student might be pregnant or have a close friend who is pregnant. Those who subscribe to misc.jobs.offered might not want their employers to know they are looking for another job. With respect to pornography, many users might be embarrassed if their subscriptions to alt.sex.bestiality or alt.sex.fetish.watersports were a matter of public knowledge. Even if other users are blocked from accessing the logfiles of their peers, it is still potentially significant that network administrators have access to these logfiles.[146] Most users appear not to know that these logfiles exist, and were they informed of their existence, might find such monitoring of their accounts objectionable. Unscrupulous network administrators are in a position to compile and sell detailed online information about each of their users to third party vendors.

F. The Marquis de Cyberspace

Law enforcement efforts have done little to discourage consumers or indeed the market leader, Robert Thomas from discussing, trading, and redistributing these images on other computer bulletin boards around the world. As of this writing, Thomas sits in jail and has complained of cockroaches, bad food, filth, and violent prisoners.[147] His wife continues to run Amateur Action BBS, and when she reports to jail, Thomas has indicated that "friends" will run his BBS. His subscriber list still contains citizens from Memphis, Tennessee, where he had been convicted of obscenity charges, and Salt Lake City, Utah, where he faces a federal indictment for distribution of child pornography. Since his trial in July 1994, his subscriptions have increased threefold, to 10,687. His downloads have more than doubled, from 1.6 million to more than 3.7 million. And millions more have likely been downloaded by Usenet aficionados and subscribers to dozens of other commercial "adult" BBS, which have pirated or licensed Amateur Action BBS images.

Thomas is a brilliant marketer, presenting himself as a modern day Marquis de Sade. It is not without irony that he promotes Amateur Action BBS as "the nastiest place on earth." Indeed, the parallels between the lives and "art" of Thomas and Sade would provide fodder for a fascinating study. Both pushed "taboos" and societal mores to the extreme; both flaunted the law and were jailed; both considered themselves renegade artists; and both revolutionized the dissemination of paraphilic pornography Sade in print, Thomas on computer networks. As historian Lynn Hunt has noted:

Virtually all of the themes of modern pornography were rehearsed by Sade; indeed, he specialized in the cataloging of pornographic effects. Rape, incest, patricide, sacrilege, sodomy and tribalism, pedophilia and all the most horrible forms of torture and murder were associated with sexual arousal in the writings of Sade. No one has ever been able to top Sade because he had, in effect, explored the ultimate logical possibility of pornography: the very annihilation of the body, the very seat of pleasure in the name of desire.[148]

The Marquis, it seems, has finally been topped.

V. Summary of Significant Results of the Carnegie Mellon Study

Category "AMAZING":

NASTY NASTY! I WILL NOT DESCRIBE THIS KINKY NASTY GIF!
Downloads: 368
filed under: AMAZING
PARAPHILIA

Category ANAL:

Brunette coed has tiny boobs! 3 inch wide cock in her ass
Downloads: 232
filed under: ANAL
HARD-CORE

Category ASIAN:

Oriental girl inserts her fist in friends pussy, Ass dildo!
Downloads: 365
filed under: ASIAN, FISTING, FOREIGN OBJECTS, ANAL
PARAPHILIA

Category BESTIALITY:

BRUNETTE SLUT TAKES A HUGE HORSE COCK IN HER TIGHT PUSSY!
Downloads: 672
filed under: BESTIALITY, WHORE/SLUT
PARAPHILIA

Category B&D/S&M:

SUPER TORTURE! PIERCED CLIT! PUSSY NAILED TO A TABLE!
Downloads: 297
filed under: B&D/S&M
PARAPHILIA

Category BREAST:

2 firm breasted fresh amateurs! Super cute and big nipples!
Downloads: 868
filed under: BREAST
SOFT-CORE

Category CUMSHOTS:

More amateur CUM! D/L this messy GIF now!
Downloads: 201
filed under: CUMSHOTS
HARD-CORE

Category CUNNILINGUS:

He pulls her panties aside and eats her hairless twat!
Downloads: 22
filed under: CUNNILINGUS
HARD-CORE

Category DOGSTYLE:

Stereo GIF! Juliet loves to take big cock doggy style!
Downloads: 42
filed under: DOGSTYLE
HARD-CORE

Category OBESE:

PRIVATE FAT CHICK GIF! THIS IS MUCH TOO NASTY TO DESCRIBE!
Downloads: 68
filed under: FAT
PARAPHILIA

Category FELLATIO:

Cum squirts from his uncut big cock into her mouth!
Downloads: 234
filed under: FELLATIO
HARD-CORE

Category FISTING:

She inserts her whole fist in her girlfriend's pussy!
Downloads: 494
filed under: FISTING, LESBIAN
PARAPHILIA

Category HOMOSEXUAL:

KINKY AND NASTY GAY BUTTFUCK GIF! D/L NOW IF YOU LIKE KINK!
Downloads: 180
filed under: HOMOSEXUAL
HARD-CORE

Category SHAVED-PUBIC:

CLOSE-UP! TINY HAIRLESS TWAT! TAKEN BY A HORNY PEDIATRICIAN!
Downloads: 178
filed under: SHAVED-PUBIC
SOFT-CORE

Category HAIR COLOR:

Brunette coed caught sunbathing on a balcony! EXCELLENT!
Downloads: 309
filed under: HAIR COLOR, PORTRAITS
SOFT-CORE

Category INCEST:

She holds the dog cock! Inserts it in her daughter's ass!
Downloads: 365
filed under: INCEST, BESTIALITY
PARAPHILIA

Category INTER-RACIAL:

New AA girl Montana! A cute black girl with big sexy boobs!
Downloads: 163
filed under: INTER-RACIAL
SOFT-CORE

Category LESBIAN:

BLONDE OPENS WIDE! HER GIRLFRIEND SHITS IN HER MOUTH! NASTY!
Downloads: 354
filed under: LESBIAN, COPROPHILIA
PARAPHILIA

Category ORGY:

Piss orgy! 3 horny guys piss all over their girlfriends!
Downloads: 141
filed under: ORGY, UROPHILIA
PARAPHILIA

Category OUTDOOR:

PRIVATE COLLECTION! SEXY COEDS CAUGHT NAKED ON THE BEACH!
Downloads: 324
filed under: OUTDOOR, PORTRAITS
SOFT-CORE

Category PANTIES:

Super cute Melinda! Nude with her panties at her knees!
Downloads: 176
filed under: PANTIES, PORTRAITS
SOFT-CORE

Category PEDO/HEBEPHILE:

Young amateur teen! No tits at all! She spreads her pussy!
Downloads: 1687
filed under: PEDO/HEBEPHILE
PEDO/HEBEPHILE

Category PETTING:

Cody is having and ORGASM! Fingers hole!
Downloads: 138
filed under: PETTING
HARD-CORE

Category PORTRAITS:

Lance White NUDE! D/L this free GIF and DISTRIBUTE FREELY!
Downloads: 301
filed under: PORTRAITS
SOFT-CORE

Category PREGNANT:

Pregnant girl, pants at ankles, spread cunt, milking tits!
Downloads: 246
filed under: PREGNANT
HARD-CORE

Category PROMOTIONAL:

Cindy! Footage from her video! Inserts HUGE 12 inch dong!
Downloads: 150
filed under: PROMOTIONAL
HARD-CORE

Category SERIES:

#1, series of nostalgia nudes! Very sexy!
Downloads: 69
filed under: SERIES
SOFT-CORE

Category SHOWER:

Jasmine and Cindy take a shower and play with each other!
Downloads: 80
filed under: SHOWER, LESBIAN
HARD-CORE

Category SIXTY NINE:

#43, They are horny! They are in the 69 position!
Downloads: 59
filed under: SIXTY-NINE
HARD-CORE

Category FOREIGN OBJECTS:

Cindy inserts fat cucumber in Jasmine's cunt! She screams!
Downloads: 266
filed under: FOREIGN OBJECTS
PARAPHILIA

Category TRANSSEXUAL:

SEXY SHE-MALES TAKE HUGE COCKS IN THEIR ASSHOLES! D/L NOW!
Downloads: 239
filed under: TRANSSEXUAL
PARAPHILIA

Category VAGINAL:

A super view of an amateur getting fucked real good!
Downloads: 233
filed under: VAGINAL
HARD-CORE

Category VOYEURISM:

College coed runs nude on a stadium field! Guys watch her!
Downloads: 139
filed under: VOYEURISM, PORTRAITS
PARAPHILIA

Category UROPHILIC:

Cindy stands over Jasmine and PISSES on her face! PEE PEE!
Downloads: 607
filed under: UROPHILIA
PARAPHILIA

Category WHORE/SLUT:

Super slut! Her girlfriend squats and pisses in her mouth!
Downloads: 319
filed under: WHORE/SLUT, UROPHILIA
PARAPHILIA

As mentioned in the introduction to this article, the bulk of the data was collected in May and June 1994, unless otherwise noted. At that time, no member of the research team had discovered any Web sites which contained sexually explicit imagery or materials. In March 1995, the research team discovered 123. The results are discussed here.

II. Methodology

Finally, the Web is browsed predominantly by males.[152]

Appendix D: