The following, provided by Mike Godwin who was instrumental in revealing the deception of the "Rimm Study," is the proposed research that Rimm and Martin Sirbu submitted to the US Department of Justice in 1995--jt ------------------ Here's the wonderful David Banks proposal to DOJ. From this we learn that the Rimm article in the Georgetown Law Journal was meant to bolster the success of the grant proposal. --Mike Proposal to Analyze Patterns of Pornography Access on Two Commercial Electronic Bulletin Boards David Banks, Department of Statistics Carnegie Mellon University, Pittsburgh, PA 15213 Martin Rimm, Department of Electrical and Computer Engineering Carnegie Mellon University, Pittsburgh, PA 15213 Marvin Sirbu, Departments of Engineering, Public Policy, Industrial Administration Carnegie Mellon University, Pittsburgh, PA 15213 Project Summary We are interested in analyzing detests which describe patterns of usage on two adult bulletin board services (BBS). One of these is the Amateur Action BBS, and the other is the Pequena Penacha BBS. The statistical analysis that is most appropriate depends upon the questions the Justice Department wants to pose and the kinds of information that the system operators have maintained in their records. Based on our understanding of the available data and the results we have obtained in previous studies, we feel that the following analyses can be performed and propose to undertake them. 1. Summary Statistics of Pornography Traffic. These analyses determine the proportions of BBS images which show 71 different categories of sexual acts (e.g., pedophilia, bestiality, etc.) and the cross-tabulations of these categories (e.g., the proportion of pictures that show both pedophilia and bestiality). A similar analysis determines the number of times each category or combination of categories is accessed by customers. This enables an assessment of both demand and supply for specific kinds of imagery on a particular BBS. 2. Trends Over Time. By examining the date on which an image is first posted, longitudinal analysis shows how the BBS grew over time, both in terms of overall size and by specific categories of pornography. For example, a system operator might begin by adding many different kinds of images, and then decide to specialize in particular imagery. 3. User Preferences. If the logfiles show which customers accessed which images, then one can make a refined analyses of consumer preferences. It may be that most customers have narrowly focused preferences. These situations can be distinguished, and profiles can be made of typical purchasing behaviors. 4. Placement on the Space of Adult BBS. Adult BBS are multivariate phenomena; key measurements include their age, size and relative levels of activity in different categories of pornography. By obtaining such measurements from the Amateur Action BBS and the Pequena Penacha BBS databases, we can use multivariate analysis to compare them to our already existing database on 66 other adult BBS that are comparable size and activity. This comparison will show whether the two BBS under scrutiny are similar to other BBS, or whether they are outliners with respect to one or more features. It is very likely that the information maintained on the databases of the Amateur Action BBS and the Pequena Penacha BBS can support all of the analyses listed above. However, we await confirmation on this point from representatives for the Department of Justice, who are currently undertaking an inventory of their data to verify the kinds if information that ware available. The analyses we describe above may have value to the Department of Justice in many ways, depending upon current prosecutorial strategies and their level of general interest in the market structure of commercial computer pornography. Specifically, we expect the results of this study will enable estimation of the volumes of pedophile activity (or similar kinds of pornography) at the two BBS under examination, determine whether particular categories of pornography have been preferentially cultivated by the system operators, assess the extent to which the customer base has specialized tastes, provide demographic profiles of typical users of pedophile or other categories of pornography, and indicate whether the BBS in question are conspicuous in comparison with other BBS. additionally, by including information from alternative data sources we have already examined, it will be possible to draw conclusions about the rate of growth of this industry, and the market factors that affect the development of individual BBS. We also note that the questions we address have academic value, to psychologists, sociologists, legal theorists, and network engineers. Methodology This section reviews the methods we intend to use in addressing each of the four analyses that are proposed, and the kinds of data that are required. Rimm (1995) reports the results of related work which essentially performs the first and fourth analyses upon slightly different databases. The development of summary statistics can be performed on almost any BBS. By examining the "allfiles" list, which provides a caption for each image, the date it was first posted, and the number of times it was downloaded, we can calculate all the quantities we seek. The key to most of this work is a linguistic parsing program that categorizes the images according to their captions; this enables us to sort, label, and tabulate the 50,000 or more images that are commonly available. Issues concerning the reliability of the parser and the validity of the captions have been assessed for other bulletin boards, and are described in Rimm (1995). The examination of trend over time requires knowledge of the dates on which images were posted (or removed). That, coupled with the linguistic parser and elementary statistical methods, should suffice for this aspect of the analysis. If desired, one can assess competing models of economic growth. The study of user preferences requires information on the download history for each customer. From contacts on the Usenet, it is our understanding that Amateur Action BBS has complete information, while the Pequena Penacha BBS maintained only credit card information. In either case, it is possible to track purchasing patterns at the individual level. For the Amateur Action BBS, it may be possible to go further, and determine simple demographics for, say, the pedophile consumers, such as age, income, and urban/rural location. This information would be useful in attempting to forecast the potential market for this pornography, as computer literacy and access spreads. additionally, the Department of Justice might want the ability to generate a list of the top 100 consumers of pedophilia on the amateur action BBS. the problem of placing Amateur Action BBS and the Pequena Penacha BBS in the space of BBS depends only upon the summary statistics from these BBS and the data we have compiled from a study of 66 other BBS. statistical techniques such as regression analysis and cluster analysis will highlight the unusual BBS, when compared against each other. Examples of such methodologies appear in Rimm (1995). Time and Money We envision a short-term project, taking less than six months to complete. Details depend upon the amount of work we are requested to perform, and the start date (summer months are much more productive than academic year months, so an early start would enable an appreciably shorter contract duration). The cost of this proposal is affected by which analyses the department of Justice wants us to perform, what information is available in the detests, and whether the contract is made directly with the principal investigators (acting as independent consultant), or through Carnegie Mellon University (which entails substantial overhead charges). In advance of specific calculations, the cost of performing the first analysis is approximately $20,000, the first, second and fourth is $75,000, and all four is $100,000. References Rimm, M. (1995). "Marketing Pornography on the Information Superhighway," to appear in the Georgetown Law Review.