Tse-Sung Wu, Measuring Environmental Performance:  Sources and implications of uncertainty.
Committee: Chris Hendrickson (CEE), Lester Lave (GSIA/EPP), Francis McMichael - chair (CEE/EPP), Indira Nair (EPP), and Mitchell Small (CEE/EPP)

     This thesis is an exploratory study on the sources and implications of measurement uncertainty in the estimation of toxic industrial emissions.  It analyzes the methods by which surveyed plants estimate their toxic releases for EPA's Toxic Release Inventory (TRI).  The TRI is a collection of environ-mental discharges and other kinds of transfers reported by some 25,000 industrial facilities in the US of more than 300 toxic chemical substances listed in the enabling legislation.  In addition, case studies of industrial plants reporting the TRI were conducted to ascertain their discharge estimate methodologies, to estimate the uncertainties of and cost of obtaining their toxic discharge data, and to describe what kinds of management decisions, if any, are supported by these data. 
     Within the small opportunity sample of case studies, there is great variability in the estimated uncertainty of emissions estimates; in the ways in which plants treat the reporting of TRI data; in their methods (and combinations thereof) to calculate discharges; and in the decisions that some use toxic emissions data to support.  Some interviewed plants are found to face a non-linear loss function regarding decisions that are supported by toxic emissions data.  Current reporting of TRI data is found to be insufficient for explicitly reporting data uncertainty, which would lead to sub-optimal firm and public policy decisions.  Finally, in an analysis of the TRI data with toxicity data, this research finds that from 1988 to 1993, while sources of toxic emissions decreased their overall emission in terms of mass, the toxicity of the discharges from a significant number increased. 
     Since emissions data are becoming more important in both firm and public policy decision making, it is recommended that EPA treat uncertainty explicitly, in order to make the TRI database more useful in support of design for the environment. 
     The work was supported by NSF grant DMI 9319731, IBM, and Carnegie Mellon academic funds. 

Linda Jean Camp, Privacy and Reliability in Internet Commerce.
Committee:  Granger Morgan (EPP/ECE/Heinz), Pam Samuelson (Univ. of Pittsburgh),  Mary Shaw (CS), Marvin Sirbu - co-chair (EPP/ECE/IA), Doug Tygar - co-chair (CS), and Bennett Yee (Univ. of California,
San Diego)

     This thesis examines the conflict between consumer privacy and data availability in electronic commerce systems designed for the Internet.  Particular attention is devoted to the relationship between anonymity and reliability.  Systems which require that the consumer have dedicated hardware, such as smart card based systems, are not included. 
     A representative set of subset of Internet commerce protocols - Digicash (Chaum, 1985), traceable Digicash (Chaum, 1985), MicroMint (Rivest, 1996), Secure Socket Layer (Freier, 1996), Secure Transactions Technology (Mastercard, 1996), Anonymous Credit Cards (Low, 1993), NetBill (Goradia, 1994), and First Virtual (First Virtual, 1995a) - are examined.  These protocols are evaluated in terms of reliability, security, privacy and regulatory fit.  A certified delivery layer for the provision of the highest degree of atomicity with anonymous currency is introduced.  After a discussion of how well the system provides consumer privacy and data for regulatory purposes, consideration is given to the ease with which changes in the regulations could be made to accommodate the protocols considered. 
     A policy discussion of the regulatory proposals is provided from three perspectives: law enforcement, data marketers and civil libertarians. 
     The work was supported by the US Postal Service. 

_____________________________________________________________________