The first titIe of my talk, as it appeared in the tentative program, was "Encryption Work at IBM." Then the second title, as it appeared in the program, was "Data Encryption."The third title appeared in the final program and was "Cryptographic Key Management."The title of the talk I will give to you today is "Approaches to Cryptographic Key Management."When you speak later in the day, you always wonder if it is an advantage or whether people are tired and bored.In my case, I'm happy it's late because so many things have been alluded to and encryption was mentioned in several talks.So I don't have to explain what the DES is and what the NSA is up to.As someone mentioned, the NSA is trying to convince various users to use algorithms that are not fully known or evaluable by users, and there is some controversy on that point.I will not discuss this issue further.Instead, the environment I am addressing myself to is one where many modern crypto-techniques operate, where the algorithm is known but the key is kept secret.I'm going to deal with three approaches to key management.Personal Key Management, where the person decides what the key is.Centralized Key Management, where the ten tral organization decides and distributes the keys.And finally, what is most interesting in the field today, a system which incorporates private keys and __-______-__

Nearly all very large data bases of the future will be built by large organizations, and will be very complex and diverse. This paper discusses two approaches to VLDB management: centralized control, where a powerful DB administrator makes all the important decisions about the contents, representation, and access authorization; and distributed control, where the authority to make those decisions is shared between an administration oriented central office and many action oriented users. It is claimed that in large complex enterprises, whether private or public, the organizational and social aspects of DB usage outweigh the technological ones. Arguments are presented to support the thesis that, in such environments, only a regime of distributed control can meet the users' needs and provide the required adaptability, perspicuity and responsiveness. A concept of a decentralized DB is then briefly outlined. An example is presented early on to stress the scope of the problem discussed, and is used later to illustrate specific points.

As a step toward the solution of the placement problem in engineering design, a procedure has been developed for detecting intersections of convex regions in 3-space by means of a pseudocharacteristic function. The mathematical techniques underlying the procedure are discussed, and a system of programs embodying these techniques is described. As a special case a solution is given for the hidden-line problem in graphic display.