Banquet Speaker

From Crossing Chasms to Climbing into Clouds

Gio WiederholdStanford University

A prime motivation in 1984 for initiating the IEEE Data Engineering Conferences was to bridge the gap from theory to practice. I believe in the theory that there should be no gap between theory and practice, in practice there still is.

DataBases. Visualize the state of the art in 1983: Ted Codd published his paper on the relational approach in 1970, Ingres appeared in 1975, ACM TODS started in 1976, Oracle’s SQL (v.2) came out in 1979 and by 1984 the 3rd version release of Oracle achieved read consistency. Papers on optimization were in vogue (13 out of 29 in 1984). Arguments by practitioners that real work had to stick with IMS and CODASYL systems were common.

Systems to process data. With costly hardware timesharing had become the accepted way to access computers, with many distinct and incompatible systems [CTSS 1961] [Plato 1961] [Dartmouth 1964] [Tymshare 1965] [UMichigan MTS] [ABC 1966] [SAIL 1966] [ACME 1967] [IBM TSS 1967] [IBM TSO 1968]. These systems provided access and flexibility, but limited data-processing to a shared central computer. Systems support for database applications consisted of traditional languages, which implied specifying requirements, design, program preparation, and testing versus the specifications. The flexibility implied by query languages was disabled. Impressive systems were built.

Client-server computing. Combining local computation with central resources had been foreseen for a long time, but grew slowly [Tymnet 1974]. Broad acceptance required the availability of adequate terminals, eventually based on the PCs [1981], Windows [1.0 1985], and of ubiquitous networks [Availability of Internet for commercial use 1991] [Gopher 1991]. X-windows provides infrastructure concepts[MIT Athena 1984]. As software and hardware became ubiquitous and relatively cheap, the greatest costs of data-processing became operations and management. This is a topic now addressed by initiatives in autonomic computing (MS). But innovation does not lend itself to an approach that depends on specifications.

Freestanding applications. Exploiting the ease and low cost of experimentation offered by microcomputers motivated development of database systems for the PC [Microrim 1983] and many others, even Oracle PC[1986], 7 years after it was available on the VAX. But limited communication and protection kept that market disjoint from mainstream IT systems. And without shared access they could not compete with spreadsheets, which are now the dominant resource for storage and computing on single-user systems. The computational capabilities of spreadsheets exceed those of databases, allowing their use in planning models that project into the future. An attempt to integrate those technologies went nowhere [Simql, 2001].

Reaching into the clouds. Ubiquitous communication enables now yet another architecture. Keeping data and computation in the clouds solves the sharing problem, As long as data in distinct clouds don’t have to be joined large data volumes can be processed with modest demands on communication and local capabilities. Systems that depend on cloud resources as Facebook applications, Google maps, etc., are precursors of more general service-oriented systems.

Economics and technology. We see that technology acceptance and dissemination is driven by consideration of economics, but not by the economics of writing and improving the software. Successful software is easy to replicate, but the cost of running a copy of the software has to weighed versus benefits it provides to the many end-users: students, researchers, accountants, manufacturers, information providers, and planners, for whom the software is merely a means to exploit information. Here is where our computer science education falls short. Economists are aware that the value of the intellectual property generated by our community is the benefit it can generate in the future. Most disciplines are sensitive to the value of their products. Engineers design to price points. Architects know if they are defining public housing or banker’s mansion, and will design and invest accordingly. Even artists know that in order to become successful they must develop a sense of what grabs an audience. As we are entering a world where computing is no longer esoteric it becomes important to assess a priori our intentions and try to assess its public value a priori. There are segments of software where free and open source can dominate. There are other segments where software has to add non-routine value to what would otherwise be commodity products. Determining the value of software is not easy, but is possible. Entire countries and regions try to distinguish themselves as being innovative, in the expectations of gaining industry and jobs. Engineering remains at the forefront of bringing innovations to practice. To remain successful the developer community has to understand what it benefits it contributes and how its results will be valued.

Gio Wiederhold is now an Emeritus Professor of Computer Science, Electrical Engineering, and Medicine at Stanford University, continuing part-time with courses seminar on `Business on the Internet’ and `Software Economics’. Gio Wiederhold was born in Italy in 1936, received a degree in Aeronautical Engineering in Holland in 1957 and a PhD in Medical Information Science from the University of California at San Francisco in 1976. Phases of his life included 16 years in industry, 22 years full-time academia, and 10 years government service, with cross-interactions throughout. Research topics addressed combustion analysis, compilers, timesharing, database technologies, knowledge-based integration of information, an algebra over ontologies, access to simulations to augment decision-making, privacy protection in collaborative settings, composition of software, and contributions to the semantic web. Wiederhold has authored and coauthored more than 400 publications and reports on computing and medicine, including an early popular Database Design textbook, now in the ACM Digital Library. Gio Wiederhold has been elected fellow of the ACMI, the IEEE, and the ACM. He has been an editor and editor-in-chief of several IEEE and ACM publications. Gio’s web page is at