Report from the 1997 International Conference on Software Engineering
A UniNews Exclusive Conference Report
by Tony Wasserman, UniNews online Conference Correspondent
More than 900 people, half from outside the US, attended the 1997 International Conference on Software Engineering (ICSE 97) in Boston during the week of May 18. The conference theme was "Pulling Together", referring to the goal of bridging the gap between the research community and professional developers. In recent years, the conference was focused primarily on technical papers and research results, with only a few presentations describing new commercial technology or industry practices. While the quality of the papers was high, there was often little awareness of the commercial state-of-the-art.
The Keynotes
Mark Weiser, from Xerox Palo Alto Research Center, spoke directly to this issue in his keynote talk. (His presentation can be viewed or downloaded.) He observed that most academics don't do software engineering, but rather do research on software engineering. Whereas software engineers produce software, research on software engineering produces an understanding of processes. As a result, he noted, academics and industry professionals might not even be talking about the same thing. He also described a form of "reality distortion" in industry, based on his observation that engineers tend to talk with some understatement, while marketeers may overstate the facts. Thus, the engineer's statement, "This software isn't too bad", might be seen as a negative comment by the marketeer, who fails to see that the engineer is quite proud of the software. (Is this why engineers are rarely allowed to talk to prospective customers?)
The organizers of ICSE 97 set out to broaden the focus of the conference and to create an event that would be of interest to the broad software engineering community, not just the researchers. The research paper sessions were augmented by tutorials, workshops, invited panels and presentations, industrial lessons and status reports, and the co-located Symposium on Software Reuse.
The conference was opened with a harsh dose of reality by keynote speaker Ed Yourdon, who talked about "Death March" projects, as described in his recent book of that title. (His presentation can be downloaded.) Yourdon described such projects as having one or more of the following characteristics: 50% or more schedule compression from "normal"; half or fewer of the staff normally needed for such a project; or budget cut to half or less of what would be expected. He placed such projects into four categories, ranging from "Mission Impossible", where an experienced team could put forth a superhuman effort to do the job, to "suicide", where the project team is likely to be extremely unhappy during and after the project and where there is little chance of success. He gave numerous suggestions for how to succeed on a "Mission Impossible" project, and noted that Death March projects are becoming a way of life in many companies, citing several firms for which it is their standard operating procedure.Yourdon gave numerous suggestions for how to succeed on a "Mission Impossible" project, and noted that Death March projects are becoming a way of life in many companies, citing several firms for which it is their standard operating procedure.
While most people enjoyed the presentation, some found it a rather depressing view of the "real world", and not what they hoped to hear in a keynote talk. Of course, with so many organizations "living on Internet time", the question is how to develop and follow good software engineering practices under severe time and staffing constraints. While Yourdon made some suggestions along these lines, these issues served as recurring themes in the conference.
Key Conference Topics
Many of the research papers focused on software quality and on best practices. While software inspections are a widely accepted "best practice", they are usually done in a group meeting. With the increase in distributed development, it is not always feasible to bring everyone together for such meetings. Stein, Riedl, Harner, and Mashayeki described an industrial study using a tool designed for distributed, asynchronous inspections. They reported on experience showing that distributed work groups could collaborate effectively using this method. A separate panel session, chaired by Kanth Miriyala of Andersen Consulting, addressed collaborative software engineering issues, including assigning and coordinating activities in a dispersed team, issue posting and resolution, and handling of cultural differences.
The topic of "software processes" was also widely discussed. A tradition of the ICSE conference series is to recognize the most influential paper from ten years ago. At ICSE 97, the award was given jointly to Prof. Leon Osterweil, from the University of Massachusetts at Amherst, who reviewed his now-classic paper, "Software Processes are Software, Too," and to Prof. M.M. Lehman, from Imperial College, London, who was a respondent to Osterweil in 1987.
Since then, there has been a vast amount of work on modeling of software processes, with extensive efforts to help organizations create better managed, repeatable processes for developing software. The Software Engineering Institute at Carnegie Mellon University (SEI) has been a leader in this effort. The SEI has supported the establishment of software engineering process groups around the world, and has put forth a five-level capability maturity model (CMM), where the lowest level describes an ad hoc, unmanaged process, and the highest level describes an optimized process, with continuous process improvement. In addition to several research papers on software processes, Marie Silverthorn, from Texas Instruments, organized a panel session on the benefits of having a high CMM level. This notion is the subject of some controversy, since having a high CMM level is no guarantee of project success.
The IEEE Computer Society presented its prestigious Computer Entrepreneur award to Dan Bricklin, co-designer of Visicalc, the first electronic spreadsheet. This award is given to people 15 years or more after their contribution. Previous recipients have all been associated with advances in hardware design, so it was appropriate that the first award for an advance in software be given at an ICSE conference. The Special Interest Group on Software Engineering (SIGSOFT) of the Association for Computing Machinery (ACM) gave its Distinguished Service Award to Dr. Peter Neumann, from SRI International, who is best known as moderator of the RISKS Newsletter on computer-related risks. The inaugural SIGSOFT Distinguished Research Award was given to Dr. Barry Boehm, of the University of Southern California, for his numerous contributions, particularly his work on software engineering economics and COCOMO cost modeling.
Conference Tutorials
The main conference was preceded by 25 tutorials over a two day period. These tutorials covered such topics as design patterns, software architecture, Java, CORBA, OO analysis and design methods, and software processes. Watts Humphrey, known for his contribution to the SEI's CMM, presented a tutorial on his most recent work, the Personal Software Process. Associated workshops included software engineering and the World Wide Web, software engineering for parallel and distributed systems configuration management. Added to all of these activities were demonstrations of research prototypes and a doctoral consortium, with graduate students presenting the results of their work.
In all, there was something for everyone interested in software engineering. Researchers were able to benefit both from reported research results and from industry presentations that provided some practical problems for future research. Industry attendees had the opportunity to learn of new advances from researchers in the field.
The conference proceedings are available from ACM (Order number 592970) and the IEEE Computer Society (Order number PR07816). ICSE 98 will be held in Kyoto, Japan, in April, 1998, and ICSE 99 will be held in Los Angeles in May, 1999.
Anthony I. Wasserman
Software Methods and Tools
tonyw@methods-tools.com