BY FRED B. SCHNEIDER

RICHARD WALTER CONWAY, who retired from Cornell University in 1999 as the Emerson Electric Company Professor of Manufacturing Management, died on March 19, 2024, at the age of 92. He is today best known for co-authoring the authoritative text on the theory of scheduling, but during his lifetime he published influential research agendas on stochastic simulation, developed early error-correcting compilers that were widely used to teach introductory programming, and co-authored a series of pioneering introductory programming textbooks.

Dick was born to Robert and Lillian Conway on Dec. 12, 1931, near Milwaukee, Wisconsin. In 1953, he married Edythe Davies Conway, who predeceased him on April 1, 2022, at the age of 89. They had three children: Kathryn, Ralph, and Evan. It was a very Cornell-connected family. Both Dick and Edythe received their undergraduate education and doctorates at Cornell, and both served on the Cornell faculty for their entire careers; the children received their undergraduate degrees from Cornell, with Kathryn later returning to Cornell to work in the College of Engineering.

While earning his bachelor’s degree in mechanical engineering (1954) at Cornell, Dick played in Cornell’s Big Red Band and in the university orchestra, he was a member of the Cornell rifle team, and he rowed “stroke” in the 150-pound crew for the spring 1952 junior varsity boat and “middle position” in the 1953 varsity boat. In 1958, Dick received Cornell’s first Ph.D. degree in operations research and industrial engineering. His thesis, “An Experimental Investigation for Single-Stage Production,” used simulation to investigate job shop scheduling, a topic that was a research focus both early and late in his career. His Ph.D. was supervised by Andrew Schultz, Cornell University School of Operations Research and Information Engineering (ORIE) department chair and later dean of the College of Engineering, who remained an influential figure throughout the early part of Dick’s career.

Dick joined the ORIE faculty as an assistant professor in 1958, and in 1964 he was promoted to full professor. While an ORIE graduate student, Dick had been drafted by Schultz to teach Cornell’s first course on digital computing (using an IBM 650). Teaching this course proved to be ideal preparation for Dick to work with Cornell faculty Anil Nerode and Bob Walker, who had been tasked by Schultz to create a blueprint for creating a department of computer science at Cornell. Not without controversy, the department was launched in 1965 with Dick and three faculty members that he helped to recruit from outside. The first mission of the department was to produce Ph.D.s to populate the many computer science departments that were expected to be formed at other universities in this new discipline. A $1M grant from the Sloan Foundation made this feasible; not until 1978 did Cornell offer an undergraduate major in computer science.

Dick was known to say that creating the computer science department was one of the highlights of his career. He remained a faculty member in the computer science department for 19 years, moving in 1983 to Cornell’s Samuel Curtis Johnson Graduate School of Management (JGSM). The move was partly to seek new challenges but also a reaction to the computer science department’s direction and culture, which had diverged from Dick’s research vision. At JGSM, where in 1993 Dick became the first Emerson Electric Company Professor of Manufacturing Management, his research returned to the development of tools to facilitate simulation — revisiting questions from the early part of his career but now leveraging opportunities afforded by modern processors and graphical displays. His XCELL Factory Modelling System, developed with Bill Maxwell and Steven Worona, was among the products of that period.

While on the JGSM faculty, Dick also created a full-semester integrated 15-credit course in manufacturing. This so-called “immersion course” was the only course a student would take that semester. It involved extensive field trips to various manufacturing sites, as well as a series of lectures where other JGSM faculty would present material that supported the course’s theme. The course was a great success, and other topic concentrations within JGSM subsequently began to offer their own immersion courses.

Among Dick’s most visible research accomplishments was the 1967 text Theory of Scheduling (Addison Wesley Longman), co-authored with his Ph.D. student, Bill Maxwell (NAE 1998), and with another Cornell ORIE Ph.D. student, Louis Miller. The book was an attempt to assemble what was then known about job shop scheduling. The text succeeded far beyond expectations. Republished in 2003 by Dover Books with no changes needed, the book continues to be highly regarded within the ORIE community. For example, the Institute of Operations Research and the Management Sciences (INFORMS) includes the book in its “top twenty-five seminal books in Operations Research,” the journal OR/MS Today cites it as “one of the 26 books that defined and established the field of Operations Research,” the book’s publication is listed in an INFORMS timeline as one of the “great moments” in operations research history,¹ and in 2002 Dick was named one of the inaugural INFORMS fellows. A Russian translation was produced in 1975.

When closed-form solutions are unavailable, simulation may be the only option for predicting the behavior of a manufacturing system. Dick’s Ph.D. thesis considered such simulations, and that was the focus of his early research while working in Cornell’s ORIE department. A 1961 leave to work at RAND Corporation gave Dick the opportunity to use the newly developed SIMSCRIPT language for describing simulations that would then be executed on an IBM 7090; this made Dick one of the first outsiders to program in SIMSCRIPT. His real impact in simulation, though, was through a pair of papers he co-authored with Bill Maxwell and Bruce Johnson on outstanding problems. These papers were published in the journal Management Science in 1959 and 1963; in 2004, the journal described them as “seminal papers” for the study of stochastic simulation.²

Once Dick moved into the computer science department, his research focus shifted to the implementation of error-correcting compilers. In those days, students would submit a program as a deck of punch cards and, after waiting for the deck to be processed, be informed that a small syntax error had caused the compiler to suspend processing of the deck. But if the compiler could detect the error, why not also attempt to repair it and thus reduce the number of iterations needed before a program would compile and could be executed? The CORC (Cornell Compiler) programming language and compiler, developed with Bill Maxwell in 1958, was an attempt at such an error-correcting compiler; it ran on a Burroughs 220 and was used from 1962-66 to teach programming at Cornell. CLP (Cornell List Processor) extended CORC for use in writing simulations. Next came CUPL (Cornell University Programming Language), which ran programs on Cornell’s CDC computer replacement for the Burroughs and was in use until 1969.

But Dick’s PL/C compiler, which processed a dialect of IBM’s PL/1 language, was the error-correcting compiler that had broad impact. IBM initially funded its development in 1968 as part of their attempt to attract Fortran users to PL/1. At its peak, PL/C was in use at over 250 universities worldwide — a huge number at the time. The compiler was not only effective at correctly repairing errors in programs, but it was also efficient, making it ideal for teaching programming to students who were using a shared mainframe computer. PL/C compilers remained in use through the early 1980s.

Interactive computing (and ultimately personal computing) diminished the need for the types of proactive error correction pioneered and implemented in the PL/C compiler. Dick’s COPE (Cornell Programming Environment) system investigated some of these issues.

Dick also co-authored with David Gries one of the first introductory textbooks that covered “structured programming,” a radically new (but today widely accepted) discipline for restricting control flow in programs. The textbook, Introduction to Programming: Structured Approach Using PL/1 and PL/C-7, was also among the first to explain “loop invariants” as a new way to reason loops. The first edition of the textbook used PL/C; it was published in time for fall 1973 classes, and it was subsequently used by Dick, Gries, and others to teach programming at Cornell to large numbers of students. The book was so successful that Dick produced eight other versions to cover other programming languages and various audiences (2nd ed., Winthrop Publishers, 1975).

Dick’s software-development efforts were not exclusively devoted to compilers for supporting instruction, though. For example, in 1970, with Bill Maxwell and Howard Morgan, he also developed the ASAP file management system. This system embodied a novel approach to authorizing fine-grained access to the fields stored in a file’s records. In particular, ASAP allowed security policies to be specified using Boolean expressions — novel for its time — and it used a compiler to enforce those policies.

Those who knew Dick personally will report that he was goal oriented and not easily distracted. Outside of work, he was passionate about sailing. He spent summers sailing his 37-foot yawl, Sea Fever, on Cayuga Lake, Lake Ontario, and on the East Coast from the Chesapeake Bay to Nova Scotia. He was also an accomplished tennis player, winning senior tournaments long after he had retired.

___________________

¹Saul I. 2002. Great moments in histORy. (Celebrating 50 Years of Operations Research). OR/MS Today 29(5):31.
²Nelson BL. 2004. Stochastic simulation research in management science. Management Science 50(7):855-68.