Office Automation or Office Information Systems was a field of HCI research that flourished for a decade and then disappeared. It attracted leading researchers and established valuable new directions, but much of its record is scattered or absent online. This is a brief account of that activity.
Advances in hardware have produced a new platform about every decade, each about two orders of magnitude smaller (Figure 1). Scientific computing based on vacuum-tube computers gave way to business computing (or "administrative data processing") on mainframes. Then came minicomputers, followed by microprocessor-based personal computers and workstations. Handheld computers such as the Palm Pilot appeared in the mid-1990s and are beginning to flourish as smart phones. And now computation is everywherefew children's toys do not have a chip lurking within.
Time spans in Figure 1 indicate substantial interest in human-computer interaction, not when a platform first or last appeared. Handheld devices are now reaching that level of maturitya decade after the Apple Newton, the form factor is finally taking shape.
Some platforms vanished. Others faded slowly. Vacuum tubes gave way quickly to solid state alternatives. Mainframes leveled off and declined more gradually. For now, personal computers, workstations, and smaller form factors are alive and well. But minicomputers disappeared, squeezed out by microcomputers. Office-systems research rose and fell with the minis.
From the mid-1960s, minicomputerscabinet-size computers that could support more than one personwere used in science and engineering. Minis could support up to half a dozen terminals for tasks such as text editing or a single more-demanding task such as computer graphics or neural networks. Some came with "dumb terminals" and handled each keystroke at the central processor; others downloaded application-specific code to a terminal that handled some interaction locally before sending a screenful of information to the processor.
Our story begins with the "super-minis" that appeared in the late 1970s. Capable of handling a relational database or a few dozen terminals, they were a major force in the marketplace. Boston-area minicomputer companies Digital Equipment Corporation, Wang Laboratories, and Data General became formidable challengers to mainframe giants such as Burroughs, Control Data, and Sperry. They even targeted IBM, though it isn't clear that anyone at IBM lost much sleep over it.
Digital's VAX 11/780 super-mini was first available in 1977 or 1978, depending on whether you wanted just the hardware or an operating system too. In 1984 Digital released the faster 11/785 and a relational database, a strong alternative to more-expensive mainframes.
Data General's 1980 Eclipse MV/8000, the subject of Tracy Kidder's book The Soul of a New Machine, produced billions of dollars in revenue. Wang Laboratories released the first widely successful word-processing software on 8-bit systems in 1976. By 1978 Wang, which placed the first Super Bowl computer commercial, led the world in CRT-based word processors and North America in small business computers. By 1982, Digital's All-in-1 and Wang's Alliance, successor to its 1979 Office Information System (OIS), provided well-integrated solutions: file sharing, word processing, calendar, spreadsheet, business graphics, and email.
Europe saw similar developments. For example, in 1970 CeBIT (Centrum der Büro- und Informationstechnik, translated as Center for Office and Information Technology) began as part of the German Hanover trade show. In the late 1970s and early 1980s it grew rapidly and in 1986 became a separate entity and the world's largest computer expo.
Those companies and the word minicomputer have disappeared. This creates a challenge in conveying how significant the industry was. Digital was the second-largest computer company in the world, with more than 100,000 employeesMicrosoft employs two-thirds that number today. An Wang was one of Forbes's five wealthiest Americans in 1984; his company's stock was the most actively traded on the American Stock Exchange.
Research is not deliberately tied to a hardware platform, but each platform enables different applications and engenders new issues. In Figure 2, major HCI research threads are color coded to indicate platforms that supported applications that drew particular attention. Human factors and ergonomics computer research started with vacuum-tube computers in the 1950s , then shifted to mainframe systems. Business computing gave rise to information systems research that included HCI as early as 1967 . Office automation research blossomed in 1980. CHI emerged a few years later as hands-on use of microcomputers spread. Small form factors are now creating a new research thrust in ubiquitous or pervasive computing.
These threads evolved, intermingled at times, and exist today, except for Office Systems. Yet OA/OIS anticipated and contributed to many important trends.
Starting in the 1950s, technology to support office work incorporated more electronics, most of it limited and very expensive. These early tools, which culminated in electronic typewriters with one-line editing capability but no visual display, were labeled office systems. By the early 1970s, computer-based editors were used by some professional programmers but were still too expensive for most secretarial work.
Prices came down and features were added. By 1977, Wang was the industry leader in standalone word processors, competing with IBM, Xerox, and others (Exxon Office Systems were marketed from 1976 to 1982). Ricoh says it introduced the term office automation (OA) in 1977 . OA and office information systems (OIS) were soon widely used.
A few research papers with Office Automation in their titles appeared between 1977 and 1979 [e.g., 4], but 1980 was the banner year. In March, the first issue of the ACM Special Interest Group on Office Automation (SIGOA) newsletter was published and two conferences were held, the Stanford Symposium on Office Automation , and in Atlanta, the first of seven annual AFIPS Office Automation meetings comprising technical and research presentations and a product exhibition. AFIPS, the American Federation of Information Processing Societies, was the parent organization of ACM, IEEE, and other groups. Six months later a major IEEE workshop on Office Information Systems was held at Lake Arrowhead, California. Figure 3 shows these and other events.
In 1981, the IEEE Computer Society formed a Technical Committee on Office Automation. The next year SIGOA initiated a biennial conference series on Office Information Systems (COIS), and the well-received independent journal Office: Technology and People appeared, followed in 1983 by ACM Transactions on Office Information Systems (TOOIS). 1981 also saw the release of the first German DIN office system standard and the formation of an ANSI "text and office systems" standards group.
Why the word office? Office systems was used for earlier technology. Also, paralleling Management Information Systems research that was focused on mainframes, Office Information Systems fit the more limited group-size scope of these systems.
Why automation? This was controversial from the outset. Automation sounded challenging and appealed to some researchers, but not allEngelbart in particular had long advocated augmentation. For the business world, though, with its keen eye for efficiency and the holy grail of a provable return on investment, automation had a consistent appeal, whether the promise was automatic programming or automated office work. Researchers were discretionary computer users. Most office workers didn't choose their technologies; automation no doubt conjured up different images for them.
The 1980 Symposium stressed that human behavior would be critical in designing and using technology in support of groups. This is unremarkable now, but it was two years before SIGCHI and the conferences seen as key CHI antecedents, and long before CSCW. Leading researchers in human factors, information systems, and AI who were interested in social and behavioral science contributions to computer design and use converged.
The initial charter of TOOIS listed database theory, artificial intelligence, behavioral studies, organizational theory, and communications. This reflected minicomputer use at the time. Minis were an accessible database research tool. Digital's PDP series was a favorite for AI researchers, notably at MIT, whose graduates fueled the minicomputer company growth around Boston. Minis were used by behavioral researchers for experiments and became interactive computers of choice for many organizations. Finally, computer-mediated communication was an intriguing new capability often employed by users at different terminals of the same mini.
Topics included technical work on database theory, some AI papers (the early 1980s were the final days of a quiet period for AI), IS research into decision support and computer-mediated communication, and behavioral studies by researchers later active in CHI. The newsletter favored IS papers, TOOIS favored technical papers and was a major outlet for behavioral studies until Human-Computer Interaction started in 1985. Early repeat authors included Douglas Engelbart, researchers from MIS (e.g., Rob Kling, Lynne Markus, Ray Panko, Murray Turoff, Starr Roxanne Hiltz), CHI and Human Factors (Clayton Lewis, John Gould), CSCW (Lucy Suchman, Thomas Malone), AI and workflow (Carl Hewitt, Clarence Ellis), and other areas including databases and information retrieval (Dennis Tsichritzis, Bruce Croft).
Authors outlined a "theory of the office." A strong emphasis on behavior was complemented by technical approaches. Office Automation: Concepts and Tools, a 1985 collection of research papers edited by Tsichritzis, described its main theme as "the application of Data Base and Artificial Intelligence concepts and techniques to the implementation of Office Automation tools."
A paper by Douglas Engelbart from the 1982 AFIPS Office Automation Conference was reprinted in Irene Greif's seminal collection Computer-Supported Cooperative Work. Lucy Suchman's 1983 TOOIS paper Office procedure as practical action: Models of work and system design profoundly affected me. Malone's How do people organize their desks? from the same volume is another influential example. Other less-cited work reads well, such as a nice overview in the SIGOA Bulletin by Clarence Ellis, Robert Johansen, and Jim Bair of a 1985 workshop with contributions from Rob Kling, Robert Johansen, Bruce Croft, Susan Ehrlich, Siegfried Treu, and others.
The OA/OIS literature also included seminal work on hypertext, workflow management, the use of object-oriented databases and programming languages, and information retrieval. So what happened?
Between 1981 and 1984, as minicomputer sales peaked and mainframes staggered, the IBM PC and the Macintosh appeared, as did Sun and Silicon Graphics workstations and Symbolics and LMI Lisp machines. Initially not networked, these machines pulled researchers oriented toward HCI and AI back to issues around single-user design and use. They also rapidly undermined sales in a minicomputer industry that had counted on expansion.
Minicomputers were largely gone by 1992 (Figure 3). Data General shifted to the Aviion microcomputer in 1989, Digital to the Alpha in 1992. These companies survived several more years in a diminished state; Wang Laboratories declared bankruptcy in 1992. All remained focused on hardware and proprietary software, a strategy that did not work out.
In addition, the slumbering giant had stirred. IBM introduced the System/36 in 1983 followed by the AS/400 in 1988. Although IBM was not intent on promoting these as an alternative to mainframes, this offered safe midrange options in an era when "you'll never get fired for recommending IBM" was common guidance.
Why did the mainframe survive? This question points to a problem with the concept of "offices" within an organization. Most groups are not clearly demarcatedtheir membership is ambiguous, elastic. Group dynamics shift radically when one person comes or goes. People in a workgroup often fall under different budget authorities, making technology acquisition tricky if it isn't organization-wide. In contrast, organizations and individuals are clearly demarcated, with missions and defined needs. An organization must handle payroll, provide email service, maintain records, and acquire technology. Large central machines remain effective for organizational purposes; the mainframe market flattened but didn't dry up. Individuals focused on PCs and workstations, which were later networked to support groups. Minis were squeezed out.
The research field came under pressure from several directions. The erosion of the mainframe market and shift in research attention to microcomputers led to the collapse of AFIPS, which was dependent on a trade show and near-monopoly on research papers at its National Computing Conferences. The intense enthusiasm for AI in the early 1980s, fed by the challenge of the Japanese Fifth Generation project, massive US and European funding, and the drama of the Lisp machine startups, wore off as AI and workflow modeling and management did not meet expectations. Offices were not being automated. In 1986, the last AFIPS OA Conference was held, and SIGOA became SIGOIS.
No sooner was automation out of favor than office lost its appeal. In 1989, TOOIS dropped office to become TOIS. In 1990, Office: Technology and People became Information Technology & People. In 1991, Office Information Systems conferences gave way to Organizational Computing Systems. (SIGOIS kept the name in the face of declining membership until 1996, and its descendant, SIGGROUP, is now gone.) Only the ANSI standards group retained the term, now office and publishing systems; it is far less active today.
Where did the researchers go? Some researchers returned to roots in Information Systems and AI, others helped form and populate CHI and CSCW, still others fed growing efforts in object-orientation and information retrieval. Work on social aspects of computing that had been fostered in this setting later flourished in digital library and social informatics efforts.
When we look at conceptual development in our field, progress seems steady, even slow. Ideas and prototypes built on research machines are slowly elaborated and eventually lead to widely used tools and applications. On closer inspection, the picture is not so tidysome themes disappear for a time, and may reappear promoted by new groups that lack awareness of the prior work. Some researchers, developers, and technology adopters are left high and dry by rapid changes in direction of research streams.
This column calls attention to a largely forgotten but significant contribution to human-computer interaction 20 years ago. It reminds us that what is solid and thriving today can unexpectedly disappear. We are well advised to pay close attention to the environment and diversify our skill set. Tomorrow's weather will not be the same as today's, and yesterday's may quickly be forgotten.
A Tip of the Hat... To Bob Allen, Dan Corwin, Michael Good, Starr Roxanne Hiltz, John King, Clayton Lewis, Dennis McNurland, Ray Panko, Trish Pendleton, John Thomas, John Whiteside, Gayna Williams, Dennis Wixon, Eleanor Wynne.
2. Rajiv D. Banker and Robert Kauffman, The Evolution of Research on Information Systems: A Fiftieth-Year Survey of the Literature in Management Science. Management Science, Vol. 50, No. 3, pp. 281-298, 2004.
5. Proceedings of the Stanford International Symposium on Office Automation, March 23-25, 1980. Extended version published as R. Landau, J. Bair and J. Siegman (eds.), Emerging Office Systems. Norwood, NJ: Ablex, 1982.
About the Author:
Jonathan Grudin is a senior researcher in the Adaptive Systems and Interaction group at Microsoft Research. From 1983 to 1986 he worked for Wang Laboratories, a leading producer of minicomputer-based office information systems. His Web page is http://research.microsoft.com/~jgrudin.
©2006 ACM 1072-5220/06/0500 $5.00
Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee.
The Digital Library is published by the Association for Computing Machinery. Copyright © 2006 ACM, Inc.