DARPA's IPTO had Formidable Reputation

By Arthur L. Norberg

One of the prime agencies for computer R&D funding in the post-1950 period was the Defense Advanced Research Projects Agency. DARPA, through its Information Processing Techniques Office (IPTO), which operated from 1962 to 1986, selected a series of specific programs in computing to solve bottlenecks in command and control systems.

IPTO's emphasis on ambitious technical objectives and its years of nurturing the institutional framework for R&D in information processing resulted in some remarkable technical achievements. As a result, the office gained a formidable reputation among its associates and many outside observers. A recent history of IPTO describes and evaluates the management practices used in IPTO that gave rise to these achievements and uses this study of management to examine the programs and accomplishments of IPTO. The history, Transforming Computer Technology: Information Processing for the Pentagon, ends with an evaluation of the influence of IPTO programs for the computing enterprise.

As one of its prime practices, IPTO promoted an array of high-risk R&D projects to expand the frontiers of computer science and engineering (CS&E) for both civilian and Defense Department needs. High-risk projects often require long-term support. Because of its place in DOD, IPTO had significant budgets to expend and, thus, could sustain these projects. IPTO employed an amalgam of approaches used by other military agencies in the 1950s and some new ones designed for the special circumstances in DARPA. Among these approaches were fast turnaround times for project approval and extensive personal initiative on the part of program personnel. Indeed, it was the program directors and managers who conceptualized, designed, organized, funded and monitored projects and who played a major role in the diffusion of results.

IPTO succeeded largely because it recruited capable and technically able members of the research community and allowed these people-within the context of DARPA's mission-an unusual amount of freedom to promote research as they saw fit, seek advice as they felt the need and manage the office as they thought the program required.

Focus on specifics

The IPTO programs were designed to focus on specific objectives, not to support general R&D in computing, but they still had a remarkable impact on computing. The academic and corporate researchers funded by IPTO sought a more flexible, capable and interactive computer. These machines set the stage for the transformation of computers from the large, alphanumeric-based machine of the 1960s to the colorful desktop computers of today.

Subsequent to the 1960s, four remarkably striking new features were added to computers:

• The interactive nature of computer use (online activities).

• The connectivity of computer systems and users (networking).

• The visual capabilities of input/output systems (graphics).

• The increased sophistication of systems (intelligent systems).

IPTO programs and contractors developed many of these new areas, including networking. Occasionally, as in the case of graphics, IPTO projects in the 1960s and 1970s laid the fundamentals for exploitation by industry in the 1980s. A few times, as in the case of time-sharing, IPTO-sponsored researchers joined in a development and stimulated rapid change. And research in artificial intelligence virtually owes its progress to IPTO support and leadership.

Emphasis on time-sharing

IPTO's first major venture in computing was to ensure greater use of time-sharing, increase its effectiveness in R&D and encourage other groups to adopt the idea. Time-sharing demonstrated the benefits of locally shared resources using identical computers. After advancing the state of the art in time-sharing and encouraging the use of the new time-sharing systems, IPTO turned to consideration of how to increase the sharing of resources among different computer systems. IPTO took technological leadership in the development of packet-switching technology and networks, especially the ARPAnet and, later, the Internet, today's driving engine of computer communications.

Officially, IPTO's networking project was the implementation of a specific tool to solve resource-sharing problems that the office encountered in administering and guiding its programs. IPTO's direct management involvement was the key that allowed the network to succeed. ARPAnet, a large-scale experiment, started to solve the problem of resource sharing between computers and among researchers. The goal was to connect computing systems and-through the systems-the researchers. Then research could be accumulated and duplication of effort avoided through the sharing of resources and improved communication. Through the use of packet switching to connect computers-and, later, computer networks-distributed computer resources could be economically shared, and access to the resources and the people using them could be improved. ARPAnet was designed inside IPTO to achieve these ends.

Introducing the Internet

IPTO made even greater contributions with the design and implementation of the Internet. The Internet grew out of DOD's need to interconnect its various research networks. The solution to the problem was a new protocol and architecture, designed by Vinton Cerf and Robert Kahn in IPTO and published in 1974. The development of internetwork protocols and their acceptance and use within DOD and the university communities made flexible, worldwide interconnection of computer users commonplace.

The history of IPTO's involvement with computer graphics illustrates the growth of a vital professional community and the importance of support for the exchange of ideas for R&D. The fundamental concepts behind the remarkable computer graphic images we encounter every day emerged primarily from research projects funded by IPTO. These projects defined the problems regarding I/O devices, the visual characteristics of computer graphic images, and the theory and techniques needed to produce aesthetically pleasing and seductive computer images. Through the commercial exploitation of these developments, graphic systems became available to everyone with access to a computer.

IPTO-supported projects produced results that could be used in any computing system for many purposes, regardless of the nature of the organization. A list of the technical leaps produced by IPTO and the research community that it supported during its 25-year history goes beyond time-sharing, wide area networking and connections across networks. The list includes interactive graphics, distributed computing, very-large-scale integrated circuit design systems, natural language communication systems for use with computers, and expert systems. The R&D programs produced broadly relevant results that found their way into both military and civilian systems.

Many of the people associated with these IPTO projects eventually started other innovative projects, such as the Alto project at the Xerox Palo Alto Research Center; Alto was the grandfather of all subsequent microcomputers. The graphics on this machine, which developed from the results of earlier IPTO projects, found their way into Apple Computer Inc. computers and Atari's game machines. In the 1980s, industry's exploitation of expert systems, local area networks, applications software packages, flexible and sophisticated chip designs, graphics, and the use of AI in design and operation of computer systems-in other words, the world of computing in the 1990s-became possible because of the deliberate efforts of IPTO to produce more capable, flexible, intelligent and interactive computer systems.

IPTO sponsored the ILLIAC IV, designed at the University of Illinois at Urbana-Champaign and built by the Burroughs Corp. ILLIAC IV was the world's first large-scale array computer and a new development in parallel processing architecture. With the advent of DARPA's Strategic Computing program in the 1980s, IPTO returned to the problem of massively parallel computers. First, there was the development of more powerful workstations. Based on government-supported work at the University of California at Berkeley and Stanford University, Sun Microsystems Inc. and Silicon Graphics Inc. developed-with further help from the government-advanced workstations using the reduced instruction-set computer architecture. RISC, proposed by IBM Corp., was brought to practice by these companies and the two universities.

IPTO's influence

While there are many challenges for computing in the 1990s, some of which (i.e., security issues and network management) have been introduced by the online and interconnected computing environment, there is no question that computing in the 1990s has overcome many of the limitations of the 1960s environment. This is in large part due to the emphasis IPTO placed on the interaction between humans and computers and on making the computing environment more responsive to human needs.

IPTO's programs achieved these results because the office focused on a four-pronged approach to computing system development: resource sharing, system development, integration of results and test beds for implementation. In addition to influencing changes in the computing environment, IPTO also stimulated the creation of closer ties among members of the CS&E community, affected military systems through the introduction of new computing techniques and systems, and had a substantial effect on the computing education community.

For more information, see Norberg's book, Transforming Computer Technology: Information Processing for the Pentagon. Johns Hopkins University Press, 1996.

Norberg is a professor of computer science at the University of Minnesota. His interests include the history of science and technology.