Paul Baran

• Born: April 29, 1926
• Birthplace: Grodno, Belarus
• Died: March 26, 2011
• Place of death: Palo Alto, California

Developer of the Internet and researcher at RAND Corporation

Primary Field: Internet

Specialty: Applications

Primary Company/Organization: RAND Corporation

Introduction

Paul Baran's research suggested a distributed network as a strategy for building communications systems that could survive serious infrastructural damage (specifically a nuclear attack), and when the Internet ancestor ARPANET was created, it was with that goal in mind. Baran also developed a key form of data transmission for networks, packet switching, which helped make the Internet feasible.

Early Life

Paul Baran was born the youngest child of a Jewish family in Grodno, Poland (now part of Belarus), on April 29, 1926, and immigrated with his family to Boston in 1928. When Paul was still young, his family relocated to Philadelphia, where he delivered groceries in the neighborhood from a small store his father, Morris, ran. He studied electrical engineering at the Drexel Institute of Technology (now Drexel University), graduating in 1949. His first serious work was for the Eckert-Mauchly Computer Corporation (EMCC), formed by the designers of the Electronic Numerical Integrator and Computer (ENIAC). EMCC had recently built the Universal Automatic Computer (UNIVAC), an ambitious improvement on ENIAC that used magnetic tape for memory storage and was employed to tabulate the 1950 U.S. Census.

After marrying in 1955, Baran relocated to Los Angeles, where he worked on radar systems at Hughes Aircraft while attending night classes at the University of California to earn his master's degree. His graduate adviser later commented that Baran was his first student to investigate the patentability of his thesis—a sign that he was not just a theorist but interested in immediately engaging with the practical applications of his field. In 1959, once he had earned his master's of science in engineering, he began work at the nonprofit RAND Corporation in the mathematics division's computer science department.

Life's Work

During the late 1950s and early 1960s, much of the public was focused on the space race as a manifestation of Cold War tensions. The threat of nuclear attack, however, was also a grave concern in the wake of the U.S. deployment of the first atomic bombs at the end of World War II and the development of the hydrogen bomb. With the threat of increasing stockpiles of nuclear weapons by the world's two superpowers, the United States and the Soviet Union, Baran focused on the problem of dealing with the aftermath of a possible nuclear attack, working under contract to the Air Force through RAND. Long-distance communication networks were still young: The first coast-to-coast telephone call had been placed in 1915, only fifty years earlier, and the greatest teledensity was found in the parts of the country most likely to be struck in a nuclear attack. Like electricity, telephone lines had not become commonplace in much of rural America until the 1930s, only a generation earlier. The technology involved in the nation's long-distance communication was also rather primitive in many respects: Touch-tone signaling had not yet been introduced, outages were a regular occurrence in severe weather, and cellular phone technology had been proposed but remained undeveloped. In short, a nuclear attack or any other serious attack on the nation's infrastructure would make an organized response and even basic continuity of government difficult if not impossible. As a result, retaliation would be more difficult, and without the threat of retaliation, the country would be more vulnerable. The doctrine of mutually assured destruction (MAD) was widely accepted: As long as each side was too intimidated by retaliation to launch a strike, peace could be assured through this deterrent effect; however, MAD worked only if that retaliatory capacity would survive a nuclear strike.

Designing a communications network that could survive at least that first strike would ensure continuity of government and a retaliatory capability because of the still-functioning command-and-control systems. For Baran, digital computers seemed to promise the solution, yet they were still quite new: ENIAC itself had been only the second electronic digital computer, and computers were still used principally for intensive computation and manipulation of numeric data for research; they were not synonymous with communications systems, as they are today; indeed, it is Baran's work that led to that alignment. Baran's colleagues did not always follow his reasoning, so he was inspired to write a critical series of technical papers, defending his ideas and forcing him to define and develop them more rigorously. Key to understanding Baran's solution to the problem of ensuring that communications networks could survive an attack is the metaphor of the human brain, which is not destroyed by damage to a single part and which is capable of resuming functions by working around the damaged portion. Baran wanted a network that would work the same way: a network that would function with whatever surviving parts (nodes) it possessed.

This idea ran counter to the model of the time: Communications networks were discussed in terms of centralization. A centralized network is one in which all nodes connect to a hub that routes data back and forth among them. This makes it easy to add new nodes, up to the capacity of the hub, but the health of the network depends on the health of the hub. A decentralized hub solves this problem; in a decentralized system, destroying one hub destroys only part of the network. Baran's idea was to develop a hubless network—that is, a distributed network, in which nodes were connected not to hubs that moderated traffic but to each other. Although a node might be stranded if enough of its neighboring nodes were destroyed, the network as a whole could survive significant damage without losing its communication capabilities.

Originally Baran approached AT&T with the idea, but decision makers at that company insisted his idea was unworkable, despite repeated pitches from Baran. Ultimately, the distributed network was developed by the Defense Department's Advanced Research Projects Agency (ARPA). Although it originally connected only four nodes, one of the important innovations of ARPANET over previous computer networks was that it was heterogeneous; computers did not have to be compatible with one another in order to be compatible with the network. ARPANET went into development in 1967, incorporating both the distributed network idea and Baran's packet-switching methodology.

Packet switching is Baran's second major contribution to the Internet: a method of delivering data across a network by reducing the original package into bundles for transmission. Baran called the bundles of data “message blocks,” which were sent across the network and reassembled at the node of their destination to form the original package of data. Because communications networks transmit data in bursts rather than in a continuous stream, packets are an especially efficient means of transmission. Baran's proposal was to use nodes that would automatically route packages as they were received, whenever bandwidth was available—taking advantage of the pauses between those bursts. Advanced computers (at least advanced by the standards of the time) could monitor network activity in order to maximize efficiency in packet routing, sending packets to nodes that had the bandwidth capacity to deal with them—an approach called “dynamic routing.” Simultaneous with Baran's work, British computer scientist Donald Davies proposed a similar system using the term packets, which became the accepted label.

Although Baran's network ideas originated in considering the nuclear attack problem, he advocated numerous applications for them. In particular, in 1966 he presented a paper on the future of marketing to the American Marketing Association, describing—before ARPANET even began development—something very similar to online shopping. He envisioned customers at home using their television sets to interact with different stores.

By the time of ARPANET's first demonstration in 1969, Baran had left RAND to cofound the Institute for the Future in 1968. The nonprofit research group focused on another intriguing and difficult problem: the science of long-term forecasts. Baran continued to consult on ARPANET's work. Originally intended for researchers to share information, ARPANET quickly became a communication network used for just about anything those with access to it could imagine. Not just researchers but also college students were granted access through their universities' computer departments, and as computers became more common, so did net users. ARPANET split off from its military component in 1983 and was renamed the Internet in 1989, around the time that modem speeds were becoming fast enough for home use of the Internet to be feasible over normal telephone lines. Because there was initially no need to install a new infrastructure or dedicated high-speed lines used to connect professional servers, general use of the Internet via the World Wide Web quickly spread among the public.

Baran was always quick to point out that the modern Internet was the result of not one or two innovations but the hard work of hundreds of people—an emergent invention. Nonetheless, his vision and practical development of both distributed networking and packet switching place him in the vanguard of those now considered responsible for creating the Internet.

Personal Life

Baran married Evelyn Murphy in 1955. They had one son, David. After Evelyn's death in 2007, he dated and later lived with Ruth Rothman, although they did not marry. He died on March 26, 2011, of complications from lung cancer.

Bibliography

Abbate, Janet. Inventing the Internet. Cambridge: MIT, 2000. Print. A history of the Internet with a focus on ARPA, including a discussion of Baran's packet-switching technology.

Hafner, Katie. Where Wizards Stay Up Late: The Origins of the Internet. New York: Simon, 1998. Print. Although old enough to be noticeably incomplete now, this history of the Internet nevertheless provides thorough coverage of its origins and early years.

Lima, Manuel. Visual Complexity: Mapping Patterns of Information. Princeton: Princeton Architectural Press, 2011. Print. Visualizations of data, including Baran's distributed network.

Salus, Peter H., ed. The ARPANET Sourcebook: The Unpublished Foundations of the Internet. New York: Peer to Peer Communications, 2008. Print. Assembles primary sources related to the ARPANET project.

Wu, Tim. The Master Switch: The Rise and Fall of Information Empires. New York: Vintage, 2011. Print. An analytical look at the Internet and its role in history.