David Clark

Former chairman of the Internet Architecture Board

• Born: April 7, 1944
• Place of Birth: Concord, Massachusetts

Primary Company/Organization: Internet Architecture Board

Introduction

In the twenty-first century, David Clark is considered one of the Internet's elder statesmen. The former chief protocol architect of the Internet, he headed the Internet Architecture Board for most of the 1980s, overseeing the transitional period after the early days of the Defense Advanced Research Projects Agency (DARPA) and before the inception of the World Wide Web.

Early Life

David Dana Clark was born on April 7, 1944, in Concord, Massachusetts. He attended Swarthmore College, graduating with distinction in 1966 with a bachelor of science degree in electrical engineering. In 1969, he earned a master's degree in electrical engineering and computer science from the Massachusetts Institute of Technology (MIT). Four years later, he earned a doctoral degree in electrical engineering and computer science from MIT. His thesis was titled “An Input/Output Architecture for Virtual Memory Computer Systems.”

During his graduate studies, Clark worked on the Multics project, which was a time-sharing operating system that is perhaps most famous as something of a precursor to Unix. It was initially the product of a joint venture between MIT, General Electric (GE), and Bell Laboratories (Bell Labs); Bell Labs pulled out in 1969, and the Bell Labs programmers who had been working on Multics switched their attention to a new operating system that became Unix, retaining some of the Multics conventions but none of its basic design. Multics, in fact, was intended as a commercial endeavor for GE, which wanted to offer a computing-power utility similar to other utilities, such as water, electricity, and gas; a computing-power utility would supply rented time on a powerful computer accessed remotely. From today's perspective, that GE was interested in spending years developing this plan says something about how abruptly the personal computer revolution began: Barely a decade after Multics started, the notion of marketing computer time was superseded by the purchase of personal computers that would become more powerful than envisioned at the time. The computing-power utility model would never become a successful commercial operation, and Multics would be put to different uses. The work on Multics was sophisticated, however, and it introduced the first hierarchical file system. It was also the first major operating system designed with security in mind.

After finishing his PhD, Clark continued to work at MIT in the Laboratory for Computer Science, which today is the MIT Computer Science and Artificial Intelligence Laboratory. He joined the TCP development team in 1975.

Life's Work

TCP stands for transmission-control protocol, which along with the Internet protocol forms the TCP/IP suite, the core protocols of the Internet. TCP is used by e-mail, the World Wide Web, and file transfer programs—the applications that function on the transport layer of the Internet (while IP is the communications protocol, routing packets from network to network). TCP works with IP, detecting problems such as lost, improperly ordered, and duplicated packets (data in the form of sequences of octets, preceded by a header describing the destination).

In a 1981 paper that he wrote with Jerome H. Saltzer and David P. Reed, Clark defined the end-to-end principle, which is the principle that communications protocol operations should be defined at the end points of the communications system. Internet architecture soon adopted the principle.

From 1981 to 1989, Clark served as the Internet's chief protocol architect and chaired the Internet Activities Board, predecessor to the Internet Architecture Board (IAB). The IAB is the steering committee that guides the development of Internet protocols. After his term as chair ended, he was the first head of the Internet Research Task Force (IRTF), a body of the IAB charged with long-term research issues (as opposed to the older Internet Engineering Task Force (IETF), which concerns itself with shorter-term issues such as developing Internet standards).

In 1991, Clark resigned from the IAB to pursue the advanced research that has occupied him since. He has performed considerable research in support for very large, very fast networks and methods to support real-time Internet traffic. He has also focused on the coming “post-PC” era and networking issues concerning mobile Internet-enabled devices, Internet-enabled appliances, and embedded computers. In his work with protocol overhead, he has helped develop a new set of protocol organization principles for high-performance systems.

Clark is also known for his talks on the future of the Internet. He has raised concerns about Internet security and the necessity of the redesign of the Internet's architecture. The variety and quantity of Internet applications have increased to a degree utterly unlike any foreseen by its designers, and as various technologies such as wireless devices and peer-to-peer protocols were developed, their developers incorporated patches and workarounds to make them possible. Incremental changes and home-brewed solutions have been wrought on what is now an old architecture. This ad hoc growth has resulted in an increasing amount of resources being spent patching holes instead of moving forward. What was once a relatively simple if decentralized communications technology has become convoluted and, in Clark's view, increasingly fragile. He has called for a rethinking of the Internet's basic architecture, with a view toward actual implementation, not merely impressive hypothetical designs. In 2006, Clark was a vocal advocate of the National Science Foundation's plan to do just that. Redesigning the architecture could incorporate what is now known about Internet traffic, prepare for scaling in ways the original designers were not in a position to do, and perhaps offer better protection from botnets, spam, viruses, and distributed denial-of-service attacks.

Clark identified four goals for the new design: security, including a way of authenticating both users and computers; protocols, such as better routing agreements between Internet service providers so that everyone benefits from a more efficient Internet; instrumentation, the requirement that every component of the Internet be able to report problems to an administrator; and mobility, specifically the assignment of IP addresses to mobile devices of all sorts in order to ensure network security.

Clark has also pointed out a compelling truth about the Internet in the twenty-first century: The rise of video, both streaming and downloaded, will change everything. In the United States, flat-fee Internet service billing quickly became the standard for nonbusiness customers. That model has survived long enough that the public is accustomed to it, but it was adopted when most forms of Internet usage were roughly equal in their resource consumption; even early exceptions to the flat-fee model, after all, charged by the hour, not by the byte. While BitTorrent and other file-sharing programs created a class of resource-intensive users, they have remained a small, if impactful, portion of the demographic. Video files, however, are much larger even than audio files and moreover have become mainstream; indeed, Netflix streaming movies, Hulu, and YouTube are popular among even “light” users of the Internet, who may access little else. As much as it is painful to accept and unpopular for service providers to admit, the “all-you-can-eat” pricing model has become untenable.

Clark has worked as a consultant for a number of companies, including Openroute, Inc. (1984–99), Bellcore (1994–95), MCI (1994–96), Time Warner Cable (1997–98), and Hewlett-Packard Labs (1997–99), and he has served on the advisory committees of Nexabit (1988–99), Broadband Access Systems (1999–2000), AT&T (1999–2001), Invisible Worlds (1998–2001), Chiaro Networks (starting in 2000), and Telcordia (starting in 1998). He is a previous chairman of the Computer Science and Telecommunications Board of the National Academies. He is a fellow of both the Institute of Electrical and Electronics Engineers (IEEE) and the Association for Computing Machinery (ACM).

Asof 2023, Clark was a senior research scientist at the MIT Computer Science and Artificial Intelligence Laboratory (CSAIL), leading the Advanced Network Architecture group and serving as technical director of the MIT Internet Policy Research Initiative. He has also served as the codirector of the MIT Communications Futures Program and the networking series editor for Morgan Kaufmann Publishers. Clark has said that the primary benefit of the Internet is its ability to connect people and to connect people to information. He believes that the primary Internet issue concerns how to police and control the bad elements on the Internet without impeding or impairing the good elements.

Personal Life

Clark is a Monty Python fan. His sense of fun may be reflected in the fact that a colleague once hung a sign on the door of his office, identifying it as the office of Albus Dumbledore, the Hogwarts headmaster and wise sage of the Harry Potter novels. The Oxford Internet Institute presented Clark with its Lifetime Achievement Award in 2011. He was inducted into the Internet Hall of Fame in 2013.

Bibliography

Carpenter, Brian, ed. Charter of the Internet Architecture Board. RFC 2850. N.p.: The Internet Society, 2000. Web. Internet Engineering Task Force. 12 July 2012.

Clark, David. “The Internet Is Broken.” Interview by David Talbot. Technology Review Dec. 2005–Jan. 2006: n. pag. Print.

Clark, David. “A Taxonomy of Internet Telephony Applications.” Telephony, the Internet, and the Media. Ed. J. Mackie-Mason and D. Waterman. Mahwah: Erlbaum, 1999. Print.

Clark, David, et al. Addressing Reality: An Architectural Response to Real-World Demands on the Evolving Internet. ACM's Special Interest Group on Data Communication (SIGCOMM) 2003 Workshops, August 25 and 27, 2003. Daedalus 140.4 (2011). Print.

“David Clark.” Massachusetts Institute of Technology, 22 Feb. 2023, www.csail.mit.edu/person/david-clark. Accessed 6 Mar. 2024.

McKnight, L. W., W. Lehr, and D. D. Clark, eds. Internet Telephony. Cambridge: MIT, 2001. Print.