JITE v35n1 - Under Review - Where Wizards Stay Up Late

Volume 35, Number 1
Fall 1997

Hafner, K. and Lyon, M. (1996). Where Wizards Stay Up Late: The Origins Of The Internet. New York: Simon & Shuster. $24.00, 304 pp. ISBN 0-684-81201-0.

Brian K. McAlister
Pittsburg State University

Where Wizards Stay Up Late is a book about frontier technology development, fate, human nature, government, and society. The cast of characters is impressive, the plot intriguing, and the outcome affects us all. Is it a spy thriller? No, but Where Wizards Stay Up Late is a must read for anyone interested in the origins of the Internet.

This topic is especially timely considering the growing popularity and accessibility of a revolutionary communication medium. The value of this book lies in the comprehensive coverage of the significant factors that lead up to the development of the Internet. History unfolds as the authors introduce the reader to theorists, computer programmers, electronic engineers, and computer gurus who had the foresight and determination to pursue their ideas and affect the future of technology and society.

The prologue and epilogue tell the story of a reunion of these players. In 1994, many of the key contributors to the development of the ARPANET, the first wide area network that eventually evolved into the Internet, were invited to celebrate their contributions to the effort. The author uses this event to introduce effectively the topic, key players, and to provide closure.

The effective use of a prologue and epilogue exemplify the fine job that the authors did in structuring their book. The prologue is followed by eight chapters, each used to organize events surrounding key developments. Most flow chronologically, except when it is more logical to follow stories of individual key players whose backgrounds overlap in their development of key concepts.

Immediately following the Epilogue, the authors included a section called Chapter Notes. These notes are organized into short 1-2 paragraph synopses of each chapter. This section provides a concise means for the reader to locate content of interest and would be most useful for the person wishing to use this book as a reference.

A complete bibliography, listing books, journal, magazine and newspaper articles, unpublished papers, interviews from secondary sources, and other documents is provided. Also, as one would expect in a book about the Internet, electronic archives are included. The authors also provide an acknowledgments section that lists almost everyone discussed in the book, and many who were not. This extensive list is due to the numerous interviews conducted while collecting information for the book. The book concludes with an Index listing key words, events, and names. This is another invaluable section for anyone using the book as a reference.

The authors did a great job organizing and structuring the book so that it can be used long after a first reading. While the book is useful as a resource, the content delivery will entice you to read it thoroughly. Much of the book reads like a mystery. The authors introduce key events and players who each play a small part in the long, often haphazard development of this technology. First, the reader follows the early history of the Advanced Research Projects Agency (ARPA); ARPA funded the project that lead to the ARPA network (ARPANET). The agency was proposed in 1957 (following Sputnik), in an attempt to (a) centralize the organization and control of major research projects, and (b) reduce rivalries between different branches of the military. Though opposed by many of the leaders in the different branches of the military, Congress funded ARPA in 1958.

The authors remind us of the key role that the military plays in technological and, more specifically, computer research. For example, the Navy supported Howard Aiken's work on the Mark I; the Army supported the University of Pennsylvania's Electronic Numerical Integrator and Calculator (ENIAC) project; and the Navy and the Air Force supported a computer called the Whirlwind at MIT. Many early computers that are listed as examples in various text books have some tie with military support. It is not surprising to learn that the origins of the Internet started at ARPA, whose offices were initially housed in the Pentagon.

It is important to note that ARPA's charge was much broader in scope than simply the Internet. Understanding its early history helps the reader get a feel for the conditions under which early work on the Internet took place. It is also important to note that although the Internet has its links to the military, it was not developed for the purpose of national security in the event of a nuclear attack. This myth that has even been printed in such widely-read publications as Time magazine.

The authors develop the plot by interweaving stories about the significant contributors to the Internet. I was pleasantly surprised when I found, among the cast of characters, well known people from various fields. I had no idea about their involvement with ARPA. For example, Joseph Carl Robert Licklider, a noted psychologist whose interests included human factors research, worked at ARPA and is credited with shifting the focus of the Command and Control Research office from a lab that played out war-game scenarios to advanced research in time-sharing computer graphics and improved programming languages. The office later became known as the Information Processing Techniques Office (IPTO), the home of the ARPANET. Another familiar name was Licklider's successor, Ivan Southerland, an expert in computer graphics. Southerland's successor was Bob Taylor, the man many consider the father of the Internet.

The list of contributors is not limited to those who worked for ARPA on the ARPANET. Some of the contributing theorists' ideas about linking computers predated the ARPANET by decades. Paul Baran, a RAND researcher, worked on theories of distributed networks and introduced the concept of message block transmission in 1959. During the early sixties, when he approached AT&T with some of his ideas, they resisted, even after the Air Force offered to pay them to get involved. Baran later tabled his networking project idea for fear that the Defense Communications Agency, the group that proposed taking the project on, "would screw it up." Stories such as this are reminders that not every theory or idea is met with support, and that sometimes technological innovation is hindered by the ingrained thoughts of our establishments.

The history of the Internet also provides examples, not that uncommon with technology, where similar ideas were pursued totally independent of each other. In 1965, Donald Watts Davies proposed a "data block" type of transmission method which became known as "packet switching." Although he knew nothing of Baran's work, Davies' research paralleled it. Coincidentally, he even picked the same packet size and data transmission rate. The only difference was that the British Post Office and National Physics Lab embraced his idea. As a result, the term "Packet Switching" was established for this type of transmission although Baran's concept of message blocks predated it by years.

Another related story involves Larry Roberts who was hired from MIT's Lincoln Laboratories by ARPA director Bob Taylor in 1966. Taylor was faced with a serious problem related to the isolated duplication of research efforts. He saw communication through a computer network as an answer to this problem and hired Larry Roberts to do the job. In one of the more interesting twists of fate, Larry Roberts attended a conference presentation by Roger Scantlebury, a representative from Britain's National Physics Lab. He was so excited by the presentation that he stayed up all night talking with Scantlebury. It was through this interaction that Roberts learned, for the first time, of Paul Baran's work at Rand. He immediately sought copies of Baran's work and used them. From that point on, Paul Baran became an informal consultant. This is a perfect example of how the authors have successfully researched the topic to expose interesting stories behind the development of the Internet.

The authors have included far too many significant breakthroughs and events to list in a review, but some of the more important and interesting are worthy of mention. One such development was the idea of Interface Message Processors (IMPs). IMPs are smaller computers designed to be inserted between host computers providing an interface between unlike mainframe computers, sparing the host from routing tasks. Wes Clark, from St. Louis University, came up with the idea while traveling with colleagues to the airport following a meeting. The interesting factor here was that he was known to dislike the idea of time sharing and did not support the idea of hooking computers together. It is insights like these that the authors include that make for such an interesting read.

Now, with key players such as Taylor and Roberts in place and ideas such as packet switching and IMPs conceptualized, it was time for the IPTO at ARPA to find a contractor to make the ARPANET a reality. It was interesting to find out that IBM and Control Data were not that different from AT&T in their views of computer networking. They both said it couldn't be done for the money proposed. Bolt, Beranek and Newman, a small Cambridge consulting firm who's previous work included acoustical work related to analysis of tapes of the JFK assassination and Watergate, was hired to establish the first major wide area computer network in the world.

The next part of the book focuses on work at Bolt, Beranek, and Newman and the challenges they faced as they linked a series of computers, one at a time. This series of computers eventually became the main backbone of the ARPANET. While it was suspenseful to see if this small firm could actually pull off their part in establishing the network under the time frame established for the project, there are times when the reading becomes bogged down, especially if the reader is not interested in the specific technical obstacles that had to be overcome.

Many chapters of the book are filled with anecdotes surrounding key developments. Probably the most significant in terms of breadth of impact was e-mail. A full chapter was devoted exclusively to e-mail developments. First it is important to acknowledge that e-mail predated the Internet. The first e-mail program that was utilized within one system was MAILBOX. It was developed in the early 1960s on the Compatible Time Sharing system at MIT. The first e-mail communication between two machines occurred in 1972 by Ray Tomlinson at the BBN. E-mail on the ARPANET was inevitable but came about with no clear planning or foresight. The significance of its use on the ARPANET was the geographic range.

While the authors did a good job including key e-mail breakthroughs, they also made it clear that e-mail on the ARPANET did not happen overnight. Just as the inventors of computers limited their visions based on their own research needs (Ceruzzi, 1997), the inventors of the Internet had similar limiting thoughts. Besides, the ARPANET was designed for resource sharing, not electronic messages. As ARPANET evolved, users at government installations were figuring out ways to use the system for e-mail and sometimes broke standing operational rules to do so. One of the more interesting stories about early intercontinental e-mail came about because Len Kleinrock, a computer researcher from Los Angeles, forgot his razor at a computer conference in Brighton, England. He remembered a way that the ARPANET could be searched to get a list of everyone who was logged on. He used this to do a "who" search to find that Larry Roberts was still logged on in England. Through a teletype connection to BBN, he was able to link to Roberts using "TALK" which allowed them to type messages back and forth on a split screen. A friend, flying out that day, brought him his razor the following day.

It was just a matter of time before researchers were communicating electronically through their computers. Eventually programmers started writing their own custom e-mail programs to make sending and receiving electronic messages more user friendly. As multiple formats flourished, disagreements erupted regarding the protocols that should be used when sending electronic messages. One of the more notable surviving format nuances was developed by Ray Tomlinson at the BBN, the first to successfully use e-mail between two systems. He used the symbol (@) to separate the name of the user and the machine the user was on. Other innovations followed. In 1975, the first electronic discussion group was formed on the ARPANET. It was called MSG Group and was used to as a mailing list to discuss new ideas.

Around the same time, Bob Kahn and Vint Cerf, who had worked together at UCLA, started to discuss the possibility of creating a "network of networks." Through years of discussions, researchers came to understand that if computers were to be networked, a more independent host-to-host network control protocol would have to be developed. The result was the development of "Transmission Control Protocol" or TCP. By 1978, researchers realized the need to break the protocol down into two parts. It was decided that if the gateway needs to know the information, it went into TCP, if not, the information went into a separate Internet Protocol (IP). Thus, by 1978 TCP/IP became the new transmission protocol.

The last chapter concludes with a discussion of the development of Ethernet for Local Area Networks and the rapid growth of the Internet. Eventually, the network that started it all, ARPANET, was dismantled and no longer serves as the backbone of the Internet. Through this discussion, the authors came full circle.

At one level they have written more than just a history of the Internet; it is a history of technological development from which many parallels can be drawn. The ARPANET is an excellent example of a technology that was created out of a need or desire. Many obstacles were overcome until the technology flourished. But in the end, as with most technologies, there came a time when it outlived its usefulness and was dismantled in favor of new developments.

Where Wizards Stay Up Late is well written. The authors did an excellent job of researching the topic. They relied heavily on primary sources in the form of documents and personal interviews. This was possible and expected because of the recent time-frame in which the origins of the Internet have taken place. It is also organized logically. Key sections such as chapter notes, a bibliography, and index make it a very useful book to have as a reference.

Where Wizards Stay Up Late is a great addition to anyone's library of reference books. It is not suited, nor was it written, to serve as a text. It is simply a discussion of the origins of the Internet. It does not discuss the future or the implications of this frontier technology. It is not suited to serve as a text unless the central topic of the course is the history of the Internet. If you are looking for something to add to a course reading list where you plan to spend one to two days discussing the history of the Internet, I suggest finding something with less breadth such as The Past and Future History of the Internet published in the February 97 issue of Communications of the ACM. (Leiner, et al., 1997).

In conclusion, Hafner and Lyon's book is packed with great information. Many excerpts could be used as a springboard for discussions about historical or sociological perspectives of technology. It is a must read for anyone interested in the origins of the Internet or the process of technological development.


Ceruzzi, P. (1997). An unforeseen revolution: Computers and expectations, 1935-1985. In A. H. Teich (Ed.), Technology and the Future (pp. 117 - 131). New York: St. Martin's Press. (Reprinted from J. J. Corn (Ed.), Imagining tomorrow: History, technology, and the American future, Cambridge, MA: MIT Press, 1986.)

Leiner, B., Cerf, V., Clark, D., Kahn, R., Kleinrock, L., Lynch, D., Postel, J., Roberts, L., & Wolff, S. (1997). The past and future history of the internet. Communications of the ACM, 40(2), 102-108.

Tracy Gilmore