Behind the Net: The untold history of the ARPANET By Michael Hauben hauben@columbia.edu The global Internet's progenitor was the Advanced Research Projects Agency Network (ARPANET) of the U.S. Department of Defense. This is important to remember, because the support and style of management by ARPA to its contractors was crucial to the success of the ARPANET. As the Internet develops and the struggle over the role it plays unfolds, it will be important to remember how the network developed and the culture with which it was connected. As a facilitator of communication, the culture of the Net is an important feature to acknowledge. The ARPANET Completion Report, as published jointly by Bolt, Beranek and Newman (BBN) of Cambridge, Mass., and ARPA concludes by stating: ...it is somewhat fitting to end on the note that the ARPANET program has had a strong and direct feedback into the support and strength of computer science, from which the network itself sprung. (Chapter III, pg.132, Section 2.3.4) In order to understand the wonder that the Internet, and various parts of the Net, represent, we need to understand why the ARPANET Completion report ends with the suggestion that the ARPANET is fundamentally connected to and born of computer science, rather than of the military. PART I: The history of ARPA leading up to the ARPANET A climate of pure research surrounded the entire history of the ARPANET. The Advanced Research Projects Agency was formed to fund research, and thus was not oriented to a military product. The formation of this agency was part of the U.S. reaction to the then Soviet Union's launch of Sputnik in 1957. (ARPA draft, III-6). ARPA was assigned to research how to utilize the mili- tary's investment in computers via Command and Control Research (CCR). Dr. J.C.R. Licklider was chosen to head this effort. Licklider came to ARPA from Bolt, Beranek and Newman, (BBN) in Cambridge, MA in October 1962. (ARPA draft, III-6) He came to ARPA from a background of combining engineering studies and physiological psychology. This provided Licklider with an unusual perspective uncommon among engineers. From Licklider's arrival, the department's contracts were shifted from independent corporations towards "the best academic computer centers" (ARPA draft, III-7). The then current method of computing was via batch processing (i.e., input via stacks of punched cards, and output the results, or lack of them, made known one or more days later.). Licklider saw improvements could be made in CCR only from work on advancing the current state of computing technology. He particularly wanted to move forward into the age of interactive computing, and the current contractors were not moving in that direction. In an Interview, Licklider told the interviewee that SDC (System Development Corporation) "was based on batch processing, and while I was interested in a new way of doing things, they [SDC] were studying how to make improvements in the ways things were done already." (An Interview with J.C.R. Licklider conducted by William Aspray and Arthur Norberg on October 28, 1988 Cambridge, Mass. CBI Univ of Minn., Madison) The office "developed into a far-reaching basic research program in advanced technology." (ARPA draft III-7) Licklider's Office was renamed Information Processing Techniques Office (IPT or IPTO) to reflect that change. The Completion report states that "Prophetically, Licklider nicknamed the group of computer specialists he gathered the 'Intergalactic Network'." (ARPA draft, III-7) Before work on the ARPANET began, the very idea of the network was planted by the creation of the Information Processing Techniques Office of ARPA. Robert Taylor, Licklider's successor at the IPTO, remembers why this was true because of Lick's interest in interconnecting communities: Lick was among the first to perceive the spirit of community created among the users of the first time-sharing systems... In pointing out the community phenomena created, in part, by the sharing of resources in one timesharing system, Lick made it easy to think about interconnecting the communities, the interconnection of interactive, on-line communities of people, ..." (ARPA draft, III-21) The "spirit of community" was related to Lick's interest in having computers help people communicate with other people (Licklider and Robert Taylor, "The Computer as a Communication Device") Licklider's vision of an "intergalactic network" con- necting people represented an important conceptional shift in computer science. This vision was also an important beginning to the ARPANET. After the ARPANET was up and running, the computer scientists using it realized that assisting human communication was a major fundamental advance that the ARPANET made possible. As early as 1963, a commonly asked question of the IPTO directors by the ARPA directors about IPTO projects was "Why don't we rely on the computer industry to do that?", or occasion- ally more strongly, "We should not support that effort because ABC (read, "computer industry") will do it - if it's worth doing!" (ARPA draft, III-23) This question leads to an important point - this ARPA research was different from what the computer industry had in mind to do - or was likely to undertake. Since Licklider's creation of the IPTO, the work supported by ARPA/IPTO continued his explicit emphasis on communications. The Completion Report explains, The ARPA theme is that the promise offered by the computer as a communication medium between people, dwarfs into rela- tive insignificance the historical beginnings of the comput- er as an arithmetic engine." (ARPA draft, III-24) The Completion Report goes on to differentiate ARPA from the computer industry: The computer industry, in the main, still thinks of the computer as an arithmetic engine. Their heritage is reflect- ed even in current designs of their communication systems. They have an economic and psychological commitment to the arithmetic engine model, and it can die only slowly..." (ARPA draft, III-24) The Completion Report further analyzes this problem by tracing it back to the nation's universities: ...furthermore, it is a view that is still reinforced by most of the nation's computer science programs. Even univer- sities, or at least parts of them, are held in the grasp of the arithmetic engine concept.... (ARPA draft, III-24) ARPA's IPTO was responsible for the research and development which led to the success of first the ARPANET, and later the Internet. Without the commitment that existed via this support, such a development might never have happened. One of ARPA's criterion for supporting research was that the research had to be of such a level as to offer an order of magnitude of advance over the current state of development. As most research and develop- ment is not immediately profitable, there is a need for organiza- tions which do not pursue profit as their goal, but rather work on furthering the state of the art. What is very telling is that computer networking spread widely without profit being involved. Others have understood the communications promise of comput- ers. For example, in RFC 1336, David Clark, senior research scientist at MIT's laboratory for computer science, is quoted, It is not proper to think of networks as connecting comput- ers. Rather, they connect people using computers to mediate. The great success of the internet is not technical, but in human impact. Electronic mail may not be a wonderful advance in Computer Science, but it is a whole new way for people to communicate. The continued growth of the Internet is a technical challenge to all of us, but we must never loose sight of where we came from, the great change we have worked on the larger computer community, and the great potential we have for future change. Various research predating the ARPANET had been done by Paul Baran, Thomas Marill and others. [End note 1] This led Lawrence Roberts and other IPTO staff to formally introduce the topic of networking computers of differing types (i.e.: incompatible hardware and software) together in order to share resources to the early 1967 meeting of ARPA's Principal Investigators (PI). In the spring of 1967 at the University of Michigan, ARPA held its yearly meeting of the Principal Investigators from each of its university and other contractors. (ARPA draft, III-25) Results from the previous year's research were summarized and future research were discussed, either introduced by ARPA or the various researchers present at the meeting. Networking was one of the topics brought up at this meeting. (ARPA draft, III-25) At that meeting, it was decided that there had to be agreement on conventions for character and block transmission, error checking and retransmission, and computer and user identification. These specifications became the contents of the inter-host communication's "protocol." Frank Westervelt was chosen to write about this protocol and a communication group was formed to study the questions. (ARPA draft, III-26) In order to develop a network of varied computers, two main problems had to be solved: 1. To construct a 'subnetwork' consisting of telephone circuits and switching nodes whose reliability, delay char- acteristics, capacity, and cost would facilitate resource sharing among computers on the network. 2. To understand, design, and implement the protocols and procedures within the operating systems of each connected computer, in order to allow the use of the new subnetwork by the computers in sharing resources. (ARPA, II-8) After one draft and additional work on this communications position paper was completed, a meeting was scheduled in early October 1967 by ARPA at which the protocol paper and specifica- tions for the Interface Message Processor (IMP) were discussed. A subnetwork of IMPs, dedicated mini-computers connected to each of the participant computers, was the method chosen to connect the participants's computers (hosts) to each other via phone lines. This standardized the subnet to which the hosts connected. Now, only the connection of the hosts to the network would depend on vendor type, etc. ARPA had picked 19 possible participants in what was now known as the "ARPA Network." From the time of the 1967 PI Meeting, various computer scientists who were ARPA contractors were busy thinking about various aspects which would be relevant to the planning and development of the ARPANET. Part of that work was a document outlining a beginning design for the IMP subnetwork. This speci- fication led to a competitive procurement for the design of the IMP subnetwork. By late 1967 ARPA had given a contract to the Stanford Research Institute (SRI) to write the specifications for the communications network they were developing. In December of 1968, SRI issued a report "A Study of Computer Network Design Parame- ters." Elmer Shapiro played an important role in the research for this report. Based on this work, Roberts and Barry Wessler of ARPA wrote the final ARPA version of the IMP specification. (ARPA draft, III-32) This specification was ready to be discussed at the June 1968 PI meeting. The Program Plan "Resource Sharing Computer Networks" was submitted June 3, 1968 by the IPTO to the ARPA Director, who approved it on June 21, 1968. It outlined the objectives of the research, and the plan of how the objectives would be fulfilled. The purposed network was impressive as it would prove useful to both the computing research centers which connected to the network and the military. The proposed requirements for the research would provide immediate benefits to the computer centers the network would connect. (ARPA draft, III-35) ARPA's stated objectives were to experiment with varied interconnections of computers and sharing resources in an attempt to improve produc- tivity of computer research. (ARPA, II-2) Justification was drawn from technical needs in both the scientific and military environ- ments. The Program Plan developed into a set of specifications. These specifications were connected to a competitive Request for Quotation (RFQ) to find an organization which would design and build the IMP subnetwork. Following the approval of the Program Plan, 140 potential bidders were mailed the Request for Quotation. After a bidders conference, 12 proposals were received and from them ARPA nar- rowed the bidders down to four. BBN was the eventual recipient of the contract. (ARPA draft, III-35) The second technical problem, as defined by the ad hoc Communications Group, still remained to be solved. The set of agreed upon communications settings (known as a protocol), which would allow the hosts to communicate with each other over the subnetwork, had to be developed. This work was left "for host sites to work out among themselves." (ARPA draft, III-67) This meant that the software necessary to connect the hosts to the IMP subnetwork had to be developed. ARPA assigned this duty to the initially designated ARPANET sites. Each of the first sites had a different type of computer to connect. ARPA trusted/knew the programmers at each site would be capable of modifying their operating systems in order to connect their systems to the subnetwork. In addition the sites needed to develop the software necessary to utilize the other hosts on the network. (ARPA draft, III-39) ARPA's assigning of responsibilities made the academic computer science community an active part of the ARPANET develop- ment team. (Interview with Alex McKenize, Nov, 1 1993) Steve Crocker, one of graduate students involved with the development of the earliest ARPANET protocols, associates the placement of the initial ARPANET sites at research institutions to the fact that the ARPANET was ground-breaking research. He wrote in a message responding to my questions on the COM-PRIV mailing list: During the initial development of the Arpanet, there was simply a limit as to how far ahead anyone could see and manage. The IMPs were placed in cooperative ARPA R&D sites with the hope that these research sites would figure out how to exploit this new communication medium. (Crocker, 1993A) The first sites of the ARPANET were picked to provide either network support services or unique resources. The key services the first four sites provided were UCLA - Network Measurement Center SRI - Network Information Center UCSB - Culler-Fried interactive mathematics UTAH - graphics (hidden line removal) (Cerf, Vinton 1993) Steve Crocker also recounts that the reason for selecting these particular four sites was because they were "existing ARPA computer science research contractors." This was important because "the research community could be counted on to take some initiative." (RFC 1000, pg 1) The very first site to receive an IMP was UCLA. Professor Leonard Kleinrock of UCLA was involved with much of the early development of the ARPANET. His work in queuing theory gave him a basis to develop measurement techniques used to monitor the ARPANET's performance. This made it natural to make sure that UCLA received one of the first nodes as it would be important to measure the network's activity from early on. In order for the statistics to have correct data and analysis purposes - one of the first two or three sites had to be the measurement site. Sure enough UCLA was assigned to be the Network Measurement Center (NMC). [END NOTE 2] Part II. The Network Working Group Once the initial sites were picked, representatives from each site gathered together to start talking about solving the technical problem of getting the hosts to communicate with each other. The ARPA Completion report tells us about this beginning: To provide the hosts with a little impetus to work on the host-to-host problems. ARPA assigned Elmer Shapiro of SRI "to make something happen", a typically vague ARPA assign- ment. Shapiro called a meeting in the summer of 1968 which was attended by programmers from several of the first hosts to be connected to the network. Individuals who were present have said that it was clear from the meeting at that time, no one had even any clear notions of what the fundamental host-to-host issues might be. (AC Draft III-67 1.4.1.7) We see that this group, which came to be known as the Network Working Group (NWG), was exploring new territory. The first meeting took place several months before the first IMP was put together and they had to think from a blank slate. Throughout the existing recollections of the important developments the NWG produced, (especially RFC 1000) the reader is reminded that the thinking involved was groundbreaking and thus exciting. Steve Crocker remembers in the RFC Reference Guide (RFC 1000) that the first meeting was chaired by Elmer Shapiro, who initiated the conversation with a list of questions. (Crocker, 1993b) Also present were Steve Carr from University of Utah, Stephen Crocker from UCLA, Jeff Rulifson from SRI, and Ron Stoughton from UCSB. These attendees are the programmers referred to in the ARPANET Completion Report. According to Steve Crocker, this was a seminal meeting. The attendees could only be but theoretical, as none of the lowest levels of communication had been developed yet. They needed a transport layer or low-level communications platform to be able to build upon. BBN would not deliver the first IMP until August 30, 1969. It was important to meet before this date, as the NWG "imagined all sorts of possibilities." (Rfc1000) Only once their thought processes started could this working group actually develop anything. These fresh thoughts from fresh minds helped to incubate new ideas. The ARPANET Completion Report properly acknowledges what this early group helped accomplished: "Their early thinking was at a very high level." (ARPA draft, III-67) A concrete decision of the first meeting was to continue holding meetings similar to the first one. This wound up setting the precedent of holding exchange meetings at each of the sites. Steve Crocker, describing the problems facing these network- ing pioneers, writes: With no specific service definition in place for what the IMPs were providing to the hosts, there wasn't any clear idea of what work the hosts had to do. Only later did we articulate the notion of building a layered set of protocols with general transport services on the bottom and multiple application-specific protocols on the top. More precisely, we understood quite early that we wanted quite a bit of generality, but we didn't have a clear idea how to achieve it. We struggled between a grand design and getting some- thing working quickly. (Crocker,1993c) The initial protocol development lead to DEL (Decode- Encode-Language) and NIL (Network Interchange Language). These languages were more advanced than what was needed or possible at the time. The basic purpose was to form an on-the-fly description that would tell the receiving end how to understand the informa- tion that would be sent. These first set of meetings were ex- tremely abstract as neither ARPA nor the universities had deemed any official charter. However, the lack of a specific charter allowed the group to think broadly and openly. BBN did submit details about the host-IMP interface specifi- cations from the IMP side. This information provided the group some definite starting points to build from. Soon after BBN provided more information, on Valentine's Day, 1969, members of the NWG, members of BBN and members of the Network Analysis Corporation (NAC) met for the first time. [Endnote 3] As all the parties had different priorities on mind, the meeting was a difficult one. BBN was interested in the lowest level of making a reliable connection. The programmers from the host sites were interested in getting the hosts to communicate with each either via various higher level programs. And BBN also did not turn out to be the "experts from the East" that Steve Crocker wrote the members of the NWG expected. He continues by writing in RFC 1000 that they constantly thought that "a professional crew would show up eventually to take over the problems we were dealing with." A step of incredible importance and openness occurred as a result from a "particularly delightful" meeting that took place a month later in Utah. (RFC1000) The participants decided it was time to start recording their meetings in a consistent fashion. What resulted was a set of informal notes titled "Request for Comments." Steve Crocker writes about their formation: I remember having great fear that we would offend whomever the official protocol designers were, and I spent a sleep- less night composing humble words for our notes. The basic ground rules were that anyone could say anything and that nothing was official. And to emphasize the point, I labeled the notes "Request for Comments." I never dreamed these notes would distributed through the very medium we were discussing in these notes. Talk about Sorcerer's Apprentice! (Crocker, RFC 1000, pg 3, 1987) Crocker replaced Shapiro as the Chairman of the NWG soon after the initial meeting. He describes how they wrestled with the creation of the host-host protocols: Over the spring and summer of 1969 we grappled with the detailed problems of protocol design. Although we had a vision of the vast potential for intercomputer communica- tion, designing usable protocols was another matter. A custom hardware interface and custom intrusion into the operating system was going to be required for anything we designed, and we anticipated serious difficulty at each of the sites. We looked for existing abstractions to use. It would have been convenient if we could have made the network simply look like a tape drive to each host, but we knew that wouldn't do. (Crocker, RFC 1000, pg. 3) The first IMP was delivered to UCLA in late August, 1969. The next was delivered to SRI a month later in October. [Endnote 4] Once more than one IMP existed, the NWG had to implement a working communications protocol. This first set of pairwise host protocols included remote login for interactive use (telnet), and a way to copy files between remote hosts (FTP). Crocker writes: In particular, only asymmetric, user-server relationships were supported. In December 1969, we met with Larry Roberts in Utah, [and he] made it abundantly clear that our first step was not big enough, and we went back to the drawing board. Over the next few months we designed a symmetric host-host protocol, and we defined an abstract implementa- tion of the protocol known as the Network Control Program. ("NCP" later came to be used as the name for the protocol, but it originally meant the program within the operating system that managed connections. The protocol itself was known blandly only as the host-host protocol.) Along with the basic host-host protocol, we also envisioned a hierarchy of protocols, with Telnet, FTP and some splinter protocols as the first examples. If we had only consulted the ancient mystics, we would have seen immediately that seven layers were required. (RFC 1000, pg 4) After Robert's guidance, the Network Working Group went forward in developing the protocols necessary to make the network viable. The group swelled in attendance as more and more sites connected to the ARPANET. The group became large enough (around 100 people) that one meeting was held in conjunction with the 1971 Spring Joint Computer Conference in Atlantic City. A major test of the NWG's work came in October 1971, when a meeting was held at MIT. Crocker continues the story, [A] major protocol "fly-off" - Representatives from each site were on hand, and everyone tried to log in to everyone else's site. With the exception of one site that was com- pletely down, the matrix was almost completely filled in, and we had reached a major milestone in connectivity. (Crocker, RFC 1000, pg. 4) The NCP was created as what was called the "host to host protocol." Explaining why this was important, the authors of the ARPA draft write: The problem is to design a host protocol which is suffi- ciently powerful for the kinds of communication that will occur and yet can be implemented in all of the various different host computer systems. The initial approach taken involved an entity called a "Network Control Program" which would typically reside in the executive of a host, such that processes within a host would communicate with the network through this Network Control Program. The primary function of the NCP is to establish connections, break connections, switch connections, and control flow. A layered approach was taken such that more complex procedures (such as File Trans- fer Procedures) were built on top of similar procedures in the host Network Control Program. (Arpa draft, II-24) As the ARPANET grew, the number of users bypassed the number of developers. This signaled the success of these networking pioneers. Steve Crocker appointed Alex McKenize and Jon Postel to replace him as Chairmen of the Network Working Group. The Comple- tion Report details how this role changed: McKenzie and Postel interpreted their task to be one of codification and coordination primarily, and after a few more spurts of activity the protocol definition process settled for the most part into a status of a maintenance effort.(ARPA draft,III-69) ARPA was a management body which funded academic computer scientists. ARPA's funding paved the way for these scientists to create the ARPANET. BBN helped via developing the packet switch- ing techniques which served as the bottom level of transmitting information between sites. The NWG provided an important develop- ment in its "Request for Comments" documentation which made possible the developing the new protocols. PART III. About RFC's as "Open" Documentation The openness initiated from the very first meeting of the Network Working Group continued on in a more informal formalized manner in the Request For Comments. As meeting notes, the RFCs were meant to keep members updated on the status of various developments and ideas by the development community. They were also meant to gather responses from people. The Documentation Conventions RFC (RFC 3) documents the "rules" governing the production of these notes. Heading the page were the open distri- bution rules: Documentation of the NWG's effort is through notes such as this. Notes may be produced at any site by anybody and included in this series. These opening sentences invite anyone willing to be helpful in the protocol definition process. This is important because all restrictions are denied by these words, allowing for the best possible developments. The guide goes on to describe the rules concerning the contents of the RFCs: The content of a NWG note may be any thought, suggestion, etc. related to the HOST software or other aspect of the network. Notes are encouraged to be timely rather than polished. Philosophical positions without examples or other specifics, specific suggestions or implementation techniques without introductory or background explication, and explicit questions without any attempted answers are all acceptable. The minimum length for a NWG note is one sentence. The RFC continues to explain the philosophy behind the perhaps unprecedented amount of openness represented: These standards (or lack of them) are stated explicitly for two reasons. First, there is a tendency to view a written statement as ipso facto authoritative, and we hope to pro- mote the exchange and discussion of considerably less than authoritative ideas. Second, there is a natural hesitancy to publish something unpolished, and we hope to ease this inhibition." (Crocker, RFC 3 - 1969) [The entire RFC is reproduced in Appendix B.] This openness led to the exchange of information. Technical development is only successful when information is allowed to flow freely and easily between the parties involved. These open principles are what made the development of the Net possible. Statements like the ones contained in RFC 3 are very pro- gressive in their openness. The late 1960's was a time alive in popular protest for freedom of speech and people demanding more of a say in how their country was run. The openness applied in trying to develop new technologies fits well with the cry for more democracy which students demanded throughout the country and the world. What is amazing is that the collaboration of the NWG (mostly graduate students) and ARPA (a component of the mili- tary), seems to be contrary to the normal atmosphere of the times. Robert Braden of the Internet Activities Board reflects on this collaboration: For me, participation in the development of the ARPAnet and the Internet protocols has been very exciting. One important reason it worked, I believe, is that there were a lot of very bright people all working more or less in the same direction, led by some very wise people in the funding agency. The result was to create a community of network researchers who believed strongly that collaboration is more powerful than competition among researchers. I don't think any other model would have gotten us where we are today. (RFC 1336) These ideas point to a reason why the work of these computer scientists founded what has led to be one of the most amazing and democratic bodies (i.e.: The Net and the culture attached to it) to emerge in a long time. The community that has developed and the tools which accompany it form an important democratic force. [See endnote 5.] The idea of calling these notes a "Request for Comment" set a fascinating tradition. It predates the Usenet Post, which in a fashion could be called a "request for comment" as it is the presentation of a particular person's ideas, questions or com- ments, to the general public (of those who read that newsgroup) for comments, criticism or suggestion, or just plain to further the readers' knowledge. Other Early RFCs echo this reality. There are plenty of RFCs which are in response to a previous RFC. Following are some examples, more are contained in the appendix. 1 Crocker, S. Host software 1969 April 7 65 Walden, D. Comments on Host/Host Protocol document #1 36 Crocker, S. Protocol notes 1970 March 16 38 Wolfe, S. Comments on network protocol from NWG/RFC #36 39 Harslem, E.; Heafner, J. Comments on protocol re: NWG/RFC#36 33 Crocker, S. New Host-Host Protocol 1970 February 12 47 Crowther, W. BBN's comments on NWG/RFC #33 1970 April 20 Part IV: Conclusion How were the developments of the ARPANET made possible? This question appears from the very problems that the various contri- butors to the ARPA project faced themselves. None of the partici- pants had the solutions to any of the tasks they approached before putting much thought and work into their research. As the resulting ARPANET was tremendously successful and fulfilled the project ARPA presented, it is important to see what can be learned from the research out of which it emerged. Bernie Cosell, who worked at BBN during this early period, describes the impor- tance of openness in a developmental situation: *no*one* had the necessary expertise [and vision] to figure any of this out on their own. The cultures among the early groups were VERY different multics, sigma-7, IBM ... at Rand, ... PDP-10s at BBN and SRI... [and possibly] UCSB and Utah had pdp-10's, too. The pie-in-the-sky applications ranged over a WIDE landscape, with no one knowing quite where it would lead. Some kind of free, cross-cultural info/idea exchange *had* to happen. (Cosell 1993) The computer scientists and others involved were encouraged in their work by ARPA's philosophy of gathering the best computer scientists working in the field and supporting them: IPT usually does little day-to-day management of its con- tractors. Especially with its research contracts, IPT would not be producing faster results with such management as research must progress at its own pace. IPT has generally adopted a mode of management which entails finding highly motivated, highly skilled contractors, giving them a task, and allowing them to proceed by themselves. (ARPA draft, III-47) The work of the Network Working Group was vital to the development of the ARPANET. Vint Cerf, another of the graduate students involved with the early protocol development and still closely connected to the Internet, echoed this sentiment when he opened his paper "An Assessment of ARPANET Protocols," by writ- ing: The history of the Advanced Research Project Agency resource sharing computer network (ARPANET) is in many ways a history of the study, development, and implementation of protocols." (Cerf, _An Assessment of ARPANET Protocols_) Cerf supports Cosell's opinion about the uncertainty and newness of the entire project when he continues in his paper by writing: The tasks facing the ARPANET design teams were often un- clear, and frequently required agreements which had never been contemplated before (e.g., common protocols to permit different operating systems and hardware to communicate). The success of the effort, seen in retrospect, is astonish- ing, and much credit is due to those who were willing to commit themselves to the job of putting the ARPANET togeth- er. (Cerf, IBID.) The NWG's work blazed the trail which the developers of the TCP/IP suite of protocols (Transport Control Protocol/ Internet Protocol) followed to success when the need to expand and include other networks based on other technologies than NCP arose. The principles embodied by RFC 3 and open RFC documentation provided a strong foundation which began with NCP and was continued by the work on TCP/IP. NCP was developed in the field and versions of it were released early in its development so various programmers could work on implementing and improving the protocol. In addi- tion all specifications were available for free and easily available for people to examine and comment on. Through this principle of early release the problems and kinks were found and worked out in a timely manner. The future developers of TCP/IP learned from the developers of NCP a practice of developing from the bottom up. The bottom-up model allows for a wide-range of people and experiences to join in and perfect the protocol and make it the best possible. The public funding of the ARPANET project allowed its documentation to be open and available. This documentation was neither restricted nor classified. The possibility of communi- cation represented by openness was necessary for these pioneers. Research of new fields of study require that researchers cooper- ate and communicate in order to share their expertise with the larger body of people conducting research. This openness is especially critical when no one person has the answers in ad- vance. Larry Roberts of ARPA explained in an article: "Since the ARPANET was a public project connecting many major universities and research institutions, the implemen- tation and performance details were widely published." ("The Evolution of Packet Switching", 267) The people at the forefront of development of these proto- cols were the members of the Network Working Group, many of whom came from academic institutions, and who therefore had the support and time needed for the research. In summing up the achievements of the process that developed the ARPANET, the ARPANET Completion Report draft explains: The ARPANET development was an extremely intense activity in which contributions were made by many of the best computer scientists in the United States. Thus, almost all of the "major technical problems" already mentioned received con- tinuing attention and the detailed approach to those prob- lems changed several times during the early years of the ARPANET effort. [II-24] Fundamental to the ARPANET, as explained by the Completion Report, was the discovery of a new way of looking at computers. The developers of the ARPANET viewed the computer as a communica- tions device rather than only as an arithmetic device. (draft, III-24) This new view made the building of the ARPANET possible. This view came from the research conducted by those in academic computer science. The shift in the understanding of the role of the computer is fundamental to advancing computer science. The ARPANET research has provided a rich legacy for the further advancement of computer science and it is important that the significant lessons learned be studied and used to further advance the study of computer science. END - NOTES 1. This history is covered well in the article "From ARPANET to USENET" by Ronda Hauben. Also in Chapter III, section 1.1.2 starting on page III-9 in the published ARPANET Completion Report. 2. These quotes show some of the perspective chosen to pick the initial ARPANET sites. III - 689 "CCN's [The Campus Computing Network of UCLA] chance to obtain a connection to the ARPANET was a result of the presence at UCLA of Professor L. Kleinrock and his students, including S. Crocker, J. Postel, and V. Cerf. This group was not only involved in the original design of the network and the Host protocols, but also was to operate the Network Measurement Center (NMC). For these reasons the first delivered IMP was installed at UCLA, and ARPA was thus able to easily offer CCN the opportunity for connection." pg II-16 " In a somewhat less structured way, the research groups receiving ARPA IPTO support were then encouraged to begin consid- ering the design and implementation of protocols and procedures and, in turn, computer program modifications, in the various host computers in order to use the subnetwork. Several specific responsibilities were arranged: UCLA was specifically asked to take on the task of a "Network Measurement Center" with the objective of studying the performance of the network as it was built, grown, and modified; SRI was specifically asked to take on the task of a "Network Information Center" with the objective of collecting information about the network, about host resources, and at the same time generating computer based tools for storing and accessing that collected information. Beyond these two specific contracts, some rather ad hoc mechanisms were pursued to reach agreement between the various research contractors about the appropriate "host protocols" for intercommunicating over the subnetwork. The "Network Working Group" of interested individuals from the various host sites was rather informally encouraged by ARPA. After a time, this Network Working Group became the forum for, and eventually a semi-official approval authority for, the discussion of and " III - 60 1.4.1.5 The Network Information Center The accessibility of distributed resources carries with it the need for an information service (either centralized or distribut- ed) that enables users to learn about those resources. This was recognized at the PI [ed. Primary Instigators] meeting in Michi- gan in the spring of 1967. At the time, Doug Engelbart and his group at the Stanford Research Institute were already involved in research and development to provide a computer-based facility to augment human interaction. Thus, it was decided that Stanford Research Institute would be a suitable place for a "Network Information Center" (NIC) to be established for the ARPANET. With the beginning of implementation of the network in 1969, construc- tion also began on the NIC at SRI." 3. The NAC was contracted by ARPA to "specify the topological design of the ARPANET and to analyze its cost, performance, and reliability characteristics. (ARPA, III-30) 4. RFC 1000 reports on the process of the installation of the first IMP. "[T]ime was pressing: The first IMP was due to be delivered to UCLA September 1, 1969, and the rest were scheduled at monthly intervals. At UCLA we scrambled to build a host-IMP interface. SDS, the builder of the Sigma 7, wanted many months and many dollars to do the job. Mike Wingfield, another grad student at UCLA, stepped in and offered to get interface built in six weeks for a few thousand dollars. He had a gorgeous, fully instrumented interface working in five and one half weeks. I was in charge of the software, and we were naturally running a bit late. September 1 was Labor Day, so I knew I had a couple of extra days to debug the software. Moreover, I had heard BBN was having some timing troubles with the software, so I had some hope they'd miss the ship date. And I figured that first some Honeywell people would install the hardware -- IMPs were built out of Honeywell 516s in those days -- and then BBN people would come in a few days later to shake down the software. An easy couple of weeks of grace. BBN fixed their timing trouble, air shipped the IMP, and it arrived on our loading dock on Saturday, August 30. They arrived with the IMP, wheeled it into our computer room, plugged it in and the software restarted from where it had been when the plug was pulled in Cambridge. Still Saturday, August 30. Panic time at UCLA. The second IMP was delivered to SRI at the beginning of October, and ARPA's interest was intense. Larry Roberts and Barry Wessler came by for a visit on November 21, and we actually managed to demonstrate a Telnet-like connection to SRI." 5. This democratic community is in danger of being fundamentally altered. This study of the history of the development of the ARPANET in conjunction with my paper, "The Social Forces Behind the Development of Usenet News" are meant to help people under- stand where the Net has come from, in order to defend it, and try to fight to keep it open and democratic - the seventh wonder of the world as a recent ad called the Internet, misdirected as it was - but correct any way. I hope to make this analysis available in RFC form as a comment on RFC 1000. Bibliography Special Thanks to Alexander McKenizie of BBN, Stephen Crocker of TIS, and Vinton Cerf of CNRI for making research materials available. ARPANET COMPLETION REPORT DRAFT , September 9, 1977, unpublished. Cerf, Vinton G., private corespondence, dated Nov 27, 1993. Subject: "Re: Early Days of the ARPANET and the NWG" Cerf, Vinton G., "An Assessment of ARPANET Protocols." Infotech Education Ltd. Stanford University, California, 21 pages Cosell, Bernie "Re: RFC1000 - Questions about the origins of ARPANET Protocols 2/2" Article: 54310 of alt.folklore.computers, Nov. 23, 193. Crocker, Stephen D., 1993A email message to Com-Priv mailing list (com-priv@psi.com) Subject "Re: RFC1000 (Partial response to part 1)" Date: Nov 27, 1993. Crocker, Stephen D., 1993B email message to Com-Priv mailing list Subject: "Re: RFC1000 (End of response to part 1)" Date: Nov 27, 1993. Crocker, Stephen D., 1993C email message to Com-Priv mailing list Subject "Subject: Re: RFC1000 (Response to part 2)" Date: Nov 27, 1993. Crocker, Stephen D., RFC 3, DOCUMENTATION CONVENTIONS. Crocker, Stephen D., RFC 1000, RFC Reference Guide. Heart, F, McKenzie, A., McQuillan, J., Walden, D., ARPANET Completion Report, Washington, 1978. Licklider, J.C.R., Interview conducted by William Aspray and Arthur Norberg on October 28, 1988 Cambridge, Mass. CBI Univ of Minn., Madison. Licklider, J.C.R. and Robert Taylor, "The Computer as a Communi- cation Device" from "In Memoriam: J.C.R. Licklider 1915-1990," Aug. 7, 1990, p. 40; reprinted by permission from Digital Research Center; originally published as "The Computer as a Communication Device," in "Science and Tech- nology", April, 1968, pg. 40 Mckenzie, Alexander, Interview with Nov 1, 1993. Roberts, Lawrence Member IEEE, Invited Paper, "The Evolution of Packet Switching", Proceedings of the IEEE Volume 66, Number 11, November 1978, pages 1307 - 1313