ARPANET, The Grandfather of the Internet

What does the threat of nuclear war, a secret government agency, the United States’ top universities, and an eccentric computer scientist who dreamed of melding man and machine together have in common? They all helped create the modern internet. 

The Need

One of the biggest drivers behind the creation of the internet was the United States military. From the late 1950s up through the 80s the threat of nuclear war was ever present, but one thing that many may not think about being lost in such an event is access to information. At the time the only way for computers to share data with one another was through the use of a central core, and if this core was destroyed in an attack then the entire communication system between computers would be taken down with it. The military wanted a system that was not reliant on this central core. 

There was also the academic side of it. Scientists and researchers needed an easier way to share information with their colleagues who were at a different university or research lab. Charles M. Herzfeld, former ARPA director, put it best by saying that it “came out of our frustration that there were only a limited number of large, powerful research computers in the country and that many research investigators who should have access were geographically separated from them.” The Advanced Research Projects Agency (ARPA), a branch of the Department of Defense, was tasked with solving this problem. 

Computer room in the 60s.

Development 

The very root of the soon to be computer network, which would be named ARPAnet, can be found in the U.S. military’s Semi Automatic Ground Environment (SAGE) system. SAGE was a network of 26 supercomputers linked together through manned switch stations that brought together data from different radar sites, mainly to track airplanes and coordinate a military response if needed. This network took 6 years and cost $61 billion to develop and implement, and it was just the sort of groundwork ARPA needed to get the national computer network on its feet. 

An officer monitoring a SAGE terminal.

In 1962 Joseph Licklider, a computer scientist, joined ARPA as director of the Information Processing Technique Office (IPTO). Licklider was highly intelligent, though a bit eccentric for the time. He fully believed that computers were the way of the future, and in one of his more famous essays said that one day humans would have computer implants that would improve their decision making. He worked to shift ARPA from working on purely military projects to allowing workers and researchers to develop and test just about any new idea they had. While he only worked at ARPA for two years his shifting of its focus helped the agency later branch out and develop many new softwares and computing techniques. 

Joseph Licklider. Courtesy of MIT.

The idea for ARPAnet came from the next director Ivan Sutherland who joined in 1964. In the IPTO office there were three teletype terminals, and each was hooked up to an ARPA time sharing mainframe. This meant that he could see when people were using the computers in other parts of the country, and watched how they connected and interacted with one another through the time-shared computers. It made him think that it was silly to require three different mainframes just to connect three computer systems because they were not compatible with one another. This gave him the idea to create one computer language protocol that all could use and connect with. He came forward with his idea and was able to secure funding for it in 1966, and the idea was presented to the public a year later. In 1968 the Department of Defence put out a competitive bid to build a network that would link together 16 ARPA sponsored universities and research centers. January of 1969 saw a $1 million contract awarded to Bold, Beranek, and Newman (BBN) to create the network. 

Ivan Sutherland. Public Domain.

The plan was to connect these computers through the already expansive telephone system, but there was a unique challenge. They did not want the system to be dependent on every single node (a node is a physical device within a network, today the personal computer is the most common form of node) because if one node went down the entire system would go down, which was very similar to the problem with using a central core. The solution came from Paul Baran, a researcher from the RAND Corporation, who came up with a way to route data transfers through different phone lines. He called the process “hot-potato” routing, though it would later be renamed to packet switching. Whenever the digital information was transferred it would be broken down into smaller “packets” so that it could move through the network faster. These packets would then be sent more or less at random through the telephone lines. Instead of them all taking the same path they would take the path of least resistance so that the information would not become jammed up on a phone line that was down or already in use. To ensure they did not become lost in the tangled network they had mathematical verifications that would go through the network of computerized switch stations that forwarded these packages to their destination. Each packet would have an identifier so that when they reached their destination the modem would know what order to reassemble them in. Baran had actually come up with this technique in 1964 but at the time he was largely ignored by the military. However ARPA was more receptive to his idea and based their entire network off of it. 

Diagram showing packet switching. Courtesy of Network Encyclopedia.

At first there were only 4 computers connected, they were from the computer research labs at UCLA, Stanford Research Institute, University of California, Santa Barbara, and the University of Utah. The first network connection and exchange happened in 1969 between computers at UCLA and Stanford. When they first tried to log into the Stanford computers from UCLA they tried to type in “log win” but the system crashed when they typed the “g”. Obviously there were a few bugs that needed to be worked out. With some fine tuning they were able to link the four computers together, but  unlike today these computers were limited to only three functions, logging in to a remote computer, printing to a remote printer, and file transfers. Still this was much better than the previous system and it was not long before other universities and research labs wanted to connect to the ARPAnet. 

Expansion and Improvements

BBN would remain busy for the next several years. By April of 1971 there were 15 nodes and 23 host terminals, by 1973 35 nodes, and by 1976 63. With increased connectivity and the change of focus thanks to Licklider, ARPA scientists ran countless tests and made many breakthroughs in computer technology, many of which we still use today. Some of these breakthroughs include e-mail, telnet (which allowed a user to control a computer in a different location from his computer), File Transfer Protocol (FTP,  which allows a user to transfer data in one large bulk package), computer graphics, parallel processing, and even computer flight simulations. Unfortunately the system began to quickly deteriorate as more and more computers of different models were connected and created compatibility issues. ARPAnet also struggled with connecting to other popular networks like SATNET (satellite networking) and ALOHANET (a packet radio system that was based in Hawaii). The solution lay in the development of an improved set of protocols. 

ARPAnet 1973.Courtesy of the Daily Mail.

In 1977 they tested a new protocol called Transmission Control Protocol/Internet Protocol (TCP/IP). When data is requested from one computer, the Transmission Control Protocol defines how the file is to be broken down into packets and what kind of channels they need to follow through the network. The Internet Protocol then addresses the packets and defines their route and ensures they make it to their destination. While TCP/IP is somewhat similar to packet switching it is a vast improvement as it allows data to be transferred to a virtually indefinite number of computers and networks, and overtime creates computer networks and information pathways that allow for faster data transfers within the larger network. 

The full ARPAnet would not be switched over to TCP/IP until January 2, 1983, and that same year would see the adoption of the term “internet.” However as computers became more available the network became less secure, and honestly computer scientists were much more interested in seeing what the network could achieve and not so much with securing it. This forced a split between ARPAnet and the military who created their own network that was aptly named MILNET. 

ARPAnet connections. Courtesy of The Daily Swig.

LAN and the Modern Internet

Along with ARPAnet, universities and research groups began using in-house networks called Local Area Networks (LAN). These LANs allowed computers in a limited geographical area to be interconnected to one another, which was useful for sharing data between researchers. Eventually LAN networks began to use the Internet Protocol to connect with other LAN networks of similar focus all across the country. In 1986 a LAN network was developed to create a new competing network called the National Science Foundation Network (NSFnet). NSFnet started out by linking together the five national supercomputers in the U.S. and then began linking every major university together, and of course it did not take long for smaller universities and research labs to want to join. Over the next few years NSFnet began to replace the slow and aging ARPAnet, and it was not long before nodes that were not research focused joined the network. 

ARPAnet was officially shut down on February 28, 1990, but by that time most people had moved on to NSFnet or another faster network. NSFnet was further improved and joined together with other networks, going on to form the backbone of the modern internet. 

Public Domain.

Sources

https://www.thoughtco.com/arpanet-the-worlds-first-internet-4072558https://www.computerhope.com/jargon/a/arpanet.htmhttps://www.britannica.com/topic/ARPANEThttps://www.sciencedirect.com/topics/computer-science/arpanethttps://www.darpa.mil/attachments/ARPANET_final.pdfhttp://iml.jou.ufl.edu/carlson/History/arpanet.htmhttps://www.youtube.com/watch?v=EXVXtIqXXw4https://www.techtarget.com/searchnetworking/definition/TCP-IP