ARPANET

Almost everyone is familiar with the internet – even those that don’t have access to it have an idea of what the global network is. But have you ever heard of Arpanet? This is what existed before the internet. Although this network ceased to exist in 1990 and has since been largely forgotten, the knowledge of how it worked and its de­vel­op­ment history enables us to gain a deeper un­der­stand­ing of the internet today.

The 1960s signaled the triumph of the computers. For private users, technical devices were usually out of reach due to their price. However, in the areas of sci­en­tif­ic research and the military, working without computers already was almost unimag­in­able. They also became in­creas­ing­ly important for large and medium-sized busi­ness­es. In order to exchange in­for­ma­tion faster, many tried to connect up their networks to exchange in­for­ma­tion as quickly as possible. In the 1960s, sci­en­tists still had to send data in the form of printouts – a laborious and time-consuming process. Fur­ther­more, the his­tor­i­cal context must be taken into account in order to be able to place the de­vel­op­ment of the Arpanet correctly: the USA was in the middle of the Cold War with the Soviet Union.

At this time, these two nations were not only engaged in a near-limitless arms race and various proxy wars, they also tried to outdo each other when it came to science, space travel in par­tic­u­lar. Con­sid­er­ing the political situation at the time, it’s hardly sur­pris­ing that the missions to set up a US computer network came from the military – more precisely, from the Advanced Research Projects Agency (ARPA). The president at the time, Dwight D. Eisen­how­er, had already founded ARPA in 1958 as an agency of the Ministry of Defense, re­spond­ing to the launch of the Soviet Sputnik satellite.

In order to be able to better co­or­di­nate sci­en­tif­ic projects and catch up to the research of the Soviet Union, this new authority had the task of or­ga­niz­ing military research projects centrally (under the name DARPA – the authority is still active today, for your in­for­ma­tion). To this end, a project was devoted to net­work­ing computers from various uni­ver­si­ty research in­sti­tu­tions. In 1962, the computer scientist and psy­chol­o­gist, J.C.R. Licklider, took over the man­age­ment position at ARPA. He already had a concrete idea for a network and was able to inspire the two IT pioneers, Robert Taylor and Ivan Suther­land, with his vision. From then on, the two computer sci­en­tists conducted research on a de­cen­tral­ized network (Licklider himself left ARPA and the project in 1964).

Initially there was no support from the Ministry of Defence, however, in 1965 the research work was very suc­cess­ful and was even­tu­al­ly completed in 1969. BBN Tech­nolo­gies – an IT company for which, in­ter­est­ing­ly enough, Linklider had worked as Vice President for a long time – was then com­mis­sioned to carry out the technical im­ple­men­ta­tion.

On 29^th October 1969, ap­prox­i­mate­ly three months after Buzz Aldrin and Neil Armstrong first set foot on the moon, the pro­gram­mer Charley Kline was able to send the first fully readable message via ARPANET: 'login'. At this time, the newly developed network consisted of exactly four computers in four different locations: the Uni­ver­si­ty of Cal­i­for­nia, Los Angeles (UCLA), the Uni­ver­si­ty of Cal­i­for­nia, Santa Barbara (UCSB), the Stanford Research Institute (SRI In­ter­na­tion­al), and the Uni­ver­si­ty of Utah (UU).

Gradually, the network was expanded and further American uni­ver­si­ties and au­thor­i­ties were connected with each other. In 1973, the network finally made the leap across the Atlantic and was also used in Europe. In 1983, the TCP/IP network protocol was also used for Arpanet, making the older network a part of the internet. In 1990, Arpanet was finally dis­con­tin­ued and replaced by the NSFNet, which had been in existence since 1985. NSFNet was dis­con­tin­ued in 1995.

Several technical de­vel­op­ments and in­for­ma­tion science concepts con­tributed to Arpanet’s creation. Some of the fun­da­men­tal in­ven­tions did not come from the com­mis­sioned sci­en­tists them­selves but were – in the spirit of an open, global knowledge culture – con­tributed by other re­searchers.

The basic idea came from J.C.R Licklider and man­i­fest­ed itself in his vision of the In­ter­galac­tic Computer Network. He wanted to develop a network that worked in­de­pen­dent­ly of the computer’s location, the systems used and – ideally – even if the user didn’t have a lot of IT knowledge. Within networks like these, each par­tic­i­pant should be able to com­mu­ni­cate with each other (this applies to both the human user and the machine user). Although nodes were (and still are) necessary for this purpose, there is no central point from which data can be dis­trib­uted.

This meant that the idea of an end-to-end con­nec­tion (such as telephone calls), which was prevalent at that time, was abandoned: telephone calls are accepted at a central interface and then forwarded by switch­board operators, i.e. a direct line is es­tab­lished to the recipient.

Data packets are sent within the de­cen­tral­ized network. However, not all the in­for­ma­tion to be trans­mit­ted is put into a packet and sent on its way. It is divided into several packets instead, known as packet switching. Each packet works in­de­pen­dent­ly of its siblings (i.e. the other packets that transmit parts of the in­for­ma­tion). This also means that the packets can take different routes because there is no end-to-end con­nec­tion. The packets are then put together once they are received. Packet switching works dif­fer­ent­ly to circuit switching, which you are probably familiar with from the phone network.

The advantage of packet switching is that the network is used much more ef­fi­cient­ly. Several par­tic­i­pants share the same channels and send their parcels using them – on ab­solute­ly equal footing. Ac­cord­ing­ly, the Arpanet also laid the foun­da­tion for what our society today is calling net neu­tral­i­ty. In addition, there’s less chance of complete network failure due to the many possible routes available.

In order for the different computers (based on different systems) to be able to com­mu­ni­cate with each other, it was necessary to create in­ter­faces to regulate com­mu­ni­ca­tion between par­tic­i­pants. For this purpose, they first looked for a software solution but finally decided to connect mini computers between the normal computers and the network. These interface message proces­sors (IMPs) were smaller (in relation to computers that took up a whole room in those days) machines that provided in­ter­op­er­abil­i­ty in the network.

Built by BBN Tech­nolo­gies, the IMPs were based on mini computers from Honeywell. Their function would nowadays be compared to that of a router: data packets are sent via them to the correct re­cip­i­ents. The receiver IMP then sends a con­fir­ma­tion, but only if the packet arrives without errors, otherwise the sender will au­to­mat­i­cal­ly re­ac­ti­vate and restart the trans­mis­sion. When Arpanet first started out, it had a speed of 50 kbit per second and was im­ple­ment­ed via telephone lines.

In order for the IMPs to com­mu­ni­cate with each other, BBN Tech­nolo­gies in­tro­duced a new standard: the 1822 protocol, named after BBN Report 1822, in which the company first described the protocol. In this protocol, the for­mal­i­ties required for com­mu­ni­ca­tion in Arpanet were specified. Since only the in­ter­me­di­ate mini computers had to comply with these reg­u­la­tions (the con­nec­tion from host to IMP had to be managed by each site itself), the protocol ensured in­ter­op­er­abil­i­ty.

The protocol required the packet to contain two sections: the in­for­ma­tion (or part of it) you want to transmit and a header. This header section also contains a numeric recipient address (com­pa­ra­ble to the IP address) with which the sender IMP can determine the receiver IMP so that the packets arrive at the correct location. Packets had a maximum size of 8159 bits, 96 of which were reserved for the header.

The Network Control Program (NCP) is also a protocol, however, it’s used in the middle level (transport layer) of com­mu­ni­ca­tion. This is higher than the 1822 protocol, which can be located on the three lowest levels of the network ar­chi­tec­ture: physical layer, data link layer, & network layer. The main tasks of the NCP are to establish, terminate, or redirect con­nec­tions between two hosts. In order to be able to establish a con­nec­tion at any time, a so-called logger was in­te­grat­ed into the protocol. This service was busy searching for login attempts from other host computers.

The Network Control program was replaced on 1^st January 1983 by the successor TCP/IP (the same family of network protocols on which the internet is still based). This day of switching known as Flag Day was com­par­a­tive­ly complex since the different protocols cannot be used side by side. It isn’t possible for hosts using NCP to com­mu­ni­cate with hosts using TCP/IP. It was, therefore, necessary to start the complete network on the cut-off date.

Although the Unix operating system is not directly connected to Arpanet, the de­vel­op­ment of both tech­nolo­gies ran parallel. Both Unix and the pro­gram­ming language C, which the operating level is based on, in­flu­enced the de­vel­op­ment of Arpanet. The software could already be used on many different platforms at that time. This made Unix a standard in the newly created network. The resulting uni­for­mi­ty sim­pli­fied the in­tro­duc­tion of mandatory ap­pli­ca­tions and protocols.

Looking back, many see the main benefit of Arpanet as having paved the way for the internet. But Arpanet even brought enormous benefits to the as­so­ci­at­ed uni­ver­si­ties in the first few decades after its de­vel­op­ment: it made it possible to run programs on remote computers. The cor­re­spond­ing service was, and is, known as 'Remote job entry' or NETRJS. In order to fully un­der­stand the sig­nif­i­cance of this, it is necessary to realize that the computing power of most computers at that time was far below what we are used to today. The fact that sci­en­tists were able to use the resources of connected computers made it possible to work much more ef­fi­cient­ly.

Two ap­pli­ca­tions that still char­ac­ter­ize the global network were already in­tro­duced in the early days of Arpanet: FTP con­nec­tions and e-mails. Both services were im­ple­ment­ed in 1972. The file transfer protocol made it possible to exchange files between two hosts. E-mail, on the other hand, is something that every internet user is familiar with. At that time, the two programs READMAIL (read mail) and SNDMG (send message) were developed, which enabled Arpanet users to send and receive text messages elec­tron­i­cal­ly. Both were pro­grammed by Ray Tomlinson, who worked for BBN Tech­nolo­gies at the time. We can also thank him for the @ sign, which is used to dis­tin­guish the name of the user from the name of the server.

Telnet is another service that is still part of the internet but has lost much of its im­por­tance in the meantime. The client-server protocol makes it possible to access another computer via a remote con­nec­tion. With the input console, users can then execute commands on other computers to search databases or start programs, for example. Nowadays, however, this service is hardly used. Although Telnet offers a number of ad­van­tages (e.g. the protocol is cross-platform), it also has an obvious dis­ad­van­tage: since the data exchange is com­plete­ly un­en­crypt­ed, the access points are entry points for cy­ber­crim­i­nals. Instead of Telnet, encrypted SSH is used nowadays.

Hardly anymore will deny that the internet as we know it today was de­ci­sive­ly shaped by Arpanet. But what exactly does that mean? After all, Arpanet was not simply the internet with a different name. Some of the in­no­va­tions that ARPA and BBN Tech­nolo­gies developed in the 1960s and 1970s are still used today – such as e-mails and the file transfer protocol. In principle, however, it was probably the concept of a de­cen­tral­ized, location-in­de­pen­dent network that ul­ti­mate­ly led to the internet.

Other aspects of com­mu­ni­ca­tion between computers have changed so much that the origins are difficult to recognize today. It is hard for many to imagine things that small routers are now capable of, were orig­i­nal­ly achieved by re­frig­er­a­tor-sized IMPs. Another decisive factor was the de­vel­op­ment of TCP/IP – a tech­nol­o­gy that was also inspired by the computer sci­en­tists at ARPA. Ethernet, a network tech­nol­o­gy, was also developed in the Arpanet en­vi­ron­ment and is still a standard when it comes to net­work­ing computers.

But what didn’t exist then – probably one of the most important dif­fer­ences between Arpanet and the internet – is the world wide web. The hypertext system was developed by Tim Berners-Lee in 1989, shortly before the Arpanet project was dis­con­tin­ued. In­for­ma­tion was pre­vi­ous­ly not ac­ces­si­ble in the network of servers available at that time, and had to be requested in­di­vid­u­al­ly from hosts – and even the first web browsers were still a long way off in 1969.