guia do novo coronavirus
Imprimir Republish


The internet is born

The scientific and technological steps that led to the great worldwide network of computers

GUILHERME LEPCAThe internet had many fathers, as it was not the work or the inspiration of just one person. It was born and developed within an academic environment with financing from the Advanced Research Projects Agency (Arpa), a military research agency connected with the US Department of Defense (DoD). Arpa was created in 1958 to face the so-called Cold War between the United States and what was then the Soviet Union. The idea was to look for technologies that did not centralize the processing and filing of information in large computers and that enabled data exchanges amongst them. In October 1962, Arpa hired Joseph Licklider, previously a researcher at the Massachusetts Institute of Technology (MIT), who was then working for Bolt Beranek and Newman (BBN), a firm formed by MIT students and professors. Licklider was researching the interaction between computers and users and had studied physics, mathematics and psychology. As early as the beginning of the 1960s, he published a series of articles in which he commented on the possibility of using interconnected computers to enable global communication with access to electronic libraries. At Arpa, his memos referring to his colleagues as “members of the intergalactic computer community” became notorious. Licklider was the first director of the Information Processing Techniques Office (IPTO), which was to become the embryo of Arpanet, the internet’s predecessor.

Arpa made some of the early and unsuccessful attempts to connect computers. The story started changing when, in May 1961, Leonard Kleinrock, a professor at the University of California at Los Angeles (UCLA), presented at MIT a doctoral thesis that put forth a theory that later would become known as packet switching, whereby information would be transformed into small electronic packets before being sent to another computer. This characterizes the internet today. At the same time, the engineer Paul Baran, from the Rand Corporation, an organization created at the end of World War II to advise the American Air Force, also showed the viability of exchanging digital electronic packets. On the other side of the Atlantic, in England, communication via packets was also being studied. Professor Donald Davies, from the National Physics Laboratory of the United Kingdom, coordinated, in the early 1960’s, a project about computer communication networks financed by the British government. He was the person who named the system “packet” in an NPL memo dated June 1966.

At this time, there was already enough theory to move ahead with the formatting of a packet switching network and in 1968, the Arpa researchers, coordinated by Lawrence Roberts and Robert Taylor, along with researchers from other universities, including Kleinrock, prepared a system called Interface Message Processor (IMP) to allow data packet switching among computers from different manufacturers. At that moment, the function of the hosts was determined. Hosts are computers that hold data and that subsequently were renamed routers. Their function is to carry messages to the correct destination. To implement the system, Arpa held a tender, won by BBN. Here, the group of researchers included Robert Kahn, who later became one of the authors of the internet network protocol, the Transmission Control Protocol-Internet Protocol (TCP/IP).

In September 1969, the first computer of the future network at UCLA was activated, at Leonard Kleinrock’s laboratory. The second computer was connected at the Stanford Research Institute (SRI), in Menlo Park, also in California, at the laboratory of Professor Douglas Engelbart, a researcher who, along with William English, created the mouse in 1968. The first connection took place that very same month, October 1969, on the 29th, at 10:30 p.m., with a message that left Kleinrock’s laboratory for SRI. At his site, he tells us how this came about: “I was supervising the student programmer Charley Kline and we transmitted the message from the Host SDS Sigma 7 computer to the Host SDS 940 computer at SRI. The aim of the transmission was merely to log onto the SRI with the team of professor Engelbart. We managed to transmit the ‘l’ and the ‘o’ and then the system crashed. Therefore, the first internet message was ‘lo’. We were able to log on properly one hour later.”

The following month, two other initial units of the Arpanet joined the network, the University of California at Santa Barbara (UCSB) and the University of Utah at Salt Lake City. It began as a network of four large computers with a speed of 2.4 kilobits per second (kbps), which soon rose to 50 Kbps, after an agreement was reached with the telephone companies that owned the lines supporting the network.

War scheme
Arpanet was then devised and tested as a decentralized communication network, without an operations center. Of the four computers, if two failed to run because of a war, for instance, the other two could still communicate. Because Arpanet involved national security, it was limited, at that time, to just a few government research institutes and universities. In December 1970, the Network Control Protocol (NCP), the predecessor of TCP/IP, became ready, making communication easier. Soon thereafter, universities such as Carnegie Mellon, Harvard and MIT, besides NASA (the US space agency), and companies such as BBN and Rand joined the network. The exchange of messages soon acquired more features, with a program dedicated to electronic mail having been developed in 1971 by the electronic engineer Raymond Tomlinson, from BBN. The year after that, Arpa was renamed Darpa, when the word “defense” was added to its name.

GUILHERME LEPCAAlthough Arpanet was created for military purposes, it mainly linked academic researchers, so that the exchange of messages started to extrapolate the military and scientific sphere, moving onto the exchange of personal messages as well. In the rest of the world, new networks such as Bitnet started to appear, but NCP was incompatible with them. To solve this problem, Robert Kahn, now working for Darpa, turned to Professor Vinton Cerf, from the Stanford Institute, to format a new intercommunication protocol. After conducting several experiment with the new protocol, the Department of Defense decided that, in January 1983, all the Arpanet computers should switch to TCP.

As Arpanet was closed and only for a few users, some US universities wanted to have their own network. In 1979, these universities obtained the support of the National Science Foundation (NSF) of the United States. Two years later, with a US$5 million budget, the Computer Science Network (CSNet) started operating, bringing together computer research groups that were not on Arpanet. The new network used TCP/IP and was the first to establish a connection with Arpanet. CSNet was maintained by the NSF for three years and then became independent.

Without CSNet, the NSF planned a new, broader network that could encompass not only computer scientists but that was also an academic research tool, supplying technology to anyone within the university, and not only to researchers, among other features. In 1986, the institution launched NSFNet in order to interconnect networks. As people wrote at that time, it was the “inter net,” a network of networks. NSFNet initiated its activities using TCP/IP at a speed of 56 kbps. It encouraged regional networks in the United States and set up a structure of internal connections in that country. In the early 1990s, most of the networks, such as Bitnet and CSNet, started being routed to the internet, which showed how versatile the system was. At this point, the military decided to create their own network, called Milnet, and Arpanet was wound up. In subsequent years, other countries joined NSFNet and an unprecedented growth took place, causing those who were not within academic networks or who were no longer at university (and who therefore missed the networks) to look upon them with hungry eyes. Therefore, in 1991, NSF allowed the use of the network for commercial purposes and, as from 1995, its structure was transferred to the private sector.

Before the internet became open to all, however, another technology arose that made it sturdier. This was the birth of the worldwide web, in the laboratories of the Cern, the European Organization for Nuclear Research. The web was devised by the English physicist Timothy Berners-Lee, a researcher at Cern, who studied a system to exchange scientific documents among institutions tied to the institution in several parts of the world. He created hypertext and the means for accessing links, besides having made the exchange and exhibition of documents easier. The term worldwide web was invented by him when he was writing the new system’s code in 1989. In 1990, Berners-Lee proposed, along with the Belgian engineer Robert Cailliau, the hypertext transfer protocol (http) and the hypertext markup language (HTML), a software program for designing web pages. Http is also a protocol based on TCP/IP. Like everything that concerns the internet, growth was very fast. What helped this expansion substantially was that Cern and Berners-Lee did not take out a patent for the invention. Once the http system became widespread, researchers from the National Center for Supercomputing Applications of the University of Illinois at Urbana-Champaign created Mosaic in 1993, the first web browser with information positioned graphically. This subsequently gave rise to Netscape, the first commercial browser that disseminated the web all over the world

Close to the end
The first generation of Internet Protocol (IP) numbers, identifying each computer, is about to reach its capacity

For the internet to work and each message to get to its destination or for a user to find a server linked to a given site, each piece of equipment in the network must have a previously encoded number. Upon connecting to the internet for the first time, the new computer and more recently a mobile phone or tablet is given an IP number, which identifies it in the network. Because it encompasses the entire world, the release and control of the numbers is centralized at the Internet Assigned Numbers Authority – IANA, in the United States. From the very start of the internet, a set of existing numbers has been used, known as IP version 4 (IPv4), with 4.3 billion numbers. However, this is reaching its capacity. Only 2.73% of the total numbers are now available for the world. The solution that is currently being implemented is IPv6, a system with a fantastic size. The new version equals the number 340 followed by another 36 digits, or 79 trillion trillion times the size of IPv4. New pieces of equipment and operating systems are ready to join the new system, which will coexist with the old one for many years.

“The problem is the service provider, if it needs a lot of numbers and if it hasn’t yet adapted itself to IPv6: it will have difficulty accessing the network in the future,” says Demi Getschko, director-president, the Center for Information and Coordination of Dot BR. This is the entity that receives blocks of numbers from IANA and is responsible for the distribution and control of IP numbers in Brazil. At the time when he was in charge of coordinating the FAPESP data processing center and the internet was starting up in Brazil, he received from IANA the first lot of IP codes for Brazil, consisting of four million numbers, back in 1994, when the internet entered its commercial phase. “Previously, we would ask them for numbers directly and once we got them we would organize them into blocks for USP, Unicamp and other academic institutions. The rest we would allocate, free of charge, by means of a justification form,” says Getschko. Only in 2005 did FAPESP stop controlling this. That year, CGI (the Internet Management Committee) took over this duty and created