Fuelled largely by the PBS television series "Triumph of the Nerds" some years ago, and by the earlier writings of Silicon Valley gossip columnist Robert Cringely in the beautifully titled "Accidental Empires - how the boys of Silicon Valley make their millions, battle foreign competition, and still can't get a date" (Penguin, 1992), a popular belief has sprung up that the Internet was invented in the Pentagon in 1969. The theory goes on to suggest that the Internet network invented in the Pentagon was designed to survive a nuclear attack.
This theory survives and is even propagated by individuals who celebrated "the Internet's 35th birthday" in 2004. However, not everyone celebrated, and not everyone agreed.
Perhaps the most serious rebuttal on the theory of Pentagon origins (otherwise known as the big bang theory of Internet origins) came from the person who was in charge of the Pentagon Arpanet project at the time when the Internet supposedly began, Bob Taylor. Writing in reference to a mailing list invitation to attend the 35th anniversary event, Bob Taylor explained.
"In February of 1966 I initiated the ARPAnet project. I was Director of ARPA's Information Processing Techniques Office (IPTO) from late '65 to late '69. There were only two people involved in the decision to launch the ARPAnet: my boss, the Director of ARPA Charles Herzfeld, and me.
Numerous untruths have been disseminated about events surrounding the origins of the ARPAnet. Here are some facts.
The creation of the ARPAnet was not motivated by considerations of war. The ARPAnet was not an internet. An internet is a connection between two or more computer networks."
So then, where and when did the Internet begin? The only thing historians seem to agree on is that it was not 1969, or the Pentagon, (or for that matter Al Gore). From there on, there is a wide divergence of views as to when, where, and by whom the Internet may have been invented.
It will help in discussing the beginnings of the Internet to define what the Internet is. Now you can get as many different definitions of what the Internet is as you can dictionaries. But for must of us, the simple description, a "worldwide system of interconnected networks and computers" is pretty good and adequate. But some people get more technical, and want to add to the definition terms such as "a network that uses the Transmission Control Protocol - Internet protocol" (or TCP/IP). But not everyone agrees that it is TCP/IP which defines the Internet or its point of origin.
We have analysed the literature on this, and over the years had correspondence with most of the pioneers whose names are mentioned in this article. At www.nethistory.info we list a number of the books devoted to early Internet history: some of the most prominent are
None of these are definitive or complete. Most focus heavily on the Arpanet developments. On further analysis we come up with at least five distinct theories, each of which can be credibly discussed. We state from the beginning that we do not personally see the theories as mutually exclusive - we have for many years believed in a multiple origins theory rather than a single point of invention one.
But the theories which need to be examined are:
1. Packet switching represents the origins of the Internet
2. The TCP/IP protocol represents the origins of the Internet
3. A range of telco-led activities from the 1960s represents the true origins
4. The birth of the Internet is best explained through a history of applications rather than the protocols
5. The range of inventions and activities emanating from Xerox Palo Alto laboratories, including Ethernet, represent the true beginnings.
Let's explore these various theories further, but first….
By what criteria would we determine what was a (primitive) Internet?
We will examine the various verifiable events according to the following criteria
If a theory passes all these tests, the final test is - was it the first
These criteria, we suggest, all have to be met before we have the event which could be called "the birthday of the Internet". If not all are met, we are dealing with a different but perhaps closely related species, or with a theory which was at that place and time untested.
So, let the games begin!
The Arpanet 1969 claim to Internet origins largely rests on acceptance of
this theory and a belief that this was the first ever packet switching exchange.
As we are told,
On October 29 1969, UCLA computer science professor Leonard Kleinrock led a team of engineers in launching the first Internet message from UCLA to Stanford Research Institute, as part of the Arpanet project. As Kleinrock is purported to have reflected on the 35th anniversary of this event in 2004,
"When we sent that first message, it marked the birth of a new method of global communications that has forever changed the course of business, politics, entertainment, education and social interaction, Now, 35 years later, the Internet has become so pervasive that even my 97-year-old mother uses it."
Important as this event was, there are several reasons not to regard it as the birth of the Internet. These include:
Let's explore the flaws in this theory in more detail.
In my "History of the Internet", I
"What Arpanet did in 1969 that was important was to develop a variation of a technique called packet switching. In 1965, before Arpanet came into existence, an Englishman called Donald Davies had proposed a similar facility to Arpanet in the United Kingdom, the NPL Data Communications Network. It never got funded; but Donald Davies did develop the concept of packet switching, a means by which messages can travel from point to point across a network. Although others in the USA were working on packet switching techniques at the same time (notably Leonard Kleinrock and Paul Baran), it was the UK version that Arpanet first adopted."
Ronda Hauben writes similarly indicating a 1966 event when Donald Davies in the UK implemented a packet switch connecting a set of host computers.
Kim Veltman goes further in exploring this
"We are almost always told that the Internet began solely in America. This is not really true. The earliest pioneers included a Frenchman, Louis Pouzin, who introduced the idea of data grams and an Englishman, Donald W. Davies, who was one of the inventors of packet-switching. Another of the great pioneers in Britain was Peter T. Kirstein, who went to America at the beginning of the Arpanet in 1969 when it was decided that Davies could not go for reasons of national security. "
And Bruce Sterling adds
"The National Physical Laboratory in Great Britain set up the first test network on these principles [of packet switching] in 1968. Shortly afterward, the Pentagon's Advanced Research Projects Agency decided to fund a larger, more ambitious project in the USA. Hence an English project of 1968 inspired the beginnings of the US Internet in 1969".
What follows from this analysis is that, if we believe that the first trials of packet switching represents the beginnings of the Internet, the Internet began in the UK, not USA. But does that theory meet all our tests?
Another reason to reject the Arpanet 1969 origins theory is that the Arpanet
was not about people communicating over distance at all - something which would
be a primary determinant of what we know as the Internet.
Arpanet was about time-sharing. Time sharing tried to make it possible for research institutions to use the processing power of other institutions computers when they had large calculations to do that required more power, or when someone else's facility might do the job better.
Although Arpanet developed packet switching, Larry Roberts (Project Manager and Architect for the Arpanet project) makes it clear that sending messages between people was "not an important motivation for a network of scientific computers".
It seems difficult to stretch belief that timesharing between mainframe computers is the Internet - a point of origin perhaps, but hardly the single point of beginnings for the Internet.
The last point we need to make in deciding that Arpanet/1969 was not the
birth of the Internet relates to the definition of a "network of networks".
What defines the Internet is the capacity to connect networks of different types. It therefore follows that Arpanet as a single network could hardly be described as an Internet - Arpanet would have to connect to something completely different before it could be part of an Internet.
According to Bob Kahn, who we will hear more of in the next section,
"What the ARPANET didn't address was the issue of interconnecting multiple networks and all the attendant issues that raised."
(Kahn, E-mail to Hauben, September 15, 2002)
Similarly, Vint Cerf, who also features strongly in our next theory, explains that the NCP protocol on which internal Arpanet connections took place was inadequate for addressing the problem of interconnecting multiple packet networks which were not identical.
For the above reasons, we do not believe that 1969 and Arpanet can claim to be the origins of the Internet. If packet switching is the origins, the Internet was invented elsewhere. If we are looking at some other role Arpanet played, the date is not 1969 and another theory will have to pass our tests.
Among today's Internet community, this is the most commonly held belief, fed
largely by the impressive and ongoing role played by Vinton Cerf in the
evolution of today's Internet and its governance structures.
TCP/IP is the backbone protocol which some people claim is the basis for determining what the Internet is. It was developed in the 1970s in California by Vinton Cerf, Bob Kahn, Bob Braden, Jon Postel and other members of the Networking Group headed by Steve Crocker. TCP/IP was developed to solve problems with earlier homogenous attempts at communication between computers undertaken by ARPANET.
Vinton Cerf had worked on the earlier Arpanet protocols while at the University of California in Los Angeles from 1968-1972. He moved to Stanford University in late 1972. At the same time Bob Kahn, who had been the chief architect of the Arpanet while working for contracting firm Bolt Beranek and Newman, left that firm and joined ARPANET.
In October 1972 ARPANET publicly demonstrated their system for the first time at the International Computer Communications Conference in Washington DC. Following that meeting, an International Networking Group chaired by Vinton Cerf was established.
Bob Kahn visited Stanford in the spring of 1973 and he and Vint Cerf discussed the problem of interconnecting multiple packet networks that were NOT identical. They developed the basic concepts of TCP at that time, and presented it to the newly established International Networking Group. This meeting and this development, this theory suggests, rates as the beginning of the Internet.
By 1975 the first prototype was being tested. A few more years were spent on technical development, and in 1978 TCP/IPv4 was released.
"We had running code by the middle of 1975 for TCP. We ran this protocol on selected nodes of the ARPANET, the packet radio net and the packet satellite net - all of which were ARPA sponsored. Xerox PARC implemented a version for their Ethernet around 1976 if memory serves - connected it to the nascent Internet by way of packet radio in the San Francisco Bay area".
Among those working on this specification were researchers from Stanford University, a range of other universities, BBN contractors, Xerox Parc employees, and researchers from the United Kingdom, France and Norway.
It would be some time before it became available to the rest of us. In fact, TCP/IP was not even added to Arpanet officially until 1983.
So which date do we celebrate if we adopt the TCP/IP origins theory?
We do not yet have a definitive date, but 1975 seems to be the definitive year in which, for the first time, networks connected to each other. But was the first connection between disparate networks a TCP/IP one? Well, you'll have to read on to find out.
Given the above, the TCP/IP origins theory and a 1975 date remains in our tests. This theory represents to us a far more credible description of the origins of the Internet, passing all of our tests to date. However, although these developments definitely led to the establishment of standards bodies such as IETF, ISOC and the emergence of ICANN (headed still by Vint Cerf), it is still a stretch to some people to believe this as the single point of origin- and was it the first connection? Let's explore other theories as well before making a final determination.
The primary proponent of this theory would be
Kim Veltman of the McLuhan Institute in Holland. As Veltman states,
" Since its beginnings …..there have been many stories about the Internet. One is that the Internet was a US invention. The story that officials in AT&T (a large American telecommunications company) were once adamantly opposed to the Internet led to a received wisdom that telephone or telecommunications companies (telcos or telecoms) and the Internet were unrelated. The telephone companies, we are told, were big monopolies, blind to innovation and the Internet was started on the sly by a few scientists and academics. The Federal Communications Commission (FCC) version is more subtle: The telcos did infrastructure, while those who developed the Internet did applications."
As Veltman points out, the AT&T Bell Labs did some of the first digital transmission and switching in 1962, seven years before the "US Internet" began. When the Department of Defense (DoD) commissioned the Advanced Research Project Agency's Network (ARPANET) to do research into networking, it was AT&T that provided 50kbps lines. In 1969, the year that Arpanet began, AT&T's Bell Labs developed Unix which was "the operating system behind the early Internet, and was one of the key operating systems in the middle and late ARPANET."
Between 1969 and 1972, Bell Labs developed the C programming language basic to much of Internet software. In 1970, AT&T installed the first cross-country link between the University of California at Los Angeles (UCLA) and Bolt, Beranek and Newman (BBN) in Boston. In 1976, AT&T's Bell Labs developed (Unix-to-Unix Copy (UUCP), which was distributed with UNIX one year later."
All of these were important points of origin of the Internet as we know it, so the telco theory, unpopular as it is in Internet circles, should perhaps be explored in more detail. Certainly the physical infrastructure created by the telcos was central, and certainly telcos had worked out protocols for sending voice data between disparate networks early in the piece. In the examples above, they added the component of computers and networked them. Can we completely eliminate the telco origins and contributions to early developments?
We `cannot altogether dismiss this theory either. We see in the telco theory
connections between networks at an infrastructure level, involvement of
computers, actual networked events earlier than TCP/IP, and at least some
indication that human communication was a motivating force.
But then, is a telephone line an Internet? The purists would argue certainly not. But by the same token, is a transport protocol the Internet? No more so than a telephone line, we would suggest, much though both were necessary. The same logic would tend to suggest that neither TCP/IP or the existence of a telephone line can be said to indicate the existence of an Internet (unless you accept the circular argument that internet=tcp/ip and try to redefine what the world has come to know as the Internet).
This set of developments is important, because it brings in commercial origins, as well as the academic research/government funded origins of the more popularly espoused theories. With Unix and UUCP, the telcos played a larger part than many would have us believe.
Perhaps answering this line of confusion as to whether the Internet can be
represented by either telephony infrastructure or any particular protocol at the
transport layer, Mitra Ardron takes it further with another theory altogether.
(Internet History newsletter, October 2004)
"I would suggest that defining the history of the internet by the particular protocol that won is only one way to do it. Ask yourself - would it still be the internet if we were using ATM, or X.25 or any of the other competing protocols? Of course it would.
An alternative view of history tracks the history of the Internet as the ubiquitous use of electronic "online" communications. The history belongs at the applications level - with the development of email, with the progression from proprietary databases to Gopher and Wais to the World Wide Web, and from newsgroups and conferencing (eg BITnet and Usenet) through mailing lists and blogs.
One very significant trend which tends to get ignored is the various online systems, the early Source, Compuserve, Dialcom, and of course APC networks, Fidonet etc. If anything, the history of the use of the Internet, at least from the point of view of the public, owes more to that stream of development than the more common version.
From that perspective, the switch from X.25 to TCP/IP around say '92 for the transport was just something that was done when cost/benefit of TCP/IP dropped below that of X.25."
Ardron refers to the existence of an emerging interconnectivity regime between various commercial, non-profit and hobbyist networks which began to emerge from the mid 1980s without necessarily using TCP/IP. During this pre-web era, email began to be exchanged freely between networks according to emerging standards and through various gateways. Newsgroups emanating from academic circles became available on various networks, with only a percentage of the people utilizing the growing global network using TCP/IP.
The "killer app" driving all of these networks was email. Email existed before the Internet, but in those days email could only be used to send messages to various users of the same computer. Once computers began to talk to each other over networks, however, the problem became a little more complex - We needed to be able to put a message in an envelope and address it. To do this, we needed a means to indicate to whom letters should go that the electronic posties understood - just like the postal system, we needed a way to indicate an address.
This is why Ray Tomlinson is credited with inventing email in 1972. Like many of the Internet pioneers, Tomlinson worked for Bolt Beranek and Newman as an Arpanet contractor. He picked the @ symbol from the computer keyboard to denote sending messages from one computer to another. So then, for anyone using Internet standards, it was simply a matter of nominating name-of-the-user@name-of-the-computer. Internet pioneer Jon Postel was one of the first users of the new system, and is credited with describing it as a "nice hack". It certainly was, and it has lasted to this day.
Email drove mass adoption in the pre world wide web era known as the "protocol wars". Governments continued to argue for some time for a completely different set of standards based on OSI; Hobbyist networks maintained the Fidonet system; various efforts such as APC and UFGate software bridged the Unix and PC based network worlds - and a host of commercial systems such as Dialcom, Compuserve, AOL and other email systems with entirely different operating systems were all seeking an answer to the connectivity crisis. But email exchanges across these systems were already working by a variety of means.
Later on, TCP/IP won the protocol wars on costs and simplicity of adoption. Soon after, the World Wide Web appeared, and the last of the laggards believing they had a separate future began to convert.
The theory is compelling because it explains clearly the reasons why people started to use the Internet. It creates a more popular understanding of what is essentially a popular phenomenon. It passes all our tests; certainly email was passed between non-homogenous networks not utilising tcp/ip.
And then we come to the theory advanced by the person who headed the Arpanet
project itself, Bob Taylor. Quoting Bob,
"I believe the first internet was created at Xerox PARC, circa '75, when we connected, via PUP, the Ethernet with the ARPAnet. PUP (PARC Universal Protocol) was instrumental later in defining TCP.
For the internet to grow, it also needed a networked personal computer, a graphical user interface with WYSIWYG properties, modern word processing, and desktop publishing. These, along with the Ethernet, all came out of my lab at Xerox PARC in the '70s, and were commercialized over the next 20 years by Adobe, Apple, Cisco, Microsoft, Novell, Sun and other companies that were necessary to the development of the Internet."
John Shoch, who worked with Robert Metcalfe on the Ethernet developments at Xerox Parc, and who is at great pains to stay out of debates about who started the Internet, has concluded that PUP (the Parc Universal Protocol) was the first complete, operational set of Internet protocols. Schoch was also involved in the development of TCP/IP at a later date. To quote Shoch,
"Starting around 1974, Xerox PARC designed and deployed an internet architecture called PUP; it was up and running on multiple machines and networks when TCP was just a design for byte stream protocols. Input from Xerox' operational experience helped convince the TCP working group to add the IP packet layer!"
This might in fact provide another answer for us - the first Internet connection may not have involved TCP/IP or government funding at all, and may be solely the result of commercial research.
Examining these various events, we come to some important findings.
These findings are important lest any institution or organisation lays claim to some proprietary ownership of the Internet. As Kim Veltman puts it,
"It seems false to claim that America invented the Internet and is simply misleading to argue that because America invented the Internet, it is their right to control its governance through organizations such as the Internet Assigned Names Authority (IANA) and more recently the Internet Corporation for Assigned Names and Numbers (ICANN)."
It would appear however that some proprietary ownership has been assumed. To quote from a US Dept of Communications (DOC) memorandum (Karen Rose)
" Purchase Order No. 40SBNT067020 provides that "[ICANN] will perform other IANA functions as needed upon request of DOC. As such, the Department of Commerce requests that, as part of the IANA functions, ICANN undertake administration of the arpa TLD in cooperation with the Internet technical community under the guidance of the IAB, as a limited use domain for Internet infrastructure applications, including the migration of Internet infrastructure applications that currently reside in the .int TLD. Further, as indicated by DARPA, the arpa TLD string should be given a different expansion such as "Address and Routing Parameter Area" to avoid any implication that DARPA has operational responsibility for the domain.
Thus, although distancing from DARPA, the right to determine priorities and request changes is definitely assumed by DOC in this memorandum.
This is not to suggest that the DOC involvement has been anything other than benign and useful. Or that Arpanet wasn't in there somewhere. But current institutions and governance structures must take their validity from the current involvement and endorsement of major stakeholders, not from any assumed proprietary history. They need to justify their future involvement in the Internet similarly, on the basis of current and future usefulness and validity, not on the basis of a Purchase Order.
Nor should this article undermine the significant contributions of a number of individuals to claims as "fathers of the Internet". Most of these individuals, particularly those who are most prominent, are at pains to point out the crucial involvement of others - however, the institutions they represent are often less careful in ensuring that widespread involvement of individuals from commercial and government funded sources in a number of countries are ultimately to be thanked for the origins of the Internet. If this paper does no more than clarify that the Internet really has no owner and no single place of origin, it will have served well.
In the end, the History of the Internet is better understood as the history of an era than that of a protocol. To draw a parallel: we are treating the Internet more like a history of transport, which obviously has several threads and origins and important developments, rather than a history of the steam engine (part of the transport history, for sure, but only a small part). This is why we have problems with the TCP/IP origins theory (or the packet switching theory or the telco/infrastructure theory) and that that alone is the history on the Internet. There is so much more to it!
Multiple events, multiple players, and multiple points of origin need to be mentioned in any sensible understanding of the emergence of the Internet. Any claim by a nation, project, person, or team of individuals, or participants in any single event to "the beginnings of the Internet" is rubbish. Further, any claim that the validity or legitimacy of any structure or arrangement can be justified as Internet governance purely because it arose from one of these events is false.
Ian Peter lays no claims to having invented the Internet, but maintains an Internet history site at www.nethistory.info and is actively involved in current Internet developments through the Internet Mark 2 Project.
Permission to re-use this material for non-commercial purposes is granted provided that
www.nethistory.info is appropriately
credited as the source.
Please feel free to link to this page!
If you've read this far, and liked the material, get yourself a copy by buying Ian Peter's History of the Internet Audio CD here