the man who created the Internet

Every Internet user knows what ARPANET is. Well, okay, not everyone, but only those who were interested in the history of the World Wide Web. The main milestones of this project are also widely known: funding from the US Defense Advanced Research Projects Agency (DARPA), the first connection of the terminals of the University of California and Stanford University, which took place on October 29, 1969… But the names of the engineers and developers who took part in the creation of the first The history of humanity's global computer network often remains behind the scenes. Today's story is about one of these engineers, Frank Hart.

Frank Hart was not the typical lone genius often portrayed in fictional novels or movies. Hart was born in 1929 in the Bronx, New York. His father worked in elevator equipment, and his mother worked as an insurance agent. After graduating from high school, Hart entered the Massachusetts Institute of Technology (MIT), where, thanks to a fortunate coincidence, he ended up in a newly created computer programming course taught by Gordon Welchman. It was thanks to this that Hart was able to get a job as a research assistant at Lincoln Laboratory, where the Whirlwind I computer was installed, which controlled the radar tracking system for aircraft.

After graduating as an engineer, Hart remained at MIT's Lincoln Laboratory, where he was appointed director of projects to create real-time computing systems in which measuring devices collected data through telephone lines connected to computers. It was not yet a full-fledged computer network, but already something vaguely similar to it. According to the recollections of his colleagues, Frank really enjoyed working at MIT, since the team of young scientists did not resemble employees of a commercial company, but rather a group of like-minded people engaged in a common interesting business and solving unusual engineering problems.

However, in 1966, Frank Hart left Lincoln Laboratory to join Bolt, Beranek and Newman (BBN). BBN was known at the time for its cutting-edge developments in electronics and defense technology. In the late 1960s, this company won a tender announced by ARPA to develop an Interface Message Processor (IMP), a specialized packet switching node. This ancestor of the router, similar in appearance to the Soviet soda machine, was later actively used as part of the ARPANET. It was here that Hart's fate intersected with the history of the Internet: he became the project manager for the creation of IMP.

IMP: the first step towards a global network

In the mid-1960s, the US Department of Defense, concerned about the possible threats and vulnerabilities of communications, decided to fund a project to create a network that could provide reliable communications even in the event of a nuclear war. The Advanced Research Projects Agency (ARPA) led this effort and selected BBN to implement one of the important components of the project – the interface message processor.

The IMP hardware was a 16-bit Honeywell DDP-516 computer, connected to the network via a special serial interface. Later, devices were built on the basis of the lighter Honeywell 316, but BBN developed the software independently. The complete code consisted of six thousand words in Assembly language, and it was written mainly on the PDP-1, after which the already assembled program was transferred to the target device. The router had an error control mechanism based on calculating a 24-bit checksum; it discarded packets with errors without acknowledging receipt. The sender of the IMP data packet, without receiving confirmation from the recipient, resent a duplicate packet. Hart and his team developed not only the firmware for IMP, but also the data transfer protocols that provided IMP-to-host and IMP-to-IMP connections, as well as the IMP-sender-IMP-receiver protocol (IMP-s-IMP-r).

Initially, it was assumed that each computer on the ARPANET would connect to the network via IMP, but in order to save money, the project's customers decided that one IMP should serve several network nodes at once.

Bolt, Beranek and Newman were contracted to build four IMPs, the first of which the contractor was to deliver to UCLA on August 30, 1969, where it was connected to a Scientific Data Systems SDS Sigma 7 computer. The company undertook to transfer the remaining three IMPs at intervals of one month. The second copy was received by Douglas Engelbart's group at the Stanford Research Institute (SRI) on October 1 of the same year, it was connected to the SDS 940 computer. The third IMP was delivered in November '69 to the University of California at Santa Barbara, and the fourth machine was installed at the University Utah in December.

The main function of IMP was to ensure uninterrupted data transfer between nodes of the ARPANET network, but the creators also included in this computer the ability to remotely control using terminal access, and developed software for remote diagnostics. In other words, IMP has become a practically autonomous ARPANET router, capable of operating for a long time without operator intervention.

IMP Development Team, Frank is sixth from left.

In addition to the basic Interface Message Processor, Frank Hart and his team built a modification of it called TIP (Terminal IMP), designed to connect teletypewriters, which played the role of terminal terminals, and other peripheral equipment to the network. This device was also based on the Honeywell 316. Much later, the first versions of IMP based on Honeywell computers were replaced by higher-performance multiprocessor BBN Pluribus computers, which were developed in-house by Bolt, Beranek and Newman.

On October 29, 1969, the first ARPANET node was launched at the University of California, Los Angeles (UCLA). Subscriber lines DS-0 with a capacity of 56 kbit/s were leased from the telephone company AT&T. This was the moment when the network came to life. The first message sent between UCLA and Stanford University was short and consisted of just a few characters – “LO”. This message was supposed to contain the “LOGIN” command, but the system crashed after sending the first two characters. However, this event marked the beginning of a new era.

The network expanded rapidly, and by the early 1970s, dozens of research institutions had joined ARPANET. IMP devices were the heart of the ARPANET until DARPA decommissioned the network in 1989. Subsequently, most IMPs were either dismantled or transferred to computer museums.

Leadership and Engineering Approach

Hart was not only an outstanding engineer, but also an excellent organizer. When he led the IMP project in 1969, no one knew how to properly organize such a network. There were no standards, guidelines, textbooks, everything had to be invented from scratch. But Hart, thanks to his engineering thinking and ability to manage a team, was able to create a team that was able to bring ARPANET to life in a few years.

One of his key tasks was to ensure network reliability. It had to survive possible disasters, such as a nuclear attack, so reliability and stability were the determining factors. For these reasons, Hart paid great attention to testing and debugging the system. Every link, every piece of code, every piece of hardware was thoroughly tested.

An interesting point: during tests of the first version of ARPANET, it turned out that the network was able to withstand much greater loads than expected. In a report written by Hart on testing of the UCSB-SRI link conducted in late 1969, based on a 27-hour period of active computer operation on the link, IMP detected approximately one erroneous packet in every 20,000 transmitted. Also, the device created by Hart worked stably for long periods of time (hours) without errors or failures, which at that time was considered a high indicator of reliability.

By 1971, Hart's IMP team had grown to 30 people. In their book Where Wizards Stay Up Late: The Origins of the Internet, Kathy Hafner and Matthew Lyon wrote that Hart became a “highly respected and valued project manager” at BBN because he teams had members “committed to a common mission rather than to personal interests, who took personal responsibility for what they did.”

Heritage

Frank Hart wasn't always in the spotlight when it came to creating the Internet. His name has often been overshadowed by other pioneers such as Vint Cerf and Bob Kahn, who developed the TCP/IP protocol. However, it was Hart and his team at BBN who laid the groundwork for these technologies to be put into practice.

ARPANET became a testing ground for ideas and concepts that would later become the global network. It was a bridge from scientific research to real life, and Frank Hart played a key role in building this bridge. While working at Lincoln Laboratory, Hart met Jane Sundgaard, one of the first women computer programmers at MIT. They married in 1959 and had three children. Jane Hart passed away in 2014, Frank himself died of melanoma at the age of 89 in a nursing home in Lexington, Massachusetts on June 24, 2018.

We rarely think about how much work and ingenious decisions went behind what has become an everyday part of our lives. But without people like Frank Hart, the Internet might have been very different—or not at all: without his contributions to technology and telecommunications, we might still be exchanging information on floppy disks.

The article is supported by the team Serverspace.

Serverspace – a cloud service provider that provides rental virtual servers with Linux and Windows OS in 8 data centers: Russia, Belarus, Kazakhstan, the Netherlands, Turkey, USA, Canada and Brazil. To build an IT infrastructure, the provider also offers: creation of networks, gateways, backups, CDN, DNS services, S3 object storage.

IT infrastructure | Double the first payment by code HABR