But a nervy group of computer scientists and engineers, backed up by a U.S. government eager to fund research into the latest technological marvels, was convinced it could be a reality. On the eve of the Internet's 30th birthday, some of the biggest luminaries of that revolution gathered at Harvard University to recount its history and swap war stories.
Vint Cerf, now the senior vice president for Internet architecture at MCI WorldCom, and co-developer of TCP/IP, the computing language on which the Internet is based, said that even in the medium's earliest days, there were hints of the global phenomenon that was to come.
With the aid of a primitive satellite technology called "packet radio," technologists in the mid 1970s were able to communicate over the Internet's precursor, the ARPANET, even though the land lines were broken, as officials assumed they would be in a nuclear attack, Cerf said.
A4-sized Internet
According to Dave Mills, a University of Delaware computer science researcher, "The enthusiasm driving the Net's development was due to the urgent need (of scientists) to communicate without wasting trees and airplane fuel."
The desire was out ahead of the scientific proof, said Larry Roberts, now president and CEO of telecommunications switch maker Packetcom, and a key ARPANET engineer in the 1960s.
"This was a fair challenge" from a hardware perspective, Roberts said. "We had to figure out how to get two computers to talk to each other. The problem was, we were using totally substandard, unreliable telephone lines."
Stunned that it worked
"We well exceeded that in the end," Roberts added.
Another challenge facing the forward-thinking researchers was that certain key computing technologies were still years in the future.
Sandy Fraser, who was a key networking engineer at AT&T Bell Laboratories starting in 1969, told how Bell Labs workers made primitive microprocessors, which they called "switching machines," at Bell Labs the following year.
"Then, the interest was in distributed computing, the idea that you could build a single distributed 'machine' out of a number of computers," Fraser said. "I was thinking about how you could have high-speed communication between these machines, but you needed a microprocessor to make this work -- and they didn't exist yet."
The sizable switching machines developed at Bell Labs helped researchers overcome the "significant hardware challenge" in computer networking at that time, he said, adding that now, tiny chips perform the same function.
'Built like a tank'
But using those primitive machines, researchers developed some of the credos that have become key to the development of the Internet, Walden said.
"The idea that a lot of different people would share the same network, and use it for different purposes, just wasn't the way things were done back then," he said. "But many of the principles we developed for the ARPANET have continued to be played out in the development of the Internet."
'It was as big as a refrigerator and built like a tank. It also had practically no memory and no operating system'
-- engineer Dave Walden recalls an ARPANET-connected computer
But the Internet's growth was slow at first, as engineers wrestled with questions that were considered hopelessly arcane outside academia and the military, according to several of the experts present.
When ARPANET engineers did the first packet switching experiment in 1966, they were stunned to see that it actually worked, Roberts said. From then on, meeting the challenge of linking computers together "was in fact survival" for the project, which had a US$3.4 million budget -- a fair amount of money in those days, he said.
Another ARPANET engineer, Dave Walden, recounted what one of the original ARPANET-connected computers looked like: "It was as big as a refrigerator and built like a tank. It also had practically no memory and no operating system."
