Page 2 of 2
ALOHAnet at Parc
Back at Parc the number of machines was growing and it was becoming clear that existing methods of connecting them, together wasn't practical. A local area network was needed although not even the terminology was invented back then. Metcalfe's thoughts were that something like ALOHAnet was needed because it provided a way of sharing the communications channel - the radio signal propagating in the "aether". The "lumininferous aether" was the imaginary medium through which radio waves and light propagated. There was no need to use radio communication within a single building and so Metcalfe started to think about ways of sharing a single cable that all of the computers could be connected to. The cable would take the place of the "aether". Sharing a cable efficiently isn't an easy problem to solve. At the time people were already thinking of ways of doing it - for example, the first token ring systems were being developed. A token ring network works by circulating a packet which acts as a key to using the network - if you have the key you can transmit if you don't you can't. Token ring networks are efficient but complicated and a break in the ring stops it working.
From ALOHAnet to Ethernet
The key to making the ALOHAnet system efficient occurred to Metcalfe in 1973. Using cable allowed each node to listen to see if the cable was free before transmitting. Each transmitting node could also detect that another was transmitting at the same time and abort before too much time was wasted. and a "back-off" method could be used to cope with congestion when the network was heavily loaded. He wrote a memo containing all of these ideas and called it the ETHER Network - this name quickly became shortened to Ethernet. All he now had to do was to make it all work.
He went down to a basement in the Parc building and started to experiment with 1000ft reel of coaxial cable. It soon became clear that he needed help. Metcalfe knew about digital electronics but the analog circuits needed to drive bits down a coaxial cable at the speeds needed were beyond him. Then he met Dave Boggs. He was an electronics student and enthusiast employed by Parc to commission minicomputers. Metcalfe quickly realised that he was being under-utilised and Boggs joined the Ethernet project.
"Dave and I were called the Bobsey Twins in 1973/74 as we designed, built and operated a 100 node experimental Ethernet. We used to work until we were exhausted and would sleep until we woke, without regard for alarm clocks or the sun."
The experimental Ethernet was fast for the time - 3Mbps compared to the ARPAnet’s 50Kbs. Another innovation was the use of "hot taps" adapted from Closed Circuit TV systems. Using a hot tap you could drill a hole in the Ethernet main cable and insert a needle probe to make a connection without stopping the network running. Eventually they managed to produce Ethernet adapters for the Alto, the standard personal computer used at Parc, but they were an optional extra. Many users saved budget by ordering cut down Altos - after all who needed a network? The Alto came with a large removable disk drive that was ideal for transferring data. As the only high level software working on the Ethernet at the time was file transfer Ethernet was up against "sneaker net" from the word go. (Sneaker net is the US jargon for a network implemented by running from machine to machine carrying floppy disks!) At around the same time Metcalfe defined a set of network protocols - Parc Universal Protocols (PUP) - this influenced the design of TCP/IP and eventually evolved into the Novell IPX protocol.
The rise of Ethernet within Parc was surprisingly slow. The killer application for it was also surprising. When the laser printer was invented, yes another first for Parc, there was suddenly a need to transfer megabytes of data and fast. Ethernet was ideal and slowly but surely it became an essential rather than an extra. It may have become an essential within Parc but outside of Parc people still asked the question "why do we need a 3Mbs network connection?" The answer seems obvious today but then there were very few computers which had large capacity storage devices - a 1MByte floppy was a typical storage medium and Ethernet could transfer its entire contents in just over 2 seconds. When Metcalfe turned up at conferences to spread the word about Ethernet people got up and said that 1200 baud modems were overkill as far as data transfer was concerned!
In 1979 Metcalfe left to found 3Com the company that finally sold Ethernet to the masses. Dave Boggs stayed on at Xerox Parc but still managed to make a million dollars speculating on the stock market.
In 1982 the board of 3Com decided that Metcalfe could no longer be the chief executive. He described his decision to stay on with the company he founded as the most difficult thing he had ever done. The temptation to storm out and cause a fuss was strong but he worked hard at seeing the sense in the fact that someone else might be better at managing the company at its new size. The company kept on growing and with Metcalfe in marketing it seemed as if Ethernet’s time had come. The final straw was when IN 1990 the top job came open again and the board once again appointed someone else.
For the next 10 years Metcalfe became a publisher and columnist, writing about the Internet every week in InfoWorld where he was also an executive director. He achieved a degree of fame for his dire prediction in 1995 that the following year the Internet would suffer a complete and catastrophic crash. When this didn't happen he kept his promise to eat his words and during his keynote speech at the Sixth WWW International Conference in 1997 took a copy of the column put it with some liquid into a blender and drank the result. Strange that now the inventor of the Ethernet should be better known as a prophet of doom for the Internet.
In November 2010 Metcalfe was selected to lead innovation initiatives at The University of Texas at Austin Cockrell School of Engineering, an appointment he took up in January 2011.