Metcalfe on Ethernet’s lessons, unsung heros

The Crane connection

Another drag, another name. Ron Crane, another Stanford grad student. Metcalfe hired Crane for his Xerox team and later recruited him as a hardware guru at 3Com, the company Metcalfe built to take Ethernet to the market. “Crane led the design of all our early Ethernet products and helped develop the IEEE standard” along with many others, Metcalfe recalled. “The list starts getting long now,” he said.


Click on image to enlarge.

Indeed, hundreds of engineers have helped drive various flavors of Ethernet through standards bodies, into products and out to the market. It all began when PARC hired Metcalfe to find a way to network the Alto microcomputers it built in the days before the PC. He wound up unseating another engineer who previously had that charter, Charles Simonyi.

In a twist of fate, “when I came Simonyi switched to developing a text editor which over time evolved into Microsoft Word and made him a billionaire who has since visited the International Space Station twice,” said Metcalfe, taking another long draw on his cigarette.

In his 1973 memo, Metcalfe outlined his thoughts on Ethernet. He conceived a variation of the Aloha Network at the University of Hawaii he studied as part of a Harvard doctoral thesis. It sported two innovations — an algorithm that uses data collisions to minimize interference and the flexibility to run over any medium, even the mythical Ether for which he named it.


Click on image to enlarge.

There were plenty of fights along the way in IEEE standards groups with giants like IBM and even General Motors who tried to define their own networks--and fights in the market trying to establish those technologies and others.

In the end, Metcalfe’s concepts became the basis for local area networks.  Ethernet won in part “due to the sincerity of our efforts,” as opposed to the motivations of other proposals trying to protect legacy businesses, Metcalfe claims.

Perhaps more importantly, “Ethernet understands its role in the comms hierarchy, it was a Layer 1-2 protocol and didn’t try to do more than that,” he said. “As a result it is simpler and cheaper than the alternatives because it only did want it needed and not more,” he added.

Ultimately Ethernet was adopted as the basis for Wi-Fi and a wide range of nets that now span everything from industrial controls to optical telco core networks where it is replacing Sonet. It has evolved from 3 Mbits/s to 100 Gbits/s. Last year, vendors shipped 1.2 billion Ethernet connections, 800 million of them wireless ones, according to International Data Corp.

“It’s not even close to the same technology — CSMA/CD [Carrier sense multiple access with collision detection, one of Metcalfe’s early innovations] isn’t even in it anymore, but the name is used to embody a bigger idea,” Metcalfe said. “It could be the packet format or the business model of a de jure standard with fierce competition and backward compatibility -- that business model is I think what people mean by Ethernet, today,” he said.

Last year, Metcalfe took up a new job as Professor of Innovation at the University of Texas in Austin.

“My mission is to help make Austin a better Silicon Valley, and the core of that is an engineering school where I’m focused on helping engineers who want to start companies,” said Metcalfe. “If you are interested in starting a company, it’s my job to show you how to do it,” he said.

We asked if there are any unsung up-and-coming engineers he wanted to name.

“I am rooting for them all. There are plenty of people who will discourage them,” he said taking a last pull on his cigarette.