Ken Olsen, the founder and long-time CEO of Digital Equipment Corporation (DEC), died Sunday. It is a sad day for hundreds of thousands of DEC alumni.
As a grad student at MIT Ken worked on Project Whirlwind, the world’s first interactive real-time computer system. It was also the first computer built by engineers rather than scientists with features like built-in test circuits and standardized modules.
In 1957, with $70,000 in VC funding, Ken started DEC. Computers were then a money-losing business, so they left the word “computer” out of the name.
My work with Ken
I joined DEC in 1981 at a great time. The VAX 32-bit supermini computer was the hot FORTRAN box – 5 MHz processor, 2 GB user address space, and a screaming 13.3 MB/sec backplane – and soon DEC announced Ethernet and then VAXclusters.
My KO time began in 1987 when I became the Manufacturing Automation Protocol (MAP) marketing manager. Three weeks later Ken told the Financial Times that MAP was “built like a Russian truck.” In short order I became the Industrial Ethernet marketing manager.
MAP: the network no one needed
MAP was a lesson in how niche markets go wrong. The vision was for a network that could cover huge factories with a deterministic protocol, driving robots across the network. And it needs to be EMI proof and handle huge amounts of factory floor data.
These perceived needs led to a new network architecture: broadband token bus. Yes, an analog CATV network – with headends and all the analog headaches – customized for the factory floor.
The industry and some VCs were going along with the MAP fantasy to the tune of about $100 million a year. Only KO had the smarts to call it out.
The fun part for me was that due to KO’s interest I was given carte blanche to market Ethernet to manufacturing. The #1 problem was that no one had validated MAP claims or Ethernet’s suitability for the factory floor.
One example: Digital promoted Ethernet by saying a single ethernet could be 13 miles long. But manufacturing didn’t care about distance: they cared about how much area a network could cover.
It took about 20 minutes to configure a maximal Ethernet network and I found it could cover an area of over 4 sq. mi. – larger than 99% of the world’s plants. In 5 years of Ethernet marketing no one had done that.
MAP’s other “advantages” had about as much substance. Factory floor data volumes were minuscule. Robots had local controllers – duh! –
and thickwire Ethernet was so over-engineered it would work anywhere EMI wasn’t burning people alive.
KO proposed a demonstration of Ethernet’s EMI resistance: use a Van de Graaff generator to generate huge sparks around an Ethernet cable. He even whiteboarded some differential equations to be sure it would work. It was a huge success: even non-techies got the message.
KO the terror
KO was kind to less senior employees. But he could be brutal with senior management.
The much-liked network group VP, Bill Johnson, was to give a conciliatory speech to the MAP User Group. But at 1AM BJ got a call from KO, who dictated the key points of a more confrontational speech.
I was tweaking the speech with BJ when the call came and I was surprised at Ken’s energy around it. And nobody in the MUG liked the new speech.
After 30 years at the helm, Ken’s vision failed him. Senior people urged him to open-source DECnet to head off TCP/IP; to move VMS to x86 to beat Microsoft; and to adopt UNIX as a core DEC technology.
But he just couldn’t see the sense of it. DEC had invested millions to create superior products and he felt they should get a return. So he moved to emulate IBM instead.
DEC hired a lot of IBM’ers who didn’t get the DEC culture. And squandered a billion dollars on the VAX 9000 mainframe fiasco.
When the PC juggernaut hit DEC, after wiping out DG, Wang and Prime, the company couldn’t reposition. KO got fired and a clueless new CEO accelerated DEC’s decline.
The StorageMojo take
But Ken’s missteps came after 30 years of leading a tiny startup to become the 2nd largest IT company in the world. He created a unique culture with a sense of mission that valued creativity and experimentation.
Very few entrepreneurs have done what Ken did in building DEC. Partly it was luck – interactive computing was and is a big trend – but he was also able to remain grounded and open to the possibilities over decades of hard work, long hours and unbelievable success.
But perhaps the greatest monument to KO is the affection for him that suffused DEC. That is why so many of us are sad that he is gone.
Courteous comments welcome, of course. Here’s the New York Times obit.