Networking: Turbulence 1981-1982
The PBX, the IBM PC and the Chaos of Competition
8.5 Concord Data Systems
In 1979, Ken Miller, Director of Modem Development for Codex - the largest modem company in the world - wondered if this was all he was meant to be, an employee addicted to the safety net of a large company, deadened to the excitement that once made coming to work meaningful. Somehow in the process of becoming large, Codex’s burgeoning bureaucracy had stifled the boldness, and yes fun, of the cash-strapped, "let's go for it" company Miller joined in 1972. He recalls:
"I got frustrated. A lot of people at Codex got frustrated then because they were used to the early days, the early go-go years when you could get things done easily."
He wanted to work in a small company culture, a culture of deciding and acting, a "make it happen" environment. He wanted to be like those who had already left Codex and successfully started their own companies: Jerry Holsinger, founder of Intertel, and Jim VanderMaye, Integral Data Systems.
Miller’s frustrations grew from his inability to convince his superiors to enter a new product market. When promoted to Director of Modem Development, Miller became Codex’s representative to the CCITT modem standards committees. There he learned of a new 1200 bps dial-up modem standard -- V.22 -- being established. Since no firm then sold a V.22 modem, Codex had as good a chance as any other firm to dominate. For over a year he had been returning from CCITT meetings, filled with entrepreneurial enthusiasm, committed to selling Codex management on the opportunity to expand beyond their exclusively lease-line modem focus. But every time his efforts were thwarted, slowly buried in successive rounds of meetings and never-ending memos. To Miller it made no sense and begged the question of why he remained at Codex.
The proximate cause to consider leaving Codex, however, was the company's recent move from Newton to Mansfield, MA. A tolerable commute had become downright ugly.
"I didn't want to move, especially if I was uncertain about Codex. But then, I mean, I always liked small companies. I had a secret wish that Dave Forney would go start a new company so I could go join him, but -- and at the time I didn't really -- I had always had a little bit of a dream of starting a company, but never really had the guts to do it!"
Then, as if a prayer had been answered, he received a phone call from a former Codex engineer, Mike Kryskow, now working at Gould-Modicon. Miller remembers the catalytic question:
"Hey, we're working on some future networking stuff, and it's really communications, and we got all these consultants in here and they don't know what they're doing and we need some help. Could you consult?"
In the past he wouldn’t have given such a proposal a moment’s thought, but then he had never considered leaving Codex before. Surprising even himself, he said: "O.K., I'll come up and look."
Miller consulted over the 1979 year-end holidays and knew he had not been misled. There just might be life outside Codex and products other than modems. After the first of the year, he made calls to friends in other companies and discovered numerous consulting opportunities and soon gave notice at Codex, leaving in the middle of March. Miller remembers:
"That was the most gut-wrenching thing for me to do, because I'm leaving a salary. I had a wife and kids and a mortgage and everything else and I mean, I would have to bring the bacon in and I didn't have a job. I had to go scrounge it up. But the opportunities that came by were just amazing."
In March 1980, Miller formed Miller Associates. He continued consulting to Gould-Modicon on their efforts to create a communication network able to meet the rigorous real-time needs of manufacturing; a network interconnecting factory floor equipment with the computers controlling and directing manufacturing. Miller also learned that Gould/Modicon was not the only company thinking about networking computers and peripherals. Kryskow, however, had devised a new scheme, to become known as token bus, and Miller, convinced of the unique needs of factory automation, became thoroughly "enamored with token bus." Token bus differed from token ring in topology: the token is broadcast over a bus instead of passed in a ring. The broadcast token contained only one address, that of the next station authorized to communicate, the token address then would be updated at the next station similar to the mechanism of token ring. Miller believed the deterministic behavior of token passing crucial to factory automation.
Within months, Miller came into contact with a whole new community of local area networking believers who reinforced his sense that something special was happening. Miller recalls:
"Now, back then, local networks were really in their infancy. So I looked around and said: 'There's opportunity here. All these people participating are computer people. This isn't a computer problem, this is a data communication problem.'“
The computer people he met kept talking about Ethernet. But for Miller, Ethernet had a serious problem: it could not guarantee an access response time, because it was based on probabilistic access. Furthermore, the longer the medium, the greater the propagation delay which aggravated indeterminate response times. An uncertain response time might be acceptable for office automation, but Miller knew it had no chance of satisfying the real time needs of factory automation. Token bus, on the other hand, could guarantee response times. Assuming this insight was one the computer-types had glossed over, Miller saw a golden opportunity to transform Miller Associates into a product company selling token bus.
However, Miller had so much consulting work he had no time to devote to token bus. Uncertain as to what to do, Miller convened a small group of friends whom he thought might be interested in joining a start-up for a planning session between Christmas and New Years, 1980, at the Colonial Inn in Concord, MA. Not attending the meeting, but critical to being seen as a credible token bus company, was Kryskow of Gould-Modicon who had committed to join Miller’s new company. He remembers:
"I mean I came up with a basic approach which we talked about. It was a three or four hour meeting. It was a basic strategy. We said: 'Hey, look. We want to do token bus LANs, but that's a long-range thing, I mean, we're committed to standards, and a standard doesn't exist, so one of the things we have to do is get it through the standards committee. There's a lot of work. It's a system-oriented product. But we also have identified this other thing that we know a hell of a lot about. It's a market that's here and now. It's dial modems. And so, let's do both. There's a short range strategy, get the company going, and the longer range one which we thought, at the time, would be a higher growth field."
They encouraged Miller to write a business plan, a prerequisite before meeting venture capitalists. After months of drafts and redrafts, Miller began meeting with anyone who would see him. He remembers one surprising presentation:
"My neighbor knew another person who knew Tom Stevenson who was President of Fidelity Ventures. I went in there. That was very interesting because I was preaching the gospel of token passing. I'll never forget that day because I went in and was telling him why token passing is better than Ethernet and this guy is giving me all these technical arguments about Ethernet. I said: ' How the hell does this guy know this?' It turns out he had just gotten aced out from investing in 3Com and he had been filled with all of Metcalfe's propaganda.”
Realizing he needed help, Miller recruited a financial consultant, John Plakans, to help him polish his plan:
"Well, it turns out that Plakans knew Kevin Landry of TA Associates, the venture capital firm. So we got an introduction through him, and Kevin and Andy McLane came out and visited. My neighbor was employed with Analog Devices Enterprises so I also talked to them, among other venture capital sources. Lo and behold, we got offers both from TA and Analog Devices.
Analog Devices said we'll only give you a certain amount of money and then you've got to have a progress payment and, if you meet some milestones you get more and I said: 'Ah, screw that shit. Who knows, this is a volatile business. You could end up changing direction. I don't want to get too constrained.' And also, it was a clear direction. If you were with Analog, you were going to get acquired. So we went with TA and closed $1.4 million in May, 1981.”
A potentially devastating confrontation at a company meeting happened soon afterwards. Miller remembers:
"I got a bit panicked because Mike Kryskow was going bananas! We had a company meeting one time and he said: 'You're not flamboyant enough to be President. You'd be a good VP of Engineering. We need to go get another President.' I said: 'God damn it, NO!' And the guy finally didn't join.”
Knowing he needed a token passing expert, Miller then recruited Tom Phinney of Honeywell to replace Kryskow. Phinney, however, would not leave Phoenix, AZ. Miller, feeling he had little choice, caved and agreed to locate Concord Data Systems’ LAN expert nearly a continent away.