Chapter 7 - Data Communications: Market Order 1973-1979
7.8 Codex: The Statistical Multiplexer and Competition 1975-1976
In early 1974, after Carr had authorized the development of a statistical multiplexer, Vander Mey recommended a multi-microprocessor architecture. While probably biased, as his intellectual interest was multi-microprocessor architectures, the recommendation fit well with Codex’s engineering culture, a culture that prized the technically challenging. Only Vander Mey changed his mind as to which microprocessors to use. This led to delays and mounting stress.50
The one who felt the most stress was Forney. With more talent than he ever imagined he would have working for him, Forney began questioning his management abilities. Carr, however, would have nothing of Forney’s self-doubts. Forney didn’t want confidence boosts, but help in bringing an end to an “extremely painful development.” By the fall of 1975, he concluded outside review might bring fresh perspective and insight no longer possible on the part of his exhausted and over-worked staff.
Knowing Wesley Chu was co-chairing the Fourth Data Communications Symposium being held in Quebec City, Canada, that October, Forney attended hoping to talk to Chu about consulting. On learning Chu’s multiplexer was much more modest than Vander Mey’s ambitious design, he reluctantly concluded they were way beyond the point where high-level consulting might help. Vander Mey’s team did lack practical engineering skills, however, like knowing how to release a product to manufacturing, or knowing how to eliminate noise from their design. So Forney hired expert consultants when possible, still hope for success fundamentally rested with Vander Mey and his team. They forecast a spring 1976 manufacturing release.51
Based on a spring release, marketing kicked off the launch of the statistical multiplexer in early 1976. Management’s first goal was to stimulate customer interest in their innovative -and expensive - new product. A second goal was to train the sales force how to sell a product requiring customization, a far more complicated sale than selling modems off a price sheet. Codex began making customer presentations, leaving customers product data sheets and other information for review. By February, TDM competitors had copies of Codex’s literature, and while the expected price of $25,000 to $50,000 meant they posed no immediate threat to the much cheaper TDM’s, many competitors recognized that they too had to offer statistical multiplexers if they were to remain competitive.
By mid-1976, Codex was no nearer to shipping product than when they began marketing their statistical multiplexer earlier that year. Frustrated, Carr sought an outside opinion he could trust and hired Lowell Benski, a former engineering manager he had worked with at 3C and who was now a consultant, to conduct a program review. Carr remembers:
He came in and his final report was: It will take another year before you introduce this thing. I told him while he sat there: ‘It’s going next month,’ and he looked at me and said: ‘You are either incredibly ignorant or you have the biggest balls I have ever encountered.’ And I said: ‘If we work at a lab pace, it’ll be another year, or maybe longer. Does it work? You tell me, yea or nay, does it work?’ ‘Yes, it works, but it doesn’t have this and that and this part of it is shaky,’ and I said: ‘Then we go, and we will work night and day in some customer facility, if that’s what it takes. And that’s what we did.
In September 1976, Codex introduced the first two models of their 6000 Series statistical multiplexers: the 6030 and 6040. The Model 6030 supported 124 terminals and cost up to $30,000. The Model 6040, to ship May 1977, supported 248 terminals and cost up to $60,000. As Forney opines: “It was a fantastic tour de force in computer architecture; but in retrospect, we undoubtedly aimed too high.”
Carr may have hoped his biggest problem was behind him when he forced the Model 6030 out-the-doors. As things turned out he but it had been swapped it for a new, no less challenging, problem. Carr explains:
We changed two out of three regional managers that one year, small measures like that. The problem that we had anyway was this cultural thing about it isn’t a modem and those are computer things. Secondly, the very pragmatic thing is that they could sell modems like falling off a log, and the SM [statistical multiplexer] needed support help and you had to work with the customer and you had to configure systems and you had to bid it and the whole nine yards, and it was just too damned hard. We tried contest incentives and I got to what I called the disincentive mode. They wanted to know what that was, and I told them that for everything below quota they were on the SM, they lost commission on the modems. And I said: “Anybody don’t want to do that,” and these guys said: “Me,” and “Me,” and I said: “Good-bye, and good-bye.” We had quite a year, a big, big field turn over year.52
By year-end competition emerged, in part because Codex had given their competition warning in their early marketing to customers. Codex would prove up to the challenge. In mastering statistical multiplexing, Codex had established a second strong product line to its new LSI modems and had anticipated the emergence of a new driving force in data communications: networks.
- [50]:
Vander Mey began with Intel 4004s, switched to 8008s, then 8080s and next to Motorola’s 6800s and finally to Advanced Micro Devices’ 2900. They stayed with the 6800s.
- [51]:
Others who worked on this project who went on to distinguish themselves both at Codex and other companies were: Steve Finn (founder of Bytex), Jim Hart, Bill Tao (founder of Ztel), and Larry Krakauer.
- [52]:
Carr: “It was interesting, some years later, Infotron, which was in the mux business, went into the modem business probably ‘79 or ‘80, and their VP called John Pugh up and said: ‘Maybe you can tell me something. You’ve got these mixed product lines. Can you tell me how to get mux salesmen to sell modems?’ And John just lost it, because we had been at the end of three years trying to get modem salesmen to sell multiplexers.”