These are further thoughts I've had on my Novell experience. They are not well organized, but they may provide more insight. Most of these thoughts cover the transition from visionary to ordinary organization. I consider myself part of the visionary camp. I left in 1989, and I was sore disappointed when I felt it was the right choice. The Novell experience of the eighties was a wonderful ride, and I wanted it to go on and on.
The following are insights that may help others facing the challenge of keeping a visionary organization visionary.
The most obvious challenge of Novell after the revolt was surviving without a visionary. There were other candidates: Drew Major, Darryl Miller and Kanwal Riche were all on board in 1989, but each had weaknesses. Drew was articulate and understood the technology, but he was in the middle of birthing Netware 386 and was indispensable for that. Darryl was articulate, but he was a self-professed information broker. He was better at picking and choosing from the stream of concepts that flowed by him than at generating his own. He could articulate, but he couldn't distinguish good ideas from bad, feasible ideas from unfeasible. Kanwal Riche was just on board and came from a very different background. He was conversant in TCP/IP issues and minicomputer industry issues, not Novell LAN issues or PC industry issues.
In the end there turned out to be no replacement for Craig. Instead the company changed so that it didn't need a visionary.
The departure of Craig and Judith was a social upheaval, a revolution, for Novell. In most social upheavals the disruption after the historic event far outweighs the disruption of the event itself. The French didn't start guillotining people until months after the storming of the Bastille. The biggest emigration of boat people from Vietnam didn't occur until years after the Viet Cong had taken Ho Chi Min City. And so it was at Novell. Craig and Judith left the first time in November of 1988 and for good in March of 1989, but it took the rest of 1989 and 1990 for Novell to define what the post-Burton/Clarke Novell would be like.
For months the top managers left at Novell set about reorganizing. The political crises leading up to Craig and Judith's departure were hard on the company and paralyzed earnings. In April Ray took this as an opportunity to be the company doctor again and trim ship. The victors took Ray's announcement as an opportunity to consolidate alliances and settle old scores. A "meeting with no agenda" for top management was held at the Homestead Resort in nearby Heber. The managers made their lists and checked them twice while humming Christmas tunes about Santa Claus coming, and in June and July about 10% of the company was laid off. Sales was hit. So were Developer Relations, Communications and Marketing.
Those things tainted with Craig's or Judith's blessings became suspect. Stratifying Netware 386, Novell taking the initiative to support Microsoft's SQL server and Macintosh support for the 386 went to the back burner.
The LAN Times was sold to McGraw Hill. This lemon-to-lemonade act solved most of the "Judith problem" in one stroke by ripping out her work and people simultaneously. The rest of her work was scattered among the various Novell groups, and Novell MarComm thereafter restricted itself to approving ads and putting out brochures and other sales aids -- all controllable, and all within budget.
One of the fundamental choices that must be made in marketing a product is whether to pursue a lower-margin, wider-market strategy or a higher-margin, narrower-market strategy. Players in the computer industry face this challenge almost daily and the "smart money" is to bet on the higher margin, narrower-market strategy. There are many reasons to support this choice: a smaller market is easier to measure and predict, if the product is more popular than anticipated there is bonus profit, and high margins look good on the balance sheet.
The only problem with this habit of seeking higher margin is that periodic "zingers" come along to prove that low-price, high-volume strategies succeed too -- and when they do they succeed in a much bigger way than the high-margin strategy. One example is personal computers: a product that brought computer appeal to a much wider audience than any computer analyst of the seventies deemed possible.
But the personal computer companies weren't immune to this pursuit of higher margin either. Once a personal computer company became established, it started "margin creeping" -- moving to higher priced, higher margin products. One way this showed up was moving from making a computer for home and business to making a predominantly business computer. Apple is an example of a company that made this transition successfully.
Novell of the 90's faced this issue and responded in an average way: most products were higher margin, but occasionally a "trend bucker" is introduced. In 1988 Novell announced that Netware would be stratified and that one of the releases would be ELS Netware (Entry Level System). ELS was a big seller, but it was also a Craig Burton idea, and that put further efforts in the low price regime into limbo for three years.
Craig and Judith's departure marked the end of Novell's cooperator-cooperator era. There had been cracks in the cooperator-cooperator facade from the beginning, but in the beginning they were small and kept small by the intensity and clarity of the company vision. The cracks grew in 1986 and 87 as the political battles intensified, but with Craig's and Judith's departure a lot of pretense at cooperation ended. The victors had won by defecting and they knew it. They now set about institutionalizing measures to see that their subordinates and associates couldn't do the same things to them.
The key tool for defectors is control of information flow, so security was tightened and surveillance was heightened. One sign of the change was a series of memos issuing from Ray in the first and second calendar quarters of 89 in which the words "grounds for immediate termination" and "immediate dismissal" appeared for the first time -- an example being a memo concerning signing on behalf of another employee issued in May. The magnitude of the change afoot was enlarged when similarly worded pronouncements, but with much broader sweeps of authority, started appearing from second and third level managers. An example being a memo issued in May by Jack Blount -- then a director of Developer Relations-- informing his Developer Relations staff that sending out a communication from the office before he had a chance to review it was grounds for immediate termination.
Another step came in July when Marcomm started insisting that all communications between with the media be done through Marcomm or in the presence of a MarComm person.
These were small signs. There were bigger ones as well -- signs that a Novell built around a vision of information control was being hammered out. In Developer Relations Jack Blount described this new vision as a department filled with director-level MBAs who would "control their accounts" by acting as liaison between internal engineers and marketing people and external developers. Whatever passed between the inside and the outside, Developer Relations people would have their fingers in.
Control is good, but it's not without it's costs. Lets take a look at these two developments in the Developer Relations situation to see the tradeoffs.
The goal of Jack's review policy was to avoid blunders: if one person -- Jack -- saw all the communications, then in theory he would know all that's happening and be able to catch blunders, miscoordinations and misstatements before they left the office walls. The implementation, on the other hand, was not so easy. Jack's business style of that time had him commuting back and forth between Dallas, Denver and Provo -- as well as calling on critical developers in the Bay Area and the east coast. This meant he was out of the office 80% of the time, and that getting out a communication involved:
This control requirement added an overhead cost to the communications process. Communications slowed and became uncertain. When Jack wasn't available -- asleep or in a plane -- communications halted entirely.
This cost changed the range of suitable communications subjects. Spontaneous and serendipitous inquiries became too high-risk to justify sending through this expensive communications process. They became "time wasters" that clogged the channels. Items of known importance -- items that Jack would process quickly -- squeezed other agendas out of the channel.
Jack's vision of the liaison process is also more cumbersome than direct communications. Since the liaison set-up is common in other companies, it means that the developer-to-developer linkage between Novell and a company such as Lotus tended to look like this:
developer <-> <-> developer
developer <-> liaison <-> liaison <-> developer
developer <-> <-> developer
This situation puts a lot of technical information flow through business people and these business people represent filters between the developers. Communications will be slow and there will be the distortion caused by a message passing through many mouths connected to many minds, each with a different frame of reference. One more thing adds to this message distortion: these liaison people are high-priced. They aren't paid to be just messengers, they're paid to be smart -- to add their input.
In the years from 86 to 90 Novell moved from being an exceptional and highly profitable organization to being an average and highly profitable organization.
The organization became decentralized. Company vision became a matter of having business-wise managers picking and choosing between various technological options.
Personal relations became "competitive" in the personnel usage. Competitive is a word used by personnel people around the nation to describe their compensation package. It means "thoroughly researched: and no better than it has to be to attract the kind of people we need." Personal relations of Novell in the 1983 to 1988 period weren't competitive, they were cooperative. They were so special that ex-Novellers of that period formed the LAN Group (Life After Novell) and called themselves alumni. The post-89 Novell offered only embers of that warmth and enthusiasm. It was now a competitive organization.
After 1989 Novell as an organization transitioned towards "gray" and tidy -- but there was still one founder left to annually pull surprises out of his hat.
In the spring of 1990 Jim Manzi and Ray Noorda committed an act of true industry leadership: To the surprise of most people in both organizations they announced an intention to merge. This was an act of leadership precisely because it was so unexpected -- neither man was "going with the flow" of consensus either within in their respective organizations or within the industry as a whole.
Some good reasons for the merger were put forward, but industry analysts, virtually to a person, predicted a rocky relationship. Jim Seymore in PC Week pointed out the problems of culture clash. Others pointed out that there were a lot more non-fitting parts than fitting parts of the two companies. Craig Burton pointed out in the Clarke Burton Report that this merger would polarize the industry and that would slow the development of intercompatibility.
For Ray watchers the announcement was less of a surprise. Lotus was a linear extension of Ray's acquisition policy. In 1985 he started acquiring small companies such as failing distributors and $6 million Cache Data. In 1986/87 he was adding larger companies such as CXI. In 1988 he added Exelan -- a company half the size of a now much bigger Novell. Lotus was another step -- this time one slightly larger than Novell. Ray was to be co-chairman and would still hold a very large chunk of the surviving company -- much larger than Jim Manzi's, his co-chairman. Lotus/Novell might not be his company to start with, but he surely had a chance to make it that way in a few years.
He might have... but as the weeks passed and the time for final signing approached, the uproar in the industry increased rather than diminishing. The polarization Burton spoke of started almost instantly. Others were still speculating darkly on how this fit would come about, and a group of stockholders put their money where there mouth was by initiating a law suit asserting that this merger wasn't feduciarily responsible for Novell to do. Novell should get more.
And finally, the issue of control raised it's ugly head once again. The weekend before the signing Ray announced to Lotus management that his board wouldn't agree to the deal unless they were equally represented on the board of the new company -- the agreement as it stood called for a five/four split. Manzi backed out at this final demand and this act of true industry leadership went sailing into oblivion.
There are still mysteries surrounding this episode. (Why did this board -- the same board that has consistently and quietly supported Ray since 1986 -- balk at this deal? Was it a face-saving measure for Ray? If so, why did he choose to use them this time? Why was Ray content with a five/four split to start with, then get cold feet? Did he discover something about Jim that raised the importance of the control issue? Was it really Jim that backed out, but somehow maneuvered it to make Ray look like the backing out party?)
One outcome of this change of plans was that it knocked Ray of off his "roll". It hadn't cost him his place in the company the way Harry's, Craig's and Judith's miscalculations had, but it was just as much a signal for future change.
Having gotten an eye blackened by being a leader in 1990, Ray was more careful in 1991. Novell, Lotus and Microsoft had also changed a lot in the 1990-91 time period. Microsoft was close to demonstrating that it was wearing the Emperor's New Clothes. It finally fell out with IBM due to the lackluster technical and market performance of Windows and OS/2. At the beginning of 1990 the workstation operating system marketplace looked like a monolithic DOS-Windows-OS/2 playground. By the beginning of 1991 it looked badly fragmented and a whole host of industry players were noticing the difference.
Digital Research wasn't as big as Lotus, but it offered a lot of experience relating to this new workstation operating system opportunity. Novell decided to use Digital Research as it's entree for providing a "complete solution" to user's networking needs -- clear from the screen to the file server.
In the short-term this merger makes sense. But as a strategy it's not quite so attractive. It's yet another step away from open systems approach, and a step closer to the old minicomputer industry practice of fortress building. Novell is building a self-sufficient software fortress. The problem isn't so much Novell's relation to Microsoft as it is to all the other operating system vendors that would pop up as the industry moved "beyond DOS". These others would see Novell as well as Microsoft as competition.