Just five years ago we buried Digital Equipment Corporation. This post-mortem looks at the incidents that led to Digital's demise, and also briefly looks at how voice processing finds it self in a similar bind. The intervening years have, of course, revealed that when Compaq ate digital, it ate poison, and Compaq itself never recovered from the downward momentum of all the lines of business it acquired with Digital.
Newsletter #14: Strategies work, until they stop working
Recent events in Washington illustrate a point: strategies appear to work until they stop working. While it did not result in a complete catastrophe, it is difficult to see boinking interns as a sustainable practice in the executive branch. Similarly, or perhaps just contemporaneously, the purchase of Digital by Compaq brings to mind once again the long and painful history of how Digital came to decline from the second largest computer company on the planet. Despite the dominance of the PC, it is startling to see Compaq, a company which “owns” very little of the technology on which it has built its tremendous prosperity buy Digital. Digital developed the computers that enabled my generation to see computers as a personal tool, cheap enough to deploy in a lab and manage with student labor. I was that student labor.
But enough nostalgia. The point is that you need to be able to tell when a strategy stops working in time to do something about it. It took Digital many years to go from the most vibrant force in computing, to a troubled giant, to a diminished once-great, to a potential alternative power in microprocessors, to their current end-game. The complex trajectory of Digital’s decline illustrates several ways in which strategies are evaluated more or less successfully. To take just two: How could Digital have prevented its greatest and ultimately fatal early mistakes, and did Digital take its best shot at renewal with Alpha and Windows NT?
Much apocrypha and irrelevant information, like Ken Olsen’s statements about personal computers, has clouded evaluation of Digital’s early errors. This is the bottom line: First, early in the game, when it would have been easy, Digital did not learn how to sell computers, PDP-11s for example, the way PCs are sold. Second, when the openness of the PDP-11, and the number of independently developed peripherals for it far outstripped what was available for the IBM PC, Digital did not enhance this openness and instead created a new, closed, patent-protected bus for its VAX line of computers, driving many potential allies in the market out of business. Finally, the failure to aggressively capitalize on and dominate networking technology, especially Internet routers, foreclosed the most likely path to renewal at Digital. It is particularly astonishing to recall that, at one time, the Internet consisted almost entirely of PDP-10s, PDP-11s, and VAXes.
It took three separate and massive errors to sink a company as big, powerful, and loaded with top talent as Digital. Two of these errors illustrate why evaluation of strategy is often difficult: Why not sell PDP-11s like PCs? When I first saw a PC, it was, by comparison with PDP-11s, a joke. Cheaply made and shoddily engineered. The motherboard was laden with “ECOs” – hand soldered connections correcting errors in the printed circuit board. Such things were rarely seen on anything Digital shipped. The 8086 processor was a caricature of the PDP-11 architecture. My first thoughts were “If DEC (that how we all though of “Digital” at the time) could make a small, cheap PDP-11, they’ll kill this PC thing.”
How naïve of me. Every mature man in a suit at Digital looked at the PC and thought “Not worth responding to.” If the IBM PC was so awful, why cut the margins made on PDP-11s? Here is the first reason why evaluating strategies in time to make a difference is so hard: every timely change means sacrificing safety. And big corporations are made up of herds of safety-seeking guys in suits who are, generally, in over their heads when it comes to seeing strategy. Not the best environment for making timely changes. The second reason it is difficult is the time-scale. All the people at Digital who made that fateful decision were probably doing something else by the time the consequences began to bite. Those consequences did not even begin to appear by the time the VAX was introduced, along with the Unibus successor, the SBI bus.
Here was a second chance: If Digital had failed to grab the emerging PC market, they could move their successful formula of open architectures up-market. Instead, they saw a chance to increase their profits and hold on to margins in the peripherals and enhancements business. Instead of opening the VAX’s SBI bus, Digital kept it for themselves, and left to independent peripheral makers only the slower Unibus. And it worked: Digital prospered while the companies that made the PDP-11 such an attractive computer for its flexibility and range of choices in peripherals died off. Nobody, I can confidently say without rummaging through Digital’s personnel records, got fired for keeping SBI bus proprietary.
And what to make of the muff in networking? Just as a cigar is sometimes just a cigar, in this case an error is an error. What can you say about blowing a chance where no safety is sacrificed, and nobody need risk their career? Digital could have been Cisco.
What can we learn from these errors? The first thing to learn is that the same sort of “situational ethics” that enable stuffed shirts to make bad decisions and then move before they are called to account can be turned against these poltroons. Use what you have at hand: Internet time-scales for example. Light a fire under your sluggish colleagues, and stoke it with fear that they will be caught out and embarrassed. Create an environment, through the use of press accounts of competitors’ announcements, where strategies are evaluated sooner rather than later. If the culture in your company is intractable, change horses or become one of the poltroons yourself. If you are in a leadership position, you have fewer excuses. The environment is what you have created, and if you have not created an environment where processes are transparent, you stand to lose exactly the sort of people who will be your ablest help in making important transitions. If your people don’t know what you are thinking, you are in trouble.
In our next example, things get more complex: Can good strategy fail? Yes, especially if starting from a hole that is too deep, into which rocks and debris are falling. So it was at Digital. Alpha is a better chip than the Pentium, and NT is destined to dominate enterprise computing. Trouble is, Microsoft did not get around to taking advantage of the 64 bit Alpha in NT, and Alpha itself could not overcome the space-warp economics of modern chip fabs. There Intel sets the pace, and sets it high enough to overcome any architectural advantage its competitors have.
What can one learn from an untenable situation? First, just as one needs to learn that apparent success may be the illusory, “locally optimal” result of a shortsighted decision, a losing position is not necessarily the result of bad decisions. One need ask what else could Digital have done? If there are no better options, they may have pursued the only reasonable strategy. Nine billion and change from Compaq may have been the best possible result. N.B.: this applies only to the question of strategy. Details of implementation, like the question of whether Digital cut back deeply enough quickly enough are outside the scope of the discussion. But to me it was evident: Driving on I-495 one could only ask “What the hell are all those people in all those Digital buildings doing? How many people does it take to make VAXes and VMS?” Now there are many fewer buildings so labeled. Now the signs say 3Com and Cisco and a large variety of smaller, nimbler companies.
I follow once again the pattern of developing a general idea and then mapping it onto telephony:. This time in focusing on the makers of voice processing cards. They face a decision point when it comes to strategy. They will feel interesting macroeconomic inputs that will warp the time-scale of decisions and results. The outcomes of decisions taken today will not be known for three years or more. Yet the fate of the voice processing card makers hangs on decisions taken today.
The first challenge voice processing card makers face is the temptation to do nothing. If nothing else happens this much is sure: the rest of the world will fill up with call centers and voice mail just as the U.S. has. This creates a huge temptation to do nothing. Do nothing and you will surely grow. Do nothing and you risk nothing. This temptation is reinforced by the face that efforts to break out of the core markets and proprietary habits of voice processing card makers have in relative terms not amounted to much. Technologies like TAPI face a credibility gap when it comes to delivering money into the pockets of voice processing card makers.
To overcome this challenge, management in voice processing companies must make the strategic process open and collaborative. Otherwise even the best lieutenant will just keep his head down and ride the “rest of world effect” to success. Only if made clear at a leadership level that the inevitable expansion of current markets reached by conventional means is not enough will a better result be reached. And only if resources are ventured on obtaining a better result will it be realized.
The second challenge voice processing card makers will face is that time frames for results from alternative strategies are largely unknowable. The right strategy may fail once or twice before it succeeds. Only by making clear at the leadership level that pursuit of the right strategy is the right thing to do even in light of uncertainty over the time frame in which it will succeed will you get anyone but the most naïve and foolhardy to implement that strategy. You can easily see the potential for failure to spiral out of control in that picture. Conversely, if a company’s leadership recognizes good strategy and good implementation, and is able to sustain efforts while making such wise evaluations without the external verification of immediate market gratification, then leadership will draw the following needed to make such strong implementations happen.
Forgive me if I do not go into the specific strategic choices facing voice processing card makers. Time and space have grown short. But I will make the following observation: it will be fascinating to watch the result. It is likely that one of the top tier companies will lose their position due to bad strategic decisions made at this point. There is also the opportunity for a second-tier company to vault up in the next few years based on the right decision. So watch the initiatives these companies make this year. This year’s decisions will determine the outcome of a shake out that will happen as telephony changes due to the impact of the Internet.
Copyright 1998 Zigurd Mednieks. May be reproduced and redistributed with attribution and this notice intact.