Ken Olsen died Sunday


He founded Digital Equipment Corporation, and was still running it when I worked there.

(ETA: Huh; posting from the phone reduced the photo and posted only that one version, rather than a smaller linked to a bigger.  Thus some of you may have missed  that I have Ken’s obituary up on the left monitor, and my old Minneapolis Branch DEC mug on the desk in front of the right monitor.)

Cannabalizing Your Current Business

In his recent article, Seth Godin quotes some executives as saying “After all, they argued, how could they wipe out their current business just to succeed online?”

His article bothers me just a little, though; it seems to take for granted that succeeding online is the only possible choice, that there is no alternative to online success. This seems to me too narrow a view, at least in general. (Seth’s article is primarily about the benefits of single-mindedness, and I have little to disagree with there; this is more a spin-off on another topic than a disagreement.)

The decision the business he was talking about was facing is in fact a hard one. Sometimes new paradigms, business models, products, marketing approaches, or whatever, succeed, pushing businesses relying on earlier approaches onto the scrapheap of history. But sometimes they fail, too, quite spectacularly. Netflix is a huge success, but DIVX (the non-returnable rental DVDs) were not.

If you have an existing business, and the new business model cannibalizes  your old business, you’re risking something when you decide to commit to the new model.  For a new startup, there’s no risk (the risk is born by the investors, really); you can go full-bore for the new model and try to make it work, knowing that that’s the only way for your business to survive. This, of course, is one of the benefits that single-mindedness can convey.

Of course, if the new business model does succeed, then you’re toast if you don’t pick it, too.

By some theories, Digital Equipment Corporation should have owned the single-user computer business. They made the best time-sharing computers (which was a way to give people the experience of using a single-user computer while actually sharing central resources; economically a good idea when computer power was terribly expensive), and their PDP-8 was largely deployed in single-user configurations in laboratories, industrial, and later business settings.  Lots of things we take for granted today, like file extensions and command-line options, I first met on DEC computers.  They had software for that world, and experience.  All they had to do was build cheaper hardware to put that software on, and move it out into the mass market.

But to seriously pursue that course, they would have had to cut the heart out of their current business.  If they failed to increase their sales by several orders of magnitude, they would have killed themselves by cutting their profit margin drastically (and to even try, they would have had to invest in high-volume manufacturing capability and re-engineer their hardware for mass-production). So, in the end, they didn’t try it. And the company which invented the minicomputer and which, when I worked for them in the 1980s, was the second-largest computer company in the world, competing head-to-head with IBM in business sales and often winning, ended by being bought by Compaq, a PC-clone maker that hadn’t existed for most of DEC’s lifetime (and which was failing itself; it ended up being owned by HP shortly after).

If DEC had made the other decision, would they have survived?  I really don’t know.  Quite possibly not; the change in engineering and manufacturing required, and hence probably in corporate culture in large parts of the organization, would have been overwhelmingly massive, and expensive. And the decision to try that would have guaranteed that they could not survive in just their existing markets.

At the time they had to make the decision, nobody really expected the microprocessor to become the main kind of computer, and nobody had any clear ideas what sort of market there would be for $5000 computers that were powerful enough for business use (PDP-8s were around $50,000 then). Visicalc was still in the future. Moore’s law was in effect, but nobody had quite made the connection that it didn’t really apply to the CPUs of old-fashioned computers built from wire-wrap modules (the interconnect speeds were going to surpass what any of the cheap reliable technologies could provide over those distances). The size of today’s computer market was simply not anticipated at that time. The PDP-8 is famous for many things, but one of them was being the first computer ever to ship more than 50,000 units (DEC distributed a poster about this at the time).

Buggy-whip manufacturers didn’t do all that well when the automobile took hold. The entire horse tack industry took a big down-turn, the total market for horse support gear collapsed when the horse population collapsed. But it was by no means obvious at first that automobiles were going to become important. That didn’t really happen until Henry Ford’s model T, at the earliest.

DEC made an attempt to play in the “workstation” market which developed in parallel with the early development of personal computers, but Apollo and later Sun pretty much kept them out of that market, and that market turned out not to last very long. They also developed one of the important RISC architectures, the Alpha, which was for a while even supported by Microsoft Windows NT (and note that the Alpha was a 64-bit architecture introduced in 1992).  Still, with slightly different luck and decisions, they might have made the transition to workstations and then to servers (where Sun is still holding on, though now as a division of Oracle; Apollo also ended up owned by HP). This was in some sense the first commercial wave of the single-user computer market, and it’s interesting that the big wave did not develop out of players in the first wave.

These strategic business decisions are often not at all obvious at the time  you have to make them.  Even some of the ones that are obvious in hindsight were quite obscure at the time they had to be made. The inevitable course of history is very often not clearly visible looking forward, only backward.