Disorderly Content


All The Wood Behind One Arrowhead

Many years ago, back when I still worked at Sun Microsystems and Scott McNealy was still in charge, he had what turned out to be a very bad idea: to turn the company that put All The Wood Behind One Arrowhead into a more decentralized set of self-contained business units. The idea was to replace supposedly strategic decisionmaking within individual units with choices that made not just sense, but dollars and cents. No more "We're doing this because it's right for the company." Instead, it was to be "We're doing it because the money we spend will come back tenfold," or more, or whatever the factor was determined to be.

Why was this such a bad idea? First, it reduced efficiency by a whole lot, as each unit had its own marketing and sales and advertising, mostly targeted at the same set of Sun customers. Second, it make decisions very narrow, as my own developer tools business unit concentrated on eliminating competition for software development on Sun, rather than on increasing the success of Sun as an application development and delivery platform. And third, it forced these new business units to take on agreements from the old days, agreements like a third party Cobol compiler that was a money loser for the development tools unit but which was intended to (and did) make lots of money from server sales.

Two years later this business unit model was undone, as were many of us who had participated in it. It was acknowledged to be a bad and very expensive brain fart on the part of our CEO, and became an entertaining part of our shared history. Buy me a drink some time and I'll tell you of some sadly funny examples of what seemed to someone like rational decisions.

I mention this because something reminded me that the business unit idea wasn't entirely a bad one. That something is Microsoft's late and not terribly lamented Kin mobile device, the last gasp of their acquisition of Danger, inventor of T-Mobile's Sidekick. The Kin took two years to develop, and lasted about a month and a half before being shot at sunrise. How, one wonders, can any organization, even one so famously screwed up as Microsoft, get a product so wrong that they have to kill it so quickly?

There are many answers, as there generally are, but one big one is that Microsoft still subscribes to Sun's All The Wood Behind One Arrowhead philosophy. With Sun that arrowhead was its SPARC processor; with Microsoft it is Windows. And one of the reasons Kin took so long to arrive, and so missed its moment when it did, was the corporate decision that running some form of Windows was more important than getting out to customers. Two years to rewrite the Java-based software that came with Danger, to fit a corporate philosophy of Windows Everywhere. Even, as in this case, if Windows provided no benefit. Even if it wasn't going to matter, or even be visible, to the device's user.

In short (yeah, I know: too late), Microsoft sacrificed its Kin on the altar of Windows. As it is doing with phones, not that anyone's left to care. Sun abandoned SPARC as its only processor, although that decision came too late, and after other developments made them less and less relevant. I expect Microsoft to come to the same conclusion about Windows, although it's likely already too late. And yes, I know that Apple has its mostly unified OS strategy with Mac OS X and iOS. Apple seems to know how much to share and how much to make different in a way Microsoft never seemed to learn. Or maybe it's a matter of taste, which Apple has.