Early in July, CEO Steve Ballmer sent out a memo to each of Microsoft's 57,000 employees stating that the company will slash annual costs by $1 billion following a sharp rise in spending over the past three years. Those of us who have bought information technologies from RCA, GE, Honeywell or Amdahl are rather sensitive about such announcements. After serving as Xerox's CIO through the ups and downs of that stellar firm, I am particularly touchy about them. Are they harbingers of decline, or are they reassertions of vigor?

One way to judge a company's viability is to check out trends. From 1999, when Microsoft's price per share peaked, through 2003, when its latest financial report was published, several of Microsoft's costs grew faster than its revenue: Overhead costs grew by 15%, the cost of goods by 34% and employment by 12%. Meanwhile, growth in profits was 37% lower than revenue growth. Actually, that lag in profits is worse than it appears because 15% of the revenue was from investment income and not from sales of software. The numbers confirm that Ballmer was right to institute massive cost reductions.

There is another way to check Microsoft's performance. The Economic Value Added reflects the enormous accumulation of cash that belongs to shareholders but awaits profitable investment. From the shareholders' standpoint, the EVA is equal to profit minus payment for shareholder equity at the shareholder's valuation of the worth of investments in Microsoft. If the EVA is negative while there is ample cash, it suggests that management has stopped innovating. Well, the EVA numbers (in millions of dollars) tell that story: It was $1,225 in 1999, -$1,788 in 2000, -$2,794 in 2001, -$176 in 2002 and -$998 in 2003. Altogether, these indicators don't tell a pretty story.

As a remedy, Microsoft has just decided to unload most of its cash as a way of improving its EVA. That's a sure sign that the company is unable to find sufficiently attractive opportunities to create added value for its customers. We shouldn't be pleased with this analysis. Microsoft has been a leader in asserting the U.S.'s global primacy in IT. Microsoft has provided the de facto standards that have made it possible for a significant share of hardware and software to become interoperable.

If Microsoft wobbles (as IBM did in the early 1980s), we will all incur heavy costs in migrating to other vendors. The cost of replacing Microsoft-embedded systems will rise as the company tries - by every legally permissible means -- to reinforce the dependency on its "monoculture" (see Microsoft: A US Security Threat).

What to do? I'm not ready to give up Word and Excel. I would struggle to function as an author and researcher if I did. However, the steady stream of security and program updates upsets me. I have made 53 updates for the Windows XP operating system in less than a year. To that I must add patches for Internet Explorer (five in the past 60 days) and several for Word. Every time I make any change, my Windows registry and program-integrity checks show errors that require me to rerun diagnostic routines and take elaborate security and backup precautions. I can only imagine what the systems operators at big organizations must suffer making patches to Exchange servers and other Microsoft software.

Confucius said, "Do not enter a house if you do not know how to exit." That's good advice for anyone who is tempted to install systems from which it would be costly to migrate. In the short run, continuing with an all-Microsoft environment will always be the most attractive option. Nevertheless, users should avoid using software components that are so tightly coupled that any future migration could require wrecking much of the systems architecture. Explore all options to avoid the monoculture perils.

I have been successfully using a non-Microsoft e-mail package for decades, and I have avoided much grief because intruders couldn't readily find my address book. I have just switched happily from Explorer to the perfectly functional Mozilla browser. That is just one easy way to preserve diversity while minimizing risk. When you re-examine your architecture policies, you might want to add a diversity advocate to your staff to examine what it would cost you to migrate from Microsoft in case you must do so.

Paul A. Strassmann (paul@strassmann.com) has recently reviewed the flexibility of a tightly coupled single-vendor system. Taking it apart will cost more than putting it in did.