(FORTUNE Magazine) – Yes, I'm still here, writing this column for Fortune. When I was whining last issue about nothing going on in the technology industry, you may have noticed my observation that Windows 95 doesn't work as well as it should for a product that has sold more than 40 million copies. Even my editors thought that might make a good column, and begged me to stick around a little longer. (How sweet it is!) The topic is even timely, since Microsoft is about to introduce a set of enhancements to Windows 95, informally referred to in the trade press as Windows 97.

Microsoft introduced Windows 95 with great fanfare in August 1995. At the time, there was much debate about which markets would adopt it. The reality is that virtually all have. Though you may see reports that corporations have been slower to take up Windows 95 than consumers and small companies, everybody I know who used Windows 3.1 and a fair number who used the Macintosh or OS/2 have already made the switch. (I do know some normal people, so this isn't as irrelevant as you might think.)

No one doubts that the number of people using Windows 95 will get bigger and bigger until it includes pretty much all of us, at least all of us knowledge workers employed by large organizations--ones with more than, say, 200 people. If we do end up making Windows 95 and its successors the default operating system for white-collar workers, eventually all or most of 100 million or so office workers will be using the system every day. That's just in the U.S.

We appear as a society to have decided that it's okay for one company to control the software that operates the machines we use to compute and communicate every day. (We've also decided to let one company control the microprocessor technology in those machines, even though we still harbor fantasies about a competitive threat from some other microprocessor. But that's wishful thinking. And maybe another column.)

The point here isn't to debate whether supporting Microsoft's de facto monopoly is a smart thing to do. No, the point is to assess the impact of Windows 95 on society and business. In other words, deciding how bad Windows 95 is ends up being a lot more important than just beefing about the stupid thing some unnamed nerd in Redmond, Washington, did at 3 a.m. after his 15th Jolt cola. We're really talking about a key to the productivity and happiness of the American (actually, the global) knowledge worker.

Bottom line: Windows 95 is pretty bad. It's bad enough to present Microsoft and its customers with something of a problem. But it's good enough for us to muddle through, which is reassuring, given the fact that we really don't have a choice.

Just so you know my bias, I used a Macintosh as my primary computer from 1985 through June of this year: nearly 12 years, more than a lifetime in computing. I pride myself on how many other operating systems I've learned at one time or another: Apple II, Amiga, OS/2, DOS, Windows 1.0, Windows 3.x, and now Windows 95 and Windows NT. (I tried learning several Unix-based systems but finally had to admit defeat.)

For a clue as to how bad Windows 95 is, look at the number of Microsoft operating systems your average PC analyst has to learn. Microsoft just can't seem to get it right and keeps making new operating systems to fix its old mistakes. Other companies, especially Apple, seem to have a handle on how to write an operating system so that it can survive the upgrade process for some longish period. Macintosh System 7.5, the current version, is a recognizable offspring of System 1.0, despite more than ten years separating their introductions. With Windows, Microsoft has replaced so many key features so aggressively that you need to count the various versions as different products. Few of the programs designed to work on previous versions of the operating system really work on the current version, forcing customers to replace their old software if they buy new versions of Windows.

Windows 95 is bad enough so that Microsoft really wants everyone to switch to Windows NT, which is technically a great operating system. The hardware companies and software developers like this idea because it will solve a lot of their problems. It's just that those damn customers, both individuals and companies, present a sticking point: They really don't want to buy new hardware and software just to solve Microsoft's problems. Windows NT needs to run on a big computer--the kind that costs at least $2,500 now--and wants to run on a really big computer, like the dual-processor machine I use to impress people who come into my office. It costs about $9,000.

Here is the problem with Windows 95: Rather than working from a theory or philosophy of system design, Microsoft created Windows over time to respond to different competitive threats. So each succeeding version of the software has focused on solving a different set of problems. Commercially, this strategy has been very successful, but it is a less than artful way to make an operating system that works smoothly and intuitively. Worse, Windows must work with a plethora of hardware, each device made by a company fighting to differentiate itself, often without regard to what would make Windows run better.

The result: Windows 95 works, but not very well. Talk to a few users. You'll start to hear stories about the thing. You'll hear how it tends to get confused. That's because it doesn't manage the computer's memory banks well, so it leaves bits and pieces of software in memory when it shouldn't and then starts talking gibberish. You'll hear how the software is sometimes confusing to use. That's because different pieces of Windows were developed at different times and work with the user interface in different ways. You'll hear that Windows is frequently stumped when you change the settings on old hardware or install new hardware, even though Microsoft has promoted its so-called plug-and-play standard very hard.

Now Microsoft is getting ready to introduce a new version, the aforementioned Windows 97. In this version--a response to a competitive threat from Netscape, a tiny company that figured out the World Wide Web faster than anyone else--Microsoft has decided that the Internet changes everything. As a tactic in its anti-Netscape campaign, Microsoft has fixed on the idea that a World Wide Web browser should not only be integrated into the operating system but should also become the primary method for viewing and retrieving information wherever it is: on the Internet, on your company's private network, or on your hard disk.

The trouble is that Microsoft wasn't thinking this way last year, when it was busily trying to finish Windows 95. So the question becomes: Given that Microsoft has never taken a breather to go back and clean up all the problems that each version of Windows inherited from the previous one, can it now change the entire theory of the user interface without messing up something else? Microsoft has that certain confidence that comes from enjoying a monopoly and being very good at its business, which leads it to believe that it can do anything. Microsoft and its employees now think it is indeed the Master of the Universe.

I have not used Windows 97. But I have used Internet Explorer, Microsoft's Web browser. Microsoft says it is going to merge Internet Explorer with Windows Explorer (its program for navigating information on your hard disk) to create a new, sensible system. I hope that when Microsoft does so, it also merges both versions of Windows Explorer--the one you get when you double-click on the My Computer icon and the one you get when you select the Windows Explorer program from the Start button.

The irony is that Windows 95 was supposed to merge all the programs Microsoft had for navigating your computer's hard disks and local-area network. That's why Windows 95 comes with icons idiotically labeled My Computer (which shows you what is on your hard disk) and Network Neighborhood (which shows you what is on a local computer network). Unfortunately, in the middle of designing the software for viewing files, Microsoft introduced two fundamentally different viewing methods and made the problem it was trying to solve even worse.

How bad is Windows 95? Bad enough that over time, users get more and more frustrated with it rather than feeling more and more comfortable. Bad enough that it is costing corporations lots of money to keep it working. Bad enough, indeed, that Microsoft believes it needs to replace a big chunk of the user interface to make Windows simpler, and ultimately wants to persuade everybody using computers in large organizations to upgrade to Windows NT.

But Windows 95 is also good enough. It is preinstalled on 75% of the computers sold today. It is the target for nearly 100% of the software developers making new programs. It is our choice as the standard interface for computing and communications worldwide.

That must make it pretty good, huh?


STEWART ALSOP is a partner with New Enterprise Associates, a venture capital firm. Neither he nor his partnership has financial interests in the companies mentioned. Alsop can be reached at stewart_alsop@fortunemail.com