At the end of March, Microsoft sources whispered that users can expect Windows 2000 (Win2K) to ship in October. (This suggestion circulated 3 weeks after those same sources predicted a release date in February 2000, so take it with a grain—or pound—of salt.) Like every other shift of Win2K's unofficial ship date, these whisperings sent rumor mills into overdrive.
"At this rate, I'll retire before Windows 2000 ships," opined one writer. "Microsoft needs to admit that it bit off more than it can chew and just ship something, adding Active Directory (AD), IntelliMirror, and other features in subsequent years."
This "dang it, just ship something" theme is popular and sometimes humorous. I couldn't control my laughter when I read one commentator's suggestion that Microsoft could just add features such as AD and Plug and Play (PnP) as part of a service pack after Win2K's release. If I didn't know better, I'd think that a bunch of journalists and pundits have run out of things to say about NT 4.0 and are ravenous for the fresh meat of Win2K. (Sources inside Microsoft say Win2K service packs won't provide new features; from here on out, service packs will contain only bug fixes.)
Reality Check
Will Win2K's final ship date really matter in October? I'm not the first person to observe that systems administrators' biggest concern at the end of 1999 will be Year 2000 (Y2K) problems, not Win2K. I can see the scenario now: The world is 87 days away from 2000. People in all walks of life hope they're Y2K-ready, but no one's sure. Companies are allocating last-minute funds for last-minute Y2K fixes. But IS departments everywhere are celebrating, because despite the Y2K fiasco, Win2K has finally gone shrink-wrap!
I realize that many corporate users have been holding their collective breath for a long time in anticipation of Win2K Server and that some of those folks bet their personal credibility on a 1998 NT 5.0 ship date. I understand and share the frustration of pinning hopes on vaporware. But consider that NT is supposedly approaching 40 million lines of code. Creating software that large is a huge undertaking. Why shouldn't Microsoft take years to do Win2K right?
What Do We Expect?
Microsoft clearly rushed NT 4.0 out the door. Service Pack 1 (SP1) appeared almost immediately to fix an egregious kernel bug, and many people felt that NT 4.0 was reliable only after the arrival of SP3. Some firms moved their domain controllers from NT 3.51 only after NT 4.0 SP4's arrival.
For many administrators, Microsoft shipped an acceptable version of NT 4.0 not in June 1996 but in October 1998. Do the media types who are encouraging Microsoft to ship an incomplete Win2K really want to depend on Microsoft's slow and unpredictable release of service packs over a period of years to gain reliability? Shouldn't NT consumers expect a quality OS from the start?
Microsoft needs to freeze the Win2K feature set, then release the new OS only when those features are reliable. When my fellow journalists crow in headlines that Microsoft's late again, the message they're sending to Redmond and to other software developers is that we don't want them to get their products right—we just want them to get products out on time.