Microsoft faces Hobson’s choice for W2K

1999-09-03

Microsoft’s most recent announcements concerning Windows 2000 release date are unsurprising. Coupled with the delays to the most recent developer release, the Seattle company are insisting that the product is on schedule. What is most telling is that nobody seems to mind.

As usual, the media campaign is a combination of promotion and expectation management. Microsoft are emphasising quality, and seem to have separated the concepts of “final code” from “released product”. But let’s add some common sense here. Rule One of software projects: you cannot compress development time. Time and again, less experienced managers try to turn eight months into four only to find that that, like a tensed spring, the required time jumps back to the initial estimates. It’s true, try it. The only thing that can reduce a development schedule is reduced functionality or, heaven forbid, reduced testing.

So, what are Microsoft’s options? The first is to remove some functionality from the product, but that would be a marketing disaster. It has been done before, of course (hence the telling reputation of Service Pack 1). The second is to risk releasing incompletely tested software onto the market. Of course, this is a strategy known to most vendors, not just Microsoft: release 1 of any product is, generally, avoided by all but the most desperate. The third is to claim a code freeze at the end of the year, then continue development until shortly before the release date. This, employing the term “code chill” rather than code freeze, is again common practice but risks reducing testing time (again) and removes any slack that may be left from the system.

Microsoft are faced with a stark set of choices. The baying wolves of the IT industry will jump on the slightest misfortune of the software giant, so it will be treading very carefully over the next few months. But what of the end user? The advice is clear, and would be the same for any new product of this scale. Take your time. Evaluate the product (and others) in a realistic environment, making sure that it works with existing and planned architectures and configurations of other software. Listen carefully to the experiences of early adopters, and learn from them. Fortunately, the signs are there that end users are adopting exactly this approach. They are “waiting and seeing,” or planning upgrades only when their existing configurations will no longer be adequate. Y2K is, funnily enough, a helpful distraction: no-one is likely to rush into anything in January 2000.

It is extremely likely that Windows 2000 arrives late, incomplete and buggy (but don’t get me wrong – I’d be delighted for this to be proved untrue). Whilst Microsoft will struggle, yet again, with the blows to its reputation that this will cause, the vast majority of potential users will be unaffected.

(First published 3 September 1999)