In this week that has been so dominated by the giving of testimony, whywas it that Bill Gates didn't want to allow the public to view hisdeposition in the pending anti-trust suit? Well I'm no lawyer, but Ihave been sued in Federal Court and deposed by attorneys for the hated"other side," and to me the answer is obvious, and it has less to dowith trade secrets than Microsoft claims. What is smart behavior in adeposition isn't at all attractive in public. In order to protect hiscompany, Gates will have to be terse, vague, and difficult, all of whichwill be interpreted by the press as indicating that he has something tohide, which he probably does. What they won't say is that EVERYBODY whoisn't an idiot sounds like that in a deposition, even if they havenothing at all to hide.

In a deposition, lawyers for the other side ask questions trying toobtain incriminating evidence. There is no way for the person beingdeposed to "win" a deposition. The best they can hope for is to notlose. This is because the lawyers do this kind of thing every day andwe don't. It's because they get to excerpt the testimony for theirwritten briefs, sometimes putting it in the worst of contexts. The bestthe witness can do to protect his interests is to say as little aspossible. So Gates is likely to come across as a contentious, difficultsmart-ass. Fortunately for Bill, the appeals court agreed this week tokeep any weaseling private.

Or he could take the high road and become a storyteller, which is wheremost CEOs screw-up in their depositions. Lawyers for the other sidelove storytellers because they voluntarily bring into evidenceincriminating facts that could never otherwise be introduced. CEOs areused to being listened-to and used to being in charge, so they often getinto trouble in depositions, telling far too much. But Gates is thedisciplined, highly intelligent son of a big-shot corporate lawyer, sohe won't make the mistake of telling stories. Instead, he'll just lookbad, which Microsoft would like to avoid. It's all about image.

Right now, though, Microsoft has an even bigger image problems withWindows NT 5.0, which seems to be in one of those perpetual holdingpatterns, never quite ready to be released. The new shipping deadlinefor NT 5.0 is summer of 1999 and impatient corporate buyers areexpressing public doubts about Microsoft even meeting that goal. JimAllchin, Microsoft's god of NT development, has been making publicapologies all over the place as well he should, since his position inthe NT hierarchy is very shaky. Not that there are lots of folksangling for Allchin's job: the NT group is short of top programmers andmanagers and losing more every day. What's amiss here is that there aremajor internal doubts emerging about NT 5.0 and nobody wants to be thechampion of a product that can't be made to work.

Desktop and home PC users may think none of this matters, but they're wrong because the grand plan at Microsoft has long been to merge Windows9X and NT. The further out NT 5 is pushed, the further out will beWindows 2000, or whatever they plan to call it. To Microsoft, NT isimportant not just because it is a very profitable multi-billion dollarbusiness, but because it has been battling with Netware and Unix and forthe most part winning. Another year of delay (and the successfulintroduction of Netware 5) could change that.

So let's take a moment to understand where Windows NT came from, why ithas been so successful, and why some of those same reasons can explainits current difficulties.

Microsoft does its best work in reaction to a real or implied threat. Tell a group of Microsoft engineers to dream up some great newtechnology and they'll almost always disappoint. Tell them to reverseengineer someone else's great new technology or to react to a specificexternal technical threat and they respond beautifully. Companies thatare aggressive yet lack imagination play good defense.

Remember that Windows NT began as the next version of OS/2, so Microsoftlearned a great deal at IBM's expense. Right from the start, Microsoftaddressed every shortcoming the world had with OS/2. Microsoft made NTdeveloper-friendly nine months BEFORE the product shipped. The 3.xversions were based on technology. Engineering goals guided thedevelopment and feature set. It is important to note the 3.1 versionwhile given a bad quality rap compared to the more mature Unix and Netwareversions, was in fact one of the highest quality first-release operatingsystems.

The 4.X version of Windows NT was the finish up, clean up, pretty upversion. A mixture of technical and business objectives resulted in anice full-featured, mature operating system.

But Windows NT 5.X is something completely different. Microsoft isattempting to add a very aggressive number of major features, with thekiller being NT's Active Directory. This is probably a situation wherethey should not try to release a single major version, but a series ofsmaller improvements.

It took Novell a long time to perfect NDS, its enterprise-qualitydirectory service. There are a lot of very subtle things that need tobe done to keep a complex distributed directory service healthy andreliable. Microsoft has a lot of experience to gain to equal Netware4.11. And with Netware 5 on the horizon, extending NDS across theInternet, Microsoft is even more vulnerable.

But even Novell is failing to see the big picture. Long file names anddistributed directory services treat a symptom but ignore the underlyingproblem. Its a human factors problem. The current approach — Novell'sapproach, now being copied by Microsoft — operates on the assumptionthat the user actually understands a computer file system. Bigmistake. As you get more and more files, and more and more servers, theapproach has been to present users with an enterprise-wide directorylisting. You can easily navigate through the haystack but there islittle or no help finding the needle.

The other area where Microsoft is especially vulnerable is complexityfor the sake of marketing. The OLE/ActiveX, ODBC, and other stuffMicrosoft is using for all their products is complex and difficult tosupport. They've wired it into their WWW server and, with Win98, intothe Internet browser as well. The performance of this stuff is at best,clunky. When you wire it into a good sized network you can easily sloweverything down. It is difficult to control access rights and stillkeep the application working. There are many new security risksdeveloping from the stuff from Redmond. But Microsoft's internal viewof this, as always, is that more cumbersome software sells new hardwarewhich sells new software. There is an end to this food chain, though,when some righteous third-party swoops in with a simple and insanelygreat alternative. I choose the words "insanely great" quitedeliberately if you know what I mean.

Could Windows NT 5.0 be Microsoft's OS/2? Has marketing and ambitionovercome technical vision and engineering practicality? Microsofthas thrived on the ability to rapidly develop, ship, improve, and shipproduct. Once the motion starts, no one has been able to keep up withthem. NT 5.0 is clearly bogged down. The risk to Microsoft is itsfocus on marketing features. To implement them they must use publicdomain technology developed outside of Microsoft, opening the door fortheir competition. Being slow, releasing immature technology, andgiving others time to adapt their mature technology, is a good way tomess up one's business.

There are certain litmus tests that can be applied to the marketing ofhigh technology and one of the most important is looking at thearguments used to keep current customers in line. This generally goesin three steps. Step one is to say bad things about competitivetechnology when you don't, yourself, have any products that can actuallygo head-to-head. Step two is to say your stuff is just as good as thecompetitor's stuff because you finally have a semi-competitive release. Note that step two requires the complete repudiation of step one, whichsomehow never seems to bother Microsoft. Finally step three — thedesperation step — is to say that switching to a competitive technologyisn't cost-effective. It's cheaper to stay with our mediocre stuff thanto switch to the better stuff coming from somewhere else.

Now look at Microsoft's claims for Windows NT versus specificallyNetware and Linux. Bill Gates has dismissed both products asgreat 80's technology, but not appropriate for the 90's. That's stepone. And with NT 5.0, Microsoft is claiming its Active Directory is theequal of Netware's NDS. That's step two. And in recent discussionswith Microsoft salespeople, some of my corporate contacts now report ashift in the anti-Linux strategy from "NT is better" to "its going tocost a lot of money to move away from NT." Step three.