We can't win, but we can't lose.
As a consultant, my biggest daily battle is convincing middle managers that a *nix box is a cleaner solution than an NT box. Unfortunately NT seems to win more often than *nix. Now this isn't due to my ability to pitch the unix argument, but a belief in the hearts and minds of a large number of people that MS is a brand that can be trusted.
And why should'nt middle managers trust the MS brand, after all they are constantly told that with Win 9x / NT support costs are normally lower, staff training costs are normally lower, and software runs out of the box (from your local retailer).
Now Unix has in the last few years made great strides in becoming accessable to the general computer public, but its still seen by the majority as a 'geeks' OS, or a back office server. Lets be honest, the average user does'nt want to trawl around various FTP sites looking for drivers or packages for the version of *nix that they've installed earlier. So without intervention we're going to remain relegated as the 'other' OS.
But intervention has come, from the enemy so to speak. With the attempts by MS to bring Office to the Linux community, and now their release of the FP98 extensions for unix, Microsoft are beginning to associate their brand name and reputation with our OS. If Microsoft ever build an NT front end on a Unix core, then the symbiosis will be complete. As Microsoft associate themselves more with our OS, then the hearts and minds of the average user will begin to take us as an alternative more seriously. Once here, it shouldn't be too long before they discover software thats just as functional (in many cases more so) as the MS offering for a nominal (or free) cost.
So where do Microsoft go from here? If they try to directly compete with us on the unix ground then they risk losing their customers to the open source movement. If they stay on their side of the trench, then we'll always dominate the large server / internet markets.