and there were some really good people working on the unix unification effort too (the OSF vs sun/at&t divide was more a contributing factor than lack of talent), yes? for instance, the openwindows UI was well-specified, consistent and fairly decent to develop on. without too much exaggeration, i can say that GNOME or KDE add more eye candy than deep UI improvements: e.g. drag-and-drop support was close to non-existant and is shaky at best now.
at the onset of the web, mosaic ran only on unix (IIRC, and i am pretty close to certain about that, since i contributed some patches) and unix looked like an attractive platform due to its networking sophistication. the windows market was hardly what it is today (and there has never really been a mac market). IMHO, OSF/CDE and the host of related efforts were indeed intended to provide a consistent UI (and engineers, at least at places like bell labs, did care to some extent about this stuff).
yes, i am no business guy, and i definitely do not want to think like one ;-), but i don't think the unix gang was disinterested in becoming a desktop leader. if they did they sure botched it big time when they had a lead: their OS did networking (unless you call appletalk networking! ;-)) when others did not (or at best did it as an add on), they had real money (not .com cash) and momentum (in the 90s unix was almost "cool"). bloody hell, they were the only real OS in town. or perhaps they did not care about the desktop at all (hey, high margins in the server market... who needs a desktop computer? thin clients will do... X terminals, java desktops, and all that good stuff). then they are even bigger fools.
about linux: SuSE got bought by Novell which by various counts is on its deathbed. VA linux, and a host of the early linux providers disappeared. last i heard, red hat wasn't the healthiest company in the world. sun: let's resell RHEL, throw out opensolaris, do a java desktop, and still try to pitch thin clients. what does that leave? ibm? if i took apache out, would linux penetration even be 50% of what it is today?
on the flip side, from joe shmoe: will my particular digital camera connect to my PC, still? will tivo desktop run on linux? how about my favourite chat client? (GAIM doesn't even do presence right for certain protocols). how about sync'ing my pocketPC? or the cool games i bought? (perhaps i have to learn how to use WinE? or run Xen!).
i agree with you: i think some very interesting things will come out in the next two years, in linux usability. redhat has done a great job (criticism from the geek gallery, notwithstanding) with their graphical install. hardware support keeps improving and a suite of tools (kpilot/gnomepilot, etc) attempt to keep up. most importantly, the linux/gnome/KDE crowd is beginning to prioritize usability.
however, i fail to see: a) why they will act faster and produce something more consistent than what microsoft already puts out, leave alone, what MS will have in the future, b) why users (individuals at home, or enterprises user desktops) will switch.
one reason i hear is security. i am yet to be convinced by that argument. windows in the enterprise (and at home) is an insecure OS primarily because of user laziness/ignorance (but as i remember someone in the unix group at bell labs, perhaps pike, used to say: unix gives you enough rope to shoot yourself in the foot! and that's true) and IT staff laxity (they are content mopping the floor and letting corner office VP click at random on the 1MB attachments that they are currently too afraid to block at the gateway).
the other is total cost of ownership. there's some legitimacy behind this claim. linux/unix is an easier and more reliable OS to run, especially on the back end -- provided all you want to do is file or compute service (or perhaps some networking services like web or FTP or DNS). if you want real enterprise services, say a groupware service, have fun integrating openLDAP with kolab or opengroupware, and pray that evolution will take less than 1 minute to auto-complete an email address from the directory (if at all it does!) or mozilla's calendar effort will some day get to release stage. or go buy the packaged stuff from novell or sun... and still not get as good integration as windows provides and potentially pay more.
in short: i remain pessimistic.
--ravi